Page 1
MEASUREMENT AND PREDICTION OF SOIL REDISTRIBUTION
ON UPLAND LANDSCAPES IN SOUTHERN ONTARIO
A Thesis
Presented to
The Faculty o f Graduate Studies
of
The University of Guelph
b y
DONALD JAMES KING
In panial fulfilmrnt of requirements
for the degree o f
Master of Science
July. 1998
S Donald J . King. 1998
Page 2
National Library Bibliothèque nationale du Canada
Acquisitions and Acquisitions et Bibliagraphic Services senrices bibliographiques
395 W M i Street 395, nie WellingtDn OüawaON K1AON4 OltawaON K1AON4 canada CaMda
The author has granted a non- exclusive licence allowing the National Library of Canada to reproduce, loan, distri-bute or sell copies of this thesis in microform, paper or electronic formais.
The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts fiom it may be printed or otherwise reproduced without the author's permission.
L'auteur a accordé une licence non exchive permettant à la Bibliothèque nationde du Canada de reproduire, prêter, disbriiuer ou vendre des copies de cette thèse sous la forme de microfiche/fïh, de reproduction sur papier ou sur format électronique.
L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent ê e imprimés ou autrement reproduits sans son autorisation.
Page 3
MEASUREMENT AND PREDlCTION OF SOIL REDISTRIBUTION ON CPLAND LANDSCAPES OF SOUTHERN ONTARIO
Donald James King University of Guelph. 1998
Co-.4dvisors: P.H. Groenevelt G.J. Wall
This thesis was conducted to measure long-tem soil erosion rates on cultiuted hillslopes
in upland landscapes of southem Ontario. Measured erosion rates and redistribution patterns
using "'CS analysis were compared to predicted water (WEPP) and tillage erosion losses.
Predictrd rates of gross soi 1 erosion best approximated measured soil erosion rates when the
water and tillage predictions were combined. Discrepancies between measured and predicted soil
redistribution patterns at uppsr landscapo positions raisod concerns about the assumptions of
uniform deposition of "?Cs ovrr the landscape implicit to measured soil redistribution values. ..\
field experiment was conducted with the use ofrainfall simulation to test the validity o f the soil
srosion measurement technique. Study results indicated that on average. one-third of the applied
cesium was removed in the ninofi mostly unanachcd to sediment. The movement of the cesium
tracer in surface runoFi from al1 plots suggest that tïeld hillslopes expenencing erosion at the time
of '"CS deposition could result in significant overestimations of soi1 loss using "'CS reference
analysis.
Page 4
ACKNOWLEDGEMENTS
i have been fortunate to have had the opportunity to study under people that approached the
subject of soi! erosion with trernendous expertise and compassion. To Dr. Greg Wall. rny sincere
gratitude for your cornmitment in support of my professionai and scientific education and for L
sharing many other life skills. To Dr. Pieter Groenevelt. thank ?ou for your unwavering
confidence. proficiencp and exuberant nature that buoysd m); own enthusiasm on more than one
occasion. To Dr. Murray Miller. who's cornmitment to the study of soi1 conservation 1 have
admired for man- years. your critical suggestions were greatly appreciated.
1 would like to thank al1 of my colleagues at Agriculture and Agri-Food Canada. Land
Resource Unit. Guelph for their tirne. support and dedication to impr0vir.g the understanding of
the land and its many h u e s . To the sroding morals who's motto remains to work hard and to play
hard. my thanks to Brian Hohner. Carolyn Miller. Irene Shelton. Brenda Grant and Peter Clarke.
This degree could not have been possible without the academic excellence and friendly
support shown by faculty. staff and students of the Land Resource Science Department. Thanks to
Petrr Smith and Glen Wilson for their advice and sincere efforts to help find solutions to my
analytical questions. To Wili Gowers. thank -ou for providing valuable guidance and
understanding to the science of atomic absorption spectrometry. To the man. LRS graduate
students that I have shared acadernic and recreational pursuits. it has been most rewardin-.
1 would like to thank my friends and fmily who continue to provide inspiration and by
their daily efforts instill in me the values and sense of achievement that are most needed in our
society. To mp wife. Michele. who has shown great patience and understanding. 1 dedicate this
work to you.
Page 5
List of Contents
Page C hapter 1 : Li terature review: Determining soil erosion processes on upland landscapes
of southem Ontario frorn empirical and dcterministic models .
1 . 0 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Soil erosion processes - 3 1.1.1 Evaluation of tield-scale soi1 crosion processes . . . . . . . . . . . . . . . . . . . . . . 3
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 -2 Field-scale erosion measurement - 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2.1 Long-term rrosion plot monitoring 5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2.2 Rainfall simulation technology 6 1 2.3 Net soi1 erosion measurement ti-orn resident cesium- 137 . . . . . . . . . . . . . . . 7
1 .2.3.1 Cesiurn- 13 7 tracer analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1 2 - 3 2 Global fallout of cesium-137 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1 2 . 3 -3 Crsium- 1 3 7 reference analysis method . . . . . . . . . . . . . . . . . . . . . 9 1.2.3.4 Field measurement o f cesium redistribution . . . . . . . . . . . . . . . I O
. . . . . . . . . . . . . . . . . . . . . 1 -3 Simulation models to predict field-based water erosion 12 1 3.1 Universal Soil Loss Equation (USLE) . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1 .3.2 Revised Universal Soi1 Loss Equation (RUSLE) . . . . . . . . . . . . . . . . . . . . 15 1.3 -3 Erosion-Productivity Impact Calculator ( EPIC ) . . . . . . . . . . . . . . . . . . . . . 17 1.3.4 Water Erosion Prediction Project (WEPP) . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.3.4.1 WEPP mode1 components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 1.3.4.2 WEPP mode1 sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.4 Tillage erosion process and prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 1.4.1 Tillage erosion measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 1-42 Tillage srosion prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.3 Soi1 redistribution prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
1.7 Study Objectives and Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 3 0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References 32
Page 6
Chapter 2: Measurement and prediction of soil erosion on an upland landscape in southern Ontario
2.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . U 2.1.1 Rockwood field site characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1.1 Land use history -45 2.1.1.3 Slope profile measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 2.1.1.3 Soil properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.1.2 Soi1 erosion measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 5 1 2.1 2.1 Period of analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
. . . . . . . . . . . . . 3.1.2.3 Soi1 redistribution estimation using cesiurn-137 52 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.2.3 1Method of anaiysis - 5 3
2.1.3 Soil erosion prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 . . . . . . . . . 2.1.3. 1 Warer Erosion Prediction Project (WEPP) mode1 53
-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.3.1.1 WEPP input files . m . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.3.1 . ? WEPP output files - 5 8
2.1.32 Tillage erosion mode1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 . . . . . . . . . . . . . . . . . . . . 2.1 3 . 3 Model cornparison testing and analysis 65
2.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 6 6 2.2.1 Slope profile analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 6 6 2 -22 Soi1 erosion measurement using Cesium- 137 . . . . . . . . . . . . . . . . . . . . . . - 6 7
2.2.2.1 Redistribution of Cesium-137 . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 . . . . . . . . . . . . . . . . . 2.2.2.3 Net soi1 loss rates from Cesium-137 values 73
2 . 3 Soi1 rrosion prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.3.1 WEPP mode1 analysis 74
. . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.3.2 Tillage erosion mode1 analysis 79 2.2.3.3 Combined WEPP/TilIage soi1 erosion mode1 . . . . . . . . - 8 1
2.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
2.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 8 9
Chapter 3 : Evaluating the use of cesium- 137 atmospheric deposition for measuring soi1 erosion rates on upland regions of southern Ontario
3.1 Literature review and hypothesis development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 3.1.1 Cesium- 1 3 7 tracer analysis assumptions . . . . . . . . . . . . . . . . . . . . . 97 3 - 1 2 Reference site deposition assumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Page 8
List of Tables
Table Page
45
56
Guelph OAC climate nomals for the 30 pr penod. 195 1 - 1 980.
WEPP soil input file parameters for the conventional and conservation tield.
Rockwood site management histoy from 1960- 199 1.
WEPP erosion rnodel parameters used for sensitivity analysis with calculated base values and the value range tested.
Measured soil IJ7Cs activity (Bq rrf') frorn the conservation field. conventional tield and forest reference area: and associated soil loss (kg rn" -il) from the cultivated fields.
Soil "'CS activity (Bq rn"). its variabiiity. and associated soi1 loss from conservation field Ap horizon averaged from 4 transects at each landscape position downslope.
Soi1 "'Cs activity (Bq rn"). its variability. and associated soi1 loss from conventional field Ap horizon averaged frorn 4 transects at each landscape position dou-nslope.
Measure of sensitivity for kry soil parameters in estimating soil loss by the WEPP rnodel.
Calibration of WEPP model soil loss parameten using average soil loss. percent deviation and Nash-Sutcli ffe coefficient of efficirncy .
Soil redistribution as estimated by "'Cs analysis and predicted bp the Tillage erosion rnodel.
Statistical summary of soi1 loss prediction for conservation and conventional field positions as they compared to the "7Cs soil loss estimation method.
Precipitation recorded at the University of Guelph. Ontario during the peak atmospheric fallout of "'CS.
Page 9
Table
3 2
Title
Soi1 characteristics at each landscape position pnor to rainfall simulation experiment.
Cesium retneved from sample portions by landscape treatment on Day 1 and Day 2.
Pedon descriptions €rom Rockwood hi 1 lslope.
Rocku-ood field site measurement of Cssium- 13 7 and related parameters by gridpoint.
Variabilie of soil "'CS activity (Bq m-') in field .4p horizon from 4 transects dowslope.
WEPP slope input file.
WEPP soil input file.
Partial WEPP climatc input file from University of Guelph climate station.
Study site soil characteristics at each landscape position taken prior to rainfàll simulation experiment.
Day 1 plot runotr volume - incremental.
Day 2 plot ninoff volume - incremental.
Day I sedimrnt concentration and sediment loadings - incrernental.
Day ? sedimenr concentration and sediment loadings - incremental.
Cesium retrieved from esperimental analysis and mass balance calcuiation.
Correlation coefficient. r. for plot soil characteristics at the site and experirnent sampling parameters.
Page
120
Page 10
List of Figures
Figure TitIe
2.1 Rockwood site contour map of conservation and conventional fields with sarnpling points identified.
Page
48
Site identification of Iandscape siope position. 67
Estimatrd soi1 losdgain rate over conservation and conventional 73 field slope profile.
Soi1 loss prediction over conservation and conventional slope profiles 78 of nonadjusted ( WEPPna) and adjusted { WEPPa) soil loss parameters.
Soi1 loss as predicted by separate WEPP and Tillage erosion mode1 82 output values and estirnated by "'Cs reference analysis with distance downslope.
Soi1 loss as predicted bq- combined WEPP and Tiilage erosion mode1 83 output values and estimated by "'Cs reference analysis with distance downslope.
Rainfall simulation total mnoff volume and sedimsnt losses measured 133 fiom each plot on Da? 1 and Day 2.
Cesiurn mass balance fiom runot't' components. 126
Page 11
Chapter 1 : Literature review: Determining soi1 erosion processes on upland landscapes
of southem Ontario from empirical and deterrninistic models
1.0 Introduction
A f m e r estimates the sevetity of the soil erosion problem on the land and decides
whether or not adoption of soi1 consemation practices is necessary. However as research s~udies
have illustrated. hm operators ofien are not sufficiently mare of the severity of the soil loss
occumng on fârms (McNairn and Mitchell. 1992). The cost of soil erosion is high (Battiston et
al.. 1987: Kachanoski et al.. 1993). Estimates for Ontario conditions are a minimum of $68
million annually (Wall and Driver. 1982) as thickness of topsoil declines. A f m e r will on[>- be
receptive to the adoption of conservation practices if he/she feeis able to alleviate the problrm
and the cost of implemrnting is wmanted by the savings. That may be direct savings of reducing
input losses from runoffor reducing the long-term economic cost of future productivity lossrs
with less topsoil.
Considerable improvements in reducing soil degradation md erosion have occurred in
rscent years with the adoption of expanded crop rotations and reduced Mage practices by many
f m opentors in southem Ontario (Wall et al.. 1995). The degree of improvemrnt in reducing
soil erosion. horvever. has been recognized on only a small fraction of the agricuitural landscape
(Shelton and Wall. 1998). The combination of intensive famiing practices. rolling topography
and moderately erodible soils in the majority of southem Ontario create high erosion potential.
Even though soil erosion in southem Ontario is the single most significant factor resulting in the
degradation of the land base. there have been few of the acnial measurements of erosion
Page 12
occumng (Kachanoski et al.. 1992). Prediction of soil erosion for soil consenation planning
purposes has relicd on smpincal relationships developed in the US. The nerd for an approach
and mrthodology for comprehensive tield-scale erosion characterization in upland topog-aphy
for southem Ontario is paramount if continued erosion losses are to be acknowledged and
reduced.
Recent computer-basrd models have been developed to predict soil erosion and soi1
redistribution at the field-scale from water erosion processes. Irnprovements to the understanding
of water erosion processes and the contributing factors have been added to thess simulation
procedures. A new erosion process has been identified as tillage erosion. The eitrent of tillage
erosion. water erosion and other erosion processes have been recently investigated using "'Cs
tracer analysis. The discovery of the cignificant impact of rrosion processes acting at diffrrrnt
positions of the landscapc: is an important consideration for f m management planning. Difkrent
areas of a field are beginning to bc managed separately with the assistance of Geographical
Positioning Systrms and digital terrain modelling technolog). Precision farming techniques. that
mainly rely on past yirld response to vary f m inputs in a field. would benrtït by recognizing the
spatial impact of signiticant rrosion processes to better manage and protrct a farm's soil
resource.
The problem of identifving soil loss rates from cultivated agricultural landscapes is being
able to recognize the contributinp processes and having available accurate methods of estimatins
their magnitude. Measuring the extent. magnitude and rate of soil erosion and its economic and
environmental consequences is dificult. Soi1 erosion is a complex relationship between climate.
soil characteristics. topography. and the influence of land cover.
Page 13
1.1 Soil erosion processes
Natural soi1 erosion processes include fluvial (water). aeolian (wind). and mass wasting.
each being dependent on the climate and physiography which act with time in developing and
modieing the landscape (Hudson. 1995). Man-made processes from land management practicss
include tillage which acts independently of natural soil erosion processes: however. tillage action
may contribute to subsequent losses by natural means (Unger and Cassell. 1990). Measurement
of natural processes of soi1 erosion have been done by various methods with limitations for each.
This inconsistency has led to data that is difticult to interpret beyond the specii'ic site
characteristics for useful extrapolation. In addition. interpretation of atimates from a single soi1
erosion process may be signifïcantly under predicting soi1 loss.
Soil erosion in Onraxio has predorninantly been associated with water (Coote et al.. 198 1 )
and wind (Nickling and FitzSimons. 1985) erosion. In the upland regions of southem Ontano of
mediurn-testured soils. uatrr erosion has been considrred a dominant soil erosion process. Soil
erosion can be considsred the interaction of precipitation induced water mnoff and the inherent
resistance of soi1 to be drtached and be transponed. The ei'fect of rain is called erosivity and the
effect of the soil is called erodibili~.. Ellison's ( 1947) concept that erosion ma' be divided into
detachment. transport and deposition may be applied to the action of an. soil erosion process.
1.1.1 Evaluation of field-scale soil erosion processes
Quantitative measures of the eirtent of soi1 erosion on a field scale is not readily available
for regions of Ontario. Soil loss estimates from water erosion using the Universal Soil Loss
Equation (USLE) have been used in Ontario to evaluate soil conservation initiatives as there has
Page 14
not been a comparable alternative. Water erosion on a complete slope sequence theoretically
assumes increased detachment and transport energ?; at lower landscape. Measurernent of water
erosion on the USLE research size plot represents the complete process of ri11 and intemll
erosion ( Wischmeier and Smith. 1978) from this single Iandscape dimension.
The relationship of water erosion processes to landscape dope units is theoretically a
linear measurement on cultivated tields ( Wischrneier and Smith. 1965). Erosion models that hakx
been used almost exclusively have been rrnpincal and more recently process-basctd in operaiion.
in simulating water erosion. At the field scale. the ESLE mode1 considers homogencous site
charactsristics of soil. slope. management. etc. (Wischrneier and Smith. 1978). Estimates of
annual soil loss rates consider soi1 movement fiom combined water detachment and runotT
transport forces. These erosion forces are greatly influenced by increasing slope angle and length
or area and result in an increase in sediment transport. InterriIl or sheet erosion is assumed to
occur throuehout the field while the ri11 erosion process. uhich contributes to sediment de l i \ q
dow-nslope. is predominant at the base of the midslope at the initiation of the concave position
(\.an Vliet and Wall. 1979).
For tkid rxperimcnts show-in3 a net soil loss. "'Cs based rstimates have bern reponed to
be greater than soi1 loss rates predicted by the USLE. Lon. correlation coefticents betkvren the
two methods of 0.39 (Bernard and Laverdiere. 1992) indicate significantly lower erosion rates
predicted by the USLE. Bernard and Laverdiere (1 992) suggest the redistribution pattern of "'Cs
from upper to mid to lower slope classes diffen from what is predicted by the LS factor of the
LISLE. which increases indefinitely with siope length. This explains the weak correlation
measured between "'Cs data and the LS îàctor. as reported by other authors (de Jong. etal..
Page 15
1983: P ~ M O C ~ and de Jong. 1990). Soil erodibility under snowmelt runoff conditions may be
responsible for a signiticant portion of the m u a i soil loss under southern Ontario climatic
conditions. This soil erosion is incorporated into the 'j7cs data but not predicted by the USLE.
Soil movement from water erosion within a season and its impact on productivity and the
environment may not be immediately apparent but the cumulative effect of susceprible conditions
afier several seasons \vil1 likely becorne significant. Consideration of immediate effects of
extraordinaq climatic events or tillage action however should be incorporated into a prediction
of the dynarnic soil redistribution process. Water srosion losses theretore are a hnction of
seasonal and multi-season variation in precipitation and field management hctors.
1.2 Field-scale erosion measurement
Measurement of watsr erosion for field-scale application in southem Ontario ha\.e reIied
on tu-O methods: long-term plot monitoring and rainfall simulation technology. Long-term plot
monitoring is subject to the inconsistencies of natural precipitation. The second method
employ ing simulated rainfa1l remedies this factor: however. it introduces an increased error in
sstrapolat ion.
1.2.1 Long-term erosion plot monitoring
The use of field-plot methods to quantifi. soi1 erosion requires several years of data
collection. Year-round monitoring is critical given that significant percentages of annual erosion
can occur in very shon periods of time ofien during the spnng thaw period (van Vliet and Wall.
198 1 ). The only long-term mrasurements in southem Ontario are from the Universit). of Guelph
Wischmeier style (USLE) runoff plots which have been monitored for approxirnately 3 7 yrs.
Page 16
Meaningful rates of soil loss arc required tiom monitoring throughout the year under many
climatic and soil management scenarios. The cost of maintaining large-scale plot monitoring
studies is prohibitivelp expensive. Unfortunately in the end, soil loss estimations fiom this
method are soil and slope speci fic (van Vliet and Wall. 1 979: 1 98 1 ).
Measured soil loss rates were collected from m o f f plots at the University of Guelph and
the Elora Research Station to evaluate the reliability of predicted soi1 erosion values in southem
Ontario as computed by the C'SLE (van Vliet and Wall. 1979). Average annual soil loss data over
a 4 to 6yr period was used. The- ranged from <! to >JO t ha-' yi'.
1.2.2 Rainfall simulation technology
Rainfall simulators have bcen developed to be utilized in the laboratory or tield to stud:.
the Factors aficting the erosion process (Wall et al.. 1988) and soil management evaluations
(Wall et al.. 199 1 ) under rnany different scenarios. The Guelph rainfall simulator (Tossell et al..
1987) has the advantases of being easily transportable. inexpensive and reliable for treatment
repl ication. Rainfall simulaton al low soil loss and runoff to be generated under controllcd and
repeatsble conditions. However. the interpretation of simulator measurements is complicated b!.
the uncertain relationship between the srosiveness of sirnulated and natural rainstorms (Nolan et
al.. 1997). The other problem exists when relating soil loss from simulated rainfall on small plots
to soil losses from larger tkld-scales with variable topographp-erosion relationships.
Soil and runofi losses from experimental plots using the Guelph rainfall simulator were
measured while investigating the benefit of intercropping red clover with silage corn for soil
erosion control (Wall et al.. 199 1 ). A rainfall at an intensity of 16 cm hr-' for 10 min was
applied. an equivalent to an annual retum period storm of approximately 25 y-. Soil loss rates
Page 17
ranged from minimal values of G O g m-' with increased residue cover to a high of > 150 g m-'.
On a field basis. a single storm rvent produced soil losses ranging from 0.3 to 1 .j t ha-'.
Estimates of annual tosses would be cumulative of a number of storms and nrnotTsvents.
1.2.3 Net soil erosion measurement from resident cesium-137
1.2.3.1 Cesium-137 tracer anaiysis
The most accurate method available to msasure net soil movernent in the landscapr is
with the use of "'Cs tracer analysis (Ritchir and Ritchie. 1997). Crsium- 137 is strongly adsorbed
to soi1 cation exchmg sites: therefore. physical processes such as naturall~ occurring soi1
rrosion and tillage redistribution of soi1 are responsible for the movernent of "'Cs h m the time
of uni t o m deposition (Kachanoski. 1987: Ritchie and McHenry. IWO).
Properties of radioactive '"Cs make it unique as a tracer for stud>-ing crosion and
sedimcntation ( Ritchir and McHenry . 1 990). There is no natural source of "'Cs in the
environment as it is derived from nuclear fission generated in testing or nuclrar reactors. "CS is
strongly adsorbed on clay and organic particles making movement by chernical or biological
procrssrs limitcd. Its gamma radiation compared to other radiotracers ( Weast. 1987). such as
*'Sr. which is a beta çmitter. make it preferable. It is easily detectûble since "'Cs emits a strong
gamma-ray (662 keV) upon degradation directly measurable by gamma spectrometry (Ritchie
and McHenry. 1973). Cesium- 137 half life of 30.2 yr ( Weast. 1987) allows measurement O\-er
several years. The soil-chernical bshaviour of "'CS has been useful in several srudies to tind a
direct correlation between "'CS movement and soi1 erosion rates (Ritchie et al.. 1974: de Jong et
al., 1982: Kachanoski. 1987).
1.2.3.2 Global fallout of cesium-137
Page 18
Cesium- 137 analysis techniques used worldwide for soi1 erosion estimation are based on
the result of global fallout of radionuclides during and immediately afier the main above-ground
nuclear bomb testing period of the 1950's and 1960's. The explosion of nuclear bombs in the
atmosphere caused some local fatlout with the remainder of the fission products being cxried
alofi to the troposphere and stratosphere where the! circuiated around the globe before depositing
on the earth (Butler. 1980). The general fallout pattern is due to the injection of air from the
stratosphere into the troposphere in the spring. and the increased precipitation in spring (Boothe
et al.. 1965).
The majority of ''7Cs came down in vec* large spring and summer rainfall rvents in storm
cells that originated in the upper atmosphere (Carnbray et al.. 198 1 ). The northem hemisphsre
received the majority of the fallout most concentrated between latitudes 40 to 50 de, arecs
(Hutchinson-Bsnson et al.. 1985) tvhere total fallout varied as a function of localized
precipitation (Kiss et al.. 1988). Xnalysis of '"Cs in precipitation in the Saskatoon area during
the major deposition Fears ( 1960- 1963 ) indicated that virtually al1 of the "'Cs drposition
occurred during the growing season (de Jong et al.. 1982).
Re_eionally. precipitation patterns have bcen used to determine the amount of '"CS input.
In southem Ontario. fallout levels were rstimated to be 2700 Bq m' in 1983 from total
accumulated atmospheric drposition comected for radioactive decay (Kachanoski. 1987).
Kachanoski in 1996 reported base-line "'CS (total deposition. Bq m-' ) values on 42 sampling
sites for south-western Ontario on the basis of availability of long-term precipitation records and
ceographic position. The "'CS values ranged from 186 1 to 30 15 Bq m-' and averaged 1420 Bq C
mm'. This interpolation mapping rnethod is usefùl in indicating the expected "'CS inventos and
Page 19
variabilin; for a region: however. site specific erosion estimation relies on baseline measurement
in close proximity to the field investigation.
The availability of detailed atmospheric radionuclide deposition from climatic data is
limited for site speciiic study. This approach however has estimated base levels of "7Cs areal
activity for various other locations in Canada (de Jong et al. 1983: Kachanoski 1987, 1996:
Brewster and Pillay. 1991 ) and in the United States (Lance et al.. 1986) relative to the a\-erage
total annual precipitation in these locations.
1.2.3.3 Cesiurn-f 37 reference anabsis method
As a substitute for sites without detailed historical clirnatic data. "7Cs reference a.nal>.sis
considers an area of undisturbed native vegetation. either grassland or forest which has not
experienced erosion. as bring a base level or reference of "'CS activity (Brown et al.. 198 1 b).
Investigation of '"Cs redistribution in a cuitivated tield is based on a cornparison of the
rneasured "'Cs ievrls and the local reference inventop (Quine and Walling. 1991 ). Areas of
rrosion are identifictd by negative drviation from the local reference. and deposition by positive
d e ~ i a t ion. The magnitude and direction of these dr'viations provide a qualitative estimate of
sedinient redistribution.
For the applicaiion ofthr "'Cs reference method. it is necessary to assume uniform
deposition across the Iandscape unit being studied (Ritchie and McHenn;. 1990). During the
period of deposition from 1 954 until 1964 (Carter and Moghissi. 1977). deposition on
agncultural land kvas therefore considered to be uniform regardless of the cropping practices and
ground cover. For much of southem Ontario. land cover during the fallout penod kvas C
predominantly permanent coïer fiom forages (Ontario Department of Agriculture and Food.
Page 20
1967). With this cover. precipitation causing mnoff losses of '"CS at the timr of deposition or
soi1 particle translocation downslope should be minimal when compared to the increaxd
potential with bare soi1 or row-cropped conditions.
Soi1 erosion rate calculations are a fünction of differences in cesium concentrations and
have for the most part used the peak radionuclide deposition period of 1963 as the starting point :
( Brown et al.. 198 1 a: Kachanoski and de Jong. 1984: Pennock and de Jong. 1990: Garcia-Oliva
et al.. 1995). blirtz and de Jong ( 1987) stated that the degree of deplstion of enrichment of soil
"'Cs (relative to atmospheric input) at a point in the landscape retlects the net loss or gain of soil
by al1 erosion processes acting at that point since "'Cs has been present in the environment
( McHenry and Ritchie. 1977: Brown et al.. 198 la: de Jong et al.. 1983: Longmore ri al.. 1983).
1.2.3.1 Field measurement of cesium redistribution
Soil erosion rate measurement in southem Ontario h a in the past been reported largely
from srnall plot rrosion studies measuring event based water runoffat diffrrent timrs of the year
(van Vliet and Wall. 1979: 198 1 : Marsh and Groenevelt- I992). The scale of these studies
hinden measuremrnt of soil rnovement at the farm management scale. where a tield oAsn
encompasses variable siope and possible soi1 and hydrologie differences. Erosion rates at this
scale c m be more accuratri? calculated by usine radioactive îàllout "'Cs as a tracer of erosion
and sedirnentation patterns throughout a field (Rogowski and Tamura. 1970: Ritchie and
McHenry. 1978: Kachanoski and de Jong. 1984: McHenry and Bubenzer. 1985: Kachanoski.
1987: Ritchie and McHenry. 1990). Since 'j7Cs is strongly adsorbed to soi1 cation eschange si tes.
combined physical processes such as naturally occumng soil erosion and tillage redistribution of
soil are incorporated into the measurement of 1'7Cs movement from the time of uniform
Page 21
deposition ( Kachanoski. 1 987: Ritchie and McHew. 1990).
The "'Cs reference analysis method measures the extent and pattern of soil erosion by the
difference in concentration of '>'CS in the samples subtracted t'rorn the concentration at the time
of uniform distribution (Ritchie et-al.. 1993). approximately 35 years ago. The variability of
127 Cs movemrnt measured over this penod will be a factor of management differences such as
increased tillage intensit?; or the proportion of monoculture cropping practices impacting on soil
quality (Unger and Cassel. 1990) and soil movement downslope. Topographic factors such as
slope steepness. shape and length h s bsen related to loss of "'Cs from upper landscape positions
and gains in depositional areas (de Jong et.al.. 1982: 1983: Lance et.aI.. 1986). An evnluation of
the spatial redistribution of "'CS within landscape units of fields. and also the total "CS loss
from fields can indicate the variabilih of soil loss and net soi1 loss at a site.
The processes responsible for soil redistribution in upland southem Ontario landscapes
are from water and tillage action (Kachanoski et-al.. 1992: Lobb et-al.. 1995). .Meamrable soil
accumulation in lower landscapr positions and depositional areas is largely due to the transport
capacity of water runofi being ttxcsedsd as concave shaped slopes slow water movement and
sedimrnt drops out of solution. Long-trrm trends from water erosion would suggest maximum
soil losses from mid slope areas imrnediatrly above lower concave shaped positions (Flanagan
and Livingston. 1995). Cesium-137 measurement of net soi1 loss. howewr. indicate maximum
soil movement to be from crest. upper to mid dope positions with the greatest loss surrounding
the conves sloped areas (de Jong et.al.. 1983: Kachanoski and de Jonp. 1984: Bernard and
Laverdiere. 1992: Kachanoski et-al.. 1992: Lobb et.al.. 1995). In topographically complex
landscaprs. the dominant erosion process of the convex area has recently been reported to be
Page 22
tillage erosion (Lobb et al., 1995).
The relative magnitude of water and tillage erosion in southem Ontario hm not been
measured. The "'Cs tracer method is the most accurate method available in measuring the
combined soil redistribution of both these processes in complex landscapes. Accurate prediction
by mathematical modelling of both processes is required for tùture f m management planning
and soil degradation prevention.
1.3 Simulation modeIs to predict field-based water erosion
A mathematical mode1 is a limited imitation of reality. the degree of limitation vacing
with its objectives. Mathematical models in agriculture have been used. cimong others. to: i )
specib areas of deficient knowledgr: i i ) spread scientific knowlsdge to the final users ( tield
agronomists. famers): iii) induce adoption of best management techniques: iv) proLide a
coherent tiamework to understand the behaviour of complex systems: v ) allow surnmarization.
interpolation and. with proper care. extrapolation of data: vi) more etlicirntly use research data
collrctrd at rver incrrasing costs: and vii) test future scenarios (France and Thomlry. 1983 ).
The divenity of modrls availablr to predict soil erosion vary in their degree of
complexi& and output products. It is important to determine the variability such models produce
if they are to satisfv the above mentioned uses. Cesium-137 analysis of a representative
landscape provides a baseline of information to compare the degree of spatial accuracy over a
variable sloping field. Erosion prediction map then be adjusted to better compute the modelling
technology of most rigour and accuracy.
Prediction with models currently used for estimating soil erosion rates van. in their spatial
Page 23
and temporal accuracy. Depending on the intention of each. rnodelling the cornples
interrelationships of topopraphy and erosion may be accomplished with empirical or process-
based structure. Models ditTer in their recognition of processes that change in magnitude and
proportion by landscape and location. Temporal variability in erosion prediction rnay be
considered important on an annual ba i s or event bp event simulation.
1.3.1 Universal Soil Loss Equation (USLE)
The development of the Universal Soil Loss Equation (LISLE) kvas as a result of
extensive natural-runotTand çrosion-plot resctarch in the United States (Musgrave. 1947). The
large dataset totalling 10000 plot years was compiled by Wischmeier and Smith ( 1965) to
correlate relationships hetween soil erosion factors including soil physical and chernical
characteristics. The USLE has become the most widely used soil çrosion model (Risse et al..
1993) due to its simplicity and apparent universa1 application (Renard et al.. 1994).
The USLE uas developed for the purpose of predicting long-trrm average erosion
amounts from cultivated tields for use in conservation planning ( Wischmeier and Smith. 1 978 ).
The model lumps the soil erosion losses tiom sheet and ri11 rrosion. Its empirical design
calculates soi1 losses based on multiplication of six independent tàctor values ( Eqn 1.1 ). The
prediction of m u a i soil loss is determinrd by:
A = R K L S C P (1 .1 )
where A is the estimated soil loss per unit area caused by rainfall and its associated overland
flow: R is the climatic erosivity factor: K is the soil erodibility factor: L is the dope length
factor: S is the dope gradient factor: C is a dimensionless tàctor for cover and management: and
P is a dimensionless factor for conservation support.
Page 24
The driving force behind the calculation is the rainfall erosivity factor ( R ) that considers
expected ninfall intensity of a region. Reasonable application to Canadian locations was limited
until the early 1980's due to the lack of regional rainfdl-runoff rclationships during winter
conditions. Erosion indices were developed for Canada (Wall et al.. 1983: 1988) to partially
overcome the di fferences tiom winter runofî conditions. Limited testing of the model u-i thout
considering soil loss from snow melt events indicated the USLE successfuil~ predicted average
annual field erosion losses in southem Ontario (van Vliet and Wall. 1979). The authors cautioncd
that where soil loss during snow melt may be significant. soi1 loss predictions with the USLE
could underestimate actuaI soil losses by 5 to 15%.
The inherent weaknesses of the mode1 for use in Ontario are man- as it is based on US
soil crosion plot studies of homogeneous soil. slope. and management ( Wall et al.. 1988). .An
adaptation of the USLE for Canadian conditions has been developed as the revised CSLE for
application in Cmada. or RUSLEFAC (Wall et al.. 1998). This rnanual contains information
pertinent to Canadian conditions with methods chat are essentially the sarne as those published b?.
Wischmeier and Smith ( 1965. 1978).
A prima- CSLE model shortcoming that has been improvrd is the necessq addition of
winter and srasonal rrodibility considerations in an anempt to reflect more nonhem conditions.
Soi1 erodibility ( K ) represents the relative inherent resistance of a soil to the detachment.
entrainment. and transport forces of rainfall-runoff(Wal1 et al.. 1988). The USLE soi1 erodibility
factor ( K ) assigns a constant value of K for a particular soil. This implirs that soil erodibility is
independent of any seasonal variation. In southern Ontario. it has been shown that soil erodibilitv
can vary considerably from season to season (Mutchler and Carter. 1983: Coote et al.. 1988: Wall
Page 25
et al.. 1988).
Higher values of soil erodibility have been observed during late winter and earl'; s p h g
than in the summer (Rudra et al.. 1998). At this time. the soils still have a fiost layer at shallow
depth and with thawing. the overlying surface layer becomes saturated and unstable and highly
susceptible to detachment by ninfall or runo& Single value indices to reflect the seasonal
variabiliv for K (Wall et al.. 1988) as well as the regional application of R (Tajek et al.. 1985)
and C factors have been included in RUSLEFAC to estimate average annual soil loss rates.
An rrror assessrnent of the USLE by Risse et al. ( 1993) compared the model prediction to
measured natural runoff plots. The model etEciency. usine the Nash-SutclitTe cormcient of
efficiency. was higher (0.75) for average annual soi1 loss predictions than for predictions on a
yearly basis (0.58). The model ovrrpredicted the plots low erosion rates and underpredicted the
plots that delivered higher rates of erosion.
The temporal variabilities as well as spatial variabilities limit the application of USLE for
soil loss prediction for other than long-term averaging. Spatial evaluation of soil loss on irregular
slopes. as Wischmrirr ( 1974) identities. is a potential source of error with the use of the USLE.
A hillslope c m be dividrd into segments and svaluated as slopes of unifonn gradient: hou-evrr.
the segments cannot be treated as independent slopes if one segment receives runoff from
another. If the intention is to evaluate the potentiai for soil redistribution on a hillslope under
varying soil/climate and management conditions. the USLE based soil loss prediction is not
suitable.
13.2 Revised Universal Soil Loss Equation (RUSLE)
The Soil and Water Consenation Society has revised the USLE (RUSLE) to improve its
Page 26
prediction of average annual soil loss rate (Renard and Foster. 1995). RUSLE retains the
regression equation structure of the USLE. but each of its factor relationships has been either
updated with recent d a t a or new relationships have k e n derived based on modem erosion
theory. The cornputer-based RUSLE includes accessibility to tiles of crop da ta ticlld operations
and climatic information for regions in the USA only. The R and C factors are given biweekl)
values that are multiplied in the mode1 and integrated over the year. This accounting of seasonal
differences in erosive power (RI of the precipitation and proiective ability (C) of the crop is a
signitïcant impro~sment in resional soi1 loss estimation over the USLE. K values computed in
RUSLE are weight-based on their temporal distribution during the yxw.
In cornparing RUSLE and "'Cs soi1 loss estimates. Montgomery et al. ( 1997) adrnitted
the difficulty in concludinp whether or not RUSLE underestirnates or overestimates soi1 loss
compared with the "'CS method. Their investigation revealed no statistical differencr. When
determinations are done over the prriod since '"Cs deposition. it depends on how RUSLE h a
besn applied. a linowledgr of tillage movement rates. and the confidence and reliabilit! of
RUSLE estimates. The reliability of RUSLE estimates depends in part on how complstr the
knowledpe is o f a faim operator's tillage and management practices. The high tillage movemrnt
in the Palouse. WA. region is ofgreater magnitude than water erosion (Montgomen. et al.. 1997)
on certain parts of the landscape and is not accounted for in RUSLE. If an average tillage soil
movement rate \vas knottn and then subtracted from the mean '"Cs soil loss rate. the authors
suggest the difference ma)- represent the water-only "'Cs soil loss rate. Whether this water-only
"'Cs soil loss rate is best rstimated by RUSLE is debatable.
The improvements to the USLE by the Revised version as it is intended. do not account
Page 27
for the spatial variability of soil movement in the landscape whereby there are net losses or
accumulations at point locations in the landscape as measured by '"CS malysis. RUSLE still
retains the limitations of the USLE-based soil Ioss estimates (Risse et al.. 1993: Renard and
Femera. 1 993 ). L'di ke a process-based approach that considers erosion processes. RUSL E has a
lumped-equation structure that does not explicitiy consider mnoff or the individual processes of
detachment. transport and deposition. The output is restricted to the average annual soil loss from
a hillslope or siope segment and can only be applied to conditions and locations govrmctd by the
RUSLE database,
1.33 Erosion-Productivity Impact Cslculator (EPIC)
The Erosion-Productivity Impact Calcuiator (EPIC) was developed for use in determining
the relationship between rrosion and soil productivity throughout the US (Williams et al.. 1984).
Its strength over the LSLE based soil erosion estimation is that it contains physically-based
components for simulating erosion. plant growth. and related processes on a continuous basis.
Crop ~ield. biomass and residue retumed to the soi1 is determined to more accuratel' estirnate
water rrosion potential over a year or many years of crop rotation. The watershed arra EPIC
considers is usually small ( - 1 ha) becausc soils and management etTects are assumed to be
spatial1 y homogeneous.
The EPIC model applies the Soi1 Conservation Service (S'S) curce number rnethod to
computc effective runoff(Soi1 Conservation Service. US. Department of Agriculture. 1964). The
c w e nurnber is a kep variable used in EPIC and is estirnated on the bais of drainage class.
textural classification and other properties of the soil. The model sensitivic to this parameter is
apparently very high (Roloff. pers. com. 1996). It also uses the original or modified USLE to
Page 28
predict erosion sstimates from rainfall and runoff(Sharpley and Williams. 1990).
EPIC was originally desigried For long-tem simulations ( Shapley and Williams. 1 990)
and does not have the capabil i~ to account for year-to-year varïability. In Canada EPIC has been
tested and used mainly ro generate yield. crop residue and soil data with limited success
(Izaurnlde et al.. 1993: Bouzaher et al.. 1993: Toure et al.. 1995: Rolloflet al.. 1997). The
results of using EPIC for yield estimation have implied that it is more appropriate for long-term
studies (Moulin and Beckie. 1993: Roloffet al.. 1998). Preliminq resulb of EPIC's sensitivity
to estimate erosion in Alberta fell within Ievels calculated by the USLE (Izaurralde et al.. 1991).
Calibration of the erosion prediction. however. was deemed necessary for Canadian conditions.
Cornparison of EPIC predicted with obsen-ed natural erosion data collected from hillslopes in
Alberta found that for three year continuous data. poor mode1 agreement of soil loss rates \vas
measured from rainfall and winter runo ff events ( k d n ch et al.. 1 995)
The erosion cornponent of EPIC has relied on the USLE and its modification of the
erosivity index of the R îàctor. MUSLE. The seasonally variable improvements now availabls in
RUSLE have not been incorpontrd into EPIC as of -et: however. this step would be a logical
development. As it remains. EPIC predicts soil erosion based on empincal relationships and can
only be applied to areas for which its parameters have been calibrated.
1.3.1 Water Erosion Prediction Project (WEPP)
Soi1 loss prediction has developed with the advent of computrr technology and a better
understanding of soi1 science to allow process-based calculations to replace empirically
developed models i-r. USLE. RUSLE. The USDA has developed a new generation of process-
based water erosion technology from the Water Erosion Prediction Project (WEPP) for use in
Page 29
soil conservation and environmental planning (Lane and Nearing. 1989)- WEPP is a continuous
simulation mode1 based on physical descriptions of d l and interril1 erosion processes and
sediment transpon rnechanics. WEPP. like EPIC. is a deterministic or mechanistic model that
c m be adapted to different environments or technological advances with relative ease.
WEPP is the most accurate model to capture the spatial and temporal variability o fa tield
within a season or seasons (Nearing et al.. 1994). The process-based model uses a continuity
equation to calculate soi1 movement by detachment and transport from one landscape position to
another on an ment basis. In this manner. soil redistribution in the landscape is simulatsd to
better evaluate theoretically t h change in soi1 characteristics with the impact of soil erosion. The
USDA has released WEPP as a conservation planning tool and for site impact assessrnent
(Flanagan et al.. 1994).
l.3A.l WEPP model cornponents
The hillslope version of the WEPP model provides detailed soil erosion mrasurement and
its spatial variability on a single slope. The model simutates a number of years of erosion and
sums the total soi1 loss over those years for each of 100 equally spaced points on the hillslopr to
obtain average annual values of erosion ( Flanagan and Livingston. 1 995 ). The model calculates
both detachment and deposition at each point. Certain points on the hillslope may cspericnce
detachment dunng sorne rainfall events and deposition during other events depending on the
event duration and intensity. The output of the continuous simulation model represents an
average over al1 the srosion events. The net soil loss estimate is most analogous to USLE
estimates and is most closely tied to on-site loss of productivity.
A second section of the model output includrs the off-site effects of erosion (Foster et al..
Page 30
1995) . Estimates of sediment loads leaving the profile are determined dong with sediment
particle size information to calculate preferential delivery to wateways of a g r i c u l t d pollutants
bound to sediment. The simulation procedure may be run a number of ways to determine values
of soil loss. sediment deposition and sediment delive- oK-site on a storm-by-storrn. monthly.
annual or average annual basis.
The model is subdivided into six components: climate generation. hydrology. plant
grouth. soils. irri~ation and erosion ( Flanagan and Livingston. 1 995 ). -4 bnef esplanation of cach
is included to undsrstand the deterministic nature of the WEPP components.
Climate component
The daily weather data for WEPP may be inserted or created by the climate generator
model. CLIGEN (Nicks et al.. 1995) In continuous mode. the rainfall is disaggrsgated into a
simple single-peak storm pattern (time-rainfall intensity format) for use by the infiltration and
srosion components of the WEPP model.
Hydrology component
The hydrology component is similar to that used in EPIC (Williams et al.. 1896) where the
daily water balance is calculated for the surface and sub-surface soil layers including infiltration.
ninoff. evapotranspiration and deep percolation ( Flanagan and Livingston. 1 995 ). WEPP's
surface hydrology determines the duration of excess rain (whsn intiltration rate is less than
rainfall rate) from rainfall intensity. runoff volume and peak discharge rate for the erosion
component. The amount of water that percoiates into the soil is determined and is used for the
water balance and crop grouth and residue decomposition calculations. These are then used to
update the infiltration. runoff routing and erosion parameten (Stone et al.. 1995).
Page 31
Infiltration subcomponent
Intiltration is calculated using a modified Green and Ampt intiltration rquation. It is one
of the most widely used equations for modeling one-dimensional vertical flow of water into soi1
due to its simplicity and versatility (Risse et al.. 1993). The WEPP mode1 uses a solution of the
Green-Ampt Mein-Larson equation for unsteady rainfall developed by Chu ( 1 978). I t is a two
stage process where initially. the infiltration rate is equal to the rainfall application rate. Alier
pondinp. the infiltration rate begins to decrease until the rate approaches a constant value or h a 1
infiltration rate (Risse et al.. 1993). Improvemrnts continue to be made fiom extensive research
(Risse et al.. 1994: 1995a: 199%: Zhang et al.. 1995: 1996) to develop time-variant infiltration
parameters for WEPP as this area is key to calculating runoff and erosion. Surface runoff is
calculated by kinematic wave overland How routing (Lane et al.. 1988) or simplitïed regression
cquations obtained for a range of rainfall intensity distributions. hydraulic roughnrsses and
inkiltration parameter values.
Plant growth component
The plant growth portion is based on an EPIC approach that predicts potential grokith
based upon daily heat unit accumulation (Williams et al.. 1989). Crop stresses such as inadequate
soil moisture or unfavourable temperature will reduce potential growth. The plant grouth
component provides information to the water balance and extraction of water in the upper la! ers
(Arnold et al.. 1995). Canopy height and cover impact on the erosion components estimate of
intemll soil detachment. Crop residue from leaf-drop. senescence or harvest is relayed to the
residue decomposition mode1 which in tum impacts on the hydrologie and erosion components.
Soi1 component
Page 32
The soil panmeters are used in the hydrology and erosion calculations which r n q change
on a daily tirne step as a nsult of tillage operations. freezing and thawing. compaction.
weathering. or history of precipitation (Flanagan and Livingston. 1995). The tillage sequrnces
identified in the management file will influence the soil bulk densitp. increase the soil porosity.
change soil roughness and ridge height. destroy riUs. increase infiltration pararneters and change
erodibility pararneters. Of these factors. the erosion process is prirnarily intluenced b!- four soil
parameters which are: random roughness. ridge height. bulk densih and effective hydnulic
conductivity ( Alberts et al.. 1995).
Intemll and riIl erodibility parameten are updated in the soil component. Intemll
rrodibility is a measure of the soils resistance to raindrop impact. Rill erodibility is a soils
susceptibility to detachment by concentnted riIl tlow. A third important soil parameter is the
critical shear stress which is defined as the threshold at which a rapid increase in soi1 detachment
occurs per unit increase in shear stress. The irrigation component is not a relevant consideration
for the purpose of the study.
Erosion component
Soi1 srosion on ovrrland tlow areas is calculated as a rrsult of ri11 detachment: transport.
and deposition: and intemll detachment by raindrop impact and sediment transport via sheet tlow
(intemll del ive^. rate) (Foster et al.. 1995). The four hydrologie variables required to drive the
erosion mode1 are peak mnoff rate (m s - ' ) . effective m o f f duration (s). etTective rainfall intensit!.
(m s-'). and effective rainfall rxcess duration (s) (Flanagan and Livingston. 1995). The process-
based computations to determine the soil redistribution and sediment delivery are man- and
cornplex. as are the modrls usrd in the other components. The basis. however. for the erosion
Page 33
computations in the erosion component is the steady state sediment continuity equation (Nearinp
et al.. 1989). The equation is given as:
dG/dx = D,. + D, ( 1.3)
where G is the sediment load (kg s-' m"): x is the distance downslope (m): D, is the riIl erosion
rate (kg s" m"): and D, is the intemll erosion rate (kg s" m"). From this. the net values of
detachment or deposition rates dong the hillslope profile is calculated. The change in sediment
load in the tlow with distance downslope ( 100 segments) is rstimated to express soil loss in
ternis of loss per unit area of the hillslope. Total load for an entire storm event is obtained b>-
multiplying the load per unit time bu the effective storm runoff duration. Detachment in each
segment is computed frorn the difference in load in the segment to that in the previous segment.
The ri11 and intemll subprocess are quantiiled by different parameters and cquations.
Intsmll erosion processes include detachment by raindrop impact as a function of rainfall
intensity. and the transport by shallow sheet flow in the intemli area to rills as a function of slopc
and surface roughness. Dstachment of soil in the nlls is predictrd io occur if the hydraulic shear
stress of the tlow ssceeds a critical value and sediment in concentrated flow is less than the
tlow-s transport capacity.
Simulation of deposition in nlls occurs when the sediment load in the Row is greater than
the capacity of the tlow to transport it. Sediment transport capacity is calculated on a ri11 width
basis as a function of dowslope distance using a simplified function (Lane and Neûring. 1989).
The intemll soil detachment rate makes adjustments for the effecrs of canopy cover and %round
cover as well as rainfall intensity and riIl information. Intemll or sheet flow detachment and
transport occurs only dunng periods of rainfall excess.
Page 34
l3.-#.2 WEPP model sensitivity
The size of the input structure and complexity of the process-based WEPP model has
been the focus of several different evaluations to look at the accuracv of the h c t i o n s within
individual subcomponents. The determination of prediction uncertains for process-based natural
resource rnodels is an important step in model prediction reliability analyses (Beck. 1983).
Nraring et al. ( 1990) performrd a linear sensitivity analysis of the WEPP model with parameter
values of widcly varying value. Doniinant factors related to model responsr were precipitation.
riIl erodibility (Y). ri11 residue cover. and riIl hydraulic friction factors. Saturated hydnul ic
conductivity (kt) and intemll rrodibilitu (6) were considered moderately sensitive parameters.
The intluence of these pararneters however depends on site conditions as K, is more important on
short. flat slopes and Kw, is more important for short. less intense storms and less important for
the larger storms.
In addition to Nearing et al. ( 1990). Chavss and Nearing ( 1991 ) and Tiscareno-Lopez et
al. ( 1994) have conductrd sensitivit~ analysis on many of the parameters uscd in the model. Thsir
results generally agrer that the rainfall pararneters (arnount. duration. and intensity ) and the
pararneters which affect infiltration (surface covrr and hydraulic conductivity) ha\-e the most
impact on the runot! predictions (Risse et al.. 1993). Of the panmeters needed to drive the
intiltration componrnt of the model. the effective hydraulic conductivity (&) is critical. This
model input parameter can be assigned to an initial baseline value (&) that will be intrmally
alterred with management effects or it can be assigned a constant value that is representative of
both the soi1 and the management practices. The latter option is similar to selecting a SCS cunre
number (Soi1 Conservation Service. United States Department of Agriculture. 1972) for
Page 35
predicting runo tT.
The bûseline hydnulic conductivity approach was developed tiorn the SCS curvr nurnber
approach. There is. however. a correction for soil moisture variation that adjusts K, intsmally
within WEPP. An equation based on percent sand. percent clay and cation eschange capacity of
the soil was derived to estimate K, (Risse et al.. 1995). WEPP predicted values of K, wrre shown
to be superior to predictions obtained from the curve nurnber approach. In a cornparison of -LI 24
selected plot runoffevents and WEPP prediction results. the use of the Green-Ampt hydnulic
conductivity estimation (EQ resulted in satisfactop coefficient of determinations (i) (Zhang et
al.. 1996). The accuracy and reliability of predictions were shown to improve frorn an event to
annual to average annual basis. Zhang et al. ( 1995) showed seasonal variations of K. and ninoff
were also adequately represented.
Further validation of the WEPP model has investigated the ri11 rrosion component
(Huang et al. 1996). The riIl erosion equation based on coupied drtachment and transport
processrs propossd by Foster and Meyer in 1972 was tested in a tield esperirnent. The
absenation of a shi fi frorn a detachmeni dominated to a transport dominatsd condition has Itd
the authors to identitj a need to rspand the WEPP database to dewlop a validatrd ri11
detachment and transport model.
Validation studies of the WEPP hydrology component (Van der Zwveep and Stone. 199 1 :
Savabi et al.. 1995: bamer and Albens. 1992) have shok~n the model can perform bettrr if
certain parameters are calibrated rather than estimared based on other propenies.
The WEPP model. as the other significant erosion models. has been developed in the U.S.
and acceptance of rhe detailrd design in other regions such as southem Ontario requires careful
Page 36
consideration. Limitations of winter mno tT routines are being worked on and the application to
southem Ontario's unique combination of complex topography. humid climate and relatively
shallow soils is a concem. However. the mode1 is process-based in nature allowing modification
to an extent and an opponunity to calibrate with a large number of input parameten applicable to
the region. More validation is needed before acceptance of WEPP is met for its intended purpose
as an on-farrn soil conservation tool.
1.4 Tillage erosion process and prediction
The water erosion processes simulated by USLE-based models and WEPP predict
increasing soil loss with increasing dope length. An assumption of pnor soil loss rats predictions
for f m management planning in Ontario was that only vmer erosion relationships be
considered. The indirect impacts of intensive tillage practices would have contnbuted to water
rrosion frorn the drcrease in soil oganic matter. infiltration and soil structure ( Crnger and
Cassell. 1990). This manncr of srodibility is a univcrsally recognized soil crosion factor:
however. rncchanical soil movrmrnt by tillage has largely been overlookçd.
1.4.1 Tillage erosion measurement
The incrrasrd use oftillage implrments in the last 2 to 3 decadrs may be responsible for a
major ponion of the movemsnt of soil within comp1e.u topopph- . Research studies have
reported significant soil translocation associated with tillage practices in North Arnenca and
Europe (Mech and Free. 1942: Sibbsson et al.. 1985: Lindstrom et.al.. 1990: 1992: Govers et-al..
1990: 1994: Quine et.al.. 1990: 1994: Kachanoski et-al.. 1992: Lobb et-al.. 1995 ). In upland
regions of southem Ontario. rates of soil loss from shoulder slope positions have been estimated.
Page 37
Baniston et al. ( 1987) used measured losses of soil depth to estimate rates of 15 kg m" $.
Studies using resident "'Cs to estimate soi1 losses found rates of 10- l 3 kg m-' y i ' (Aspinall et
al.. 1988) and > 10 kg m" y r-' ( Kachanoski et al.. 1992).
It is these upper slope landscape positions exhibiting severe soil loss (>3.3 kg m-' yi') on
which major yield reductions occur (Stone et al.. 1985: Battiston et al.. 1987: Aspinall et al.,
1989). The degradation of soil properties on eroded upper slopr landscapr positions contributed
to 40-50% yield reductions. When taking into account that 75% of the southwestern Ontario land
area is rolling uplands. this yield loss translated into a 1.3-1.7% loss in total crop productivity in
southwestern Ontario ( Kachanoski et al.. 1992).
Research at the University of Guelph has investigated the movement of soil off of
convex. or shoulder landscape positions and have concluded that processes other than aater-
based erosion are responsible (Kachanoski et al-- 1992). Signiticant redistribution of soil
downsiopc due to tillage practices tiom thrse upper areas to areas of concave dimension are
likely prevalrnt throughout cornples landscapes of southwestern Ontario (Kachanoski et al..
1 Lobb et-al.. 1995). Tillage erosion concentrates its cffect at crest and upper dope positions
uith decreasing impact domslope but contributing to soil accumulation in concave areas.
1 A.2 Tillage erosion prediction
The redistribution of soil in the landscape from tillage has been recently investigated and
modelled for southern Ontario conditions (Lobb. 1998). The movement of soil is considered a
simple input/output spstem. similar to the continuity equation. The difference in the mass of soil
translocated into a point on the landscape and the mass translocated out from that point is the net
translocation. The continuity equation to desci be soil redistribution from tillage has also been
Page 38
used by Govers et al. ( 1 994).
Conccptually. the process of tillage erosion is a funcrion of the rrosivity of tillage
operations and the erodibility of the cultivated landscape (Lobb. 1998). The tillage erosivity is
defined as the propensity of a tillage operation. or a sequence of operations. to erodr soil. The
design of the tillage implement detemines its ability to translocate soil as a result of the
combination. mangement and shape of the tillage tools. Landscape rrodibility is drtined as the
propensity of a landscape to be sroded by tillage. The hillslope gradient determines the
magnitude of the rtfect of gravi- on the mass of the soi1 displaced by tillage.
The tillage erosion process has yet to be fully characterized and quantifird. Attempts to
mode1 soil redistribution fiom tillage have rrnphasized the slopr angle and dope gradient as
critical components (Govers txal.. 1991: Lindstrom et.al.. 1992: Quine et-al., 1994: Lobb et.al..
1995). Increasing these components increases tillage translocation linearly. Espenmental studies
on soil movement from tillage have suggested soil Ioss rates in cscess of water erosion processes
on complrx topograph?. to orders of magnitude greater (Quine et-al.. 1991: Lobb et-al.. 1995 ).
The arnount of soil rnoved rnrchanicall y may be affected by tillage drpth and tillage ground
speed and several other related hctors (Kachanoski et-al., 1992).
The influence of slope gradient however is the predominant factor for tillage erosion
potential. It is reasonable to conclude that the more topographicallp cornplex landscapes will be
more prone to soi1 Ioss tiom convex areas. Soi1 erosion observations of random distribution in
the landscape (Daniels et al.. 1983) however have not always confirmed this. The variability in
the degree of erosion obsenred on convex slope positions (Battiston et al.. 1987: Kachanoski et
al.. 1991: Lobb et al.. 1995) ma! be esplained by differences in past tillage patterns.
Page 39
The direct contribution to soil movement by measurable mechanical means characterizes
soil losses tiom tillage as a significant erosion process. Prediction of net soil movement in a
field must therefore consider past. present and future tillage practices and the respective tillage
implement action as w l l as the soil redistribution processes of watsr erosion in any
comprehensive study of soil erosion.
1.5 Soi1 redistribution prediction
Computer models which simulate soil erosion from a hillslope have been developed but
quantitative relationships between different erosion processes occumng over a period of timr are
not well understood. Extensive empincal measurements for water erosion have related soil
erosivity erodi bili tu. slopr characteristics and land management for reg ions across the United
States (Renard and Foster. 1994). Adaptations have been developed for application of this type of
empirical model in many parts of Canada (Wall et al.. 1998). The process-based water erosion
model WEPP drtvrloped more recently in the US has the tlexibilitp for adaptation of rnost tield
scale investigations (Flanagan and Livingston. 1995). The more detailed approach to watrr
erosion modelling oftaking soil redistribution into account. it would appear. would improve the
predictability .
Watsr erosion as the dominant process in southern Ontario has been recently contested
with the measurement of tillage translocation ( Lobb et al.. 1 995). When evaluating hillslope soi1
movement. the relative magnitude of water and tillage erosion processes is not well understood
and needs further evaluation. For upland regions of variable topographp. the mathematical
representation of erosion bp water ( WEPP) and/or tillage (Lobb. 1 998) over the last 30 years may
Page 40
be directly comparable to quantitative estimates of net soil movement as determined from "'Cs
analysis (Kachanoski. 1993). An accompanying assumption is thar "'CS analysis is an accuratr
reflection of what soil translocation has occurred sincr the time of peak nuclear fallout.
1.6 Summary
Soil erosion measurement and prediction are as accurate as the method used. In southem
Ontario. little Iiterature is availahle that reports soi1 erosion measurement and what is
documented is site and situation specitic. Field scale soil erosion information has largely relied
on USLE based technology (Wischmeier and Smith. 1978) for prediction rather than
measurement. The accuracy rrquired to rvaluate soil redistribution over a variable landscape
limits the USLE model for effective use. Erosion prediction currently availablr: that considers
water erosion processes over a spatiaily variable landscape is the WEPP model (Flanagan and
Livingston. 1995). Soil redistribution dong a changing hillslope is predicted as uell as sedimrnt
delivery from a hillslope. An additional soil redistribution process recently documented in upland
southem Ontario topognphp is Mage erosion (Kachanoski et al.. 1992: Lobb et al.. 1995). The
tillage erosion mode1 ( Lobb. 1998) is driven by the dope gradient and simulates soil loss from
convex positions and deposition in concave areas of the landscape. Net soil movemrnt
throughout the landscape from al1 soil erosion processes can be measured using the "'CS
technique (Kachanoski. 1993) for soi1 redistribution analysis.
The three rneans of estimating field scale soil erosion implicitly recognizing soil
redistribution at several positions of a changing landscape are the WEPP model. the tillage
erosion model. and "'Cs analysis. The Rockwood study location \vas chosen for its single. simple
Page 41
S-shaped profile with minimal surface variation. The variabilities to consider between soil loss
measurernents and model predictions were therefore kept to a minimum.
1.7 Study Objectives and Hypothesis
The study will evaluate soil redistribution over a 29 year period at an upland cultivatcd
site in southem Ontario bp studying the soil erosion processes responsible and the utility of
rrosion prediction methods.
Objectives:
1 ) to measurr soil redistribution with resident IZ7cs tracer:
2) to assess the sufticiency of the water erosion rnodel (WEPP) in predicting the soil
redistribution of the experiment site:
3 ) to assess the sufticiency of the Tillage erosion rnodel in prçdicting the soil redistribution of the
esperirnrnt site:
4) to assess the sutficirncy of combining the \vater and tillage erosion mode1 predictions to
improvr prediction of the soil redistribution of the expenment site:
5 ) to evaluate the critical assumptions implicit to the measuremenr and prediction o f soil Ioss
rates.
Hypothesis:
Soi1 redistribution in upland southrm Ontario landscapes as measured with "'Cs can be more
accuratel>. predicted with a combinrd tillage and water erosion model than either model alone.
Page 42
References
Alberts. E.E.. M A . Nearing. M.A. Weltz. L.M. Risse. F.B. Pierson. X.C. Zhang. J.M. Latlen and
J.R. Simanton. 1993. Soil component. In: CSDA-Water Erosion Prediction Project:
Hillslope profile and watenhed mode1 documentation. NSERL Rep. No. i O. USDA-ARS
Natl. Soi1 Erosion Res. Lab.. West Lafayette. IN.
Aspinall. J.D.. R.C. Kachanoski and H. Lang. 1988. Tillage-2000 soil consemation: 1988 annual
report. Ont. Min. of Agr. and Food. Univ. of Guelph. Guelph. ON. 34 pp.
Aspinall. J.D.. R.G. Kachanoski and H. Lang. 1989. Tillage-2000 soil conservation: 1989 annual
report. Ont. Min. of -4gr. and Food. Univ. of Guelph. Guelph. ON. 24 pp.
Battiston. L A . . M.H. Miller and I.J. Sheiton. 1987. Soil erosion and corn yield in Ontario. 1.
Field evaluation. Cm. J. Soil Sci. 67:73 1-745.
Beck. M.B. 1983. A procedure for modeling. In: Mathematical modeling of water quality:
streams. lakes. and resen-oirs. 1 1-41. cd.. G.T. Orlob. New York: John Wilq- & Sons.
Bernard. C . and M. R. Laverdiere. 1992. Spatial redistribution of Cs- 117 and soil srosion on
Orleans Island. Quebec. Can. J. Soil Sci. 72:543-554.
Boothe. 4 . H . . E.R. Samuels and V.K. Mohindra. 1965. Fallout levels in Canada during 1964.
Radiation Protection Di~ision. Ann. Rep. for 1964. Dept. of National Health and Welfare.
Ottawa ON.
Bouzaher. A.. J.F. Shogren. D. Holtkamp. P. Gassman. D. Archer. P. Lakshminarayan. A.
Carriquis. R. Reese. W.H. Furtan. R.C. Izaurralde. and J.R. Kiniry. 1993. .4gricultural
policies and soil degradation in western Canada: An agro-ecological econornic
Page 43
assessment (Rrp. 2: The environmental modelling spstem). Tech. Rep. 5!93. .4gric.
Canada- Policp Br.. Ottawa ON.
Brewster. G.R. and V. Pillay. 1991. Lrsing Cs-137 to assess soi1 erosion in the Cobequid Bay area
of Nova Scotia. Can. J. Soil Sci. 71 :26 1-24? (Abstr. ).
Brown. R.B.. N.H. Cutshall and G.F. Kling. 1981a. Agricultural erosion indicated by ' ' 7 ~ s
redistribution: 1. Ltvels and distribution of "'Cs activity in soils. Soil Sci-Soc. ..\m. J.
45: 1 184- 1 190.
Brown. R.B.. G.F. Kling. and N.H. Cutshall. 198 1 b. i\gricultural erosion indicated by ''?Cs
redistribution: II. Estimates of erosion rates. Soil Sci. Soc. Am. J. 45: 1 19 1 - 1 197.
Butler. G.C. 1980. Radioactivity in the Canadian environment. NRCC No. 18 134. Dept. Energ!.
Mines and Natural Resources. Ottawa Ont.
Cambray- R.S.. E.M.R. Fisher. K. Playtord. J.D. Eakins and D.H. Peirson. 198 1. Radioactive
tàllout in air and min: results to end of 1980. United Kingdom Atomic Energy Authority.
Publ. XERE-R 1 O O S 8 . Hantell. Oxon.. England.
Carter. M.W. and A.A. bloghissi. 1977.Three decades of nuclear testing. Health Phys. 3355-7 1 .
Chaves. H.M.L. and MA. Nearing. 199 1. Uncenainh. analysis of the WEPP soi1 srosion rnodel.
Trans. Am. Soc. Agric. Eng. S ( 6 ) :Y37-2W.
Chu. S.T. 1978. Infiltration during an unstrady rain. Water Resources Res. 1 4 3 ):46 1466.
Coote. D.R.. J. Dumanski and J.F. Rarnsey (ed.). 198 1. .h assessment of the degradation of
agricultural lands in Canada. Land Resource Research Institute- Cont. No. 11 8. Res. Br.
Agric. Canada Ottawa ON.
DanieIs. R.B.. J.W. Gilliam. D.K. Cassel1 and L.A. Nelson. 1985. Soil erosion ciass and
Page 44
landscape position in the north Carolina Piedmont. Soil Sci. Soc. Am. J. 49:99 1-995.
de Jong. E.. C.B.M. Begg and R.G. Kachanoski. 1983. Estimates of soil erosion and drposition
for some Saskatchewan soils. C m . J. Soil Sci. 63607-0 17.
de Jong. E.. H. Villar and J.R. Bettany. 1982. Preliminary investigations on the use of "'CS to
estimate erosion in Saskatchewan. Can. J. Soil Sci. 62673-683.
Ellison. W.D. 1947. Soil erosion studirs. Part II. Soil detachment hazard bu raindrop splash.
.4gncultural Engineering 28: 1 9720 1 .
Flanagan. D.C. and S.J. Livingston. 1995. WEPP user su rnmq. NSERL Rep. No. 1 1. USDA-
ARS. Natl. Soil Erosion Res. Lab.. West Lafayette. IN.
Flanagan. D.C. (ed. ). 1 994. Water Erosion Prediction Project: ver. 94.7 User Summap. NSERL
Rrp. No. 9. USDA-ARS Natl. Soil Erosion Res. Lab.. West Laîàyette. IN.
Foster. G.R. and L.D. ,Meyer. 1972. Transport of soil particles bp shallow flow. Trans. ASAE
15( 1 ):99- 102.
Foster. G.R.. D.C. Flanagan. M.A. Nearing. L.J. Lane. L.M. Risse. S.C. Finkncr. 1995. Hillslopr
erosion cornponent. In: USDA-Watrr Erosion Prediction Project: Hillslop<: profile and
watershed mode1 documentation. NSERL Rep. No. 10. USDA-ARS Natl. Soil Erosion
Res. Lab.. West Lafayette. IN.
France. J. and J.H.M. Thomley . 1 983. Mathematical modeis in agriculture. Buttenvorths Publ..
London. UK. 335 pp.
Garcia-Oliva. F.. R. Martinez Lugo and J-J Maass. 1995. Long-term soil erosion as detmnined
by "'Cs redistribution in an undisturbed and perturbed tropical deciduous forest
ecosystem. Geoderma 68: 135- 147.
Page 45
Govers. G.. K. Vandaele. P.J.J. Desmet, J. Poesen and K Bunte. 1994. The role of tiIlage in soil
redistribution on hillslopes. EUT. J. Soil Sci. 45A69-478.
Huang. C.H.. J.M. Bradford and J.M. Laflen. 1996. Evaluation of the detachment-transport
coupling concept in the WEPP ri11 erosion equation. Soi1 Sci. Soc. Am. .J. 60(3):734-739.
Hudson, N. 1995. Soil Conservation. 3rd edn. Iowa State Unit.. Press.
Hutchinson-Benson. E.. J. Svoboda and H.W. Taylor. 1985. The latitudinal inventon; of l37Cs
in vegetation and topsoi1 invorhtner Cndaa 1980
Izaurralde. R.C.. J. Tajek. F. Larney and D. Dzikowski. 1991. Evaluation of the suitabilin- of the
EPIC mode1 as a tool to estirnate rrosion from selected landscapes in Alberta. -4pric.
Canada-Alta. Agric. Edmonton. -48. 67 pp.
Jedrych. A.T.. C.R. Wright and D.S. Vandenvel. 1995. CAESA-soi1 quality. water erosion
annual report. Xlta. ..\gric. Food and Rural Dev't. Edmonton. AB. 78 pp.
Kachanoski. R.G. 1987. Cornparison of measured soi1 137-cesium losses and erosion rates. Can.
J. Soil Sci. 67: 199-203.
Kachanoski. R.G. 1993. Estimating soi1 loss frorn changes in soil cesium-137. Cm. J. Soil Sci.
73:629-632.
Kachanoski. R.G. and E. de Jong. 1984. Predicting the temporal relationship betueen soi1
cesium-137 and erosion rate. J. Environ. Qual. 1330 1-304.
Kachanoski. R.G. and A.P. von Benoldi. 1996. Monitoring soil loss and redistribution using
"'Cs. Green Plan Report. Agriculture and Agri-Food Canada. Guelph. ON. 29 pp.
Kachanoski. R.G.. M.H. Miller and D.A. Lobb. 1992. Management of farm field vanabilin-: 1.
Quantification of soil loss in complex topography. 11. Soi1 erosion processes on shoulder
Page 46
dope landscape positions. SWXEP Rep. 38. Agric. Canada Harrow. ON. 155 pp.
Kiss. J.J.. E. de Jong and L. W. Martz. 1988. The distribution of fallout Cesiurn- 137 in southem
Saskatchewan. Canada. Environ. Quai. 1 7:&5-452.
Krarner. L.J. and E.E. Alberts 1992. Frequency distributions of WEPP 92.24 predicted soil loss.
ASAE Paper No. 92-264 1.
Lance. J-C.. S.C. Mclntyre. J. W. Naney and S.S. Rousseva. 1986. Mrasuring sedimrnt
movement at Iow erosion rates using cesium- 137. Soil Sci. Soc. Am. J. 50: 1303-1 309.
Lane. L.J. and M.A. Nearing. 1989. Water Erosion Prediction Project: profile rnodel
documentation. YSERL Rsp. 2. Natl. Soil Erosion Res. Lab.. USDA-ARS. Purdue Univ..
West Laîàyette. IN.
Lane. L.J.. E.D. Shirley and V.P. Singh. 1988. Modeling erosion on hillslopes. In: Modeling
Geomorphic Systems. Ed. M.G. Anderson. John Wiley and Sons. New York.
Lindstrom. M.J.. W.W. Nelson. T.E. Schurnacher and G.D. Lemme. 1990. Soil movement by
tillage as atrécted by slope. Soi1 Tillage Res. 17:355-264.
Lobb. DA.. R.G. Kachanoski and M.H. Miller. 1995. Tillage translocation and tillage rrosion on
shoutder dope landscape positions measured using 1 3 7Cs as a tracer. Cm. J. Soil Sci. 75 :
21 1-218.
Lobb. D.A. 1998. Tillage translocation and tillage erosion in southuestem Ontario. Unpublished
Ph.D. Dissertation. Univ. of Guelph. Guelph. ON. 12 1 pp.
Longmore. M.E.. B.M. O'Lrary. C. W. Rose and A.L. Chandia. 1983. Mapping soil erosion and
accumulation with the îàllout isotope caesium-137. Aust. J. Soil Res. 2 1 373-385.
Marsh. M.H. and P.H. Groenevelt. 1 992. Effect of surfafe application of polyvinyl alcohol on
Page 47
phosphorus losses in runoff and on corn grow-th. J. Environ. Qual. 2 1 ( 1 ):36-10.
Mark L. W. and E. de Jong. 1987. Using cesium- 137 to assess the variability of net soi1 erosion
and its association with topography in a Canadian prairie landscape. Catena 14:43945 1.
McHenry. J.R. and J-C. Ritchis. 1985. Field erosion estimated fiom "'Cs activity measurements.
Trans. Am. Soc. Agric. Eng. 38:480-483.
McNairn. H. and Mitchel 1. LM. 1992. Locus of control and famer orientation: efiects on
consenation adoption. J. Agric. Environ. Ethics 5( 1 ):87- 10 1.
Mech. S.J. and G.R. Free. 1942. Movement of soi1 during tillage operations. .4griculture
Engineering. 23:3 79-38?.
Montgomery. J.A.. A.J. Busacca. B.E. Frazirr and D.K. McCool. 1 997. Evaluating soi 1
movemrnt using cesium- 137 and the Revised Universal Soi1 Loss Equation. Soil Sci.
Soc. Am. J. 6 1 5 7 1-579.
Moulin. ..\.P. and H.J. Beckir. 1993. Evaluation of the CERES and EPIC models for predicting
spring wheat grain yield over time. Cm. J. Plant Sci. 73713-71 9.
Musgrave. G.W. 1947. The quantitative evaluation of factores in w t e r erosion. a tirst
approximation. J. Soil Water Cons. 1(3): 133- 138.
Nearing. M A . . L. Deer-Ascough and J. M. Latlen. 1990. Sensitivity analysis of the WEPP
hillslope profile erosion rnodel. ASAE. 33(3):839-849.
Nearing. M.A.. G.R. Foster. L.J. Lane and S.C. Finknsr. 1989. A process-based soil erosion
model for USDA-Water Erosion Prediction Project technology. ASAE. 32(5): 1 587- 1 593.
Nearing. M. A.. L.J. Lane and V.L. Lopes. 1994. Modeiing soil erosion. In: Soil erosion research
methods. 2nd ed. (ed.) R. Lal. Soil Water Cons. Soc.. Ankeny. 10. 127-1 58 pp.
Page 48
Nickling. W.G. and J.G. FitzSimons. 1983. Relationship of soi1 type and agncultural systems to
wind erosion in southwestern Ontario. In. (ed.) S.A. El-Swaifq.. W.C. Moldenhauer and
A.Lo. Soil erosion and conservation. Soi1 Water Cons. Soc. Ankeny. 01. 34-50 pp.
Nicks. AD.. L.J. Lane and G A . Gander. 1 995. Weather generator. In: US DA- Water Erosion
Prediction Project: Hillslope profile and watershrd mode1 documentation. NS ERL Rep.
No. IO. USDA-ARS Natl. Soil Erosion Res. Lab.. West Lafayette. IN.
Nolan. S C . L.J.P. van Vliet. T.W. Goddard and T.K. Flesch. 1997. Estimating storm erosion
with a raintàll simulator. C m . J. Soil Sci. 77569-676.
Ontario Department of r\griculturr and Food. 1967. Agricultural statis~ics for Ontario. Publn 20.
Statistics Section. F m Economics. Co-operatices and Statistics Branch. Ontario
Department of .Agriculture and Food. Toronto. ON. 1 1 1 pp.
Pennock. D.J. and de Jong. E.. 1990. Spatial panern of soil redistribution in boroll landscapss.
southern saskatchewan. Canada. Soil Science. 1 SO(6): 867-873.
Quine. TA. and Walline D.E.. 199 1. Rates of soil erosion on arable fields in Britian:
quantitative data from cassium- 1 3 7 measurements. Soil Use And Management. 7(4): 1 69- 1 76.
Quine. T.A.. P.J.J. Desmet. G-Govers. K. Vandaele and D.E. Walling. 1994. -4 cornparison of the
roles of tillage and u-ater erosion in landtom development and sediment export on
agncultural land near Lruven. Belgium. Variability in Stream erosion and sediment
transport. IAHS Publ. no. 224. 77-86 pp.
Renard. KG.. G.R. Foster. D.C. Yoder and D.K. McCool. 1994. RUSLE revisited: Status.
questions. answers and the future. J. Soil Watrr Cons. N(3 ) : I 13-210.
Renard. KG. and G.R. Foster. 1993. RUSLE user guide. ver. 1.04. Soil and Water Consenation
Page 49
Society. hkeny . IO.
Risse. L.M.- MA. Nearïng. A.D. Nicks and LM. Laflen. 1993. Error assessrnent of the Universai
Soil Loss Equation. Soil Sci. Soc. Am. J. 57(3):825-833.
Risse. L.M.. M.A. Nearing and M.R. Savabi. 1994. Determining the Green-Ampt effective
hqdrautic conductivih from raintàll-mnot'f data for the WEPP model. T m s . .;\m. Soc.
Agric. Eng. V(?):-C 1 14 18.
Risse. L. M.. B.Y. Liu. md M A . . Nearing. 1995a. Using cun+e numbers to determine baseline
values of Green-Ampt effective hydnulic conductivities. Water Res. Bulletin 3 I ( 1 ): 147-
159.
Risse. LM.. M.A. Nearing and X.C. Zhang. 199%. Variabiiity in Green-.4mpt effrctive
hydnulic conductivi@ under fallow conditions. J. Hydrology. 169: 1-14.
Ritchir. J.C.. J.L\- Sprabem and J.R. McHenry. 1974. Estimating soil erosion Iiom the
redistribution of fallout ':'Cs . Soil Sci. Soc. Am. Proc. 38: 137-1 39.
Ritchie J-C. and J.R. .LlcHenry. 1973. Vertical distribution of fallout cesium-137 in cultivated
soils. Radiat. Data Rsp. 1 -k727-738.
Ritchie. J.C. and J.R. McHenn. 1978. Fallout "'Cs in cultivated and noncultivaied Xorth
Central U.S. waersheds. .i. Environ. Quai. 7:4O-U.
Ritchie. J-C. and J.R. McHenry. 1990. Application of radioactive fallout Cesium- 137 for
measuring soil erosion and sediment accumulation rates and patterns: A review. J.
Environ. Qual. 1% 15-233.
Ritchie. J.C. and C.A. Ritchie. 1997. Bibliography of publications of "'~esiurn studies related to
erosion and sedirnent deposition. International Atomic Energy Agency-Techdoc.
Page 50
Ropo~vski. .4.S.. and T. Tamura. 1970. Environmental rnobiliv of cesium- 137. Radiat. Bot.
1 O:3S-l5.
RoloK G.. R. de Jong and M.C. Nolin. 1998. Crop yield. soil temperature and sensitvity of EPIC
under central-eastem Canadian conditions. Can. J. Soil Sci. (in press).
Rudra R.P.- Dickinson- W.T.. Wall. G.J. 1998. Problems regarding the use of soi1 erosion
models. In. J. Boardman and D. Favis-Mortlock (Editors). Modelling Soil Erosion try
Water. NATO AS1 Series- Chapter 13. Vol. I 55.
Savabi. M.R.. D.C. Flanagm B. Hebel and B.A. Engrl. 1995. Application of WEPP and GIS-
GRASS to a srnaIl watershed in Indiana, J. Soil Water Cons. 50(5):477483.
Sharpley. A N . and J.R. Williams (ed). 1990. EPIC-ErosionlProductivity Impact Calculator: 1.
Mode1 documentation. USDA Tech. Bull. No. 1768. 235 pp.
Shelton. I.J. and G.J. Wall. 1998. tndicator of risk of soil degradation: erosion componcnt. the
risk of soil erosion in Canada. .Agi-En~ironrnental Indicator Projec t. Report no. 25.
Agriculture and Agri-Food Canada. GPCRC. Research Branch. Agriculture and .Agi -
Food Canada- Guelph. ON. 59 pp.
Sibbesrn. E.. C.E. Andersen. S. Andersen and M. Flenstrd-Jensen. 1985. Soil movement in long-
term field esperirnents as a result of cultivations: 1. -4 mode1 for approximating soi1
movement in one horizontal dimension by repeated tillage. Esperimental Agric. 3 1 : 10 1 -
107.
Soil Consemat ion Senfice. LX. Department of Agriculture. 1 964. National Engineering
Handbook. Section 4 Hydrology . Washington. DC.
Stone. J.J.. L.J. Lane. E.D. Shirley and M. Hemandez. 1995. Hillslope surface hydrology. Ln:
Page 51
USDA-Water Erosion Prediction Project: Hillslope profile and watershed model
documentation. NSERL Rep. No. 10. USDA-ARS Natl. Soil Erosion Res. Lab.. West
Lafayette. IN.
Tajek. J.. W.W. Pettapiece and J.A. Toogood. 1985. Water erosion potential of soils of
Xlbertxestimates using a moditied USLE. Agric. Canada Tech. Bull. no. 1985-29.
Ottawa ON. 35 pp.
Tiscareno-Lopez M.. V.L. Lopes. J.J. Stone and L.J. Lane. 1994. Sensitivity analpis of the
WEPP watershed model for rangeland applications - II. Channel Processes. Trms XS.;\E
37(1):151-158.
Tossell. R.W.. W.T. Dickinson. R.P. Rudra and G.J. Wall. 1987. 4 portable rainfall simulator.
Cm. Agric. Eng. 29: 1 55- 162.
Toure. A.. D.J. Major and C.W. Lindwall. 1993. Cornparison of five wheat simulation models in
southern Alberta. Cm. J. Plant Sci, 75:61-68.
Unger. P. W. and D.K. Cassel. 199 1. Tillage implement disturbance effscts on soii propenies
relatrd to soi1 and water conservation: a litrrature revirw. Soil Tillagr Rrs.. 1 9363-382.
Van der Zwerp. and J.J. Stone. 1 99 1. Evaluation of the WEPP hillslope profile hydrology
component on a semi-arid rangcland watershed. Paper. Am. Soc. Agric. Eng.. St. Joseph.
MI (912552).
Van Vliet. L.J.P. and G.J. Wall. 1 979. Cornparison of predicted and measured sheet and ri11
erosion Losses in southem Ontario. Can. J. Soil Sci. 592 1 1-2 13.
Van Vliet. L.J.P. and G.J. Wall. 1 98 1. Soil erosion losses fiom winter runofT in southern Ontario.
Can. J. Soil Sci. 61 :45 1454.
Page 52
Wall. G.J. and G. Driver. 1982. Cropland soil erosion. Estimated cost to agriculture in Ontario.
Ont. Min. of Agr. and Food. Guelph. ON. W pp.
Wall. G.J.. W.T. Dickinson and J. Greuel. 1983. Rainfall rrosion indices for Canada east of the
Rocky Mountains. C m . J. Soil Sci. 6 W 7 1-280.
Wall. G.J.. W.T. Dickinson. R.P Rudra and D.R. Coote. 1 988. Seasonal soil erodibili~p variation
in southwestern Ontario. Can. J. Soil Sci. 68:-Il7-W.
Wall. G.J.. E.A. Pringle and R.W. Sheard. 1991. Intercropping red clover with silagr corn for soil
erosion control. Cm. J. Soi! Sci. 7 1 : 137- 145.
Wall. G.J.. E.A. Pringlz. G.A. Padbury. H.W. Rees. J. Tajek. L.J.P. van Vliet. C.T. Stushnott:
R.G. Eilers and J.M. Cossette. 1995. Erosion. pp. 60-76 In: D.F. Acton and L.J.
Gregorich (eds.) The hcalth of our soils-ton-ards sustainable agriculture in Canada.
CL BRR. Research Branc h. Agriculture and .Agi-Food Canada. Ottawa. ON.
Weast. R.C. (rd.). 1987. Handbook of chernistry and physics. 1987- 1988. 68th cd. Chemical
Rubber Company. Boca Raton. FA.
Williams. J.R.. C.A. Jones. P.T. Dykc. 1984. -4 modrling approach to determing th srelationhip
between erosin an dsoil productivity. Trans. Am. Soc. Agric. Eng. 27: 129-14-1.
Williams. AR.. C.A. Jones. J.R. Kiniry and D.A. Spanri. 1989. The EPIC crop grouth modrl.
Trans. Am. Soc. .ALgric. Eng. 32(2):497-5 1 1 .
Wischmeier. W.H. 1974. New developments in estimating water erosion. 29th annual mtg.. Soil
Cons. Soc. Am. Proc. pp. 1 79- 1 86.
Wischmeier. W H . and D.D. Smith. 1965. Predicting rainfall-erosion losses from cropland east of
the Rocky Mountains-Guide for selection of practices for soil and water conservation.
Page 53
USDA Agnc. Hand. No. 281. US. Goy. Print. Office. Washington. DC.
Wischmeier. W.H. and D.D. Smith. 1978. Predicting rainfall-erosion losses. A guide <O
conservation planning. LYS. Dep. .4gric.. Agnc. Hand. No. 537. Washington. DC.
Zhang. X.C.. MA. Nearing and L.M. Risse. 1995. Estimation of Green-Ampt conductivity
parameters. 1. Row crops. Trans. Am. Soc. Agnc. Eng. 38(4): 1069-1 077.
Zhang. X.C.. M A . Ncaring. L M . Risse and K.C. McGregor. 1996. Evaluation of WEPP runotT
and soi1 loss predictions using natural motTplot data. Trans. .\m. Soc. Agric. Eng.
39(3 ):855-863.
Page 54
Chapter 2 : Measurement and prediction of soil erosion on an upland landscape in
southern Ontario
2.1 Methods
2.1. l Rockwood field site characterization
The studg sitr \-as selected to represent a laqe portion of the medium trxtured soils and
topography of the glaciated upland regions of southem Ontario. Complex and simple slope
topography dominate the upland region with slope lengths of usually less than lOOm and slope
uradients tvpicallp of 2 to 16 96 ( Shelton et al.. 199 1 ). Thr landform of the site area was a gent l~ =
sloping drumlin deteloped on glacial stony till with a loarn soil cap.
The study field location was 7 h northeast of Guelph. Ontario. within gently
(undulating) rolling topography located on the side slope of a drumlin of simple slope from crcst
to lower-depressional area. Two adjacent cultivated fields used in the study had bren cropped up
and d o m the dope with contrasting management systems providing both a consenmion and
conventional tillagr history. A narrow native forest stnp bisecting the two fields allowcd
sampling of background Isvels of soil parameters unaffected by cultivation. The geographic
location was 43"37'50" N latitude (Northing 4830950 UTM) and 80"11'0OW iV longitude (Easting
565880 LrTM).
The f m sitr was likrly cleared in the early 1800's prior to the establishment of
communities nearby such as Guelph in 1821. The area has been in mixed agricultural farming
systems (cereal crops and forage or pasture) up until the middle of this centun;. After the second
world war. ownership of the properties aas transkrred and the conventional field came undsr
more intensive cultivation. Both ilelds reportedly received livestock manure and both had
Page 55
forage/grass plantings until the late 1 950's.
The humid temperate climate of the southem Ontario region provides adequate
precipitation for agncultural production with uniform precipitation throughout the year (Table
2.1 ). Climatic records ftom the University o f Guelph 7km southwest tiom the site indicate annual
precipitation of 83 Jmm ( Atmospheric Environment Senrice. 1 98 53.
Table 2.1 : Guelph OAC clirnatr normals for the 30 y period. 195 1 - 1980.
Tempe rature Preci pitation Sunshine
b1a.s ("C) Min ("Cl Rain (mm) Snow (cm) Total (mm) (hours )
Jartua- -3 -4 - 1 1.0 2 1.3 36.3 57.6
Februarq -2.3 - 10.6 2 1.6 29.3 50.9
Marc h 2. I -5.9 37.3 25.1 62.4
April 10.9 0.7 67.5 6.2 73 -7
Mal.
June
JuIq
August
Septem ber
October
Novem ber
Decem ber
Y e a r
2.1.1.1 Land use history
Interviews with former property ouners and nrighbouring f m owners were used to
establish land use pnctices from about 1950 to the present. At the time of peak nuclear ';'CS
45
Page 56
fallout. site farm owners believed the consenation field was under forage cover for the early
portion of the period with the field being faIl mouldboard ploughed in 1961 and planted to mixed
cereal -min the followin two years as part of a forage-cereal crop rotation. The conventional
field reportedly was cropped to cereals and corn during this period of significant "'CS deposition
as part of a newly established wheat-corn-barley crop rotation. During the year of peak
radionuclide deposition in the northrm hemisphere in 1963 (de long et al.. 1982: Prnnocli.
1990). the field %as reponedly planted to winter wheat. a crop that provides cover çarlp in the
season and into the summer period. Tillage practice at the tirne was to moldboard plough in the
faIl to completely incorporate surface crop residue. In 1964. the field was reportedly planted to
corn in rou-s up and down the dope.
Tillage practices of thrse fields would have with time thoroughlg mixed "'CS deposited
and entrained at the surface into the surface soi1 horizon. The conservation field would not have
incorporatsd '"Cs depositrd on the surface until the fall tillage of 1962. Fa11 tillage on the
conventional tirld would have incorporated the surface entrained "'Cs deposited on the surtacs
at sach ysar end.
Subsequrnt cropping and tillage pncticrs on the two fields were of contrastinz
management. Crop rotation on the conservation tield maintained a significant forage proportion
and included a 14 y penod of continuous forage between 1 978 to 199 1 with two interruptions of
faIl ploughing and spnng reseeding. Both farm operaton continued to apply manure sporadically.
The proportion of row crops increased on the conventional field fiom the time of '"CS deposition
until a change in ownsrship in 1975. This coincided with a trend to a rnonocuiture system of
continuous corn production in man- areas of the province. The 17 years from 1975 to 199 1 were
Page 57
under fa11 mouldboard plough. corn crop management. In the 34 years fiorn 1 938 to 1 99 1. the
conventional field had primary tillage each year in the fa11 with few exceptions. The
conservation managed field site in contrast had 12 years of primary tillage in the &II.
The forest area between the two tields situated perpendicular to the slope has remained
uncleared because of its early designation as a township right-of-way. This native area-
unaffected by agricultural production systems. \vas considered a control site for background soil
characterization and '"Cs reference tsvels.
2.1.1.2 SIope profile measurement
The simple. single slope relief of the field site supported a gridpoint sarnpling schemr to
characterize the landscape features and soil properties and for investigating soil redistribution
over the hillslope. Elevation measurements were taken for both fields bu topographie s u n q in
the spring and summer of 1992 using a total station instrument (Theodolite). It recorded data
rlectronical1~- msasuring angles and distances to a visible target (rod and prisrn). A grid pattern
was rneasured coinçiding with soil sarnpling locations that were on a 20m x 20m spacing. At
rach of the sarnple gridpoints. 8 additional survry points surrounding the gridpoint in a square of
approximately 1 m spacing were used to detemine plan (across slope) and profile tdownslope)
cun-ature measurements ( Kachanoski and von Bertoldi, 1996).
The detailed elevation information provided landscape position identification to classifp
into the following iandform elemenis (Pennock et al. 1987): divergent back slope (DBS).
convergent back dope (CBS). level (L). divergent shoulder (DSH). convergent shoulder (CSH).
divergent foot-slope (DFS). and convergent foot-dope (CFS). Landscape position was also
identified by soil survey (Denholm. and Schut. 1993) at each gridpoint as crest. upper. mid. lower
Page 58
or depressional slope area. In addition to the 10m spaced measuremrnts taken at the consen-ation
tield. single points were suri-eyed in between the gridpoints to create a 10rn s 10m coverage.
Data from both field sun-eys wcre used to genente a three-dimensional map of the site (Figure
2.1) using SURFER cornputer software (Golden Software Inc.. 1997). The location of t h 20m x
20m sanipling grid is included on the hillslope surface. Individual gridpoint location and
identification is included in Appendix A.
Figure 2.1 : Rocku-ood site contour map of consen-ation and conventional fields with
sampling points identitied.
x = grid sampling points
Page 59
2.1.1.3 Soil properties
Field sampling methods
Soil samples were taken in the Ml of 1991 from the conservation field and the s p n n of
1992 from the conventional field in the rectangular grid pattern spaced IOm apart in 1 transects
up and doun the dominant slope. The prid sampling method was used since there \vas no
signiîïcant surtàce relief on site.
At each sridpoint. rrpresentativc grab samples were talien from the full thickness of the
Ap horizon. Of the 40 gridpoints on the conservation field ( 10 slope positions by 4 transecis).
additional samples were collected of the B horizon at every fourth sampling point and at every
other of these locations. a further sampling of the C horizon was collected. The thicknrss of each
horizon \\-as recorded. Thosr gridpoints sarnpled for B and C horizons had depths recorded as
determined by the consistent pressnce of calcium carbonates in the C horizon. The same
sampling design was followed on the conventional field: however. there were 1 8 gridpoints ( II
dope positions by 4 transects) of Ap sampled as it was a slightly longer hillslope.
Tn O soil pedons were selccted that represented the well and poorly dnined soils of the
hillslope. Pits of about I m s 1 m by 1 rn depth were sampled. The soil horizons were identitied
and described according to Day ( 1983 ). (Appendi'c A). About 1 kg of loose soil \ a s collected
from each horizon.
Bulk density measurements were taken usin9 an Uhland bulk core sarnpler of 1 k m depth
by 7cm diameter (456cm' \-olume) fiom the middle of the A horizon at every sampling point.
Soil moisture conditions were adequate to avoid unnecessary compaction. If samples did not fiIl
Page 60
the core resenoir area a dot opening on the side of the core could be used to measure the hright
of sarnple and adjust the volume calculation.
Sarnples for soil "'CS analysis were collected at each gridpoint of IOm spacings to the
drpth of the Ap surface horizon which equalled or excreded the depth of "'Cs enriched soil (-4p
depth varied from 1 k m to 38crn). The sarnpling method in the cultivatrd fields involved cutting
a representative slice d o m a small pit face which was removed u s i n a shovel to pro\.idr at least
1 kg of dried sieved soil for analysis.
In the forest area. the Uhland bulk density core sampler was used with the assumption
that the rnajority of the '"Cs drposited would be within a few cm of the surface (Ritchie et al..
1972: Nolin et al.. 1993) with perhaps a srnaIl amount of mixing within the surface l a y r dur to
bioturbation (VandenBygaart. 1998). Nine sites were sampled 1.5m apart in a 3x3 square grid at
a crest position and a depression location. The 1 8 sites sampled exceed a suggested minimum
(Nolin et al.. 1993) of 13 sarnpling sites in a forested site to give m estimation of the msan "'CS
with over 9096 of precision. There \vas \..env little Ah littrr layer (approxirnately 1 cm) on the
forest tloor and any lraf or twig material that had not begun decomposition \vas brushed aside. .A
second 1 5cm depth ( 15-30 cm) was taken from thrre sample points on a diagonal ~ . i t h i n both
squares to measure an'. further vertical movrmrnt of " 7 ~ s .
Laboratory analytical methods
Soi1 samples for chemical. physical and "7Cs analysis were air-dried and roller-ground to
separate the fine earth fraction (Qmrn) from coarse fragments. Pedon and field samples were
analyzed for soil reaction (pH). cation exchange capacity. total carbon. oganic carbon and
calcium carbonate equivalent. and particle size distribution (McKeague. 1978). Pedon
Page 61
descriptions are reported in Appendix A.
Soil "'Cs was rneasured using analytical procedures as described by de Jong et a1. ( 1982)
and Lobb et al. ( 199 1 ). -4 gamma ray spectrometer \vas used to count the release of emitted
gamma particles with a high resolution Germanium-Lithium detector in a known volume and
mass of each soi1 sample. Coupled to a multichannel analyzer. the particle photoelectric rnergy
of the particles erninrd was rneasured from the area under the gamma peak (662 L V ) and in
conjunction with computer software. identified the net counts of "'CS (Bq) in the linown soi1
mass of approxirnatelv 1000g. The analysis k v a s combined with the h o w n mass and depth of
each instrument sample to obtain the total specific mass of "'Cs in Bq kg". Potential detection
error or "'Cs drtection variability was discussed in Lobb et al. ( 1991 ). These m o r s howevcr are
less than the variability found in tield sampling and were considered insigni ficant.
Concentrations of "'Cs were corrected for radioactive decay (Lobb et al.. 199 1 ) between the tirne
of sûinpling and the da'. of labontory analysis using 30.17yrs as the half life of radioactive '"'Cs
( Weast. 1987).
2.1.2 Soil erosion measurement
2.1.2.1 Period of anabsis
The first substantial deposition of "'Cs occurred in 1958 followed by lesser annual
deposition rates until highest levels were deposited during i 962 to 1964. The Test Ban Treaty of
1963 implemented a moratorium on nuclear testing afier which deposition rates declined to much
lesser levels (Carter and Moghissi. 1977). The peak "'Cs fallout in 1963 was considered the
starting point or year 1 for the modeling simulations. The 29 yr penod from 1963 until the
Page 62
sampling of "7Cs in hl1 1991 and spring 1992 on the consenation and conventional fields
respectivelp. was the time frame for analping the contributing erosion processes.
2.1.2.2 Soi1 redistribution estimation using cesium-137
The study used the close relationship between the redistribution of "'CS and the
movement of sroding soil to detennine rates of soil loss. Two methods ofestimation were
available. The linear proportional mrthod of de Jong et al. ( 1983) directly relates the fraction of
'"CS lost to be proponional to the soil loss in the till laver containing the "'Cs. This rnethod is
most applicable where water erosion is dominant and there is no tillage dilution of " 7 ~ s
concentrations from the incorporation of subsoil. For the management systems studicd that
included mouldboard plough and secondary tillage over the simulation period. an rquation that
was more appropriate was the power-function relationship (Kachanoski 1993 ). The rate of soil
loss calculated by the power method for this study used the equation:
A = p , D R" (1- (C, /C, ) '" ) (2.1 )
rvhrre A is the average annual erosion rare (kg m" ).il): p, is the bulk densit). of the soil la).cr in
which "'Cs is present (kg m*'): D is the depth of soil in which the "'Cs is prcisrnt (m): R is the
cmichrnent ratio of the "?Cs concentration in the eroding sediment to that in the plow layer: C , is
the estimated referencç level of "'Cs (Bq m-'): C , is the measured mass of ':?Cs (Bq m.') present
in the soi1 at the time of sampling: and n is the number of years of soil loss undrr consideration
(yr). It was assumed that R=l for soil redistribution. The radioactive decay of the mass of "'CS
during the period considered M a s assumed to be offçet by an average annual estimated gain of I O
Bq m" through atmosphenc deposition for southem Ontario (Kachanoski and von Bertoidi.
1996). Sampled points calculated to have positive or negative values were interpreted as being
Page 63
areas of net soil gain or loss. respectively.
2.1.2.3 Method of analysis
The variabilin; of "'Cs activity h m the forest. conservation and conventional sites was
first determined to establish the expected drgree of soil redistribution. Net "'Cs loss from each
cultivated field \vas determined by first finding the difference in tield concentration ( Bq m") at
each gridpoint from the reference value. averaging the dit'ferences. and then dividing bl- the
reference value. Variation in "'CS levels at the field landscape positions were establishrd from
the mean of 4 transrcts. This position average over these points on the hillslope was the basis for
establishing measured soi1 loss rates that were later used to assess agreement from predicted
model values.
2.1.3 Soi1 erosion prediction
2.1.3. t Water Erosion Prediction Project (WEPP) model
The simulation experiment \vas designrd to fa11 within the boundq conditions of the
WEPP (Flanagan and Livingston. 1995) model operation. The hillslope model version \vas
designed to svaluatr soil redistribution from watrr erosion over a single. varying topographie
pro ti le.
2.1.3.1.1 WEPP input files
The input data required to execute the WEPP model were collected from the Rockwood
site and from climatic data recorded at the University of Guelph. The input requirements for the
process-based mode1 are drmanding and the development of the slope. soil. manasement and
climate data files were cornpleted to the extent possible from measured data.
Page 64
Slope data
The WEPP dope input file required slope gradient data at up to 20 relatiw positions
down the hillslope profile (.-\ppendix B). The positions w r e entered as a percentage of the
hillslope distance. The single. simple S-shaped profile of the field site was considrred for
modelling purposes as one overland tlow element (OFE) having no change in management or
soi1 type. .\ maximum OFE slope length restriction of 100m in the mode1 file buildrr was
overcome by creating the slope file with 200m lsngth in a tsxt rditor. .\nu discontinuities
between OFE boundaries were thus overcome. Since there wsre no signiticant tlon-
concentrations obsenred on the slope other than sheet and riIl erosion. this method was
acceptable by the model cieceloper ( Flmagan. 1997. pers.com.).
The geometry OF the tisld \vas determined from elevation measurements fiom the
topographical sune?. At each intenval down the slope. the average elevation of the 4 transects
across the slopr was used. The conservation fields surveyed pattern of 10m spacing pro\-idrd the
maximum allowabls 20 devation points for generating its 200m slopr profile. The 30m slope of
the conventionally managed tield had been sunq-ed at 20m spacings with additional points at
top and bottorn providing I - l elevation points for generating its slope proille. The other input
requirement for the dope input file was slope aspect in degrees frorn true nonh. This uas
measured on site with a compas.
Soi1 data
The WEPP soil input data tile (Appendix B). requinng physical and chemical soil
parameters. were taken from the gridpoint soil sampling results and two pedon descriptions. The
soil in the area was a Guelph (Humic Gray Brown Luvisol) loarn developed from glacial till
Page 65
parent matenal (Hoffman et al.. 1963: Soil Classification Working Group. 1998). Soil properties
from each horizon averaged fiom the site measurements were: depth fkom the surface. % sand. */O
cl-. % oganic matter. cation exchange capacity (CEC) and % grave1 (Table 2.2).
Critical soil properties in the uppermost layer required for the mode1 include soil albedo.
initial soi1 saturation. baseline intemll soil erodibility (K.,). baseline riIl erodibility (Y). baselinr
critical sheer stress ( r,) and baseline effective conductivity ( K,,). These soil properties are used to
initiate the mater balance and infiltration characteristics of the soil at mode1 initiation tirne of
J a n u q 1 of Year 1.
lnitial saturation was assumed to be 80% for the Rockwood site (Flanagan and
Livingston. 1 993). Soil albedo was determined from the equation:
Albedo = 0.6issp(O.-l*OM) (-.- 3 3 )
where OM is soil organic matter of the surface layer (Flanasan and Livingston. 1995). The
parameters K,. y. T, and K, uere also derived from cmpirical relationships that considrr soil
texture content (Flanagan and Livingston. 1995). For cropiand soils containing greater than 30%
sand. these parameters were calculared by the following rquat ions:
K, = 2728000 - 192 100 * VFS kg s rn4 (2.2 )
K, = 0.00 197 - 0.00030 * VFS + 0.0386e"' '7'('C" s m" (2.4)
z, = 2.67 -i- 0.065 * C - 0.058 * VFS N n i J (2.5)
K, = -0.165 + 0.0086 * S * 1.8 + 1 1.46 * CEC * (-0.75) mm hr" (2.6)
where VFS refers to % vep fine sand. OC is % organic matter. C is O h c l q content S is YO sand
content and CEC is the cation exchange capacity in meq 100gl. The soil data determined for the
WEPP mode1 soil file for the conservation and conventional fields are iisted Table 2.2.
Page 66
Table 3.2: WEPP soi1 input fiie parameters for the conventional and conservation field.
Soi1 Conservation field Conventional field parameter L'nit Horizon Horizon
Depth Sand Cla! OM
CEC Gravel
OC v tSand A 1 bedo
K, Kt - L .
K,
mm
O10
Oh
Y0
meq I00g" O'o
O/o
9 B
kg s m"
s m.'
N m"
mm hi'
Management data
The cropping.'management input tiic requires detailed crop information. tillage type and
sequrncr. and management practice information. Of the man? parameters required for the crop
and implement information. most input data was dependent on the WEPP management database.
Crop and management practices penaining to the simulation period of 1 963- 1 99 1 (Table 2.3 )
were obtained from interviews with current and past landoumers of this period. The management
file was compiled using rhese annuai management records. Values for barley were not availablr
from the WEPP crop database: however. they were substituted by spring wheat which has similar
management. crop grouth characteristics and rvater balance influence.
Page 67
Table 2.3 : Roc kwood site management history from 1 960- 1 99 1.
Conventional field management histop Conservation field management histon
Year Crop Ti 1 lage sequence Year Crop Tillage sequence Spnng Fall Spring Fal l
w .w heat corn bar le). IV .w heat
corn barley w.~vheat
corn corn barie y ui .u heat
corn corn barley w-wheat
corn corn corn corn corn corn corn corn corn corn corn corn
corn corn corn corn
corn
Mplough Disc-2sCult. blplough Disc-ZxCult. Disc-2sC ult blplough Disc-ZsCult. Mpiough Disc-LsCult. Disc-ZsCult Mplough Disc-2sCult. Mplough Disc-ZxCuIt, Mplough Disc-2sCult. Disc-ZsCuIt bl plough Disc-2sCult.blplough Disc-ZsCul t. Mplough Disc-ZsCult. Disc-ZsCult
Mplough Disc-LxCult. Mplough Disc-ZsCult. Mplough
Disc-ZsCult. Mplough Dix-2sCult. blplough Disc-ZsCult. blplough Disc-2sCult. blplough Disc-2xCult. blplough Disc-2sCult. Mplough Disc-LsCuIt, Mplough Disc-ZxCult. Mplough Disc-2sCult. blplough Disc-ZsCult, Mplough Disc-ZxCult. Mplough Disc-ZsCult. blplough Disc-ZsCult. Mplougt!
Disc-ZsCult. Mplough Disc-ZsCult. Mplough
forage forage forage m-grain m.grain forage forage forage forage rn-grain m .grai n
forage forage forage
forage Corn Corn
m-grain forage forage fonge forage forage forage forage forage forage forage forage forage forage
forage
blplou~h Disc-ZsC ult,Mptough
Disc-ZsCult. kl plough Disc-ZxCult
Mplough Disc-2sCult. blplough Disc-ZsCult. Mplough Disc-ZsCuIt
Mplough Disc-Hasr-2sCult. Mplough Disc-Ham2sCult. Mplough
Disc-HamZsCuI t. Mplough Disc-Harr-ZsC ult
Tillage implements listed are: Disc-single gang disc. Cult-cultivator with narrow teeth. Mplough-rnouldboard
plough. Ham-smal l toothed drag harro~v. with spring and fall tillage sequence separated bq comma.
Page 68
The initial condition tiles beginning January 1. 1960 were created for both tirlds to allow
two years of mode1 cdibration prior to the period of most significant fallout. 1961-64.
Management practices including fertilization level. selected as medium. were based on the
reported agricultural management practices of the f m operators.
Climate data
Climatic data for the Rockwood site was obtained from data records of Agriculture and
Agi-Food Canada in Ottawa WEPP climate input files had been created and generated for
several monitoring sites across the country. The University of Guelph OAC climate station site
located 7km south West fiom the Rockwood site was one of the WEPP climate input tiles which
included a climatic record €rom 1960. This dataset was input into the climate input file structure
of CLIGEN. the clirnate file generator for WEPP. The structure option was for continuous
simulation in the mode1 using a daiI'; time step. The geographic location. monthly average
masimum and minimum temperatures. monthly average daily solar radiation. monthly average
precipitation totals were included. The climate input file. without breakpoint data. included the
daily precipitation parameters. temperature parameters. solar radiation. and wind chancteristics.
A partial view of the file structure and input values is provided in Appendis B.
2.1.3.1.2 WEPP output files
The WEPP mode1 produced several simulation results. The tiles of interest for mode1
evaluation were the mnot'rand erosion summary information. The output in the summary tilr
\vas segregated into on-site and off-site effects of soil erosion. Off-site impacts of erosion
included the estimated average annual sediment delivery from the hillslope. The on-site effects
contained average annuai soil loss estimates over the areas of the hillslope experiencing soil loss
Page 69
and deposition along a hillslope. giving a table of detachment/deposition at 100 points dong a
hillslope. Predicted soi1 loss rates from the same sampling locations used for "'Cs anal~sis were
extracted and averaged to iïnd the mean soil Ioss rate from each WEPP run.
An initial sensitivity analysis was conducted to determine the model output response with
changes in selected parameter values. This step was to veri@ the high sensitivity of those
parameters identified in the Iitenture that affect soil loss with the model inputs used for the study
simulations. Sensitivity of a model is the evaluation of the relative magnitude of changes in the
model response as a function of relative changes of mode1 input parameters. In equation form.
the sensitivity parametsr S (Nearing et al.. 1990). is given by:
S = [(O? - 01) ! 012] / [ ( I? - 1 , ) / 1,J (2.7)
where 1, and 1, are the least and greatest values of input used. respectively . I l , is the average of 1,
and 1,. 0, and 0, are the output for the two input values. and O,, is the average of the two
outputs. The parameter S represents a relative normalized change in output to a normalized
change in input. which allows a means of comparing sensitivities for input parameters which ma!
have diftèrent ordsrs of magnitude.
The linear ssnsitivity analysis. used by Nearing et al. ( 1990) to evaluate the WEPP model.
has some limitations ( McCuen and Snyder. 1983). including: i ) the linear form of the sensitivitu
parameter does not fully characterize non-linear response. i i ) the linear analysis is univariate.
whereas sensitivity of the model to a variable is dependent upon the magnitude of other
variables. iii) the sensitivity parameter is single-valued whereas inputs are actually random
variables with distributions associated with them. However. as McCuen and Snyder ( 1983) point
out. the sensitivity which represents the extrernes of the physical conditions is often of p r i m q
Page 70
interest. This screening method provides some direction of what pararneters to change and by
how much when calibrating the model for their impact on runoff and soil loss predictions.
For the purposes of this study. those parametee that reponedly have a significant impact
on soil loss were chosen to compare with the rneasured values of soil loss by "'Cs analysis. The
parameter inputs and their ranges tested for mode1 sensitivity are reponrd in Table 2.4.
Table 7.4: WEPP rrosion rnodel pararneters used for sensitivity analysis with calculated base
values and the value range tested.
Parameter Cnits Base value Range of test
Conservatio Conventiona
critical shear stress ( t; ) N m.' 2.7 1 2 .S J 0.20 - 10.00
baseline hydraulic conductivit? (K,) mm hr'; 5.88 7.33 O . 10 - 10.00
The highly sensitive pararnetsrs identified with the sensitivity analysis u-ere usrd for
model adjustmrnt to achiew the lowrst percent dçviation between rneasured ( "'Cs) and
predicted ( WEPP) values of average annual soil loss over the simulation penod. The deviation
was determined bu the equation:
% deviation = [(predicted value - rneasured value) ! mrasured value] x 100 (2.8)
The calibration was conducted in a sequence of steps. Of the selected parameters. hydraulic
conductivity most influences infiltration and amount of runoff. An initial hydraulic conductivit).
(K,) was selected that did not vary throughout the simulation and was tint adjusted to best match
60
Page 71
the measured soil loss. Without runo t'f data from the site to calibratr hydraulic conductivih. the
comparison of soil loss o\zr rhe entire simulation penod \as assumed to be closely related to
runoffover rhis sarne period. M e n the minimum % deviation was artained between measured
and predicted soil loss. the calibrated K, was then used for predicting soil loss results for each of
the soil erosion panmeters independentlu. To calibrate for Y. for example. the equations 2.4 and
2.5 were applied for inputting values of y and r,. To calibratr for Y. the equations 2.3 and 3.5
were applird for values of K, and r,. The calibrated soil parameters for the rwo management
fields were then considered as input values for the WEPP mode1 prediction assessment.
The comparison of average soil loss rates predicted by WEPP to measured "'Cs values
narrowed the focus before simulation outputs of highrst model agreement were identiikd using
the predictrd position values dong the hillslope. This level of agreement was computed by the
Nash-Sutcliffe coefficient of efficiency. R'. (Nash and SutclifTe. I W O ) . This mode1 rfficiency
test was used to cvaluate the goodness of fit betucen model predicted and measured outputs. It is
d e h e d as:
R' = I - [mm( Y,, - Y,, )' ! sum( Y,, - Y,,)'] (2.9)
where R' is model efficiency: Y,, is measured output: Y,, is predicted output by the model:
andY,, is the mcan measured output. The numerator term represents the initial or rneasured
variation and the denominator term represents the unrxplained or residual variation. The model
efficiency is similar to the regression coefficient. i: however. the residual variation is calculated
using the mean of actual observations rather than values from the best regression line between
observed and predicted values (Risse et al.. 19953. The Nash-Sutcliffe coefticient of efficiency
also indicates a perfect fit with a value of 1 and decreasing values indicates less correlation. A
Page 72
value of less than zero indicates that using the mode1 resuits are worse than using the average
measured value.
Mode1 simulation used for the model cornparison was determined afier O/o deviation and
mode1 efficiency calculations were both optimized. This allowed model agreement to bs bassd
on the mean soil loss from the field as well as how well this number was representrd over the
field. Further statistical svaluations of the mode1 agreement were cornpletsd upon cornparison
with tillage erosion prediction.
2.1.3.2 Tillage erosion mode1
Predicted values of soil erosion from tillage were calculated using the linear relationship
developed by Lobb et al. ( 1998) where tillage translocation is a function of tillap implement
erosivity and landscape shape and gradient erodibility. The general rquation can br expressrd as:
A, = f'(E, . E,) (2.10)
where A, is the net fonvard soi1 translocation by tillage for the year: E, is the tillage erosivity for
the year: and E, is the landscape erodibili~..
For the simulation period of the study- tillage was assumed to be evenly distributed
between number of upslope and dounslope passes for al1 implernent types. The Tillage crosion
model also does not consider the contribution of lateral soil transiocation. tillage depth and
speed. and tractor-implement factors as the. have not been hlly developed and characterized to
date. Information regarding the effect of soil properties. which may affect the resistance of soil to
displacement. was insufficirnt and was not included. The tillage erosivity term considered the
conventional Mage sequence measurements of Lobb ( 1998). who determined the erosion
potential of tillage implements as influenced from gradient and cunature landscape factors.
Page 73
Landscape erodibility is a function of dope gradient and slopr curvature ( Lobb. 1 998).
Slope gradient. which determines the magnitude of the effect of gravie on the mass of the soil
displaced by tillage. was determined frorn titting a spline curve function to the elevation points
using Mathcad cornputer software (MathSofi. Inc.. 1992). The tirst derivative of the hillslope
elevation points determined dope gradient as the rise over run. and the second derivative of the
elevations calculated the cunature at each of the hillslope points. A change of dope cunature is
required to create either soil loss or gain.
Experimental results of Lobb ( 1 998). taken from upland landscapes of southem Ontario
similar to this studp. determined tillage rrosivity coefficients describing the relative impact of
curvature to gradient. The cun.ature erosivity value \vas appro'timately twice the gradient
erosivity valus. The range of values for slope gradient obserced in a field. however. exceeds
those of slope cunrature and consequently. dope gradient was the major driving force behind
tillage erosion. The relative influence of cunature was investigated to determine its impact on
soil loss prediction.
The tillage erosion mode1 components (Lobb. 1998) include empirical-based coefficients
of tillage erosivity for gradient. P. and cuneature. y: and the landscape rrodibility values of
gradient. 0. and curvature. y. The gradient and curvature contributions are surnmed and C
espressed over dope length. In equation form. the annual net soi1 loss From tillage erosion under
conventional management is expressed by:
A , = ( p 6 û T y 8 4 ) 1 6 s (2.1 1 )
where A, is the net fonvard soil translocation by tillage for the year (kg m-' y r ' ): P (kg m-' %-' )
and y (kg rn-' 96") are erosivity coetticirnts which describe net tillage translocation from slopr
Page 74
gradient and slope cwature. respectively 0 is the gradient of the dope area measured ("O): 4 is L
the cwature of the slope area measured (O10 m"): and s is the distance represented by the gradient
and curvature (m). The equation for the conventional field was adjusted for the consenation field
because of the reduced number of tillage passes during the simulation period (Table 2.3). The P
and y coefficients were proportionately lowered by one-third. The tillage erosivity parameters
were therefore a retlection of the tillage intensiv. The landscape erodibility was a function of
slope gradient and slope curvature.
The Tillage erosion model was applied at each hiilslope point for model comparison. A
central difference calculation was used to determine the change in calculated slope gradient and
slope curvature from the two points on either side of a point and the separation distance between
these points t i.e. 4Om) for the landscape erodibility computation at each point on the hillslope.
The change in gradient. cunature and distance introduced in Eqn. 2.1 1 was calculated by:
A4[ = [(p (el-: - 01)) A (y (@,-? - OL)] ' (SI-: - SI) (2.12)
where i and i+2 are a position on the hillslope. and a position two points downslope. rrspectivrl?-.
The annual tillage erosivity contribution for P and y were assigned values of 6 and 17.
respectively. as measured by Lobb. ( 1998) for the conventionalIy managed field. and 2 and 4 for
the conservation field.
The calculated tillage erosion and translocation prediction (kg mm' yi') at each hillslope
point \vas compared to the corresponding soi1 loss measuremsnts. The mode1 a, areement to
measured soil loss rates was determinrd by % deviation. Student's t-test. regression analysis and
the Nash-Sutcliffe coeacient of efticiency (R'). The direct comparison of tillage translocation of
soil to "'Cs estimated soil redistribution assurned that soil loss on the convex area would equal
Page 75
soil accumulations within the field boundary: however. tillage translocation at slope end of the
conventional field may have continued beyond the sampled area.
2.1.33 Mode1 cornparison testing and anaîysis
The soil loss prediction results from each of WEPP and Tillage erosion models and thrir
combination were compared to the ''7Cs sstimated soil loss rates b- a number of mçthods. Initial
assessrnent of the modelling results from both the calculated and calibrated procedures were
compared to the average annual soil losses measured using deviation (Eqn. 2.8). and the
Students t-test. The t-test (Snedecor and Cochrane. 1989) indicates the variations in the means
and is represented by the following equation:
t =.Y! (S ,' nt ' ) (2.13)
where .r is the mean of difkrences: s is the standard deviation of the differences: n is the sample
size (number of slope positions for either tield): and the signiticance level. a. was chosen to bc
o. 10.
Rsgression analysis \vas applied to the three mode1 scenarios as a measure of conformit>
between the modellrd soi1 loss rates and the "'Cs estimated soil loss rates. The data analysis of
the 1 O and f 2 Iandscape position transects of the conservation and conventional fields.
rcsprcti\-çly. wrre used as the input for the best-fit regrrssion line. The coefficient of
determination. ?. expresses the proportion of the total variation of estimated values which can be
accounted for or explained by a linear relationship with the measured values (Snedecor and
Cochran. 1989).
The final estimator of model performance in predicting values of soil loss t u s a measure
of model effciency as tested by the Nash-Sutcliffe coefficient of eficiency (Nash and Sutcliffe.
Page 76
1970). This method is recommended to objectively assess performance of continuous simulation
watershed models (ASCE. 1993). The sfficiency estimator. R'. was applied to all calculated and
calibrated predictions on an average annual soil Ioss basis.
2.2 Results and Discussion
2.2.1 SIope profile analysis
The digital terrain analysis used suri-ey data from a 3x3 grid at I m spacing around rach
sample gridpoint to detemine the surface geornetry (conves or concave) in hoth do~wslope and
across dope directions. Label variables of the results are included in the site sridpoint summary
table of Appendix A. Landfom designation with digital terrain analysis did not appear
reasonable as many upper slope positions previously identified by soil s m e y methods ( Denholm
and Schut. 1993) were given footslope designations and many points in the lower positions were
given shoulder slope desipations. Surface variability from implement trafic on hoth tields rnay
have contributed to the falsc rlevation changes within a few rnetres upon accurate measurernent
with the sensitive s u n q instrument.
The lack of confidence in the analysis precluded their use for identic-ing convrsit?' and
concavity on the two cultivated tields. Identification of landscape positions were surveycd by
professional pedologists and given crest. upper. mid. lower or depressional slope designations
(Figure 2.2). The two hillslopes were simple. gently sloping with no significant microrelief at this
scale. The conventional field was slightly longer than the consenation field with a well-defined
upper-shoulder position: however. it did not have a depressional position within the available
sampling area whereas the conservation field did.
Page 77
Figure 2.2: Site identification of landscape slopr position.
!St , LTpper (shoulder)
\ Mid (back)
2.2.2 Soi1 erosion measurement using Cesium-137
2.2.2.1 Redistribution of Cesium-137
The "'Cs lsvel of the uneroded forest site from al1 samples was 2599 Bq m" with a
standard de\-iation of 601 (Table 2.5). This value is similar to measured forest reference values
takrn in the area: 2 6 2 Bq m.' at Guelph (Kachanoski. 1987): and 2490 Bq m': at Georgetown
(Kachanoski et al.. 1996). Variability sithin the uneroded site (CV of 33.1% is likelg a rcsult of
variation in deposition from vegetative cover type and redistribution from bioturbation
(VandenBygaart. 1998). The value of 2599 Bq m-' was used as the baseline or reference for
determining soil loss or soil gain ovrr the conservation and conventional fields.
Mean value of "'Cs from the consenation field was 2348 Bq rn*' with a coefficient of
variation 0~29 .3% (Table 2.5). The values at the 40 sample gridpoints over the field ranged fiom
Page 78
1302 to 41 83 Bq m-' (Appendix A). The deletion of one outlier at the upper southeast corner was
accounted for as this gridpoint value of 4776 Bq rn': was due to preferential soil deposition from
outside the field. Over the entire hillslope of the conservation field. a high redistribution of soil
from al1 processes as estimated by "'Cs \vas indicated after accounting for 90.6% of the rspected
deposited "'Cs when compared to the reference level.
The conventional field mean "'Cs inventory was 1930 Bq m" with a coetEcient of
variation of 23.3% (Table 2.5) . The range of values from the 48 gridpoints were lower than the
conservation field adjacent and fiIl between 1125 and 2966 Bq me'. When compared to the
fores1 reference inventory. the conventional field site was observed to have a net loss over the 19
yr period of 23.2% of the expected dsposited '"Cs.
The variability of "'CS levels in the two fields was considerable with a general pattern of
lower levels at upper portions of the slope and greater Ievels at lower portions of the slopr (Table
2.6). Over the length of the conservation hillslope. "'Cs losses of between 16.3 to 42.-I0/a uere
measured at upprr to upper-mid dope regions with Iwels declining until accumulation ab0l.e
baseline levels occurred in the depressional axa to a maximum of 18OA. Net "'Cs losses oF>30°/o
were observed over a 100m lsngth of the conventional hillslopr beginning at the crest and
moving downward until loss rates declined below 30% at the mid-lower dope region. The lower
slope positions however did not indicate on average an accumulation of '"CS. The lack of
depressional area at the end of the conventional hillslope hindered significant accumulation of
"'Cs redistri bution from upslope.
Page 79
Table 2.5: Measured soil "'Cs activity (Bq rn-') from the conservation Md. conventional
field and foresr reference area: and associated soil loss (kg m" ~ i ' ) from the
cultivated tlelds.
Analysis rnethod Conservation Conventionai Forest
' "Cs (Bq m-') :
Mean
SD
c v (96)
High value
Low value
'iet loss (%)
Soil loss (kg m" 4.r" ) :
Mean
SD
c v (%)
H igli value
LON value
* Mean value and number of sarnple points in parenthesis.
Page 80
Table 2.6 a: Soil '"CS activity (Bq m-'). its variability. and associated soi1 loss from
Conservation field Ap horizon averaged from 4 transects at each landscape
position dowmslope.
Analysis Landscape position
of '::Cs c-u* t r U bI M L L t D D
Mean (4)**
( Bq m-')
SD (1)
(Bq m.')
cv
(?/O)
96 loss+
or gain-
Soil loss
(kg m.' >,r-' )
* Landscape position labels are C=crest. U=upper. kl=mid. L=lower. D=depressionaI.
** Mean value determined from number of sample points in paranrhesis.
Page 81
Table 2.6 b: Soi1 "'Cs activity (Bq m"). its variability. and associated soi1 loss from
Conventional field Ap horizon averaged from 1 transects at each landscapr
position downslope.
Analysis Landscape position
blean(4) 1717 1499 1741 1730 1659 1760 2252 1959 2340 2099 2065 2336
(Bq m*')
SD(J)** 368 241 170 259 197 l 1 1 304 220 397 509 599 344
(Bq m")
CV 1 . 4 16.1 9.8 O 11.9 6.3 9.1 11.2 17.0 24.3 29.0 11.7
(%)
% loss+ 33.9 4 33.0 3 . 4 36.2 32.3 13.4 24.6 10.0 19.2 20.5
or gain-
Soilloss 4.88 6.65 4.88 5.24 6.02 1 2.01 3.95 1.17 2-42 3.05
(kg m-' ur-' )
* Landscape position labels are C=crest. Lr=upper. %l=mid. L-louer.
** Mean talus deteminrd from number ofsarnpls points in paranthesis.
Page 82
2.2.2.2 Net soil Ioss rates from Cesium-137 values
The estimation of net soi1 loss fiom rneasurement of soil '"CS content \vas calculated
from equation 2.1 and reported for each gridpoint in Appendix A. The large decrease in "'Cs
levels from upper regions of the dope profile was evidence of significant soil redistribution over
both fields. Estirnates from the conservation and conventional fields were of average annual soil
lossss of 1 .O9 and 3.88 kg m-' yr-' ( 10.9 and 38.8 t ha-' !-il). respectively (Table 2.5 ). The
variability was closely related ro topography as Table 2.6 indicates. Net soil redistribution on the
conservation field were between an averaged midslope position loss of 5.05 kr m-' yi' and net
gain of 2-38 kg rn-: yi' on an averaged lower dope position. The conventional Reld did not show - an- net deposition of wil with net soil losses ranging from a high of 6-65 kg m" y r ' at the crest-
upper or shoulder position to a low of 1.17 kg m-' yr-' at a lower slopr position. Figure 1.3 shows
the rstimated soil loss rates as the! relate to topography.
In summarizing the extent of the soil erosion. the rates of soil erosion were grouped into
t k e srosion classes as outlined in Wall et al. ( 1998):
1 Nrgligible < 0.60 kg m" )-il
7 - L O 0.60 - 1 .O kg rn-' yi'
3 Moderate 1.10 - 2.19 kg rn" y*'
4 High - ' .- '0 - 3 2 9 kg m': yi'
5 Severe >= 3 -30 kg m" yr-'
and reported for al1 gridpoint locations in Appendix -4. The conservation Field had net soil loss
over 59% of the area with severe erosion estimated at over 28%. 15% was high erosion and 10%
of the area sutfered moderate erosion. Classi@ing erosion on the conventional field found 92%
Page 83
I Conservation field
Distance downslope (m)
Convenuonal tleld
O 10 30 O 70 90 110 130 150 170 190 210 230
Distance downslope (m)
- Slope profile "'Cs measurement
Figure 2.3: Estimated soi1 lossigain rate over conservation and conventional field siope
profile.
Page 84
of the tield area suffered from erosion of which the severe. hi& and moderate rates acre 630'0.
14% and 69'0 of the area respectively.
The large general differences in soil losses found between the consenration and
conventional field are a reflrction of the slope and management history. The dope of the
conservation field is not as long which if considering water erosion as a dominant rrosion
process. would be reason for a reduced potential for net soil transport from mid and lowsr
positions. The depressional area at end slopr has provided an alrnost enclosrd systcm: liowttver.
lateral tlow off-site from this area r n q have occurrer' from severe storm and ponding scrnarios to
account for some soil loss and a portion of the measured 9.4% '''Cs loss. The conventional field
slope protlle adjacent is vep similar but decreases to an end slope of 1 % without levelling out.
Soil transport from water erosion processes would have a greater potential for deposition O ff-site.
The other significant landscape difference was the higher soil loss estimates throughout
the convex area of the conventional field whereas the srnaller area of severe soil losses on the
conservation managed tisld were at upper to mid slope areas. This difference in extent of soil
crosion is likcly a reflection of the more intensive cropping and tillage practiccts on the
conventional management Iield.
2.2.3 Soil erosion prediction
Over the period of investigation. the relationship of the variable soil loss rates estimated
using ' ' 7 ~ s analysis was compared with the predicted soil loss rates using the WEPP model. the
Tiilage model and a combination of both.
2.7.1 WEPP model anaiysis
Page 85
The WEPP model run in continuous mode \vas very responsive to changes in the input
parametes influencing soil loss and sedirnent dclivery from the field. X sensitivity analpsis using
the input data from the site revealed the ranking of most to least sensitive parameter for both
fields were K,. T,. Kr, and K, (Table 2.7). Sensitivity to the baseline hydraulic conductivity (iQ
of approsimately -1 .O suggests the large intluence diat an adjustment of this parameter would
have on infiltration- and indirectly soil loss. The resulting m g e of sediment deli\.ery wsre
magnitudes greater than round with the next most sensitive pararnrter. The critical shtrar stress
(t,.) also proved to be very sensitive. The level ofresistance to detachment from nll rrosion
highly intluenced soil loss prediction with a sensitivity of approxirnately -0.8. Ri11 erodibility ( &)
Table 2.7: Meaçure of sensitivity for key soil parameters in sstimating soil loss b'- the WEPP
model.
blodel input Mode1 output
blrinagement CaIculated variable Sediment deliven i kg m" \.r" ) Sensitivity* and parameter value Lou H igh blean Lon High Mean
* S =[(output, - output:) 1 mean of out put^,^] / [(variable, - variable,) 1 mean o f variable~!~]
Page 86
was found to be approximately half as sensitive a parameter as &. Of the soil loss parameters.
intemll erodibility (K.) adjustment was relatively ineffective in intluencing sediment delivery
rates.
The predicted level of sediment delivery from each field site was not directly comparable
with estimated soil loss rates from "'Cs. A preliminq obsenration however between the fields
indicated sediment deliveq- of several magnitudes greater from the conventional field over the
conservation field. The degree of difference was higher than the three to four fold difference
found between the average soil loss estimates (Table 2.5) using "'Cs analysis: however. both
output parameters indicate the same high conmt . This cornparison however was preliminary to
calibrating the model.
Calibration steps in comparing the predicted average annual soil loss rates to the
estimated "'Cs rates were summarized in Table 2.8. The uncalibrated WEPP simulation from
the consen-ation field using the calcuiated soil loss input parameter values had a deviation of -
82%. Attempts to irnprove the prediction were based on the generated results of the adjustsd
parameter values used in the sensitivie analysis. Direction aas given in attempting to match soil
loss rates: however. these soil loss simulations were al1 signiticantly lower than the "'Cs
measured soil loss rates ( 1 .O6 kg m" yr-') and ranged from 0.02 to 0.56 kg rn" y-". deviations of
between 4 5 and -9g0/o. Higher average soil loss rates were obtainable by adjusting the &. s,. and
Kr: however. the model efficiency was signiticantly lowered as a result. By adjusting the K,
lower (3.664). the r, lower ( 1 -72) and the K, higher (0.01 58) from the calculated values. a
compromise of the lowest deviation (-50%) with the highest R' (0.49) was obtained and used as
the calibrated model for the conservation field site.
Page 87
Table 2.8: Calibration of WEPP model soi1 loss parameters using average soil loss. percent
deviation and Nash-Sutcliffe coefficient of eficiency .
Soi1 loss parameter adiustment Analysis Measured WEPP K b Kb Kb Kb tc tc Kr K r Ki Ki WEPP
method "'Cs uncal. I O M ; high' low' high' low high low high O high calib.
Conservation field - - ivean 1.06 O . 0 . x 0.02 0.56 0.09 0.44 0.04 0.07 0 . 3 0.10 0.16 0.50 (kg mm' > r-' Deviation - 82 - 48 - 98 5 - 92 - 56 - 96 - 93 - 77 -90 - 84 - 50
(%)
R' 0.05 0.10 -0.18 0.09 -0.02. 0.72 -0.14 -0.08 0.55 0.01 0.07 0.49
- - - - - - - - - - - -
Conventional tield Mean 3-88 4 6.74 0.59 6.1 1 1.41 1.99 0.18 0.57 1.85 1 - 1 1 9 4.27 (kg rn-' yr" Deviat ion -46 74 -41 57 - 6 1 -49 -95 - 8 -52 - 6 1 -62 10
(%)
R ' -2.28 -20.64 -3.43 -20.53 -2.30 - 3 . 5 -4.33 - 6 -4.42 -2.38 -2.22 -15.17
h-mdraulic conductivit? (K+) remained constant: : hydraulic conductivity (K,,) varied.
Results in matching the conventional field average soil loss rate of 3.88 kg rn-' yi' werr
closer than the conservation field. The accompanying model agreement ( R2). howewr. computsd
for al1 simulations were much lower. With the adjustment of lowenn_e the K, parameter to
increase runoff. WEPP could in fact match the average soi1 loss estimated by "'Cs. If considering
this minimal deviation analysis as the criteria for model accrptancr however. the rrtreme
variation of soil loss simulated would not be considered. At crest positions. minimal soil loss was
predicted. At mid slope areas. 3 to 5 fold overestimations were simulated and at the dope end.
soil deposition \vas predicted where none was measured (Figure 2.1). Adjustments to K,,. r, and
y to reduce % deviation as was done with the conservation field. resulted in an R~ of - 15.1 7.
Page 88
12 *
Conservation field
l. O 10 30 50 70 90 110 130 150 170 190
Distance downstope (m)
Disrance downslope (m)
Figure 2.4 : Soi1 loss prediction over conservation and conventional dope profiles of
nonadjusted ( WEPPna) and adjusted (WEPPa) soil loss parameters.
Page 89
indicating unacceptable model agreement. It was found that minimal deviation in average soil
loss estimates could not be accurately simulated over the hillslope.
The uncaiibrated WEPP simulation where the calculated soil loss parameters were used
provided close to the optimum R' value obtainable of -2.28 liom the conventional field. The high
negative R' value still indicated poor model agreement. The interpretation kvas that the awragr
measured value kvas a better estimate than the model prediction (Risse et al.. 1995 1.
The option of allowing WEPP to adjust hydraulic conductivity with clianges in
management and moisture From the baseline hgdrauiic conductivity (&) input value. or selecting
the value to be constant. resulted in a relatively small change in output (Table 2.8). The model
therefore considered the input to be representative for the year and tluctuated the effrctive
hydraulic conductivity (Y) around the initial value given (&). The setting of the initial value of
K, in WEPP is very important as the variable & does not appear to vary signiiicantly from its
starting value.
2.2.3.2 Tillage erosion model analysis
The Tillage erosion model (Eqn. 2.12) was applied at each hillslope position on the
cultivated tields. Rates of soi1 loss were obsemed from the convex areas above the intlection
point and redistnbuted in accumulations ovrr the remainder of the hillslope. The conservation
tield's crest-upper and top upper position were computed to sustain 0.36 and 0.15 kg m" yi' of
soil loss. respectively (Table 2.9). The intlection point of the hillslope was at 41 m downslope
after which soil accumulation occurred. Cornparison tu the minimal net soil movement estimated
by '''Cs at the crest-upper position (10m dounslope) may not be representative of potential
tillage erosion losses. A single gridpoint '''Cs value at this position indicated severe erosion loss
Page 90
had occurred. The next convex position downslope howeïer did indicate uni for mi^ (CV of
6.8%) of '"Cs estimated soil loss of which the tillage erosion prediction equated to 4.9%.
Table 2.9: Soi1 redistribution as estimated by "'CS analysis and predicted by the Tillage
erosion model.
-- - - -- -- - - - - - -
Conservation field Conventional field f ;-CS Ti i lage ' T s Ti 1 lage
sstimated model est i mated model Distance Eievation soi1 loss soi1 loss Distance Elevation soi1 loss soi1 loss
On the convex area of the conventional field. the rate of soil loss decreased as the change
in gradient decreased over the shoulder slope with a range fiom 0.85 to 0.19 kg m-' y*' over the
60m length. An average rate of the crest and upper points \vas 0.41 kg m-' yi'. a predicted
portion of the average estimated rate of soil loss from "'Cs analysis of 7.6%.
The contribution of curvature to tillage erosion compared to gradient on this landscape
Page 91
was minimal. At the 10m position on the conservation tield. the positive change in cunature
caused an additional soil loss of 2.9?'0 of the gradient soil loss. A negatite change in cun-aturc
over the upper position (3Om) area resulted in an accumulation of soil at a rate of 16.7% of the
rate o f soil loss from the change in gradient. As the change in gradient diminished closer to the
intlection point (-Il m). the importance of cunature increased at the 5Om point. The rate of soil
accumulation here from curvature was 2 1 J0/0 of the rate of soil loss from gradient. Beyond this
point on the hillslops. cunraturc effects were approximately 3% of the gradient effect.
The findings from the conventional tield were similar: whereby. the effect of cumature
was 12.8% of the gradient at 3Om and approximately 5% of the gradient effect elsewhere on the
hillslops except near the inflection point. The 90m position was just below the inflection point of
88m so that minimal gradient changes at the midslopr resulted in equal rates of soil accumulation
( 10 1 %) from cun-ature to gradient parameters.
2.2.3.3 Combined WEPP/Tillage soi1 erosion model
Soi1 redistribution in the landscape as sstimated by "'Cs was direct[>. compared to
predicted soil redistribution by WEPP and Tillage erosion models at points equidistant down the
hillslope profile. Figure 2.5 shows the relationship of each model prediction scsnario to
estimated soil loss rates. Tillage translocation prediction was minimal compared to WEPP
predictions with the on1y exception at the crest of the slope where tillage erosion \vas dominant.
The intluence of each soil erosion process was s h o w to be more significant at ditferent areas of
the hillslope. When in combination. the relationship to net soil loss appean to be improved.
The combined WEPPfTïllage graphic (Figure 2.6) fiom the conservation field displays an
underestimation of predicted soil loss over '>'CS measured soil loss at the upper and upper-mid
Page 92
12 '
Consenration field
D isrance downslope (m)
Conventional field
O 10 30 50 70 90 110 13C 150 179 190 Z1C 230
Distance downslope (m)
- Profile
Figure 2.5: Soi1 loss as predicted by separate WEPP and Tillage erosion mode1 output values
and estimated by "'CS reference analysis with distance domnslope.
Page 93
Conservation field 8
Conventional tield 8
Distance downslope (m)
- Pro file
Figure 2.6: Soi1 loss as predicted by combined WEPP and Tillage erosion mode1 output
values and estimated by '"CS reference analysis with distance downslope.
Page 94
slope areas. Predicted values tiom the conventional tield also show a signitïcant underestimation
at crest-~pper and upper designated slope positions of the hillslope by both models. On the
conservation and conventional tields, average predicted values over midslope positions. howsver.
were 80% and 72% respectively. of average rneasured values of soil loss. The higher predicted
rates of the WEPP model oc-erestimated deposition at the dope end for both tïelds.
Statistical analysis of combining WEPP and tillage erosion calculations shows a limi ted
improvement tiom either prediction method alone (Table 2.10). On the conservation tield. the
lower soil redistribution values predicted from the Tillage erosion rnodel than frorn WEPP gave a
mean soil loss value of -1029'0 deviation from '"CS estimated soil loss rates. The mean soil loss
estimate from WEPP and combined WEPP and tillage values for the consenation tield was not
improved with a -52% to -54% deviation. respectively. The t-test value. however. did not
indicatr a statistical di fference in agreement berween W'EPP. Tillage and the WEPP'Til l a s
rnodel prediction and '"Cs estirnated soil loss rates. due to the hi@ variabiiih- within the points.
Regression analy sis betwern WEPP and WEPP/Tillage predicred and "'Cs estimated soi 1 loss
retained an i of 0.6 1. Mode1 sficiency of predicted agreement with the measured sarnple points
over the field remained at a coefficient of efficiency (R') of 0.49 for WEPP and WEPP/Tiilage
models.
The fit of the model prediction on the conventional tirld overall was poor. The addition
of the tillage soil erosion predictions to WEPP values resulted in little change in using the %
deviation cornparison. T-test values indicated general agreement at the p<0.10 level for both
WEPP and WEPP.'Tillage mode1 simulations. The reeression analysis showed an improvement
from a i of 0.08 to 0.16. The most conclusive statistic of mode1 agreement with "'CS measured
Page 95
soi1 loss rates was the coeficient of eficiency which indicated an improvement fiom -2.28 to -
0.78: however. the low R' value was indicative of poor mode1 prediction in this case.
Table 1.10: Statistical summ- of soil loss prediction for Conservation and Conventional
field positions as the- compared to the "'CS soil loss estimation method.
Conservation tleld Conventional field Distance Predicted soi1 loss (kg m.' yi' ) Distance Predicted soi1 loss (kg m" yr" r
dùwnslope ' : - C S WEPP Tillage WEPP/ dowmlope ''-Cs WEPP Tillage WEPP!
(ml Tillage (ml Ti 1 lape
Msan 1 .O6 0.5 l -0.02 SD 2.28 2-34 O. 14 % dev'n -32 - 102 t(O.10.10) u=u .P=P
? 0.6 1 0.00
R' 0.49 -0.23
10 4.85 0.18 30 6.65 0.24 50 4.88 0.97 70 5 . 4 2.3 1
90 6.02 3.35 110 5.0 1 3.63 130 2.0 1 3 .O3 150 3.95 2.38
1 70 1.17 1.60 190 2.42 1.94 210 3-03 0.86 130 1.37 -2.96
.Mean 3 -88 1-46 SD 1.77 1.73 4'0 dev'n -62 t(O.10.12) u =u r 0.08 R ' -2.28
Page 96
2.3 Discussion
The rneasurement of "'Cs frorn the conservation and conventionally managed hillslopes
varied significantly around the reference level in the forest. Emplopment of the ' ' 7 ~ s reference
method of analysis indicated substantial soil redistribution. Cornparison at upper and mid slope
landscape positions indicatrd severe soil losses from both fields at the rate of 39 to 67 t ha-' ~ i '
(3.9 to 6.7 kg m-' yr-') u-ith the greater soil losses tiorn the conventional field. These crosion rates
were not unreasonable when compxed with data of Kachanoski et al. ( 1992). who found soil loss
rates on cultivated creçt and shoulder slopes in southem Ontario exceeded 100 t ha-' y r ' . At the
two upper and highrst mid landscape position. the corresponding soil loss rates frorn the
consetvation field were 62 to 84% of the conventional field. The corresponding soil losses hy
landscape position measured over the remainder of the twvo hillslopes were twice or more from
the conventional field. An absence of soil deposition at the slope end of the conventional field
was indicative of the lack of depressional area and greater combined water and tillage soi1
translocation than obsened from the consen-ation field.
The evaluation of WEPP to predict water erosion from the site confirmed K, to be the
most sensitive parameter in determining soil loss. The assessrnent of the mode1 accuracy wvas
hampered by the lack of measured runoîT data or hydraulic conductivity . It was detennined that
soil loss prediction afier assigning a constant hydraulic conductivity value did not significantlv
alter the result from varying y within a season. The apparently small modification of sirnulated
K, at critical periods of the year ma); be underestimating soil loss.
Rudra et al. ( 1998) reported that hydraulic conductivity does vary by orders of magnitude
within a year in southem Ontario and this variation should be incorporated into runoff rnodellinp.
Page 97
Gupta et al. ( 1994) have s h o w that the rnean value of saturated hydnulic conductivity for fa11
and spring conditions can be one and a half to two times greater than summer season values due
to changes in the soil surface. Tillage operations and climatic characteristics such as freeze-thaw
cycles create considerable spatial and temporal variation in physical and hydraulic characteristics
of southern Ontario soils (Rudra et al.. 1998). The adjustment to the Green-Ampt equation used
in the WEPP model may not be fullp taking into account these management and clirnatic effects.
The two hillslopes in the study provided a simple model simulation cornparison as they
were situated irnmediately adjacent with significant relief to provide variable erosion potential
under two management systems. WEPP erosion di fferences between fields were not likely ri
result of the slope characteristics. Nearing et al. ( 1990) round the inflection point position to be
of low sensitivity when the end dope was <6%. The ditferences in slope length of the fields
would also not signiticantly influence detachment and soil loss: however. high ssdiment deliveq.
differences round in the study sensitivity analysis may have been a result of this ~ariable. The
average slope gradient on the two fields being similar would not markedly influence soi1 loss.
The off-site effect of sediment delivery. however. is determined heavily by slops at end (Nearing
et al.. 1990). These reasons mal- sxplain the higher differences of sediment delivel than soil loss
between the two fields.
The rates of soil erosion as predicted by the Tillage erosion model were substantially
lower than values mrasured during the development of the model (Lobb. 1998)The
rneasurement of tillage erosivity coefficients fiom a conventional tillage sequence by Lobb
( 1998) of mouldboard plough. two passes of tandem disc and one cultivator pass was applied to
the conventional field but represented a higher intensity than reponedly occurred (Table 2.3).
Page 98
Unless a significant arnount of tillage occwed that LW not accountrd for. a greater tillage
erosivity value for the site u-ould not be warranted. Subtle changes in the landscape topograph?
at the site may be a more likely explanation for lower predicted tillage erosion.
Landscape erodibility of tillage erosion. dependent on the rate of change of gradient and
cunaturs. was greater from plot measurements by Lobb ( 1998). The model site topngnphy was
designated as being moderately to strongly rolling with mavimum gradients of 10 to 16%. The
application to the site of maximum gradient around 7% resulted in measured changes in gradient
of approximately one-half. The impact on the change in curvature between the model site and the
Rockwood site was in the order of a factor often loiver. It is difficult to determine the accuncy
of the predicted tillage rrosion rates: however. the influence on erosion bp the landscape
parameters would appear to be reasonable for the profile of this gently rolling landscape.
The Tillage erosion model needs further improvement and calibration for accuratr
application outside the conditions of which it \vas developed. Factors including tillase depth.
speed. and Iandscape influence have not been adequatel- tested. Conceptually. the pattern of soi]
redistribution produced by tillage erosion accounts for severe soil loss observed on conve'c
landscapr positions. The conclusions of other studies support tillage erosion as a major cause of
soil redistribution and the major cause of severe soil loss within topographically comples
landscapes of southem Ontario. With the application of the Tillage erosion model at the study
site. however. the findings do not account for the severe soil loss as measured tiom "'CS
analysis.
The influence of differences in hillslope geometry between the consenration and
conventional field are likely of less consequence than the management history in soil
Page 99
redistribution levels measured. Assurning water and tillagr erosion is negligible when alfalfa is
grown. the conservation field was prone to erosion during approximately one-third of the
simulation penod. The conventional tield. however. received annual primary tillage which
contributed up to three times more tillage erosion and secondly. esposed bare soil to water
erosion oveminter and dunng the cntical early spnng period.
The relative ditTerences in soil losses from the consenation and conventional hillslopes
do not appear to be a direct relationship with tillage and wvater erosion potential. Soil erosion
measurements from the "'Cs reference rnethod at the upper and mid slope area retlect an
additional factor or process unesplained by Mage or water erosion processes. Funher
esplmation is ditricult without considering the accuracy of the "'Cs reference method for
rneasuring net soil loss at upper and mid positions on variable topography.
2.4 Conclusions
In this study. the pattern of soil redistribution from an upland landscapr of southem
Ontario was measured using the "'Cs reference method of analysis. Measurements of cultivated
fields of contrasting management indicated severe soi1 losses from the crest-uppcr to midslope
area and accumulations in the depressional area. The findings are supponed from recent studies
investigating soi1 erosion on cornplex topography in southem Ontario and elsewhere. in which.
similar patterns of soi1 redistribution soil cannot be attributed to wind and water erosion
processes.
Soil redistribution from upland landscapes of southem Ontario ws assumed to be a result
of nvo dominant erosion processes. water erosion and tillage erosion. Upon combining values of
Page 100
cvater erosion prediction by WEPP and tillage erosion prediction into one model. the estimatrd
net soil redistribution on a hillslope was marginally improved. Based on measuremrnts of "'CS
reference analysis. the WEPP model did not characterize soil losses measured on the convex area
of the landscape: however. the addition of the Tillage erosion model partialiy accounted for these
losses. The accumulation of translocated soil by tillage. however. throughout the concave area
reduced modcl agreement with measured soi1 redistribution.
The tw-O fields investigated were of contrasting management history and similar
topognphy. Their cropping and Mage history likcly caused the signiiicant differencr in sxtent o f
soil losses measured. The conservation managed field showed positive mode1 agreement cvith
measured soil loss suggesting that the processes modelled reflected a significant amount of the
actual soil loss: however. differencrs at the convex area could not be matched. The poor model
representation on the conventional field. in particular the convex a r a indicated two possible
explanations. There m q have bren fault in the modelling capability under thesr circumstances.
or that the estimated soi1 loss rates are not a true retlection of erosion rates occumng on this
landscape area.
The WEPP mode1 requires îùrther cali bration for southem Ontario conditions ( Rudra et
al.. 1998): however. the model's intention is to simulate water erosion processes. Ri11 erosion
largely determines the potential transport of soil downslope and occurs afier surface runoff f o m s
concentrated flow. typically not found on convex areas. The tillage erosion prediction is in its
infancy and the model used may be underestimating the amount of soil loss from this dominant
soi1 erosion process in southem Ontario (Lobb et al.. 1995). One observation remains that on the
conservation field that had about one-third of the tillage passes as the conventional side. there
Page 101
were two-thirds of the soil Ioss at mid to upper positions as estimated using "'Cs analysis.
Considering the possibilih that "'CS erosion estimates are overpredicting rates at the
upper landscape areas. the question of how this could occur using the reference mrthod of '"Cs
analysis is an area of study that needs to be addressed. Implicit in the measuremrnt of soil
erosion using "?CS are a nurnber of assurnptions. An experimental evaluation of the cntical
assurnptions for the measurement of soi1 loss rates by the '"Cs reference method has not been
investigated.
Page 102
References
Atrnospheric Environrnent Service. 1985. Canadian clirnate normals 195 1 - 1980. Temperature
and precipitation- Ontario. Environrnent Canada.
ACSE Task Committee on Definition of Criteria for Evaluation of Watershed Models. 1993.
Criteria for rvaluation of watrrshed models. Watershed Management Committee.
Imgation and Drainage Division. J. of [mg. and Drain. Eng.. 1 19(3):429-442.
Carter. M. W. and A.A. Moghissi. 1977.Three decades of nuclear testing. Health Phys. 33 55-7 1.
Day. J.H. (ed). 1983. The Canada Soil Information System (CanSiS). Manual for describing soi1
in the field. Expert Committer on Soil Survey. 1982 rev. ed. .Agriculture Canada.
Rrsearch Branch LRFU Cont. No. 82-52. 97 pp.
Denholm. K.A. and L. W. Schut. 1993. Field manual for describing soils in Ontario. Ontario
Centre for Soil Resource Evaluation. Guelph. ON. 62 pp.
de Jong. E.. H. Villa and J.R. Bettany. 1987. Prelirninary investigations on the use of 1 37-Cs to
estimate erosion in Saskatchewan. Cm. J. Soil Sci. 63573-682.
Flanagan. D.C. and S.J. Livingston. 1995. WEPP user surnmary. NSERL Resrarch Rrpon No.
1 1. USDA-ARS. Natl. Soil Erosion Res. Lab.. West Lafayette. IN.
Golden Software. Inc. 1997. SURFER graphics software. ver. 6.04 Golden. CO.
Gupta. R.K.. R.P. Rudra. W.T. Dickinson and G.J. Wall. 1994. Spatial and temporal variations in
hydraulic conductivity in relation to four detemination techniques. Can. Water Res. J..
19(2):1-11.
Hoffman. D.W.. B.C. Matthews and R.E. Wicklund. 1963. Soil survey of Wellington County.
Page 103
Ontario. Repon no. 35. Ontario soil s w e y . Research Branch. Canada Dept. Agric..
Ottawa: Ont. Dept. Agric.. Toronto. ON.
Kachanoski. R.G. 1987. Cornparison of measured soil 137-Cesium losses and erosion rates. Can.
J. Soi1 Sci.. 67: 199-203.
Kachanoski. R.G. 1993. Estimating soil loss fiom changes in soil cesium-137. Cm. J. Soil
Sci.73 :629-632.
Kachanoski. R.G. and A.P. von Bertoldi. 1996. Monitoring soil loss and redistribution using
"'Cs. Green Plan Report. Agriculture and Agi-Food Canada. Guelph. ON. 29 pp.
Kachanoski. R.G.. M.H. Miller and D.A. Lobb. 1992. Management of farm field variabili~: 1 .
Quantiilcation of soil loss in complex topography. II. Soil erosion processes on shoulder
dope landscape positions. SWEEP Rrp. 38. Agric. Canada Harrow. ON. 155 pp.
Lobb. D.A. 199 1. Soil erosion processcs on shouider slope landscape positions. Unpubiishrd
MSc. Thesis. L'nivrrsity of Guelph. Guelph. ON.
Lobb. D A . R.G. Kachanoski and M.H. Miller. 1995. Tillage translocation and tillagc srosion on
shoulder dope landscapr positions measured using "'Cs as a tracer. Cm. J. Soil Sci.
75(3):21 1-318.
MathSofi. Inc. 1991. Mathcad 3.1 user's guide. Cambridge. MA.
McKcague. LA. 1978. Manual on soil sarnpling and methods of analysis. 2nd ed. C m . Soil Sci.
Soc.
Nash. LE. and J.V. Sutcliff'. 1970. River flow forecasting through conceptual models. Pan I - .\
discussion of principles. J. Hpdrol. 1 O(3 1282-290.
Nearing. M.A.. L. Deer-Ascough and J. M. Laflen. 1990. Sensitivity analysis of the WEPP
Page 104
hillslope profils erosion model. ASAE. 3(3):839-849.
Nolin. 1M.C.. Y.Z. Cao. D.R. Coote and C. Wang. 1993. Short-range variabilic of fallout "'Cs in
an uneroded forest soil. Cm. J. Soil Sci. 7?38 1-385.
Risse. L.M.. M A . Nearing and X.C. Zhang. 1995. Variability in Green-Ampt effective hydnulic
conductivity under fallow conditions. J. Hydrology. 169: 1-24.
Ritchie. K.. J.R. .McHenry and A C . Gill. 1972. The distribution of '"Cs in the litter and upper
1 O cm of soil under diffcrent cover t y e s in Northem Mississippi. Hcalth Phys. 22: 198-
201.
Rudm R.P.. W.T. Dickinson and G.J. Wall. 1998. Problerns regarding the use of soil erosion
models. In: Modeling soil erosion by water. Ed. J. Boardman and D. Favis-Mortlock.
NATO series. vol. 1. 5 5 .
Shelton. I.J.. G.J. Wall and D.R. Coote. 1991. Water rrosion risk. Ontario south. ..\griculture
Canada. Rssearch Branch. LRRC. Cont. No. 90-7 1.
Snedecor. G.W. and W.G. Cochran. 1989. Statistical methods. 8th rd. Iowa St. Uni\.ersit>. Press.
Ames. IO.
Soil Classification Working Group. 1998. The Canadian System of Soil Classification.
Agriculture and Agi-Food Canada. Publ. 1 646 (Revised). 1 87 pp.
VandenBygaart. X.J. i 998. Changes in soil morphology on a chronosequence of no-till
agricultural soils. Unpublished Ph.D. Dissertation. University of Guelph. Guelph. ON.
130 pp.
Wall. G.J.. D.R. Coote. E.A. Pringle and I.J. Shelton. 1998. RUSLEFAC: Revised Universal Soil
Loss Equation for Application in Canada. Agriculture and Agi-Food Canada. Rssearch
Page 105
Branch. ECORC. Ottawa. ON.
Wçast. R.C. (Ed.). 1987. Handbook of Chernisu). and Physics. 1987- 1988. 68th rd. Chemical
Rubber Co.. Boca Raton. FA.
Page 106
Chapter 3: Evaluating the use of cesium-137 atmosphenc deposition for measuring soil
erosion rates on upland regions of southern Ontario
3.1 Literature review and hypothesis develo pment
Soi1 erosion is a rneasure of soil movement bp physical forces. The use of '"CS tracer
analysis to measure soi! srosion incorporates al1 processes or mrthods of soil transport by water.
wind or tillage. A direct estimate of soil rrosion (that has occurred over the past 35 yrars) is
achieved by determining the amount of soil-attached "'Cs distributrd over a landscapc. The
distribution is a reflection of the magnitude of al1 erosional and depositional proccssrs sincr the
period of major radioactive fallout occurred.
The use of "'CS analysis as an erosion estimation tool has been extensively usrd for
several direct applications of determining soil movement. It has been applied at varying spatial
scales and uses from large agricultural watcrsheds ( Brown et al.. 198 1 a b ) . landscape rrosion
interpretation ( Prnnock and de Jong. 1990). off-site sediment accumulation (Walling and
Bradley. 1988). hillslope mode1 verification (Montgomery et al.. 1997: Bernard and Laverdirre.
1992) . and to sub-pedon measurement of the tillage erosion process (Lobb et al.. 1995). Plsntiful
documentation of soil "'Cs interpretations has given risc to wide acceptance of the erosion
prediction method (Ritchie and Ritchie. 1997).
Support for the "'Cs technique is due in combination to the need for better erosion
information. and its unobtrusive. immediate and relative ease of measurernent compared to
conventional methods. Little long-term soil erosion data are available making it di fficuit to
dispute the accuracy of the method (de Jong et al.. 1983). Kachanoski (1 987) found a reasonable
Page 107
linear relationship betwsen historical soil erosion plot measurements and "7Cs losses over the
sarne period. With these findings. it kvas suggested that data tiom the technique could be a
replacement for costly. labour intensive and time consurning mrthods. Testing of predictivs
rnodels has also been limited in sstablishing (independent) correlations between erosion and
'"Cs soi 1 concentrations.
3.1.1 Cesium-137 tracer analysis assumptions
Spatial measurements of "'CS to provide quantitative data on rates and patterns of erosion
and sedimentation are not without doubt. Brown et al. ( 198 1 b) and de Jong et al. ( 1982)
identified several basic assumptions made in the erosion interpretation from this technique:
1 ) bllout of "'Cs has been occumng since 1954:
1) "'CS is rapidly and tightl- iked in the surface soils:
3 ) uniform distribution of "'Cs over the landscape:
4) depietion or enrichment of "7Cs in soils is proportional to srosion or deposition of soil:
5 ) removal of "'Cs in crops is uniform and approxirnately 1 %:
6) fallout as redistnbuted snowfall accounting for losses of up to 4%:
7 ) "'Cs was svenl?. mixrd through the A p horizon by culti\.ation before any
redistribution:
8) loss of Fallout "'CS in runoff before it could be tixed in soil is minimal - a srnall
amount over the yrars that would occur uniformly across the landscape in large rainfall
events.
Atrnosphenc deposition of ''7Cs in precipitation has been assumed to be uniformly
distributed over the landscape. v q i n g linearly within latitudinal zones (Davis. 1963). The
Page 108
deposition pattern at a field or hillslope scale where most erosion estimations have been
conducted. is assumed to be uniform and equal to proximal reference areas of native vegetative
cover. In dry or arid climates. this assumption ma). be valid: however. in humid climates with
high arnounts of precipitation (min or snow) the assurnption may require testing. ;\ny runoff of
"'CS i ) in solution. or i i ) attached to soil particles during the depositional event. or iii) from
subsequent erosion prior io stabilizing the radionuclide in place by tillage mixing. has been
considered insigniticant. Confirmation of these three assurnptions cannot be directly assessed and
has not been reported. These assurnptions and their implications will be discussed and H-il1
subsequently be tested by an experimental procedure to determine if there could have bern
movement of "'Cs during the tirne of deposition or subsequent to deposition.
In using fallout "'Cs to measure erosion with a baseline input of a noneroded site. an
inherent assumption is in~olvrd. The assumption is that the baseline input to the rnvironment is
due to fallout from the atmosphenc nuclear testing rather than natural sources ( Davis . 1963:
Wise. 1980) and its initial distribution across the landscape is unifonn or c m be related to
measunblr environmental gradients (i-e. precipitation). This rrosion estimation technique
suggests it is not critical to have direct measurements of the actual "'CS deposited in a
watrrshed. since local input of ""Cs can be estimated in soil profiles where neithsr erosion nor
soil deposition has occurred (Ritchie et al.. 1974: Campbell. 1983: Martz and de Jong. 1987). In
their revicw. Ritchie and McHenry ( 1990) state that for the application of "'Cs for rneasuring
erosion. it is important to assume a uniform deposition pattern across the Iandscape unit being
studied or to be able to estimate the distribution pattern across a large watrrshed based on an
environmental gradient so that measurements made at noneroding sites can be used to determine
Page 109
fallout input to the studp area. The validity of this assurnption has not been well tested.
3.1.2 Reference site deposition assumption
Uncultivated areas of natural vegetatiw cover. versus cultivated fields. are not considrred
susceptible to "7Cs losses. Native growth of the Mixed woods plains ecozone (Ecological
Stratification Working Group. 1996). in the southem Ontario region. is dominatrd bx deciduous
forest stands with lesser amounts of coniferous trees. A native forest provides a multi-layer of
surface soil protection fram leaf canopy. branches. trees. bmsh. undrcomposed leaf litter and a
highly absorptive LFH Iayer. As a result. the potential for suface runoffand erosion is low and
the expected distribution of "'CS would be high in the liner and in the upper 2.5cm of soil
(Ritchie et al.. 1973). "'Cs variability however cm still be signiticant within undisturbed areas.
A standard deviation of approsimately 20% was found within small sample grids at a Quebec
forest site (Nolin a al.. 1993 1. Measured soil " 7 ~ s under permanent vegetation at the Chivsrsity
of Guelph vaned betwren three slope position means by 1 ~ S O for forest cover and by 35% under
grass cover ( Kachanoski 1987). The apparent increase in "7Cs accumulation do\~nslope in the
erassed site and to a lesser estent in the forest may be associated to surtàce runot~accumulation. C
de Jong et al. ( 1983) did not find "'Cs accumulation on the lower slope areas of permanent
native grassland basins in Saskatchewan which may retlect the lower surface ninoff potential on
the more arid Prairies.
Under southem Ontario forest cover. variability in rainfall deposition on the surface ma!
be caused by stem tlow effects of vegetation. as foound elsewhere (Gersper. 1970). Nolin et al.
( 1993) suggested a satisfactory estirnate of the mean "'Cs inventop- at reference sites can be
obtained if at least five subsamples. 1-2 m apart are composited into a single sample for analusis.
Page 110
The selection of reference sites of native vegetative cover with a noneroded past. has oRen meant
sampling on Hat crests or bonom lands in close proximity to a-picultural areas under
investigation. Consideration should be given to reference sites that do not act os catch basins for
off-tield sediment accumulation.
3.13 Agricultural field deposition variability
The variation of "'Cs concentration now found in soil and sediment is dependent on
erosion potential of the srudy area potential for adsorption of "'Cs and input of "'Cs (Brown et
al.. 198Ib) at the time of deposition. Field variability of "'CS may be considered a result of
overland transport of rainfall runoff in concentrated flows before incorporation. Deposition in
snowtall has been considered small (de Jong et al.. 1982). In temperate regions however. duhg
early spring. large sno~-tàll events occur with accompanying variable snowpack accumulations
increasing the possibility for nonuniform '"Cs deposition. In southem Ontario. the potential for
"'Cs movement in subsequent runofievents is high as spring melt over frozen and ofien
unprotected ground results in up to 60% of annual soil losses (van Vliet and Wall. 198 1 ).
3.1.3.1 Cesiurn-137 runoff in water during deposition
The prak atmospheric deposition or input of "'Cs delivered in large rainfail ryents is
considered to be relatively uniform within an area. The potential of surface runoff . ho\vever.
could influence "'Cs deposition and depends on maqr factors including soi1 infiltration capacity.
rainfall intensity. vegetative cover and topography. #en considenng the many factors and the
measurements conducted at a field or large plot scale. the opportunity for at tachent of these
radionuclide particles to the soi1 surface is debatable. In a rainstorm. as infiltration capacity is
exceeded. the radionuclide particles not immediately in contact with the soi\ surface would
Page 111
theretore be carried in suspension downslope in surtàce runot'f.
The spatial variability of "'Cs concentrations in soil is also likrly due to variabilin: in soil
infiltration rates at the time of deposition (Lance et al.. 1986). Initial infiltration rate of a medium
textured soil in southem Ontario is approximately jcm hr-'. For a surface horizon of 1.3- cm-'
bulk densi'; and 50% porosity. this initial rate of infiltration would be able to accommodate an
intensity of rainfall of-lmm in the first I O min (21mm hr-'). significantly less than annual
statistical highs (Atmospheric Enviromnent Service. 1993 for the Guelph area. In southem
Ontario. precipitation is relatively uniform throughout the year. The greatest runoff losses in
southern Ontario however occur in spring and early surnmer (van Vliet and Wall. 198 1 ) and are
expected annually as the region receives an excess of precipitation versus rvapontion. Variable
soil moisture content at this time causes variable infiltration potential and aould reflect ditrirent
runoff potentials on contrasting soils. landforms and dope positions.
Management conditions of cultivated fields at the time of "'Cs deposition are also
important in determining the likelihood of surface runoff before "'Cs a t t achen t and possible
transport downslope. -\gncultural land cover in upland regions of southem Ontario during the
period of major deposition was that of a mixed f m i n g management (Ontario Departrnrnt of
Agriculture. 1967). Corn. cereals and foragelpasture \vas a typical cropping mis. Interception of a
portion of "'Cs laden rainfall by live and decaying crop matter would be expected. more so in the
surnmer months. Most adsorbed '"Cs. however. was washed from vegetation and moved to the
soi1 (Davis. 1963: Dahlman et al.. 1 975). Rogowski and Tamura ( 1 970) found that 939/0 of the
'''Cs applied to grass washed off during the iïrst year and my absorbed "'CS would be released
to soils when vegetation dies and decays. Uptake by vegetation from soils or water (Eyman and
Page 112
Kevem. 1975) is low. The subsequent removal of crop matter. as with grain or ha- or silage corn.
containing an?; cesium would be a potentiai loss from initiai deposition levels: however. this
removal pathwap is veq small (Brown et al.. 198 la). less than 1 O/' (de Jong et al.. 1981). and
would be relatively unitom over a field area.
For upland agricultural areas of southem Ontario of complex or rolling topography and
medium textured soils. landscape profile development would be an underlying factor to the
differential ability of a field to infiltrate rainfall and allow runotT. The three dimensional shapc: of
the hillslope has been shown to be important in detsrmining soil redistribution (Pennock and de
Jong. 1990) more so than simple gradient length relationships in man? natural landscapes (MW
and de Jong. 1987). The natunl tendency of topography -erosion relationships would also affect
cesium's differential adsorption and transport. Topography containing slopes of signiticant
length and steepness are most susceptible to surface runoff. The upper to mid positions of
greatest dope gradient where runoff potential is high would experience preferential losses of *
runotTconstituents. Losses from the upper areas would be transponed downslope and would
partially compensate runoîT losses at lower positions. Therefore "'Cs deposited in ninfall
on a hillslopr would ha\-e the least opportunit- for maintaining equal soi1 adsorption
concentrations at the upper dope surface.
3.1.3.2 Cesium-137 runoff on sediment during deposition
Soil characteristics would likely have resulted in varying degrees of adsorption of crsium
to the soil surface. Soils of highrr clap contents or organic maner would positively affect
cesium's preferential adsorption to these constituents (Davis. 1963). Adsorption on soils and
sediments is reported to be rapid (Eyman and Kevern. 1975) with distribution in undisturbed soil
Page 113
profiles showing an exponential decrease with soil depth ( Ritchie et al.. 1970). In some areas.
high sand content and cracking clay soils allow geater penetntion emphasizing the importance
of the amount and type of clay (Lance et al.. 1986).
Infiltration capability of a soil would determine the arnount of precipitation that can enter
the soil before surface saturation occurs. and runoff results. Once mno ff begins. the potcntial
movement of soil will be determined by particle detachment snergy. Fine clay particles clnriched
mith cesium orice detachrd will readily remain in suspension. Clay particle transport with
cesiurn is potentially very high (Wise. 1980) as the majority of cesium deposition will be on clax
particles at the very surface. ~Maule and Dudas ( 1989) fractionated soil sarnples and drtermined
that the oganic fraction containsd nearly half of the total ' 3 7 ~ s activip-. Therefore '"CS \vas not
solely associated with the mineral fraction. Erosion potential increases by a soi 1s erodibility.
lower surface roughness and seasonal conditions that influence antecedent moisture. rainfall
intensity and vegetative çover. Under southem Ontario climatic conditions. spring runoff losses
o f soil are highrst compared with winter losses that are low (estimated to be onl?. 10% of total
annuaI srosion) (van Vliet and Wall. 198 1 ).
3.1.3.3 Cesium- 137 runoff prior to incorporation
Depositional ' ' 7 ~ s is reported to be srrongly reactive and rapidly adsorbs ont0 negatively
charged surface particles forming colloid cornpleses (Pemock. 1990). It has the ability to
displace other cations frorn clay minerals and once adsorbed has vrry limited mobility by
chernical processes. virtually non-exchangeable. As a result. spring and summer '''Cs deposition
not imrnediately transponed in rainfall runoff would have resided for the most part at the prima-
particle laver at the surface until mixing by cultivation in the fa11 or the follow-ing spring. This
Page 114
extended period of up to several months would leave the surface particle laver wlnerable to
physical transport processes and redistribution downslope.
Erosion events by water or possibly wind pnor to cultivation may preferentially move
tine particles and associated "'CS downslope or to field boundarics. Surface soil particles of high
surface area clays. light humic matrrial and small aggregates would be the first to br tnnsportrd.
DitYrrent soils would have a signiticant influence on the enrichment ratio and rrosion rate error.
Incorporation by tillage results in uniform "'Cs distribution in the plough layer ( Ritchir
and McHenr';. 1973). If erosion occurred prior to cultivation. important potential errors in
erosion estimation from "'CS are likely. The erosion calculations would senously overestimate
the amount of soil lost (de Jong et al.. 1982: Bremrr et al.. 1995). and overestimate deposition.
Redistribution of "'CS attached to surface soil particles during nuclear fallout from crest
or sloping areas to areas of deposition could result in a significant overestimation of erosion at
upprr and mid slopes using the "'Cs reference analysis. A consistent themr that has emergrd
from existing research using "'Cs analysis is that the actual amount of erosion occumng is much
higher than previously thought (Pennock. 1990). High soi1 losses from crest and convrx positions
of landscapes in Canada have been ohen reported (de Jong et al.. 1983: Martz and de Jong. 1987:
Pennock and de Jong. IWO: Bernard and Laverdirre. 1992: Cao et al.. 1993) contra- to erosion
models that associate incrrases in soil loss with increases in dope length. The high lossrs on
upper positions combined with the high variability of "'Cs found in lower positions in many
studies. may be attributed to two possibilitics: 1 ) upper slope losses are the result of physical
erosion processes like wind or tillage (Pemock and de Jong. 1990). or 2) the variability in "'Cs
concentrations must have been introduced during deposition and accumulation of the "'Cs in the
Page 115
soil (Lance et al., 1986). The relative importance can only be speculated fiom the research
available.
3.1.4 Nonuniform deposition of cesium-137
It has bern uidely accepted and considered important to assume uniform distribution of
' 3 7 ~ s deposition when estimating soil redistribution. Regionally. it is weI1 noted. precipitation
patterns determine the amount of cesium deposited. However locally on a field scalr or small
watershed where most soil erosion measurements and interpretations have been made- crsium
adsorption is considrred complete at the point of contact without significant movement in ninoff
\vater or transported sediment. Few papers have considered the potentially significant
overestimation of erosion and al1 have accepted this aspect of the l'?Cs reference technique in
determining soil movsment and redistribution. de Jong et al. ( 1982) and Bremer et al. ( 1995)
considered rrror in tirld erosion measurement using "'Cs and acknowledge the serious
overestirnation of soil loss if "'Cs was not conîined to the surtàce of the soil until cultivation
rnixed it with the whote Xp.
The consequencr of nonuniform deposition levels of "'CS at the field scale directly
impacts the estimation of soil redistribution. If the baseline levels of "7Cs at upper or mid slope
positions were reduccd from runoff losses at deposition. they cannot be assumed to be equal to
the reference levels mcasured in the forest cover. As upper and mid dope positions ma- have
brrn reduced. the resultant increase of "'Cs in areas of lower slope and depression at the timr
will be reflected as additional soi1 gains above actual soil accumulations. The initial imbalance
would be incorporated within any future Ij7Cs inventon; cornparison as present calculations
assume no initial difference.
Page 116
The rnethod of reporting soi1 movement would be affected from measurernents
incorporating an initial imbalance of soil "'CS. When considering the entire landscape profile.
the measurement of gross soi1 loss Ieaving a field wil1 not be signiticantlp alterred. The
1 )7 measured net soil loss using Cs however ma- need to be reduced as a result of a correction for
this initial imbalance. The rate of erosion or soil loss per year will be reduced due to the
overestimation of soil movement. The mean rate of rrosion. teferring to a fields croding areas.
would be seriously affected. The actual erosion rate would be lower at areas that oncountered
initial reductions in "'Cs inventon at the time of deposition.
3.1.5 Two sample method of cesium-137 analysis
nie alternative technique to assuming uniform fallout patterns is to monitor and mesure
actual changes in "'Cs with time at a site (Kachanoski. 1987: Kachanoski and de Jong. 1984).
This allows a cornparison ofactual measurements in determining erosion rates over time. The
accuracy of the estimated erosion rates improves with time. For Saskatchewan soils whrrr soi1
rrosion rates are relativsly low. Kachanoski and de Jong ( 1984) sugested the minimum
sampling intemal to br approximately 15 y. in order for the relative erron to be <=:O%. The
lengthy sampling inteneal nec r s sq to reduce these error calculations in estimating rrosion rates
by the two sample method discourages its use and encourages the use and acceptance of the '"Cs
reference method with the assumptions implied. As Ritchie and M c H e q ( 1985) have suggested
and a large volume of literature supports. "'Cs reference analysis is the rnethod of choice when
quick and accurate estimatrs of recent sediment accumulation are needed.
3.1.6 Testing the assumption of uniform cesium-137 deposition
Attempting to account for conditions at the time of deposition to determine potential
Page 117
runot-relies on accurate historical information more than 30 years ago. Bremer et al. ( 1995)
acknowledge the dificulties in obtaining an accurate correction because it requires detailed
information on the timing and intensitp of "'CS deposition. tillage operations. and crosion events.
The degree of impact is only increased with greater sloping fields. bare soil and greater natural
erodibility. For cesium measurements to be representative. the question of whethsr cesium could
have moved with runoffunng deposition eçents should be explored.
To explore the question of whether cesiurn \vas removrd frorn the soil surface during the
period of deposition. a study must sirnulate the "'CS deliven; procsss and field conditions at that
time using an accurate and pnctical method. The deposition of "'Cs in precipitation c m be
realistically and accurately simulated with a portable rainfall simulator. The ninfall simulator
c m be employed in the tield to deliver close to natural raindrop distribution and intensit!.
(Tosse11 et al., 1987).
To mrasure the ninoff potentiai and distribution characteristics of cesium- 137 as it would
have been deposited. the application of a nonradioactive cesium isotope would br prefenble to
eliminate the personal and environmental safsty risk. Cesium- 1 33 is nonradioactiw and 1 OO0o
natural yet not found in soils (Christian and Feldman. 1970). The natural cesiums' large ionic
radius. high solubility. large disassociation constant. and ion exchangeability ( Weast. 1987) are
charactenstic to the element and are no different than the radioactive isotope making it a suitable
tracer for investigating the possible redistribution of "'CS. The identical physical and chemical
characteristics to "'Cs allou. natural cesium to act with the same adsorption and bonding
capability upon delivery in the rainwatsr. upon contact to the soil surface and in rainfall runoff.
3.1.6.1 Field experiment Objectives and Hypothesis
Page 118
.A rainfall simulation rxperiment was conducted to test the assumption that Cs'" fallout
from a rainfall event would result in a uni f o m distribution of CS"' over the soil landscape.
Tosse11 et al. (1987) developed a multiple drop rainfall simulator (Guelph Rainfall Simulator)
that is capable of reproducing storm intensities ranging from 17.5mm hi' to in excess of 2OOrnm
hr-'. Employing the small plot rainfall simulator technology was appropriate for rvaluating
cesium redistribution on a hillslopr because of its reproducibility of the climatic characreristics
on field conditions at the time of deposition. The srnall plot scale provides an accurate rneasure
of the interril1 erosion processes to determine the potential movement of cesium immediately
upon deposition.
The study used rainfall simulation technology to measure the partitioning of the applied
cesium in min water benveen small plot surface runoff. plot soil surface and plot surface residue
on a hillslope in a cultivatrd field. This study was intrnded to simulate agronomie. landform and
rainfall conditions pressnt at the time of "'Cs deposition near Guelph. Ontario to evaluatr
potential variability of "Cs application to landscapes in this temperate-humid climate.
Objectives:
1 ) to evaluate the use of residrnt "'Cs in soil as a tracer for estimating soil redistribution rates on
upland landscapes in southern Ontario:
2) to determine whether landscape position of contrasting dope gradient affected the rate of '"CS
deposition.
Hypothesis:
Atmospheric "'Cs fallout was uniformly deposited and adsorbed b- soi1 on upland landscapes in
southern Ontario.
Page 119
3.2 Methods
3.2. l Site characteristics
The rainfall simulation experiment was established on a field site approxirnatrly 4 km
West of Rockwood. Ontario within 0.5 km of the hillslope f m site (Chapter 2). The soil \vas a
Guelph (Humic Gray Brown Luvisol) loam to t h e smdy loam developed tiom glacial till parent
matenal. Average suface sand. silt and clay contents Lvere 53%. 36% and 1 1%. respectively-
The study site kvas located on a simple slope on the West aspect of a drumlin face.
h u a 1 precipitation of the study area averages 834mrn tiom 30-year ( 1950- 1980) records
of the University of Guelph climate station (Atmospheric Environment Service. 1985). located 7
km south of the study site. The precipitation recorded during the spring. summer and early fa11 of
the depositional period of 1962 and 1964 is within 10% of the 30-year average (Table 3.1 )
and within 3O?6 for 1963. The 7-month penod included indicated 88% of al1 higher intensity
storms (N Omm day-' ) occurred during the Apnl to October cropping season. Erosivity or
intensity of the rvents in mm hi' is not provided: however. greater rainfall events during the late
spring and summer for the region are chancteristically of high intensity and short duration.
Crop management at the study site had been a conventional cash-crop system of corn.
soybeans and wheat for approsimately the last 20 years. The rainfall simulation rxperimrnt was
camïed out on Xugust 12 and 14. 1997 while the field was in soybeans. Live soybean plants rvere
cut at the soil surface and removed without disrupting the plot surface. Spring tillage included
chisel plough and two cultivation passes. Surface residue cover was estimated ( Denholm and
Schut. 1993) at 15% cover consisting of recent leafdrop and residue from the previous years
soy bean crop.
Page 120
Table 3.1 : Precipitation recorded at the University of Guelph. Ontario dunng the peak
atmospheric fallout of "'Cs.
Guelph OAC Precipitation (mm) Rainfail events of high accumulation (mm day" )
Period 1962 1963 1964 30-yr 1962 1963 1964
A pri l 57.7
May 23.9
June 85.5
July 77.4
A ugus t 50.9
September 68.1
October 1 08.9
Season total 472.4 36 1.3 9 5 18 16 4 14 - 3 18 7
Annual total 685.3 564.7 825.9 833 19 5 16 - 7 19 8 *Number of ciail:. rainfall accumulations of IO to 20 mm and >20 mm.
3.2.2 Rainfall simulation field experiment
The portable Guelph rainfall simulator II as descnbed by Tosse11 et al.. ( 1987) was
employed to delivrr a ninfall event. Three landscape positions of the hillslope were chosen of
contrasting dope gradients. The plots ( 1 m s I rn square with sheet metal borders) were installrd
at crest ( 1.5%). upper (jo4 and mid (6.5%) slope with three replicates spaced approximately 5m
apart along each position treatment for a total of 9 plots.
Fifty grarns of cesium chioride (CsCl) was added to the rain water source ( I-IOO L in a
plastic tank) and mi'ted to a concentration of 78 ppm disassociated cesium ('"Cs-). An); cesium
mnoff concentrations were anticipated to be less than the applied concentration and within the
Page 121
ideal working range of 0.06-24 ppm of analytical detection using atomic absorption
spectrometry. U S (Varian. 1990).
The water for both days of simulation was of drinking water quality containing dissolved
minerais. The water source for the rainfall on Day 1 was the Davis farm well which had present
Ca ( 100ppm). Mg (52ppm). Na ( 14pprn). K ( 1 Oppm) and Fe (<O. 1 pprn). Concrms of low
groundwater Ièvels prompted the water source on Day 2 to corne from the University of Guelph. C
Land Resource Science Department containing Ca ( l9ppm). Mg (7ppm). Na ( 132ppm). K
( 1 pprn) and Fe (<O. 1 pprn). The presence of exchangeable ions in the rain watrr was presurnrd to
cause little primary attachment competition for cesium as concentrations of these ions on the soi1
surface would be significantly greater. The water kvas transported to the field site in 2 10L barrels
and transferred bp sump pump into the 1500L rainfall sirnulator source tank adjacent to the plots.
The rainfall simulation runs began at the mid slope position and moved to the upper and
then crest position afier completing rach of threr slope rrplicates. Two simulators uere used in
alternation. Each plot required a positioning of the novle directly above and perpendicular to the
plot centre 1 . h from the surface. a set-up procedure done sirnultaneous to an ongoing
simulation run. On Day 1 of the experiment. each plot kvas rained on with water containing the
cesium . At the cornpletion of the simulation runs. the tank was rmptied and rinsed with clean
water. Atier refilling the tank on Day 2 approximately 48 hours later. the sarne min faIl duration
and intensity was used without cesium added to the min water. Behveen Day 1 and Dap 2. the
plots remained covered with wood boards to discount an); interim cesium losses from rainfall or
surface disturbance.
Rainfall simulation \vas conducted for a 10 minute period using a 1/2HH stainless steel
Page 122
nonle (Sprayer Sy stems Co.. Wheaton. Illinois) which delivered an intensity of 1 -82 mm min-'
or 1 10 mm h-' on the replicated plots at the three slope positions. This rainfall event is
equivalent to less than a 1 : 10 year return period storm in the study area (Department of Supply
and Services. 1978). Plot ninoff was collected ont0 a flat. triangle shaped trough that drained
into a hose connected to a kracuum systrm. The runoff kvas drawn immediately into graduated
cylinders used to record volumes at 1 minute intervals. up to the 10 minutes of rainfall. The
circular shaped rainfall pattern that fell beyond the plot b o u n d q required runotT outside the
plots to be redirected away from the collection area and a transparent plexiglass sheet was used to
cover the collection trough.
In the Guelph area the frequency of rainfall events of a 10 minute duration at an intensity
of: 70 mm h" is once even 2 years: 107 mm h-' once eveq 5 years: 130 mm h-' once cvep 1 O
years: and 150 mm h-' once every 25 years (Department of Supply and Services. 1978). The
rainfall sirnulator delivery kvas determined in the field. A plot area was covered with a plastic
tarp anached to the metal borders with clothes pins to allow any w t e r hitting the tarp surface to
run onto the collection trough. Ten minute mns n-ere completed with mnot'fvolume recorded at I
minute intervals as well as total rainfall deiivery. The rainfall amount kvas used to determine the
total cesium mass input.
3.2.2.1 Field sample collection
Prior to rainfall. loose soi1 samples were taken at a depth of O-km from the surrounding
area of the plots to determine soil texture. particle size distribution. pH. organic matter. calcium
carbonate and soil moisture (McKeague. 1978). Shalloa soil samples of less than lcm depth
were taken fiom surface areas surrounding the plots and analyzed by AAS to determine anp
Page 123
positive background readings of cesiurn. Soi1 core rings (hm x h m ) removed from the plot
surface afier rainfall simulation to measure residual cesium were also used to determine post
esperiment soil bulk density and soil moisture at the near surface of the plot.
The plot runotT volumes collected d e r each minute were composited into a pail from
which two IOOml samples were taken afier thorough mixing. One plot runoff sample uas
col lected to detemine ssdiment concentration and cesium concentration. The second runo fi
sample was considered a backup. Subsequent runoffafter rainfall had stopprd was collected
separatel-. recorded and sampled. In addition. the sedimrnt remaininp on the trough surface was
washed down into the vacuum hose using a known volume ofdeionized luter. collscted and
sampled. Plastic collection cups were placed adjacent to the plot border duting each simulation
and sampled to check actual cesium concentrations delivered in the rainfall. Al1 samplrs from
the simulation experiment were stored the sarne da? at 4 degrees C until removed for analysis of
sediment concentration and latrr cesium content. An). equilibrium reactions b c t ~ s s n crsium in
suspension or attachment to organic matter and soil particles \vas irrelevant as total cesium
content of the runoff sample mas to be determined.
Cesium rrmaining on or nrar the soil surface was measured afier each simulation day
using threr soil cores (5.7cm K k m ) sarnpled diagonally across rach plot about 1 hour aHrr the
rainfall event. The O-;cm soi1 depth sampled was assumed to contain the large majority of
applied cesium (Rogowski and Tamura. 1970). Prior to the second simulation day. core ring
plugs covered in impermeable tape were placed in the sampling holes. Afier the rainfall. the
surface crop residue \vas removed in its entirety from each plot for cesium analysis.
Page 124
Mass balance sampling methodology
The mass balance of cesium afier Da? 1 for each plot is expressed in equation 3. l where
the total applied cesium is equivdent to the surn of cesium found in the runoffand cesium found
in the surface soil. Cesium runo tT totais combine each of the 10 minute mnotT intervals during
the rainfall. the runoff into the trough afier the rainfall is stopped. and from the wash containing
(m'): Vir = volume of samplr tiltrate for each minute of ntnotT(m'): C, = concentntion of Cs in
sarnple tiltrate for each minute of runoff (g m"): W,, = weight of sample sediment for rach
minute of runoff (g,): C,, = concentration of Cs in sarnple srdiment for each minute of runoff(g
L~'' 1: V,, = volume of plot runoffwater for rach minute (m"): the above parameters are repeatsd
for the trough and wash portions collcctrd after the simulation: W, = total weight of the 3 soil
sample cores (g,,): C, = average concentration of Cs in the 3 sample cores (g &'): A, = area of
plot (m2): and A, = area of 3 soil cores (m').
In the above espression for tracking Day 1 cesium. cesiurn kvas an input with the rain
water. On Day 2. cesium was not added to the rain water but originated from cesium adsorbed on
the soil on each plot surface as detemined from soil cores extracted at the end of Day 1 :
W2 = [(WCSC,) * (4, - A,)/ A,)] (3 2 )
Page 125
where W, = weight of C s le% in surface soil for day 2 (g) : WC = total weight of the 3 soil sarnple
cores (_%): C, = average concentration of Cs in the 3 sample cores (g &'): -4, - A, = area of
plot (m') less area of 3 cores (m') from Da! 1.
The available cesium for ninoffon Day 2 is equal to the amount remaining on the surface
of the plots after Day 1. Equation 3.1 is modified slightly for Da? 1 mass balance to
acknorvledge the soil residual staning point and the Ioss of plot area removed in D e 1 soil cores:
where W, = weight of Cs leli in surface soil for Day 2 (g): WC, = total weight of the 3 soil
sarnple cores on da>- 2 (L~): Cc, = average concentration of Cs in 3 sample cores on Day 2 (g &.,-
I ): A,, = area of 3 soil cores removed in Day 2 (m'): Cr = average concentration of Cs in the
surface crop residue sampir (g L.," 1: W, = average weight of the surface crop residur sample
(g ) : W, = total weight of the surface crop residue (g): and the remaining parameters are drtïned
in equation 3.1.
The mass balance of cesium was determinrd afier Da- 2. The total applied cesium being
rquivalent to the sum of cesium found in the runoff from Day 1 and Da! 2. cesium remaining in
the surface soil atier Da? 2. cesium found in the surface residue alier Dap 2 and an' cesium loss
in the system. Surface residue was not collected afier Day 1 to avoid disturbance of the surface
but would have potentially removed or released cesium during the two rainfall events. Afier the
addition of the sample components. any difference from the input was artributed to
Page 126
unaccountable losses in the system fiom the point of deli~ery on the plot surface to the ninoff
sarnpling and to the laboratory analysis. Cesiurn would not be expected to readily adsorb to the
plot runoffcollection material or the plastic sample vials which were refrigerated the same da! of
collection. Precautions for ensuring sample integrity such as the use of new collection material to
avoid contamination and refrigeration to aven moisture loss were necessary to avoid changing
the characteristics of the sarnple.
3.2.2.2 Laboratory analvsis
Sample digestion
Sarnple digestions used a strong acid mix as per Method 4 of the Ontario Geological
Survey manual (Ontario Geological Survey. 1995) which included hydro tlouric acid ( HF).
perchloric acid (HCIO,) and nitnc acid (HNO,) in an 8 2 : 1 ratio. An initial 15 min reflusing strp
using 10ml of HNO, was included to digest the organic material in the soi1 sample and filter
paper. Al1 solutions produced were filtered with Watman's no.41 filter paper to dari6 and the
volume brought to 3 ml using Nanopure water. The sarnples were filtered directly into
polypropylene centrifuge tubes for storagr. Quality control was carefully monitored throughout
the procedure as al1 acids were of trace metal grade and laboratop material was new or the
material was cleaned with soap. soaksd for a minimum of three hours in an acid bath and rinsed
using Nanopure water before use.
Atomic absorption analysis
A tlarne atomic absorption spectrometer ( U S ) using air-acetylene was used to atomize
the sample solution containing cesiurn (Varian. 1990). Operating procedures and samplr
quantification followed the Amencan Public Health Association's standard methods for the
Page 127
examination of water and wastewater ( 1992). Detection limits used for calibration were
considered to be between minimum detectable (0.01 ppm) levels and the concentration of cesiurn
applied ((28ppm). This range closely enveloped the optimum instrument working range for
measuring cesium in solution of 0.06-24ppm at the 894.5nm wavelenpth.
The method used to measure cesium in solution required the addition of potassium to al1
samples at high concentrations (Grobenski et al.. 1983: Goeuel. 1985: Chapman et al.. 1986 ) of
approximately 2000pprn K to act as an ionization buffer. Al1 standards and samples analyzed
were subsarnpled to 10ml in test tubes containing approximately 0.05Ig of KNO: and stirred
with a sonicator.
3.2.2.3 Mass balance sample anaiysis
In completing the rnass balance of cesium. the laboratory analysis considered each
medium of the sarnpling components. The runofi portion \\;as analyzed for sediment
concentration by dmwing a representative subsample of 50 ml from the 100 ml collection vials
and filtering through a O.45pm filter paper. The preweighed tilter and the rntrapped sediment
were oven-dried ( 105°C) for a minimum of 6 hours before deterrnining the sediment weight and
calculating a sedirnrnt concentration in the runoffof mg 1-'. The remaining 50 ml of the samplr
was analyzed for total cesium using AAS.
Since the cesium could be analyzed in the clear filtrate portion untreated. the easiest
method of analysis was to fraction the sample into filtrate (which still contained clay particles
less than 0.45 um ennched with cesium). and the sediment portion requiring complete digestion
before AAS analysis. The rainfall collected was analyzed untreated to ver@ the cesium
concentration input. This clear liquid sample was used in establishing the upper range of working
Page 128
standards for ;\AS analysis. The total cesium removed in the runofl'was determined by summing
filtrate and solution portions that comprise the total runoffvolurne collected. The amount of
cesium removed in the runofi as a fraction of what was applied was determined.
Cesium remaining on the soil surface was sampled liorn soi1 cores afier each simulation
day. and h m crop residue taken afier the plot surface had dned from simulation on Day 2. The
soil cores were weighed wet. then oven-dried. the soil sieved through a ?mm sieve (wrighing any
gravel) and weighed. The three cores from each plot were composited and a 1.00 g subsarnple
\vas taken. digested completely (Ontario Geologicai Survey. 1990) and analyzed on the atomic
absorption spectrometer. Crop residue afier air drying and weighing was ground through a 1 mm
sieve using a plant grinder to ensure a hornogenous digestion smple. Represrntativr 1 -00g
residue subsarnples were taken in triplicate from each plot sample. digested completely using the
sarne strong acid mis because of the sediment attached to the residue. and analyzed for cesium
concentration using the XAS. Blanks were carrird through ail digestion and fiIrration steps:
howewr. corrections were not necessq . .&Il the components analyzed were then combined to
determine the relative proportion of crsium residue and detemine differences with slope
position.
3.2.3 Statistical analysis
Rainfall simulation mnoK soil loss. crsiurn loss. and plot soi1 data over the two days
were analyzed for differences between the landscape position treatment using analpsis of
variance (ANOVA). The SAS statistical package was used for the analysis (SAS Institute. Inc..
1998). The least squares difference test taas used to determine treatment mean differences.
Simple correlation analysis between al1 sampling parameters waas conducted afier pooling
Page 129
treatment data. Ml analysis was at the 9j0% probability level unlsss othenvise notrd.
3.3 Results
3.3.1 Soil properties
Soil property measurements taken fiom each of the plots and statistically compared by
landscape position indicated some signiticant variation over a relaticely short hillslope distance
(Table 3.2). Tssture analysis did not indicate a signiticant difference (~-4.05) in clay content
(average of 1 O J o h rneasursd gravimetricallp): however. the crest position was significantl>- louer
in sand content (50.9%) than upper (53.7%) or mid (55.4%) position. Coarse fragment content
or grave1 found at the mid slope surface of 8.6% was significantly greater than upper or crest
positions.
The pH of the crest area of 7.3 \vas signiticantly lower than mid dope (7.5) which \vas
supponed by signiticantly greater calcium carbonate (CaCO.) activity levels at the mid dope
position (Table 3.3). The CaCO,. oripinating tiorn the calcareous parent material. mrasured at
the soi1 surtacs of the three positions (mid-7.456. upper-3.9?/0. crest- 1 -9%) indicated past srosion
and subsoil incorporation into the Ap horizon. Organic matter in the surface soi1 of the mid dope
of 1.7% has been reduced significantlp from upper and crest areas of 1.2% and 2.3%
respectively. The status of the meûsured soi1 parameters: organic maner. calcium carbonate and
grave1 content near the surface indicatr an increased state of degradation at the mid slopr
position over the upper and crest positions. Nonhomogeneous soi! conditions across the site
merits investigation of any statistical relationship with measured expenmental parameters and is
discussed in section 3 2.5.
Page 130
Table 3.2: Soi1 characteristics at each landscape position pnor to ninfall simulation experiment.
Landscape Total Total Gravel pH Calcium Organic Moisture content Moisture content position clay sand >?mm ( CaCIT) carbonate matter Da! i Da! 2
(5%) c ?6) (%) (%) (%) (%) (%)
Mid 1 1.0a* 55.la 8.6a 7.5a 7.4a 1.7a 439a I1.9Ia
tipper 10.6a 53.7a 4.4b 7.4ab 3.9b 2.26 4.86a 13.89a
Crest 10.6a 50.9b 1-86 7.3b 1.9b 2.3 b 1.46a 12.61a
* Mean values for the same parameter are not significantly (p<O.OS) different if followed by the sarne letter.
Antecedent soil moisture content at the soil surface adjacent to the plots did not v q e
statistically (Table 3 .2 ) and ,-as very Iow (4.6%) prior to rainfdl simulation on Da! 1 indicative
of below average July and earl!. August precipitation. M e r the plots had recrik-ed simulated
rainfall and covsred for 48 hrs. the plot surface moisture content before raining on Da). 2 Lvere
sirnilar averaging 12.j0'a Tor al1 locations pnor to simulated ninfall. The soi1 cores (depth of O-
h m ) removed afier the rainfat 1 for residual plot cesium analysis were also measured at the tirne
for soi1 moisture comparisons brtween positions. A significantly higher moisture content at the
crest position of 5.0?/0 over the mid position \vas obsswed on both days suggesting greatsr
rainfall infiltration occurred at the crest surface. Bulk density mesurement taken from the same
soil cores found levels indicative of dry seasonal conditions with measurements of I .-!-tg cm-3
t?om Da! 1 and a slight increase of 3.5% a%er raintàll and extraction on Da? 2.
3.3.2 Runoff water and sediment
3.3.2.1 Infiltration and mnoff
R~not~volurnes collected every minute of the rainfall simulation were recorded
(Appendix C) on both Day 1 and Day 2 of the experiment. Runoff was not measurable until afier
three minutes of simulation on Day 1. uith the exception of Plot 2. With higher antecedent soil
Page 131
moisture levels on Da! 2. however. measurable runoff occurred afier two minutes. DitTerrnces in
runoff voiume totals between the slope position treatment was not significant (p<O.Oj) for rither
day. Mean runoff volume was lO.08.9.jO. and 10.1 1 L m" for the mid. upper and crest slope
respectively. The average runoff coefficients calculated as plot runotT volume divided by plot
rainfall volume were 0.56 for Da? 1 and 0.72 for Day 2.
Comparison of runotf'\-olume within groups on Da); I of the rxperirnent indicatcd high
L-ariability between plors (Figure 3.2). Coefficients of variation (CV) (Snedecor and Cochran.
1989) for the three field treatments were still below 20% for the field sxpenment. The individual
position CV's were 19.9% for the mid. 2.j0/0 for the upper and 12.6% for the crest position
treatment. The differences found in runoff parameters between simulation runs were not due to
variable rainfall intensities. Rainfall sample volumes collected in small c a t chen t pails located
adjacent to the plot border were compared over the two days of simulation and had a CV of
6.7%.
3.3.2.2 Sediment loss
The rainfall application within treatments generated a l aqe variability of soil movement
off the plots (Figure 3.2). The CV for sedirnrnt Ioss from the mid position was 99.5%. the upper
position 60.7% and crest position 79.4%. Characteristic differences in sediment loss bet~t'een
plots were etident on bot11 simulation days (Figure 3.2) as the correlation coefficient \vas r =
0.947 for Da): 1 and Day 2 srdiment loss (Appendix C). The equivalent soil erosion rates
associated with the sediment loss from the 1 m x I m plots ranged between 0.30 to 1.94 t ha" on
Day 1 and 0.16 to 2.43 t ha-' on Dap 2.
Page 132
Runoff volume fiom plots
1 U P P ~ ~ I Crest
Landscape position
Day 1 [7 Day 2
Sediment losses from plots
kl id 1 U P P ~ ~ I C rest Landscape position
Figure 3.2: Rainfall simulation total runoff volume and sediment losses measured from each
plot on Da? 1 and Day 2.
Page 133
3.3.3 Cesium runoff and mass balance
33.3.1 Cesium losses from Day 1
On Dq 1. the rainfall runoffcontained significant leve ds of the app lied cesium as
determined from runoff water and runoff sediment portions. When combined, the total runol'f
loss of cesium from Day 1 was 38.8% of the applied amount. Almost the entire amount of this
cesium loss was detected in the water portion of runoff (90.8% of cesium runoff). Less than 1 0°/0
of applied cesium was measured fiom the filtered sediment portion. Treaunent mean staristical
cornparisons (p<O.Oj) dstermined that cesiurn losses bp dope positions were not sipniticantly
different (Table 3.3). Mid. upper and crest position runoff losses of cesiurn on Day 1 nere
39.3%. 35.2% and 42.0% of cesiurn applied respectively. Crsium was retrieved in al1 sample
types measured ( -4ppendi.u C ).
The amount of cesiurn remaining on the soil and on the surface residue was assumed ro
be the residual portion of cesium applied. Sample analysis of soil taken from the surface one cm
of ninr areas surrounding the plots was considered for background levels. These were low and
w-ithin the .%AS instrument variabi lin- readings for blank sample mrasuremrnts and theretors
considered to be nrgligible. The surface residue portion kvas not removed and analyzed afier Day
1 to avoid disturbance of the surtace. The cesium adsorbed to residue was accounted tor afier
Day 2 rainfall simulation.
Cesiurn measured from soil cores (depth of 0-3cm) taken from the plot surface afier Dq.
1 simulation revealrd a residual amount equal to what was retneved in the runo& an a\-erage of
39.3% over ail plots. The soil levels differed significantly however between positions with an
increasing amount of cesium measured from the plot soil as dope gradient was reduced. Mid.
Page 134
upper and crest position soil surface levels were 30.2%. 38.6% and 48.9% of cesium applied
respectively. The cornbined lrvels of cesium retrieved fiom the hillslope as a result were
significantly diRerent. The runoff and soil portions accounted for 69.5% of the mid dope. 73.goh
of the upper dope and the crest cesiurn recovered accounted for 90.9% of applied cesium on Day
1.
Table 3.3: Cesiurn retrieved tiorn sample portions by landscape treatment on Day 1 and Day 2.
Landscape Da! 1 Dav 2
position Runoff Soil RunotT Soit Residue Retrieved Unretrieved total (9/0) O - k m (%) total (%) O-3cm (%) matter (%) total (%) total (96)
b1 id 39 .h* 3 0 2 3.2a 32.5a O.5a 75.0a 25.0a
Upper 3 5 3 38.6ab 3. la 33.9a O.4a 72.3a 27-73 Crest -iZ.Oa 48.9b 3 .?a 4 l.3a 0.6a 86.8a 1 3 3
* Mean values for the same parameter are not significantly (p<O.Oj) different if followed bq' the same lener.
3.3.3.2 Cesium losses from Day 2
On Da)- 2. the applied rainfall was round to contain residual amounts of cesium remaining
in the water source from the applied arnount of Day I (Appendix C). Plot I rainfall contained the
greatest level of 1.796 of applied concentration from Da? 1 and the rernaining plots received less C
than 196 of Day 1 application levels. The concentration of cesium From each rainfall \vas sampled
and considered as additional input into the m a s balance equation. Day 2 results included this
correction.
Cesium retrieved from Day 2 rainfall \vas measured in al1 sarnple types (Appendix C) but
at substantially reduced levels. The rainfall runoffcontained a total of 3.1 % of the applied cesium
Page 135
as determined from ninotT water and runoff sediment portions. Again. the greater amount of this
cesium Ioss kvas detected in the uater portion (78 .O% of cesium runoff). RunotT treatment mean
statistical cornparisons (p<O.Oj) determined the slope positions were not significantly ditferent
(Table 3.3). Mid. upper and crest position runoff losses were 3.20/6.3.1 O h and 3.2% of cesium
applied respectivelp. The cesium runoff losses from Da- 2 were approximately 8.2% of m o f f
losses from Day 1 .
The amount of cesium rsmaining within the plot including the residue portion aArr Day I
and Day 2 rainfalls \vas assumed to be the remaining portion of the total cesiurn amount applied.
The residue was removed and analyzed and accounted for an average of 0.5% of cesium applied.
The low concentration levrls were uniforrn between plots and an) plot variability was a
reflection of the residue arnount recovered fiom each plot. Cesium measured tiom soil cores
(depth of 0-km) after Da! 2 rainfall simulation had an average reduction in levels from 38.60;0
from D q 1 to 3j . joh of applied cesium alier Day 2. The reduction in surtàce soil cesium levels
by 3.1 ?b rquals the arnount ofcesium loss in the runofi retrieved from Da? 2 . The soil cesium
levels were not signiticantly different bstween positions but increased in the amount of cesium
measured from the mid (32.5%) to upper (33.9%) to crest (-11 3%) slope position. The cesium
mass balance including the rneasurements of al1 portion quantiries is shorvn raphically in Figure
3.3.
The mass balance of cesiurn retneved from the hillslope afier Day 2 was determined from
the addition of cesium amounts found in Day 1 ninoff. Day 2 runon Day 2 soil and Da? 2
residue (Table 3.3). The runoff. soi1 and residue portions accounted for 75.0?40 of the mid dope
cesium applied. 72.3% of the iipper slope cesium applied and 86.8% of the crest slope cesium
Page 136
) 1 Dl water 1 1 D7 water 1 1 D2 soil
1 O0
80
60
40
--
Dl sediment II/ D1 ilsdirnent Residue FI DitErence
20
Figure 3.3: Cesiurn m a s balance from runoff components where: D 1 watrr and D? lvairr is
the water ninoff portion from Day 1 and Day 1 respectivrly : D 1 sediment and D1
sediment is the sediment runoff portion from Day 1 and Day 2 respectively D2
soil is the soil O-3cm portion from Day 2: Residuc is the surface crop residue
removrd on Day 2: Di tkence is the cesium balance not detected.
'
O ' Mid U P P ~ ~ Crsst
Landscape position average
Page 137
applied. The average recovery rate of the cesium from the experiment was 78.0%.
33.4 Relationship between soil erosion and cesium loss at the time of deposition
The arnount of cesium loss as it related to soil erosion was found to be more closely
associated with the rvater runoff portion than the arnount of soil leaving a plot. Since the runotf
volumes were not statistically different. data comparisons were made from al1 nine plots. The
rate ofcesium loss from the plots on the day of application was highly correlated kvith plot runoff
with a r = 0.987 (Xppendix C). Day 1 cesium loss was positively correlated with sediment
loading. but with rveaker correlation. r = 0.671. The arnount of sediment removed in relation to
the runotTvolume was not as strong a relationship â = 0.771 1 as the cesium removed on Da? 1.
Ovrnll loss of cesium in relation to runoffvolumes from both the day of application and
the subsequent rainfall svent was highly correlated. r = 0.864. Over both rainiàll svents. the
reiationship between total cesium loss and total sediment tvas weaker with a correlation
coetXcisnt of r = 0.702. The relationship between cornbined runoff volumes and sediment losses
over both days was not as strongly related. r = 0.696. The cesium applied in the rainwater did not
readily adsorb onto particles that were transponed in overland flou-. Plots that deh-ered higher
sediment losses did not correspond to proportionally higher losses of cesium in runot't:
3.3.5 Relationship behveen soi1 erosion, cesium l o s and soil properties
The stronger relationship between runoff rates and cesium loss versus the rate of soi1
movement and cesiurn loss suggests that factors influencing infiltration and o~erland tlow. i.e.
soil properties. ma!* have intluenced the potential rate of cesiurn loss during deposition.
Relationships between the rneasured plot soil properties indicated correlation of r > ! 0.750 1
(Appendix C) with the arnount of cesium remaining on the plot soil sarnpled aRer Day 1
Page 138
application. Sand content. pH and CaCO, varied inversely to the concentration of cesium
remaining on the soil afier Day 1 r = -0.816. -0.78 1. -0.766 respectivelp). n e soi1 organic matter
levels were positively correlated with Day I soil cesium levels r = 0.938) possibly the result of
high adsorption of cesium to organic fractions within the soil. Cesium was not preferentially
adsorbed to organic crop residue sincr only 0.5% of the applied cesiurn \vas measured on the
15% crop residue cover.
3.4 Discussion and conclusions
The unrecoverrd amount of cesiurn may be atuibuted to rapid infiltration belo~v the
sampling depth of the O-;cm cores. During heak?. rainfall events. preferential flow including
rnacropore flow through the soil profile can contribute significant amounts ot'rapid downward
movement of ninfall watrr and accompanying contaminants (Edrvards et al.. 1992) to depths
wrll below the Ap (Wall et al.. 1997). This pathway could deposit cesium. if not immediatrly
attached to the surface particles. to depths below the 3cm depth of the soil corr san~ple. Visual
obsenation of the unifoml>- wetted soil surface below the removed cores suppon the possibilitv
of percolation beloa. the k m depth. The concentrations of any subsurface deposition rvould be
roughlp equivalent to application concentrations and couid represent a significant portion of the
applied cesiurn.
The proportion of rainfall infiltration in macropore and micropore openings is dependent
on the rainfall intensity and duration. the degree of macropore density in the surface soi1 protile
and soil infiltration characteristics. Soil conditions at the time of the cesium application were
extremely d q (4.6% soil moisture content) at the surface before simulated rainfall on Day I
Page 139
indicating ven; linle soil moisture other than in the smallest pore spaces. Edwards et al. ( 1992)
reported percolate tlow occurring to a depth of 3Ocm after 2.2 minutes of a "high" intensity 2mm
min-' simulated rainfall on a relatively dry antecedent soil condition. Macropore tlow was found
to increase with increased Storm intensity. In the large soil block (30 s 30 i( 3Ocm) esperiments
( Edwards et al.. 1993.34 to 60% of percolate collected drained from 1 of 64 possible
percolation sampling cells emphasizing the lack of homogeneous tlow through soils. The
wriability of macropore tlow betwern plot sizes of 1 m x 1 m area is likely similad>- large.
Diiuv et al. ( 1998) developed a simple mode1 to predict the contribution of macropore
flow to infiltration at a site of silt-loam soil texture in southern Ontario. Measurements were used
of intiltration intercepted at a depth of 5Ocm below small plots using the same rainfall simulation
technolog) to applg a uniform rainfall to an undisturbed surface. Upwards of 50% of total
infiltration during a storm of I in 10 year intensity was reportedly due to macropore tlow. This
method provided an estimate of the potential transport and unrecovered cesium in the study
through this prekrentiat pathway.
The 50% infiltration ratio of macropore tlow to total infiltration was applisd to al1
experimental infiltration volumes for the site and compared to the measured differencr in cesium
retrieved. Day I infiltration and cesium input amounts were considered on&. as Day 2 cesium
losses u-ere accounted for in Day 2 runof'tl The unrecovered amounts from cesium mass balances
of the plots indicated greatest retrieval at the crest position and lowest at the upper position
(Table 3.3). Estimated amounts of unrecovered cesium as dctermined by the 50°h macropore
infiltration ratio followed a similar trend as the measured amounts. The crest position (20.7%)
and mid position (20.9%) were lower and the greatest amount of cesiurn unaccounted for \vas at
Page 140
the upper position where 23.9% of the applird cesium was not retrieved. When considering
macropore flow. the results of cesiurn loss estimation when avengrd over the site (2 1.9%)
accounted for the entire amount of cesium unretrieved fiom the mass balance calculation
(32.0%).
No anempt was made to mesure cesium deposition by any process below the ;cm depth.
The problem of sampling preferential cesiurn transpon below this depth mAes it difficult.
Macropore density and continuity is a factor of past cropping and tillage history and sampling a
representative area for macropore distribution to determine flow and downward cesium
movement under this erperimental method is impractical.
Other potential losses of cesium in the m a s balance other than by macropore tlou-
include: 1 ) overestimation of cesium input in rainfall. 2) underestimation of cesium in runot'f
frorn sampling error. 3 ) underestimation of cesium in runo fi from storage losses. 4)
underestimation of cesium in runo ff from analytical procedure. and 5 ) underestimation of cesium
in runoff from equipment and material losses. To consider the first possibility. baseline cesium
application measured from containers within the spray pattern adjacent to the plot border would
have to receivc consistrntly highsr concentrations than the concentration received on the plot
surface. The close proximity of sampling to the plot and consistent concentrations over al1
simulations makes this possibility unlikely. The second possibility involved sampling of ninotT
into representative 100ml vials donr each minute immediately after thorough mixing of the entire
runoff interval volume. The coarsest sediment may have dropped out of suspension before
sampling: however. this would not significantly alter crsium concentrations predominantly in the
water portion. An? error associated with storage is unlikely since al1 samples were kept in
Page 141
polyethylrne containers under refngeration CO avoid altemng cesiurn concentrations due to
surtace adsorption or evaporation.
The steps taken in sample preparation and analytical procedures involved the sediment
sarnples undergoing a strong acid digestion designed to release organics and dissolve particlç
structure without evolution otàny trace metal. Ail digested samples were filtered upon
completing the digestion. Samplrs that included the 0.45pm filter papa within the digestion
consistently le% a small residue that required filtering betore using the atomic absorption
spectrorneter. Cesium however would not be associated with this material afier digestion and
cesium being in suspension would likely quickly pass through the filter openings.
The use of the atornic absorption spectrometer to analyze al1 sample concentrations was
chosen because of its rigorous and reliable analysis. The range of sample concentrations
corresponded to the optimum operating range of the instrument. The possibility of cesiurn
detection bring overestimated by the presence of K in sample matrices has been reponed:
however. with the addition of KNO, as an ionization buffer in the preparatory step. the rffçct of
possible rnhancrd or suppressrd signal detection was negated by adequately buffering al1
standards. blanks and samplrs. Ccsiurn detection using emission flarne spectrometry has besn
reported to be supprcssed afier the use of HCIO, in soi1 sample digestion methods (Otmba and
Kalacek. 1993) but not reponed for atomic absorption spectrometry.
With the findings indicating significant cesium movemenr off the plots in the water
soluble fraction and limited preferential adsorption to clay soi1 and organic niaterial. losses to a n ~
cation exchangeable surfaces or unclean materials in the handling and sampling steps usrd arc
even lrss likely. Cesium rneasurement losses fiom sampling techniques. sample contamination.
Page 142
analyticai procedures or rlsewhere in the system were considered to be minimal.
The expenmental method developed provides a successful procedure to evaluate the
potential for mowment of cesiurn from soil surfaces in runoff and associated sediment. The
extent of cesium translocation \vas found to be dependrnt on the runoff7rainhll ratio instead of a
sedimentirunot'f relationship. Localized storms wvithin a region wvould therefore determine the
uniformity of deposition. The potential may be increased for cesiurn depositrd in extendcd
storms of reduced intensity. or shorter high intensity events. to be redistributed in overland tlow
without the accompanimrnt of high soil transport. Hiph adsorption to surface particles kvas
evidently not immediate as cesium may have only brkf contact with pnmarily soil aggregates of
low surface charge. Macropore tlow may movs cesium to depth since the process occurs with
little soil contact. Rrsults in the experiment however indicated that within 48 hours of deposition.
cesium in contact with the soil surface did not become readiIy mobile in uater runoff. This tirne
period ma? have allowed surface absorption properties to inhibit subsequent release. The
potential for redistribution in subsequent rvents therefore becomes more dependent on the
potential for soi1 redistribution. Incorporation of the soil surface ends an. preferential loss of
cesium downslope.
Cssium mass found in any of the ninoff components raise concerns for emsion estimation
using the "'Cs reference analysis technique. Transport of "'CS downslope at the time of
deposition primarily in water runoff and to a lesser extent in sediment. or dunng subsequent
rainfall events pnor to mising through the Ap as is simulated in this experiment. would result in
serious oversstimates of soi1 loss. The experimental method used for cesium runoff
rneasurement indicates a correction for cesium reference levels is warranted in cuitivatsd fields
Page 143
when compared to covered. undisturbed reference areas. Tle lrvel of significance due to position
in the landscape was not made clear in the cxperiment: however. the importance of realizing the
potential impact that variable soil surface conditions have on potential runofiand therefore
cesiurn transport downslope is clearly apparent.
For field hillslopes which experienced runoff during '"CS drposition. an overestimation
of soil erosion losses using the "'Cs reference method is likely. The "7Cs reference analysis
technique based on the assumption of uniform distribution over the landscape at the timr of
deposition is questionable for areas that wrre prone at the sarne time to soi1 erosion runott:
Atternpting to account for conditions at the time of deposition to determine potential runoff relies
on accurate historical information.
Application of the "'Cs reference method to the study of soi1 erosion has increased
linearly over the last 30 years (Ritchie and Ritchie. 1997). In 1993. the authors reported a
maximum mnual number of approximately 1 30 publications relating the "'Cs technique to the
measure of soi1 rrosion and sediment deposition on the landscape. Without accounting for the
potential redistribution of "'Cs at the timr of deposition. the possibility of soil loss
overestimations in a large number of tliese publications poses a serious question of their validit)..
A discussion of the rolr of severe storms in soil erosion \vas recently publishrd to
highlight the possible weaknesses in design of conservation managements systems using GSLE
based calculations (Larson et al.. 1 997). Dunng a 10 year penod from 1962- 197 1. approximately
60% of the soil loss by erosion occurred during the 1961 cropping season. This example of
severe runoff losses during the predominant "'CS deposition period emphasizes the considerable
potential for overestimation of "'Cs and thus soil movement when compared to reference levels
Page 144
of "'CS found presently.
In conclusion. the results of this study do not support the use of "'CS for accurately
estimating soil erosion using the reference method in the hurnid upland regions of southem
Ontario. The "'Cs technique for estimating soil erosion is most accurately applied by the
cornparison method using two sample dates. In this way the accuncy of the "'CS measurement
may be applied for deterrnining actual values of soil redistribution between two time periods.
Future "'Cs analysis for erosion estimation should consider the two sample mrthod first. If rhis is
not possible. the "'Cs refirencr method of aialysis shouid be applied with caution and only afirr
careful consideration and knowledge of site conditions at the time of "'CS deposition.
Page 145
References
American Public Health Association. 1992. Standard methods for the examination of water and
wastewater, 18th edition. Arnencan Public Health Association. Washington. DC.
Atmospheric Environmeni Service. 1985. Canadian climate normals: Temperature and
precipitation 195 1 - 1980. Ontario. Suppiy Sçnr. Canada. Ottawa, ON.
Atmosphenc Environment Semice. 1993. Climatological station report - Guelph Arboreturn
1987- 1993. Environment Canada. University of Guelph, Guelph. ON.
Bernard. C. and M.R. Laverdiere. 1992. Spatial redistribution of Cs-137 and soil erosion on
Orleans Island. Quebec. Cm. J. Soil Sci. 72:j-U-554.
Boothe. A.H.. E.R. Samuels and V.K. Mohindra. 1965. Fallout levels in Canada during 1964.
Radiation Protection Division. Ann. Rep. for 1964. Dept. of National Health and Welfare.
Ottawa, ON.
Bremer. E.. E. de Jong and H.H. Janzen. 1995. Difficulties in using "'Cs to masure erosion in
stubble-mulched soi!. Can. J. Soil Sci. 75557-359.
Brewster. G.R. and V. Pilla.. 1991. Using Cs-1 37 to assess soil erosion in the Cobequid Bay area
of Nova Scotia. Can. J. Soil Sci. (Abstr. ) 7 1 :26 1-262.
Brow-n. R.B.. N.H. Cutshall and G.F. Kling. 198 la. Agricultural erosion indicated by 137 Cs
redistribution: 1. Levels and distribution of '37Cs activity in soils. Soil SciSoc. Am. J.
45: 1 184- 1 190.
Brown. R.B.. G.F. Kling. and N.H. Cutshall. 198 1 b. Agricultural erosion indicated by "7Cs
redistribution: II. Estimates of erosion rates. Soil Sci. Soc. Am. J. 45: 1 19 1 - 1 197.
Page 146
Butler. G.C. 1980. Radioactivity in the Canadian environment. NRCC No. 18 134. Dept. Energy.
Mines and Natural Resources. Ottawa Ont.
Carnbny. R.S.. E.M.R. Fisher. K. PIa?;ford. J.D. Eakins and D.H. Peirson. 198 1. Radioactive
fallout in air and min: results to end of 1980. United Kingdom Atomic Energy Authorih.
Publ. AERE-R10088. Hanvell. Oxon.. England.
Campbell- B.L. 1983. Applications of environmental caesium- 137 for the determination of
sedimentation rates in reservoin and lakes and related catchment studies in developing
countries. In: Radioisotopes in sediment studirs. 7-30. International -4tomic Energy
Tschnical Document no. 298.
Cao. Y.Z.. D.R.Coote. M C . Nolin and C. Wang. 1993. Usinp "'CS to investigate net soil rrosion
at two soil benchmark sites in Quebec. Can. J. Soi1 Sci. 735 15-526.
Chapman. J.F. and L .S. Dale. 1 986. Improved cesium sensitivity in electrothermal atomic
absorption spectrometry. Analyt. Chim. Acta. 187307-3 1 1.
Christian. G.D. and F.J. Feldman. 1970. -4tomic absorption spectroscopy: applications in
agriculture. biolog. and medicine. Wiley-Intsrscience. New York. NY.
Dahiman. R.C.. C.W. Francis and T. Tamura. 1 975. Radiocesium cpcling in vegetation and soil.
In: F.G. Howell et al. (ed. ) Mineral cycling in southeastem ecosystems. ERDA S' mp.
Ser. CONF-7405 13. US ERDA. Washington. DC.
Davis. J.J. 1963. Cesiurn and its relationship to potassium in ecology. Irz V. Schultz and A.W.
Klement. Jr. (ed.) Radioecology. Reinhold. New York. p. 539-556.
de Jong. E.. C.B.M. Begg and R.G. Kachanoski. 1983. Estimates of soil erosion and deposition
for some Saskatchewan soils. Can. J. Soi1 Sci. 63607-6 17.
Page 147
de Jong. E.. H. Villar and J.R. Bettany. 1982. Preliminary investigations on the use of I3'Cs to
estirnate srosion in Saskatchewan. Cm. J. Soil Sci. 62:673-683.
Denholm. K.X. and L.W. Schut. 1993. Field manual for describing soils in Ontario. Ontario
Centre for Soil Resource Evaluation. Guelph. ON. 62 pp.
Department of Supply and Services. 1 978. The hydrologie atlas of Canada. Depaxtment of
Energ. Mines and Resources. Ottawa ON.
Diiwu. J.Y.. R.P. Rudra W.T. Dickinson and G.J. Wall. 1998. AnaIysis of flou. through
macroporous soil by a modified hydrograph separation technique. C m . J. -4gr. Eng. (In
revieu-).
Ecological stratitication working group. 1995. A national ecological framework for Canada.
Agricul turr and Agi-Food Canada Research Branch. CL B RR and Environment Canada.
State of the Environment Directorate. Ecozone Analysis Branch. Ottawa/Hull.
Edwards. U34.. M.J. Shipitalo. W..\. Dick and L.B. Owens. 1992. Rainfall intrnsity affects
transport of watrr and chemicals through macropores in no-till soil. Soil Sci. Soc. .Am. J.
56:52-58.
Euman. L.D. and N.R. Kevem. 1975. Crsium- 137 and stable cesium in a hypereutrophic lake.
Health Ph>x 28549-555.
Flanagan. D.C. and S.J. Livingston. 1995. WEPP user summary. NSERL Research Report No.
1 1. USDA-ARS. National Soil Erosion Research Laboratory. West Lafayette. IN.
Garcia-Oliva. F.. R. Martinez Lugo and J.M. Maass. 1995. Long-term net soil erosion as
drtermined by "Cs redistribution in an undisturbed and perturbed tropical deciduous
forest ecosystem. Geoderma 68: 13 5- 147.
Page 148
Gersper. P.L. 1970. Effect of Amencan beech trees on the gamma radioactivity of soils. Soil Sci.
Soc. Am. J. 3 4 3 18-333.
Goguel. R. 1985. Hydrothermal extraction of potassium. sodium. rubidium and crsium from
rocks by lithium hydroxide and determination at very low natural levels. Anal-t. Chim.
Acta. 169: 1 79- 1 93.
Grobenski. 2.. D. Weber. B. WeIz and J. Wolff. 1983. Determination of caesiurn and rubidium
by Ilarne and fmace atomic-absorption spectrometry. Analyst- 108:925-932.
Hutchinson-Benson. E.. J. Svoboda and H.W. Taylor. 1985. The latitudinal inventor). of '"CS in
vegetation and topsoil in northern Canada 1980. Cm. J. Bot. 63:784-791.
Kachanoski. R.G. 1 987. Cornparison of measured soil 1 3 7-cesium losses and erosion rates. Cm.
J. Soil Sci. 67: 199-203.
Kachanoski. R.G. and E. de Jong. 1984. Predicting the temporal relationship betwern soil
cesium- 137 and erosion rate. J. Entiron. Qual. 1 3:30 1-304.
Kachanoski. R.G. and ..\.P. von Benoldi. 1996. Monitoring soi! loss and redistribution using
"CS. Green Plan Report. Agriculture and Agi-Food Canada. Guelph. ON. 19 pp.
Kiss. J.J.. E. de Jong and L. W. Martz. 1988. The distribution of fallout Cesium- 137 in southern
Saskatchewan, Canada. J. Environ. Qual. 1 7:U5-452.
Lance. J-C.. S.C. McIntye. J.W. Naney and S.S. Rousseva. 1986. Measuring sediment
movement at low erosion rates using cesium-137. Soil Sci. Soc. Am. J. 50: 1303- 1309.
Larson. W.E.. M.J. Lindstrorn and T.E. Schumacher. 1997. The role of severe storms in soil
erosion: A problem needing consideration. J. Soil Water Cons. 5290-95.
Lobb. D.A.. R.G. Kachanoski and M.H. Miller. 1995. Tillage translocation and tillage erosion on
Page 149
shoulder slope landscape positions measured uçing l37Cs as a tracer. Cm. J. Soil Sci. 75:
21 1-218.
Lobinski. R. and Z. Marczenko. 1996. Spectrochemical trace anaiysis for metals and metalloids.
In: S.G. Weber (ed.). Wilson and Wilson's comprehensive analytical chemistry. Elsevier
Science B.V. Amsterdam, Netherlands-
Longmore. M.E.. B.M. O'Lsary. C.W.Rose and AL . Chandia. 1983. Mapping soil erosion and
accumulation with the fallout isotope caesium-137. Aust. J. Soil Res. 21 373-38s'.
M a r t ~ L. W. and E. de Jong. 1987. Using cesiurn-137 to assess the variability of net soil erosion
and its association rvith topography in a Canadian prairie landscape. Catena 14:43945 1.
Maule. C.P. and M.J. Dudas. 1989. Preliminary identification of soil separates associated with
fallour '"Cs. Cm. J. Soil Sci, 69: l7I -1 75.
McHenrq.. J.R. and J.C. Ritchie. 1977. Estimating field erosion losses from tàllout cesium- 137
rneasurernents. IAHS-AISH Publ. 122. I.4HS Press. Wallingford. U.K.. p. 26-53.
McKeague. J.X. 1978. Manual on soil sarnpling and methods of analysis. 2nd ed. Cm. Soi1 Sci.
Soc.
-Montgomery. J.A. . -4.J. Busacca. B.E. Frazier and D.K. McCool. 1997. Evaluating soil
movsment using cesiurn-137 and the Revised Universal Soil Loss Equation. Soil Sci.
Soc. Am. J. 6 1 ( 2 ) :Y 1-579.
Nolin. MC.. Y.Z. Cao. D.R. Coote and C. Wang. 1993. Short-range variability of fallout "'CS in
an uneroded forest soil. Can. J. Soil Sci. 73:38 1-385.
Ontario Department of Agriculture and Food. 1967. Agncultural statistics for Ontario. Publn 20.
Statistics Section. F m Economics. Co-operatives and Statistics Branch. Ontario
Page 150
Department of Agriculture and Food. Toronto. ON. 1 1 1 pp.
Ontario Geological Sun-ey. 1990. The analysis of geological matenals. volume LI: a manuai of
methods: Ontario Geological Survey. Misc. Paper 119.
O m b a V. and J. KaIacek. 1993. Determination of lithium, rubidium and caesium traces in soil
and plants by cmission flarne spectrometn. Chemicke-Lisp. 87:64-70.
Pemock. D.J. 1990. Redistribution of cesium-137 as an example of a chemical indicator of
environmental stress. Env. Mon. Asse~s. 15265-27 1.
P e ~ o c k . D.J. and E. de Jong. 1990. Spatial pattern of soi1 redistribution in Boroll iandscapes.
southern Saskatchewan. Canada. Soil Sci. 150(6):867-873.
Quine. T.A. and D.E. Walling. 199 1. Rates of soil erosion on arable fields in Britain: quantitative
data from caesium- 1 3 7 mrasurements. Soil Use Mgt. 7(4): 169- 1 76.
Ritchie. J-C.. E.E.C. Clebsch and W.K. Rudolph. 1970. Distribution of fallout and natural
gamma radionuclide in litter. humus. and surface mineral soils under natural vegetation in
the Great Smoky Mountains. Nonh Carolina-Tennessee. Health P hys. 1 8 A7949 1 .
Ritchie J.C. and J.R. McHrnry. 1973. Vertical distribution of fallout cesium-137 in culti\.atrd
soils. Radiat. Data Rep. 14:727-728.
Ritchie J-C. and J.R. McHenn-. 1 985. .A comparison of three rnethods of rneasuring recent rates
OF sediment accumulation. Water Resour. Res. 2 1 :99- 103.
Ritchie. J-C. and J.R. iMcHenry. 1990. .4pplication of radioactive tallout Cesium- 137 for
rneasuring soil erosion and sediment accumulation rates and patterns: A review. J.
Environ. Qual. 1 9:2 1 5-233.
Ritchie. K.. J.R. McHenry and A.C. Gill. 1972. The distribution of '"CS in the litter and upper
Page 151
10 cm of soil under different cover types in Northem Mississippi. Health Phys. 27: 198-
201.
Ritchie. J.C. and C.A. Ritchie. 1997. Bibliography of publications of '"~esiurn studies related to
erosion and sediment deposition. International Atomic Energ Agency-Tschdoc.
Ritchir. J-C.. L.4. Spnberry and J.R. McHenr';. 1974. Estimating soil erosion from the
redistribution of fallout '"CS . Soil Sci. Soc. Am. Proc. 38: 137-1 39.
Rogowski. ;\.S.. and T. Tarnura. 1 970. Environmental mobili- of cesium- 1 3 7. Radiat. Bot.
1 O : S - l j .
SAS Institute. Inc. 1998. SAS statistical package. ver. 6.3. SAS Institure. Inc.. Raleigh. NC.
Snedecor. G.W. and W.G. Cochran. 1989. Statistical methods. 8th edition. Iowa St. Uni. Press.
Ames. IO.
TosseIl. R.W.. W.T. Dickinson. R.P. Rudra and G.J. Wall. 1987. A portable rainfall simulator.
Cm. Agric. Eng. 29: 155- 162.
Van Vliet. L.J.P. and G.J. Wall. 198 1. Soil erosion losses frorn winter runoff in southem Ontario.
Cm. J. Soil Sci. 61 :45 1-454.
Varian. 1990. Flarne atomic absorption spectrometry. analytical methods. Varian Aus. Pty. Ltd.
Mulgrave. Aust.
Wall. G.J.. B.A. Grant. D.J. King and N.B. McLaughlin. 1997. The effects of livestock rnanure
application rnethods on water quality. focusing on nitrogen and bacteria transport in soil.
Agriculture and Agn-Food Canada Green Plan Report no.5. London. ON.
Walling. D.E. and S.B. Bradley- 1988. The use of caesium-137 measurements to investigate
sediment delivery fiom cultivated areas in Devon. U.K. IAHS Publ. 174:325-335.
Page 152
Weast. R.C. (ed.). 1987. Handbook of Chemistry and Physics. 1987- l988.68th cd. Chernical
Rubber Co. Boca Raton. FA.
Wise. S.M. 1980. Caesium- 137 and lead-210: A review of techniques and some applications in
peomorphology. In: R.A. Cullingford et al.. (ed). Timescalrs in geomorphology. John
Wiley and Sons. New York.
Page 154
Location: Rockwood. Erarnosa Township. Wellington Counh
S lopr position: Depressionai
Landform and parent matsrials: Nearly level ti l l plain. dorninantly loamy textures
Slope: 1% simple
Drainage: Poorly drained
Soi1 type: Parkhill silt loam
Classification: Orthic Humic Gleysol. loamy . mildly al kaiine. moderatel- calcareous
- -- - - - -
Horizon Depth Colour Texture Priman; Structure Consistence Mottles (cm)
AP 0-34 10 YR S I L L\ eak. fine. v. friable --
su bangular bloc ky
Bgl 31-16 10YR L structureless fiable 25Y7/2. 1 O
Bg2 46-66 10 YR L massive friable 15Y R7;2
Ck 66 - 10 Y R SL weak. tine. t? rm 25Y R7/2
subangular block.
- -
Horizon Depth Gravel Sand Silt Clay PH OM CaCO, ( c m ) (>?mm)% ?/O ?40 O h CaCI: O/O Equiv.
4 4 P 0-3 1 -- 26 54 30 7.0 - 7 .- 7 ; 2.30
Bg 1 34-16 - 36 19 15 7.1 0.93 5.52
B g 46-66 -- 5 l 35 1 1 7.3 0.41 3 2 0
Ck 66 + -- 57 31 9 7.5 -- 9.76
Page 155
Table M: Rockwood field site measurement of Cesium- 137 and related parameters by gridpoint.
Conservation field
A p Bulk Specific Guelph lab Field Net loss Erosion Erosion Sample Landscape depth density m a s conctn conc'n of Cs rate class
gridpoint position (cm) ( g cm-') (kg rn") (Bq kg") (Bq m") (Bq m-'1 (kg m-> r-' ) label D BD M s C g Cf Ep
DSH
DSH
DSH CFS CFS CSH CFS CFS CFS CFS CFS CFS
CSH
CFS CBS
CFS DFS DSH
CFS DSH DFS
CFS CSH
DFS DFS
DSH CFS CSH
Page 156
CFS DSH DFS CFS CSH DFS DSH DFS CSH DFS DSH DFS
Conservation Ap average 24 5 318.0 7.45 2409 190 0.780
Conventional tield
Ap Bulk SpecitÏc Guelph lab Field Net loss Erosion Erosion Sample Landscape depth den si^ m a s conc'n conc'n of Cs rafe c l a s . ,
gridpoinr position (cm) (g cm" (kg mm') (Bq kg-' 1 (Bq m") (Bq m.') (kg m.- 1 . r ' ) label D BD FLiIs CE Cf Ep
DSH
DFS DFS
DFS DSH CFS DFS
CFS
DBS
DSH
CFS
CFS
DFS
DSH
CFS CSH
Page 157
S200 E200 SI80 El00
SI60 ES00 SI40 E200 SI00 El80 SI80 El80 SI60 El80 SI40 El80
SZOO El60 SI80 El60
SI60 El60 SI40 El60 S200 El40 SI80 El40
S160 El40 SI40 El40 SZOO El20 SI80 El20 5160 €120
SI40 El20
SZOO El00 SI80 El00 Si60 El00 SI40 El00 S200 ES0 SIS0 E8O SI60 E80 SI40 ES0
S200 E60 SI80 €60 SI60 E6O SI40 E60
CSH CSH
CSH CFS DSH DSH DSH CFS DSH
CFS
DSH
CBS CSH CSH
DFS CFS CSH
DFS CSH
CFS
CSH DFS
DFS CFS
DFS CFS DSH DSH DBS
CSH DSH CFS
Conventional Ap average 29 1.32 376.1 5.13 1930 669 3.884
Page 158
Forest Sarnple Bulk Specific Guelph lab Field Adjusted
Site depth densip mass conc'n conc'n field conc'n
(cm) (gcm-'1 (kgm-'1 (Bqkg") (Bqm") ( B q m " ) D BD M s Cr Cf Cf
Crest 0- 15cm average
Depression O- l5cm average
Forest 0- 1 5cm average
Page 159
Table A 3 Variabili- of soi1 "'Cs activity (Bq m") in field Ap horizon from 4 transects downslope.
Conservation
Landscape Sample transect number position 1 - 7 3 4 Average SD CV(%i
Bq m.:
I C-U 1664 362 1 2745 4776 3202 1143 35.7
2U 1981 2066 1898 1713 1915 13 l 6.8
3 U 2067 1501 1610 1552 1682 325 13.4
4 M 1789 1547 1302 1349 1497 192 12.9
5M 3046 2777 2035 1584 2360 58 1 24.6
6 L 2664 2732 2209 1 664 2317 427 18.5
7 L 2407 2292 1990 2637 233 1 233 10.0
8 L 4183 2472 2176 3430 3065 795 3 . 9
9 D 2182 280 1 2490 3143 2654 357 13-5
10D 366 1 26 13 3097 289 1 3066 3 83 12.5
Average 2564 2 332 3 155 2474
SD 790 600 498 1019
CV (%) 30.8 24.6 23.1 42.4
C-LI = Crest to Upper dope position: U = L'pper dope position: M = Mid dope position: L = Lo~ver dope position: D = Depressional area.
Page 160
Conventional field Landscape Sample transect nurnber Position 1 - 'T 3 4 Average SD CV(%)
1 c 2C-U 3U 4U
5M 6 b1 7M 8L 9 L
I OL I I L 1 ?T
Average SD
C = Crest dope position: C-U = Crest to Upper slope position: U = Upper slope position: b1 = Mid dope position: L = Lower slope position: T = Toe siope position.
Page 161
Landscape position and nvo depth sampling method
Repl icate Depression Crest Depression* Crest*
Bq m.'
9
Average SD c v (96)
Forest average O- I Scrn 2397 0- 1 5&30crn 2599* SD 527 60 1 CV (%) 21.1 23.1
* ' '-Cs a c t i v i ~ from lower depth ( 15-3Ocm) at rhree points \vas added to corresponding upper depth (O- I5cm i and incorporated into average of nine grid points.
Page 162
Tabie B 1 : WEPP slo
Appendix B
Soi1 erosion mode1 predicrion input data
Conservation field WEPP dope input file
95.7 s if Created on 37.4~198 bu ' WSLP'. (Ver. 1 jApr95) # Author: djk ff 1 ' 7 5 80 -- 20 200 0.0.01 0.05.0.034 0.1.0.057 0.1 5.0.076 0.2.0.074 0.25.0.07 0.3.0.062 0.35.0.057 0.4.0.053 0.45.0.047 0.5.0.046 0.55.0.043 0.6.0.037 0.65.0.032 0.7.0.025 0.75.0.01 8 0.8.0.0 15 0.85.0.0 13 0.9.0 1 .O
Conventional field WEPP soi1 input file
95.7 $
# Created on 17Apr98 by ' WSLP'. (Ver. 15Ap1-95) # Author: djk !# 1 225 80 14 240 0.0.017 0.04.0.035 0.13.0.055 0.21.0.061 0.29.0.069 0.38.0.071 0.46.0.066 0.54.0.056 0.63.0.047 0.71.0.039 0.79.0.039 0.88.0.03 0.96.0.016 1.0.01
Page 163
Table B2: WEPP soil input file
Conservation tieId WEPP soi1 input tile
97.3 # # Created on 27Apr98 b'; ' WSOL'. (Ver. 15Ap1-95) # Author: djk #
wepp soi l 1 O 'Guelph' 'Ioam' 3 0.1 1 0.8 6.10896e+006 0.0073 2.7 1 5.88 340 36 16.3 4.19 10.01 0.88 550 47.2 1 . 3 1.28 3.89 6.1 750 56.2 8.5 O 0.85 30.2
Conventional tield WEPP soi1 input tile
97.3 # ff Created on 27Apr98 by ' WSOL'. (Ver. 1 jAp~-95) I+ .Ailthor: dj k # wepp soil 1 O 'Guelph' 'loam' 3 0.15 0.8 6.493 l6e+OO6 0.0079 2.44 7.33 285 43.6 13.9 3.47 8.33 2.33 320 55.4 1 . 4 1.19 3.63 6.1 750 60.5 6.9 0 0.69 20.3
Page 164
tic 1
1 'L- OLC E'8 666'666 8-L - 6'5- 66-66 66'66 66 O 09 1 0; 5-5- S l 6 Z 9'8 666'666 8'L- 93- 66'66 66b6 66 Z 09 1 6 1
CE- 06 1 1 1 666'666 1'9- W- 66'66 66'66 66 6'8 09 1 8 1 8'5- OLC 8 3 666'666 5.5- L' 1 - 66-66 66-66 66 O 09 1 L 1
5-9- OLZ 9'8 666'666 8 2 8 7 66'66 66'66 66 8'0 09 1 9 1 Sel- S 1 C 6'8 666'666 C'Z- 1' 1- 66'66 66'66 66 c'O1 09 1 Z 1
1 'C- 06 C C 666'666 8 2 1 ' 1- 66'66 66'66 66 9'L 09 1 t. 1 8'0- 06 1'9 666'666 Z- 1'1- 66'66 66'66 66 5-Z 09 1 5 1
L'E- 06 L'9 666'666 9'01- O 66'66 66'66 66 L'Z 1 09 1 SI 6'9- 06 L ' t 666'666 1 1- - 66'66 66'66 66 O 09 1 1 1 t- O6 9'5 666'666 Z'ZI- 6-C- 66'66 66'66 66 O 09 1 O1
9'6- 06 5'8 666'666 8'ZI- 8 2 66'66 66'66 66 8'0 09 1 6 t't- C'LtC 9'8 666'666 L'9- O 66'66 66'66 66 1'1 09 1 8 t't- S X 6'8 666'666 9'01- L'1 66'66 66'66 66 5-0 09 1 L
9'0 1 - 5-LtC 6'9 666'666 C I - 9'5- 66'66 66'66 66 1' 1 09 1 9 8'; 1 - OLC CL 666'666 1 * 1 1 - 8 2 66'66 66-66 66 5' 1 09 1 5
6- OLZ C l 666'666 9'5- S- 66'66 66'66 66 8'1 09 1 t 9' 1- OLZ 1'8 666'666 8 7 O 66'66 66'66 66 C' 1 09 1 C
8' 1 - O8 1 t'9 666'666 O 1- 1 1 66-66 66'66 66 L ' t l 09 1 C 9'8- 06 L ' t 666'666 8'L 1- O 66'66 66'66 66 O 09 1 1 ( 3 ) @ a a ) (s,W(p/1) ( 3 ) ( 3 ) ( y ) ( u w
.wapl np-.M p - . ~ pw U ~ J I XTUII di di mp d3~d na oui ep 0'65 O'Z9 0'1L 0'59 O'Z8 O'C8 O'IL O'CL 0'89 O'LE O'E O ' I Z
(WU) uoyei!d!xud ane .i(yluow pauasqo 1'901 6'91 1 6 ' Z Z 8'5CE O'SSP 8'LZS t ' i t S 6 ' S t t'065 S'8 1 1 S'WZ 6'5t1
( .kp/s.ial3uel) uoywpw q o s ane .$puoui pauasqo 9 2 0' 1 - O ' t 5'6 8'Z 1 9': 1 9' 1 1 6's L'O 6'5- 9'0 1 - 0' 1 1 -
(3) a~niwadwai u!ui a-ie Âlyiuoui pa,uasqo L'O- 0'9 t'tI L'OZ L ' K L'SZ C'5S S'L 1 6'01 I ' C 5';- t 'E -
(3) amlmiaduiai ir-ui aAr) L~qiuow pahiasqo 1 C 09 S(: t55 00'0% 00'Et
paiqnuqs snaA mai àu!uu!%~ smaA .sqo ( w ) uoymsl3 apni!a'uol spnirie? O ' t NOISN3A N39113 ONVLNO '3VO :u*!lElS
0 0 1 c
Page 171
z e a = - - 'P. T Y t r , r~ IC. - r? t ? % v ! 3 + y y * -
n - Z ~ C I r - v i s - - Ir. m rci ;7 Cl ci 3C
Page 174
IMAGE EVALUATION TEST TARGET (QA-3)
O 1993. A&md Image. lm.. AU Rqh(s Reserved