Modeling Phosphate Adsorption for South Carolina Soils

Clemson UniversityTigerPrints

All Theses Theses

5-2010

Modeling Phosphate Adsorption for SouthCarolina SoilsJesse CannonClemson University jessewittcannongmailcom

Follow this and additional works at httpstigerprintsclemsoneduall_theses

Part of the Environmental Engineering Commons

This Thesis is brought to you for free and open access by the Theses at TigerPrints It has been accepted for inclusion in All Theses by an authorizedadministrator of TigerPrints For more information please contact kokeefeclemsonedu

Recommended CitationCannon Jesse Modeling Phosphate Adsorption for South Carolina Soils (2010) All Theses 829httpstigerprintsclemsoneduall_theses829

MODELING PHOSPHATE ADSORPTION

FOR SOUTH CAROLINA SOILS

A Thesis Presented to

the Graduate School of Clemson University

In Partial Fulfillment of the Requirements for the Degree

Master of Science Environmental Engineering and Science

by Jesse Witt Cannon

May 2010

Accepted by Dr Mark A Schlautman

Dr John C Hayes Dr Fred J Molz III

ABSTRACT

Eroded sediment and the pollutants it transports are problems in water bodies in

South Carolina (SC) and the United States as a whole Current regulations and engineering

practice attempt to remedy this problem by trapping sediment according to settling velocity

and thus particle size However relatively little is known about most eroded soils In

most cases little experimental data are available to describe a soilrsquos ability to adsorb a

pollutant of interest More-effective design tools are necessary if design engineers and

regulators are to be successful in reducing the amount of sediment and sediment-bound

pollutants in water bodies This study will attempt to develop such a tool for phosphate

adsorption since phosphate is the dominant form of phosphorus found in the environment

Eroded particle size distributions have been developed by previous researchers for

thirty-four soils from across South Carolina (Price 1994) Soil characterizations relating

to phosphate adsorption were conducted for these soils including phosphate adsorption

isotherms These isotherms were developed in the current study using the Langmuir

isotherm equation which fits adsorption data using parameters Qmax and kl Three different

approaches for determining previously-adsorbed phosphate (Q0) were evaluated and used

to create Langmuir isotherms One approach involved a least squares linear regression

among the lowest aqueous phosphate concentrations as endorsed by the Southern

Cooperative Series (Graetz and Nair 2009) The other two approaches involved direct

fitting of a superposition term for Q0 using the least squares nonlinear regression tool in

Microcal Origin and user-defined functions for the one- and two-surface Langmuir

isotherms

Isotherm parameters developed for the modified one-surface Langmuir were

compared geographically and correlated with soil properties in order to provide a

predictive model of phosphate adsorption These properties include specific surface area

(SSA) iron content and aluminum content as well as properties which were already

available in the literature such as clay content and properties that were accessible at

relatively low cost such as organic matter content and standard soil tests Alternate

adsorption normalizations demonstrated that across most of SC surface area-related

measurements SSA and clay content were the most important factors driving phosphate

adsorption Geographic groupings of adsorption data and isotherm parameters were also

evaluated for predictive power

Langmuir parameter Qmax was strongly related (p lt 005) to SSA clay content

organic matter (OM) content and dithionite-citrate-bicarbonate extracted iron (FeDCB)

Multilinear regressions involving SSA and either OM or FeDCB provided the strongest

estimates of Qmax (R2adj = 087) for the soils analyzed in this study An equation involving

the clay-OM product is suggested for use (R2adj = 080) as both clay and OM analysis are

economical and readily-available

Langmuir parameter kl was not strongly related to soil characteristics other than

clay although inclusion of OM and FeDCB (p lt 010) improved fit (R2adj = 024-025) An

estimate of FeDCB (p lt 010) based on OM and carbon (Cb) content also improved fit (R2adj

= 023) an equation involving clay and estimated FeDCB is recommended as clay OM and

Cb analyses are economical and readily-available Also as kl was not normally distributed

descriptive statistics for topsoil and subsoil kl were developed The arithmetic mean of kl

for topsoils was 033 and the trimmed mean of kl for subsoils was 091 These estimates of

kl were nearly as strong as for the regression equation so they may be used in the absence

of site-specific soil characterization data

Geographic groupings of adsorption data and isotherm parameters did not provide

particularly strong estimates of site-specific phosphate adsorption Due to subsoil

enrichment of Fe and clay caused by leaching through the soil column geography-based

estimates must differentiate between top- and subsoils Even so they are not

recommended over estimates based on site-specific soil characterization data

Standard soil test data developed using the Mehlich-1 procedure were not related to

phosphate adsorption Also soil texture data from the literature were compared to

site-specific data as determined by sieve and hydrometer analysis Literature values were

not strongly related to site-specific data use of these values should be avoided

DEDICATION I am blessed to come from a large strong family who possess a sense of wonder for

Godrsquos Creation a commitment to stewardship a love of learning and an interest in

virtually everything I dedicate this thesis to them They have encouraged and supported

me through their constant love and the example of their lives In this a thesis on soils of

South Carolina it might be said of them as Ben Robertson said of his father in the

dedication of Red Hills and Cotton (1942) they are the salt of the Southern earth

I To my father Frank Cannon through whom I learned of vocation and calling

II To my mother Penny Cannon a model of faith hope and love

III To my sister Blake Rogers for her constant support and for making me laugh

IV To my late grandfather W Bruce Ezell for setting the bar high

To my late grandmother Floride Ezell who demonstrated that itrsquos never too late for

God to use you and restore your life

VI To Elizabeth the love of my life

To special members of my extended family To John Drummond for helping me

maintain an interest in the outdoors and for his confidence in me and to Susan

Jackson and Jay Hudson for their encouragement and interest in me as an employee

and as a person

Finally I dedicate this work to the glory of God who sustained my life forgave my

sin healed my disease and renewed my strength Soli Deo Gloria

ACKNOWLEDGMENTS I have many people to thank and acknowledge for their support in completing this

project First and foremost I acknowledge my thesis co-advisors Drs MA Schlautman

and JC Hayes and committee member Dr FJ Molz III for their expertise guidance

encouragement and patience I am deeply grateful to all of them but especially to Dr

Schlautman for giving me the opportunity both to start and to finish this project through

lab difficulties illness and recovery I would also like to thank the Department of

Environmental Engineering and Earth Sciences (EEES) at Clemson University for

providing me the opportunity to pursue my Master of Science degree I appreciate the

facultyrsquos dedication in teaching and research and the staffrsquos support and encouragement I

also thank and acknowledge the Natural Resource Conservation Service for funding my

research through the Changing Land Use and the Environment (CLUE) project

I acknowledge James Price and JP Johns who collected the soils used in this work

and performed many textural analyses cited here in previous theses I would also like to

thank Jan Young for her assistance as I completed this project from a distance Kathy

Moore of the Clemson Agricultural Service Laboratory for her help and expertise Dr

Charles Privette for helping me to conduct pipette analyses Yanping Guo for conducting

the specific surface area analysis and Dr Tanju Karanfil for allowing her to do so

Finally I am grateful for Drs Hayslip Ploch and Sutton of Moncks Corner and

North Charleston SC for their care and attention during my diagnosis illness treatment

and recovery I am keenly aware that without them this study would not have been

completed

Table of Contents (Continued)

TABLE OF CONTENTS

TITLE PAGE i ABSTRACT ii DEDICATIONv ACKNOWLEDGMENTS vi LIST OF TABLES ix LIST OF FIGURES xi LIST OF SYMBOLS AND ABBREVIATIONS xiii CHAPTER

1 INTRODUCTION 1

2 LITERATURE REVIEW PROCESSES AND FACTORS CONTROLLING PHOSPHATE ADSORPTION AND RELEASE 4

3 OBJECTIVES19 4 MATERIALS AND METHODS20 5 RESULTS AND DISCUSSION ISOTHERM DEVELOPMENT29 6 RESULTS AND DISCUSSION SOIL CHARACTERIZATION 40 7 RESULTS AND DISCUSSION PREDICTING ISOTHEM

PARAMETERS 54

8 CONCLUSIONS AND RECOMMENDATIONS 93 9 RECOMMENDATIONS FOR FURTHER RESEARCH99

APPENDICES 101 A Isotherm Data102 B Soil Characterization Data 148 C Additional Scatter Plots 163 D Sediments and Eroded Particle Size Distributions198 E BMP Study226 REFERENCES 241

LIST OF TABLES

Table Page 5-1 Comparison of Q0 Estimates for a Subset of Soils37

5-2 Regression Statistics Relating Methods for Estimating Q0 to One Another38

6-1 Coefficients of Determination (R2) Between Hydrometer Data (Price 1994) and Soil Texture Data from the Literature43 6-2 Regression Statistics Relating Specific Surface Area (SSA) to Soil Texture and

Organic Matter (OM) 46 6-3 Regression Statistics Relating Organic Matter (OM) to Other Soil

Characteristics48 6-4 Regression Statistics Comparing Mehlich-1 to DCB-Released Analyses of Iron

and Aluminum Content49 6-5 Relationship of PICP to PIC 51

6-6 Regression Statistics Relating PO4DCB PO4 Me-1 and PO4 H2O Desorbed52

7-1 Subsoil Enrichment Ratios for Soil Characteristics and Isotherm Parameters55

7-2 Regression Statistics Between Subsoil Enrichment Ratios 56

7-3 Isotherm Parameters and Error Associated with Location Groupings of Soils59

7-4 Isotherm Parameters and Error Associated with Physiographic Groupings of Soils59 7-5 Isotherm Parameters and Error Associated with MLRA Groupings of Soils60

7-6 Isotherm Parameters and Error Associated with Level III Ecoregion Groupings of Soils 60 7-7 Isotherm Parameters and Error Associated with Level IV Ecoregion Groupings

of Soils 61

List of Tables (Continued)

Table Page 7-8 Relationship of Phosphate Adsorption Capacity (Qmax) to Other Soil

Characteristics65 7-9 Relationship of the Affinity Constant (kl) to Other Soil Characteristics73

7-10 kl Regression Statistics All Topsoils 80

7-11 Regression Statistics Low kl Topsoils 80

7-12 Regression Statistics High kl Topsoils 81

7-13 kl Regression Statistics Subsoils81

7-14 Descriptive Statistics for kl 82

7-15 Comparison of Predicted Values for kl84

7-16 Relationship of the Previously-Adsorbed Phosphate (Q0) Found By the 3-Parameter Isotherm Technique to Other Measures of Soil Phosphorus85 7-17 Qmax Variation Based on Location and Alternate Normalizations89

7-18 kl Variation Based on Location 90

7-19 Qmax Regression Based on Location and Alternate Normalizations91

7-20 kl Regression Based on Location and Alternate Normalizations 92

8-1 Study Detection Limits and Data Range 97

LIST OF FIGURES

Figure Page

1-1 Contribution of Nutrients to Surface Water Impairments in the US 1

4-1 Locations of Clemson University Experiment Station and Research and Education Centers 21

5-1 Sample Plot of Raw Isotherm Data 29

5-2 Using the SCS Method to Determine Previously-Adsorbed Phosphate31

5-3 Isotherm Developed Using the SCS Method to Determine Previously-Adsorbed Phosphate (Q0) 32 5-4 3-Parameter Fit 33

5-5 Comparison of 1- and 2-Surface Isotherms Developed for a Vaucluse Topsoil Using the SCS Method 34 5-6 Comparison of Methods for Estimating Q0 Pelion Topsoil35

5-7 Comparison of Methods for Estimating Q0 Dothan Topsoil 36

5-8 Comparison of Methods for Estimating Q0 Vaucluse Topsoil 36

5-9 Regression Plot Comparing 5-Parameter Method to SCS Method Estimates for Q038 6-1 Clay Content vs Specific Surface Area 45

6-2 Predicted Specific Surface Area Using Clay and Organic Matter Content vs Measured Specific Surface Area45 6-3 Predicted OM Using FeDCB Content and Cb vs Measured OM47

6-4 ICP- Measured P vs IC-Measured PO4 With 11 Line 51

7-1 Coverage Area of Sampled Soils54

7-2 Predicted Qmax Using SSA and FeDCB Content vs Measured Qmax 64

List of Figures (Continued)

Figure Page

7-3 Dot Plot of Measured Qmax 68

7-4 Histogram of Measured Qmax68

7-5 Dot Plot of Measured Qmax Normalized by SSA and FeDCB69

7-6 Histogram of Measured Qmax Normalized by SSA and FeDCB 69

7-7 Dot Plot of Qmax as Predicted by Regression Equation for SSA and FeDCB 70

7-8 Histogram of Qmax as Predicted by Regression Equation for SSA and FeDCB70

7-9 Dot Plot of Measured Qmax Normalized by Clay 71

7-10 Histogram of Measured Qmax Normalized by Clay 71

7-11 Dot Plot of Measured Qmax Normalized by Clay and OM72

7-12 Histogram of Measured Qmax Normalized by Clay and OM 72

7-13 Predicted kl Using Clay Content vs Measured kl75

7-14 Predicted kl Using Clay Content and FeDCB vs Measured kl 76

7-15 Dot Plot of Measured kl For All Soils 77

7-16 Histogram of Measured kl For All Soils77

7-17 Dot Plot of Measured kl For Topsoils78

7-18 Histogram of Measured kl For Topsoils 78

7-19 Dot Plot of Measured kl for Subsoils 79

7-20 Histogram of Measured kl for Subsoils 79

8-1 Predicted Qmax Based on the Clay-OM Product vs Measured Qmax 96

8-2 Predicted kl Based on Clay and Estimated FeDCB vs Measured kl 96

LIST OF SYMBOLS AND ABBREVIATIONS

Greek Symbols

α Proportion of Phosphate Present as HPO4-2

γ Activity Coefficient of HPO4-2 Ions in Solution

π Pi ρ Density Ψa Adsorption Plane Surface-Specific Electrostatic Potential X2 Chi-Squared

Abbreviations

3-Param 3-Parameter Method 5-Param 5-Parameter Method Al Aluminum BET Brunauer Emmet and Teller BMP(s) Best Management Practice(s) c Total Concentration of Phosphate in Solution C Aqueous Phosphate Cb Carbon CaCO3 Calcium Carbonate CU ASL Clemson University Agricultural Service Laboratory D Diameter DCB Dithionite-Citrate-Bicarbonate EPC0 Equilibrium Phosphate Concentration ES Experiment Station Fe Iron Fcl Fraction of Eroded Primary Clay Particles Flg Fraction of Eroded Large Aggregate Particles Fsa Fraction of Eroded Primary Sand Particles Fsg Fraction of Eroded Small Aggregate Particles Fsi Fraction of Eroded Primary Silt Particles H+ Proton IC Ion Chromatography ICP-AES Inductively Coupled Plasma Atomic Emission Spectroscopy IDEAL Integrated Design and Evaluation Assessment of Loadings IQR Interquartile Range Ki Ion- and Surface-Specific Binding Constant kl Affinity Constant

List of Symbols and Abbreviations (Continued)

Abbreviations (Continued)

LG Large Aggregate m Median macute Trimmed Median M Metal Ms Mass Me-1 Mehlich-1 MLRA Major Land Resource Areas n Number of Measurements NRCS Natural Resources Conservation Service NTU(s) Nephelometeric Turbidity Unit(s) Ocl Fraction of Dispersed Clay in Parent Soil Osa Fraction of Dispersed Sand in Parent Soil Osi Fraction of Dispersed Silt in Parent Soil OH Hydroxide OM Organic Matter p Number of Fitting Parameters p-value Probability relating to statistical significance P Phosphorus PO4 Phosphate PO4 H2ODesorbed Amount of Phosphate Desorbed in Water Q0 Previously-Adsorbed PhosphatePhosphorus Qmax Maximum Phosphate Adsorption Capacity R2 Coefficient of Determination REC(s) Research and Education Center(s) SA Surface Area SD Standard Deviation SE Standard Error SCLRC South Carolina Land Resources Commission SCS Southern Cooperative Series SG Small Aggregate SSA Specific Surface Area SU Standard Unit pH TSS Total Suspended Solids US United States US EPA Environmental Protection Agency WSP Water Soluble Phosphorus xg Geometric Mean xm Arithmetic Mean xmacute Trimmed Arithmetic Mean zi Valence State

CHAPTER 1

INTRODUCTION

Nutrient-based pollution is pervasive in the United States consistently ranking

among the highest contributors to surface water quality impairment (Figure 1-1) according

to the Environmental Protection Agency (US EPA 2002 2007 2009) Phosphorus (P) is

one such nutrient In the natural environment it is a nutrient which primarily occurs in the

form of phosphate (PO43-) Soil P whether occurring naturally or enriched by man tends

to strongly bind to surfaces in soil When this soil is mobilized by erosion it serves as the

vehicle by which P is transported to surface waters as a form of non-point source pollution

Therefore total P and total suspended solids (TSS) concentration are often strongly

correlated with one another (Reid 2008) In fact upland erosion of soil is the

2000 2002 2004

Lakes and Ponds Rivers and Streams

Figure 1-1 Contribution of Nutrients to Surface Water Impairments in the US (US EPA 2002 2007 2009)1

1 Decrease with time due to addition of parameters for which water bodies can be impaired and removal from list due to TMDL implementation

primary route by which P finds its way into the aquatic environment (Bennett et al 2001)

Weld et al (2002) concurred reporting that non-point sources such as agriculture

construction projects lawns and other stormwater drainages contribute 84 percent of P to

surface waters in the United States mostly as a result of eroded P-laden soil

The nutrient enrichment that results from P transport to surface waters can lead to

abnormally productive waters a condition known as eutrophication As a result of

increased biological productivity eutrophic waters experience abnormally low levels of

dissolved oxygen and a resulting decrease in biodiversity Eutrophication also interferes

with human recreational pursuits (Carpenter et al 1998) and can be an expensive burden

on local economies that depend on tourism Damages resulting from eutrophication have

been estimated to range from 22 to 70 billion dollars annually due to agriculture alone

(Lovejoy et al 1997)

As the primary limiting nutrient in most freshwater lakes and surface waters P is an

important contributor to eutrophication in the United States (Schindler 1977) Only 001

to 003 mg L-1 is necessary for P to contribute to eutrophication (McDowell and Sharpley

2001) The US EPA has established a eutrophication-based water quality goal of 01 mgP

L-1 for surface waters in the US Based on this goal more than one-half of sampled US

streams exceed the P concentration required for eutrophication according to the United

States Geologic Survey (USGS 1999) With this in mind attempts to limit P discharges

into receiving water bodies are very important Doing so requires an understanding of the

factors affecting P transport and adsorption

P (primarily PO4) adsorption is impacted by soil mineralogy and morphology but

generally is controlled by the presence of iron (Fe) and aluminum (Al) oxides Soil history

including land use and fertilization also plays a role as does soil pH surface coatings

organic matter and particle size While PO4 is considered to be adsorbed by both fast

reversible reactions and slow ldquoirreversiblerdquo reactions desorption is typically assumed to

correspond only with the fast reactions Therefore complete desorption is likely to occur

after a short contact period between soil and a high concentration of PO4 in solution

(McGechan and Lewis 2002) However once in a reducing environment all PO4 attached

to iron-containing sediment is likely to be released after the particle undergoes

oxidation-reduction reactions This is the final link between PO4-enriched upland soils and

eutrophic water bodies (Hesse 1973)

This study will produce PO4 adsorption isotherms for South Carolina soils and seek

to relate Langmuir isotherm parameters to soil characteristics It is predicted that PO4

adsorption parameters will be strongly correlated with specific surface area (SSA) clay

content Fe content and Al content A positive result will provide a means for predicting

isotherm parameters using easily available data and thus allow engineers and regulators to

predict a disturbed soilrsquos PO4 -adsorbing potential perhaps using an advanced stormwater

model such as IDEAL (Integrated Design and Evaluation Assessment of Loadings GVL

CO SC 2010a) This is of interest for two reasons first is the soilrsquos ability to trap PO4 that

might otherwise escape from a developing site (so long as the soil itself is trapped) and

second is the soilrsquos ability to potentially transport PO4 throughout the watershed causing

localized episodes of high PO4 concentrations when the nutrient is released to solution

CHAPTER 2

LITERATURE REVIEW PROCESSES AND FACTORS CONTROLLING PHOSPHATE ADSORPTION AND RELEASE

Sources of Soil Phosphorus

Soils naturally contain between 100 and 3000 mgP kg-1(Frossard et al 1995) This

P occurs almost entirely as orthophosphate (PO43-) Regardless of source different types

of orthophosphate are rarely distinguished in the literature About 29 to 65 of the total

soil PO4 is present as part of a variety of organic compounds From these soils phosphorus

can be released during the weathering of primary and secondary minerals and because of

active solubilization by plants and microorganisms (Frossard et al 1995)

Humans largely impact P cycling through agriculture When P is mined and

transported for agriculture either as fertilizer or as feed upland soils are enriched This

practice has proceeded at a tremendous rate for many years so that annual excess P

accumulation has increased by 75 since preindustrial times (Bennett et al 2001) Just as

important is the human role in increased erosion By exposing large plots of land erosion

of enriched soils is accelerated In addition such activities also result in increased

weathering of primary and secondary P-containing minerals releasing P to the larger

environment

Dissolution and Precipitation

While adsorption reactions should be considered the primary link between upland P

applications and surface water eutrophication a number of other reactions also play an

important role in P mobilization Dissolution of mineral P should be considered an

important source of soil P in the natural environment Likewise chemical precipitation

(that is formation of solid precipitates at adequately high aqueous concentrations) is an

important sink However precipitates often form within soil particles as part of the

naturally present PO4 which may later be eroded and must be accounted for and

precipitates themselves can be transported by surface runoff With this in mind it is

important to remember that precipitation should rarely be considered a terminal sink

Rather it should be thought of as an additional source of complexity that must be included

when modeling the P budget of a watershed

Dissolution Reactions

In the natural environment apatite is the most common primary P mineral It can

occur as individual granules or be occluded in other minerals such as quartz (Frossard et

al 1995) It can also occur in several different chemical forms Apatite is always of the

form α10β2γ6 but the elements involved can change While calcium is the most common

element present as α sodium and magnesium can sometimes take its place Likewise PO4

is the most common component for γ but carbonate can sometimes be present instead

Finally β can be present either as a hydroxide ion or a fluoride ion

Regardless of its form without the dissolution of apatite P would rarely be present

at all in natural environments Apatite dissolution requires a source of hydrogen ions and

sinks for calcium and P In addition the rate of dissolution is dependent upon the apatite

particlersquos location within the soil (and thus its accessibility) the particlersquos morphology

and the rate at which carbonate can replace PO4 within the mineralrsquos crystal lattice

(Frossard et al 1995) Besides apatite other P-bearing minerals are also important

sources of PO4 in the natural environment in some sodium dominated soils researchers

have attributed P release to dissolution of calcium phosphates at high alkalinity and pH

(Frossard et al 1995)

Precipitation Reactions

P precipitation is controlled by the soil system in which the reaction takes place In

calcium systems P adsorbs to calcite Over time calcium phosphates form by

precipitation eventually leading to hydroxyapatite [Ca10(PO4)6(OH)2] This mineral has

the lowest solubility of the calcium phosphates so it should generally control P

concentration in calcareous soils

While calcium systems tend to produce well-crystralized minerals aluminum and

iron systems tend to produce amorphous aluminum- and iron phosphates However when

given an opportunity to react with organized aluminum (III) and iron (III) oxides

organized phases such as sterretite [(Al(OH)2)3HPO4H2PO4] tinticite

[Fe6(PO4)4(OH)6middot7H2O] or griphite [Fe3Mn2(PO4)25(OH)2] can also form Other

P-bearing minerals including those from the crandallite group wavellite and barrandite

have been identified in some soils but even when they occur these crystalline minerals are

far outweighed by amorphous aluminum- and iron-phosphates (Frossard et al 1995)

Adsorption and Desorption Reactions

Adsorption-desorption reactions serve as the primary link between P contained in

upland soils and P that makes its way into water bodies This is because eroded soil

particles are the primary vehicle that carries P into surface waters Primary factors

affecting adsorption-desorption reactions are binding sites available on the particle surface

and the type of reaction involved (fast versus slow reversible versus irreversible)

Secondary factors relate to the characteristics of specific soil systems these factors will be

considered in a later section

Adsorption Reactions Binding Sites

Because energy levels vary between different binding sites on solid surfaces the

extent to which P is adsorbed relative to the P in solution is highly non-linear (McGechan

and Lewis 2002) In spite of this a study of binding sites provides some insights into the

way P reacts with surfaces and with particles likely to be found in soils Binding sites

differ to some extent between minerals and bulk soils

There are three primary factors which affect P adsorption to mineral surfaces

(usually to iron and aluminum oxides and hydrous oxides) These are the presence of

ligand exchange sites Lewis acid sites and the effects of surface charge Ligand

exchange occurs because the surfaces of iron and aluminum oxides and hydrous oxides are

generally composed of hydroxide ions and water molecules The water molecules are

directly coordinated to Fe3+ or Al3+ ions just beneath the surfaces of these minerals Only

one-coordinated hydroxide ions are able to participate in ligand exchange reactions This

only occurs under acidic conditions when the hydroxide ion (OH-) adsorbs protons (H+)

producing a positively charged OH2+ ion which serves as the ligand and may be exchanged

with PO4 Sites where this occurs are known as ligand exchange sites (Parfitt 1978)

Another important type of adsorption site on minerals is the Lewis acid site At

these sites water molecules are coordinated to exposed metal (M) ions In conditions of

high pH OH- ions may pull a proton (H+) off the mineral surface This leaves MmiddotOH- at the

surface (Parfitt 1978) The hydroxyl ion is then easily replaced with the PO4 ion when

Fe3+ and Al3+ are the coordinated metals which serve as Lewis acids (Sposito 1984)

Since the most important sites for phosphorus adsorption are the MmiddotOH- and

MmiddotOH2+ sites anything that alters these sites also alters the way adsorption will proceed

These sites can become charged in the presence of excess H+ or OH- and are thus described

as being pH-dependant This is important because adsorption changes with charge When

conditions are more acidic than the point of zero charge H+ ions will be adsorbed on the

oxide surface of the first layer of coordinated hydroxide or oxide ions with counter-ions

(anions) adsorbed in the diffuse double layer Likewise when conditions are more basic

than the point of zero charge H+ ions are desorbed from the first coordination shell and

counter-ions (cations in this case) are in the diffuse double layer (Parfitt 1978) Many

clay minerals adsorb phosphates according to such a pH dependant charge Here

adsorption proceeds when PO4 is exchanged with AlmiddotOH groups on the edges of clay

minerals (Parfitt 1978) However in general clay minerals adsorb less PO4 than oxides

Bulk soils also have binding sites that must be considered However these natural

soils are heterogeneous mixtures unlike the pure minerals considered previously Natural

soils are constantly changed by pedochemical weathering due to biological geological

and hydrological agents A particular soilrsquos surface chemistry is dependent upon the nature

of its weathering which alters the nature and reactivity of binding sites and surface

functional groups As a result natural bulk soils are more complex than pure minerals

(Sposito 1984)

While P adsorption in bulk soils involves complexities not seen when considering

pure minerals many of the same generalizations hold true Recall that reactive sites in pure

systems are often singly coordinated AlmiddotOH or FemiddotOH groups These groups are

particularly exposed on the edges of clay materials and on the surfaces of hydrous oxides

So they too are present in most soils (Parfitt 1978) However even in calcareous soils Al

and Fe oxides are probably the most important components determining the soil PO4

adsorbing capacity (although some have found that calcium carbonate CaCO3 controls in

calcareous soils) (Frossard et al 1995) In tropical soils and soils of the temperate

semiarid regions Fe oxyhydroxides and other Fe-bearing constituents are important

P-adsorbing components even in calcareous soils (Sayin et al 1990) One of the reasons

for this relates to the surface charge phenomena described previously Al and Fe oxides

and similar molecules usually have a point of zero charge of about 8 ndash therefore they are

positively charged in the normal pH range of most soils (Barrow 1984)

While Al and Fe oxides remain the most important factor in P adsorption to bulk

soils other factors must also be considered Surface coatings including metal oxides

(especially Al and Fe oxides) can be formed on the external surfaces of clay minerals

These coatings promote anion adsorption (Parfitt 1978) In addition it must be

remembered that bulk soils contain some material which is not of geologic origin In the

case of organometallic complexes like those formed from humic and fulvic acids these

substances can adsorb PO4 themselves and can also serve as vehicles for its transport In

these situations humic and fulvic acids form complexes with Al and Fe to which PO4 can

later be adsorbed However organic material can also compete with PO4 for binding sites

on Fe and Al hydrous oxides In such an instance organic matter would actually impede P

adsorption (Parfitt 1978 Kaiser and Guggenberger 2003)

Adsorption Reactions

Fast Reversible Reactions Versus Slow Irreversible Reactions The complexities of modeling P adsorption to bulk soils have led researchers to do

so using isotherm experiments of a representative volume of soil Such work led to the

conclusion that two reactions take place when PO4 is applied to soil The first type of

reaction is considered fast and reversible It is nearly instantaneous and can easily be

modeled using isotherm equations (McGechan and Lewis 2002) Fast reactions are well

described by Barrow (1983) who developed the following equation which describes the

proportion of sites occupied by PO4 ions (θ) by considering the chemical affinity between

PO4 ions and surface ions and an electrostatic component

)exp(1)exp(

RTFzcKRTFzcK

ψγαψγα

θminus+

minus= (2-1)

Barrowrsquos equation for fast reactions was developed using only HPO4

-2 as a source of PO4

Ki is a binding constant characteristic of the ion and surface in question zi is the valence

state Ψa is the electrostatic potential in the plane of adsorption α is the proportion of

phosphate present as HPO4-2 γ is the activity coefficient of HPO4

-2 ions in solution and c

is the total concentration of PO4 in solution

Originally it was thought that PO4 adsorption and desorption could be described

completely using simple isotherm equations with parameters estimated after conducting

adsorption experiments However differing contact times and temperatures were observed

to cause these parameters to change thus researchers must be careful to control these

variables when conducting laboratory experiments Increased contact time has been found

to cause a reduction in dissolved P levels Such a process can be described by adding a

time dependent term to the isotherm equations for adsorption However while this

modification describes adsorption well reversing this process alone does not provide a

suitable description of desorption as a result of dilution (McGechan and Lewis 2002)

Empirical equations describing the slow reaction process have been developed by

Barrow and Sharpley (McGechan and Lewis 2002) While neither set of equations is

entirely suitable a reasonable explanation for the slow irreversible reactions is not so

difficult It has been found that PO4 added to soils is initially exchangeable with

32P-labeled PO4 Gradually the amount of exchangeable PO4 in soil drops so that

eventually only 40 remains exchangeable This suggests that some of the adsorbed PO4

is no longer exposed It has been suggested that this may be because of chemical

precipitation of PO4 penetration of the surface by PO4 or by diffusion into pores (Parfitt

Barrow (1983) later developed equations for this slow process based on the idea

that slow reactions were really a process of solid state diffusion within the soil particle

Others have described the slow reaction as a liquid state diffusion process (Frossard et al

1995) or using an unreacted shrinking core model (McGechan and Lewis 2002) which

would involve incorporation of the PO4 ion deeper within the soil particle as time increases

While there is still disagreement over exactly how to model and think of the slow reactions

some factors have been confirmed The process is time- and temperature-dependent but

does not seem to be affected by differences between soil characteristics water content or

rate of PO4 application This suggests that the reaction through solution is either not

rate-limited or that the solution is completely uninvolved Since slow-reaction-deposited

PO4 is effective at blocking adsorption of additional PO4 it is believed that PO4 is still

available at the surface (and is still occupying binding sites) but that it is in a form that is

not exchangeable Another possibility is that the PO4 could have changed from a

monodentate form and instead be retained by bidentate binding (Sposito 1984) or bridging

(Parfitt 1978)

Desorption

Desorption occurs when the soil-water mixture is diluted after a period of contact

with PO4 Experiments with desorption first proved that slow reactions occurred and were

practically irreversible (McGechan and Lewis 2002) This became evident when it was

found that desorption was rarely the exact opposite of adsorption

Dilution of dissolved PO4 after long incubation periods does not yield the same

amount of PO4 that was known to have adsorbed ndash it yields much less especially in the

case of long contact times (McGechan and Lewis 2002) Parfitt (1978) estimated 60 less

Barrowrsquos model for fast adsorption does give a good fit to the experimental data for

desorption and short incubation periods This suggests that desorption can only occur as

the reverse of the fast reactions (Barrow 1983) Various empirical equations have been

developed to describe this process some of which are useful to describe desorption from

eroded soil particles (McGechan and Lewis 2002)

Soil Factors Controlling Phosphate Adsorption and Desorption

While binding sites and the adsorption-desorption reactions are the fundamental

factors involved in PO4 adsorption other secondary factors often play important roles in

given soil systems In general these factors include various bulk soil characteristics

including pH soil mineralogy surface coatings organic matter particle size surface area

and previous land use

Influence of pH

PO4 is retained by reaction with variable charge minerals in the soil The charges

on these minerals and their electrostatic potentials decrease with increasing pH Therefore

adsorption will generally decrease with increasing pH (Barrow 1984) However caution

must be used when applying this generalization since changing pH results in changes in

PO4 speciation too If not accounted for this can offset the effects of decreased

electrostatic potentials

In addition it should be remembered that PO4 adsorption itself changes the soil pH

This is because the charge conveyed to the surface by PO4 adsorption varies with pH

(Barrow 1984) If this charge is less than the average charge on PO4 ions in solution

adsorption increases the soil pH (that is PO4 adsorption releases hydroxide ions) If the

charge conveyed to the surface is greater than the average charge on the ions in solution

adsorption decreases pH This happens at high pH which helps prevent hydroxide ions

from escaping (Barrow 1984)

While pH plays an important role in PO4 adsorption other variables affect the

relationship between pH and adsorption One is salt concentration PO4 adsorption is more

responsive to changes in pH if salt concentrations are very low or if salts are monovalent

rather than divalent (Barrow 1984) Another such variable is precipitation-dissolution

reactions In general precipitation only occurs at higher pHs and high concentrations of

PO4 Still this variable is important in determining the role of pH in research relating to P

adsorption A final consideration is the amount of desorbable PO4 present in the soil and

the changes resulting from its release Release of PO4 occurs mainly at low pH perhaps

because some of the PO4-retaining material was decomposed by the acidity

Correspondingly adding lime seems to decrease desorption This implies that PO4

desorption involves a reverse diffusion of PO4 from the particlesrsquo interior back to the

surface Surface charges increase with decreasing pH potentially pulling PO4 deposited

by the slow reactions back toward the surface (Barrow 1984)

Influence of Soil Minerals

Unique soils are derived from differing parent materials Therefore they contain

different minerals In calcareous soils CaCO3 can control PO4 adsorption However in

general it is thought that Fe and Al oxides control (Parfitt 1978) Such minerals are

present in differing amounts in different soils this is a complicating factor when dealing

with bulk soils which is often accounted for with various measurements of Fe and Al

(Wiriyakitnateekul et al 2005)

Influence of Surface Coatings Many soils adsorb PO4 not due to the presence of Fe and Al oxides but due to the

presence of Fe and Al contained in surface coatings Such coatings have been shown to be

very important in orthophosphate adsorption to soil and sediment particles (Chen et al

Influence of Organic Matter

Organic matter (OM) in the form of humic and fulvic acids can contain Fe and Al to

which PO4 can adsorb (Parfitt 1978) However OM can also compete with PO4 for

binding sites on a particle (Chen et al 2000 Kaiser and Guggenberger 2003 Parfitt 1978

Hiemstra et al 2010a Hiemstra et al 2010b)

Influence of Particle Size

Decreasing particle size results in a greater specific surface area Also in the fast

adsorption reactions PO4 is adsorbed to surfaces This suggests that if all other factors are

the same a soil with a greater surface area will adsorb more PO4 than a soil with lower

surface area The influence of particle size especially the fact that smaller particles are

most important to adsorption has been proven experimentally in a study which

fractionated larger soil particles by size and measured adsorption (Atalay 2001)

Influence of Previous Land Use

Previous land use can affect P content and P adsorption capacity in several ways

Most obviously previous fertilization might have introduced a P concentration to the soil

that is higher than background levels (Rongzhong et al 2009) Type of fertilization is

another important variable (Herrera 2003) In addition heavily-eroded soils would have

an altered particle size distribution compared to uneroded soils especially for topsoils in

turn this would effect specific surface area (SSA) and thus the quantity of available

adsorption sites (Wang et al 2009) Land use also alters soil OM content and soil

aggregation These impacts are reflected in geographic patterns of PO4 concentration in

surface waters which show higher PO4 concentrations in streams draining agricultural

areas (Mueller and Spahr 2006)

Phosphorus Release

If the P attached to eroded soil particles stayed there eutrophication might never

occur in many surface waters However once eroded soil particles are deposited in the

anoxic lower depths of large bodies of surface water P may be released due to seasonal

decreases in stream P content (Jarvie et al 2006) and via mineral reduction reactions

(Hesse 1973) This release is the final link in the chain of events that leads from a

P-enriched upland soil to a nutrient-enriched water body

Release Due to Changes in Phosphorus Concentration of Surface Water

P exchange between bed sediments and surface waters are governed by equilibrium

reactions If a change occurs which upsets this equilibrium sedimentsrsquo role as a source or

a sink of P can change Thus sediment with a high amount of adsorbed P will act as a

source if located in a low-P aquatic environment The point at which such a change occurs

is known as the Equilibrium Phosphate Concentration (EPC0) EPC0 is the amount of PO4

in solution where no dosed PO4 has yet been adsorbed so it is driven by

previously-adsorbed PO4 (Q0) Therefore it may be determined using an isotherm

equation which includes a term for Q0 by solving for the amount of PO4 in solution when

adsorbed PO4 is set to zero When the EPC0 is greater than the amount of reactive PO4 in

solution release from sediment to solution will gradually occur (Jarvie et al 2005)

However because EPC0 is related to Q0 this approach requires a unique isotherm

experiment for each location as EPC0 will not be solely determined by a soilrsquos type or

physical-chemical characteristics

Release Due to Reducing Conditions

Waterlogged soil is oxygen deficient This includes soils and sediments at the

bottoms of lakes ponds and other bodies of water Oxygen deficient conditions encourage

the dominance of facultative and obligate anaerobes These microorganisms utilize

oxidized substances from their environment as electron acceptors Thus as the anaerobes

live grow and reproduce the system becomes increasingly reducing

Oxidation-reduction reactions do not directly impact calcium and aluminum

phosphates They do impact iron components of sediment though Unfortunately Fe

oxides are the predominant fraction which adsorbs P in most soils Eventually the system

will reduce any Fe held in exposed sediment particles within the zone of reducing

oxygen-deficient conditions This occurs when oxidation-reduction potentials fall below

the limiting value of +200 mV When this happens Fe (III) becomes Fe(II) a dissolved

phase not capable of retaining adsorbed P At this point free exchange of P between water

and bottom sediment takes place The inorganic P is freed and made available for uptake

by algae and plants (Hesse 1973)

Describing Phosphate Adsorption with the Langmuir Isotherm The Langmuir isotherm equation (equation 2-2) is commonly used

(Wiriyakitnateekul et al 2005) to relate equilibrium concentrations of adsorbed PO4 to

aqueous PO4

⎥⎦

⎤⎢⎣

Here Q is adsorbed PO4 in mg kgsoil-1 Qmax is the maximum adsorption capacity kl is a

coefficient called the affinity constant related to the bonding energy (L mgPO4-1) and C is

the concentration of PO4 in solution (McGechan and Lewis 2002) Implicit in this

equation is the amount of PO4 that is previously adsorbed (Graetz and Nair 2009) This

value can be determined experimentally or estimated from the rest of the data More

complex forms of the Langmuir equation account for the influence of multiple surfaces on

adsorption The two-surface Langmuir equation is written with the numeric subscripts

indicating surfaces 1 and 2 respectively (equation 2-3)

⎥⎦

⎤⎢⎣

+⎥⎦

⎤⎢⎣

222max

111max 11 Ck

l(2-3)

CHAPTER 3

OBJECTIVES

The goal of this project was to provide improved design tools for engineers and

regulators concerned with control of sediment-bound PO4 In order to accomplish this the

following specific objectives were pursued

1 Develop PO4 adsorption isotherms for the soils used in the earlier eroded size

distributions

2 Further characterize soils for specific surface area (SSA) organic matter (OM) content

iron (Fe) content and aluminum (Al) content

3 Relate PO4 adsorption isotherm parameters to soil characteristics including some which

are available to design engineers in the field

4 An approach similar to the Revised CREAMS approach for estimating eroded size

distributions from parent soil texture was developed and evaluated The Revised

CREAMS equations were also evaluated for uncertainty following difficulties in

estimating eroded size distributions using these equations in previous studies (Price

1994 and Johns 1998) Given the length of this document results of this study effort are

presented in Appendix D

5 Evaluate a structural BMP to observe its effect on TSS concentration particle size PO4

adsorbing potential and previously-adsorbed PO4 Given the length of this document

results of this study effort are presented in Appendix E

CHAPTER 4

MATERIALS AND METHODS

Soils to be used for this study included twenty-nine topsoils and subsoils

commonly found in the southeastern US These soils had been previously collected from

Clemson University Research and Education Centers (RECs) located across South

Carolina in the statersquos distinct soil and climate regions (Figure 4-1) The collected soils

had been identified using Natural Resources Conservation Service (NRCS) county soil

surveys Additional characterization data (soil textural data normal pH range erosion

factors permeability available water capacity etc) is available from these publications

although not all such data are available for all soils in all counties Soil texture and eroded

particle size distributions for these soils had also been previously determined (Price 1994)

Phosphate Adsorption Analysis

Five grams of soil were allowed to equilibrate in 1 L of weak electrolyte 001 M

KCl in order to counteract the soilrsquos natural tendency to disperse when filtering and

centrifuging Regular pH adjustments using HCl or NaOH were used to bring the solution

pH to approximately 65 over the course of several weeks using an Orion 420A pH meter

with an Accumet epoxy pHATC combination pH electrode (AgAgCl reference) pH 65

was chosen based on its distance from the pKa of 72 recently collected data from the area

indicating a pH range in streams of 65-75 (Hur et al 2007) and the fact that both

rainwater and soil pH tend to be slightly acidic At the end of this time 50-mL samples

were withdrawn from the larger volume at a constant depth approximately 1 cm from the

bottom of the 1 L bottle while stirring with a magnetic stir bar Duplicates were drawn

sequentially To ensure samples had similar particle size distributions and soil

concentrations turbidity and total suspended solids were measured at the beginning

middle and end of an isotherm experiment for a selected soil

Figure 4-1 Locations of Clemson University Experiment Station (ES)

and Research and Education Centers (RECs)

Samples were placed in twelve 50-mL centrifuge tubes They were spiked

gravimetrically using a balance and micropipette in duplicate with stock solutions of

pH-adjusted phosphate made from sodium monobasic phosphate and sodium dibasic

phosphate to six known starting concentrations of PO4 (eg of approximately 0 1 5 10

25 50 mg L-1 as PO43-)

Samples were then allowed to equilibrate on an end-over-end tumbler for 84 hours

based on the logistics of experiment batching necessary pH adjustments and on a 6-day

adsorption kinetics study for three soils from across the state which found that 90 of

adsorption took place within 48 hours of dosing with PO4 In addition this was thought to

be an appropriately intermediate timescale for native soil in the field sediment

encountering best management practices (BMPs) and soil and P transport through a

watershed This supports the approach used by Graetz and Nair (2009) which used a

1-day equilibration time

pH checks were conducted daily and pH adjustments were made as-needed all

recorded gravimetrically After equilibration samples were centrifuged at 5000 rpm for 12

minutes and filtered using 020 μm (or smaller) syringe filters Filtrate was analyzed for P

content using the Spectro CIROS VISION Inductively Coupled Plasma Atomic Emission

Spectrometer (ICP-AES) at the Clemson Agricultural Service Lab an instrument which

quantifies elemental concentrations in solution Results were processed by converting P

concentrations to PO4 concentrations by multiplying by 3065 evaluating the amount of

PO4 lost from solution to adsorption by mass balance and normalizing by the soilsrsquo

concentration to develop a value for adsorbed concentration (mgPO4 kgSoil-1) This process

is defined by equation 4-1 where CDose is the concentration resulting from the mass of

dosed PO4 Ms is the mass of soil in solution and C is the aqueous PO4 concentration at

equilibrium as determined by ICP-AES

Qminus

= (4-1)

This adsorbed concentration (Q) was plotted against the measured equilibrium

concentration in the aqueous phase (C) to develop the isotherm Stray data points were

discarded as being unreliable based upon propagation of errors if less than 2 of dosed

PO4 appeared to have adsorbed to the soil Finally the Langmuir isotherm parameters

were determined using the non-linear regression tool with user-defined Langmuir

functions in Microcal Origin 60 which solves for the coefficients of interest by

minimizing chi-squared (X2) a measure of goodness-of-fit in an iterative process this

process is described in the next chapter

Total Suspended Solids

Total suspended solids (TSS) content was analyzed using pre-weighed glass fiber

filters that met the requirements of Standard Method 2540D (Greenberg et al 1992) 50

mL of composite solution was withdrawn at the beginning end and middle of an isotherm

withdrawal filtered and dried at approximately 100˚C to constant weight Across the

experiment TSS content varied by lt5 with lt3 variation from the mean

Turbidity Analysis

Turbidity analysis was conducted to ensure that individual isotherm samples had a

similar particle composition As with TSS samples were withdrawn at the beginning

middle and end of an isotherm withdrawal and analyzed 30-mL aliquots were withdrawn

Turbidity analysis was conducted with the Hach 2100N in the LG Rich Environmental

Research Laboratory with turbidity measured in nephelometric turbidity units (NTUs)

Both standards and samples were shaken prior to placement inside the machinersquos analysis

chamber then readings were taken at 30- and 60-second intervals Across the experiment

turbidity varied by lt5 with lt3 variation from the mean

Specific Surface Area

Specific surface area (SSA) determinations of parent and eroded soils were

conducted at the LG Rich Environmental Research Laboratory using a Micromeritics

ASAP 2010 surface area analyzer using an eight-point Brunauer Emmet and Teller (BET)

nitrogen gas adsorption method Each sample was accurately weighted and degassed at

100degC prior to measurement Other researchers have degassed at 200degC and achieved

good results (Johnson 1995) but the lower temperature helps to ensure that soil surface

area is not altered due to heat

Organic Matter and Carbon Content

Soil samples were taken to the Clemson Agricultural Service Laboratory for

organic matter analysis using the Thermolyne Type 6000 Furnace a loss-on-ignition

technique Approximately 5 g soil was placed in a numbered pre-weighed ldquoHigh Formrdquo

porcelain crucible Crucible and soil were placed in the furnace which was then set to

105degC to allow the soil to dry After a 30-minute period required for the furnace to reach

105degC samples were allowed to dry for 2 hours The sample was then carefully placed in

a desiccator to cool and massed to 4 decimal places before being placed back in the furnace

Next samples were combusted at 360degC for 2 hours after a 1-hour warm-up period

Samples were placed in a desiccator to cool and weighted again to 4 decimal places OM

was then calculated as the difference between the soilrsquos dry weight and the percentage of

total dry soil mass lost during the 360degC phase of the experiment (CU ASL 2000)

Carbon (Cb) content was also analyzed Approximately 05 g of ground and dried

soil was taken to the Clemson Agricultural Service Laboratory for analysis using an

Elemantar Vario Macro which analyzes by separating and analyzing gases resulting from

combustion of the soil CO2(g) is produced when Cb in OM is oxidized CO2(g) is detected

by an infrared adsorption detector which measures relative thermal conductivities for

quantification against standards in order to determine Cb content (CU ASL 2009)

Mehlich-1 Analysis (Standard Soil Test)

nutrient analysis using the Mehlich-1 (Me-1) analysis This is the standard soil test

administered by the Clemson Agricultural Extension Service and if well-correlated with

Langmuir parameters it could provide engineers a quick economical tool with which to

estimate PO4 isotherm parameters According to the procedure (CU ASL 2000)

approximately 5 g of soil is placed in a polyethylene cup Then 20 mL of Mehlich-1

solution (005 N HCl and 0025 N H2SO4) was added The samples were then shaken for 5

minutes on a reciprocating shaker at 180 oscillations per minute with a 4 cm stroke The

Mehlich-1 soil solution was then filtered with the extract saved for analysis by ICP-AES

Leftover extract was then taken back to the LG Rich Environmental Laboratory for

analysis of PO4 concentration using ion chromatography (IC)

Dithionite-Citrate-Bicarbonate (DCB) Iron Aluminum and Phosphate This analysis removes amorphous Fe and Al surface coatings by reducing them

thus releasing any other chemicals (including PO4) which had previously been bound to the

coatings As such it would seem to provide a good indication of the amount of PO4that is

likely to be released by Fe reduction The dithionite-citrate-bicarbonate (DCB) experiment

uses sodium dithionite (Na2S2O4) to reduce the surface coatings sodium citrate

(C6H5N3O4 2H2O) to chelate the iron and sodium bicarbonate (NaHCO3) to buffer the

system

Approximately 05 g of soil was added to a pre-weighed 50-mL centrifuge tube To

this was added approximately 225 mL of 03 M Sodium citrate solution and 25 mL of 1 M

sodium bicarbonate solution all added gravimetrically The tubes prepared in triplicate

were then placed in an 80˚C water bath and covered with aluminum foil to minimize

evaporative losses until temperature increased to 75-80˚C when 05 g of pre-weighed

sodium dithionite was added to each tube Tubes were stirred thoroughly by hand for 60

seconds and then for about 60 seconds every 5 minutes over the next 15 minutes A

second portion of pre-weighed sodium dithionite was added and the procedure continued

for another ten minutes If brown or red residues remained in the tube sodium dithionite

was added again gravimetrically until all the soil was a white gray or black color

At this point 1-2 mL of concentrated HCl was added gravimetrically to maintain a

pH of about 4 and to prevent the reoxidation of any reduced iron The sample tubes were

weighed again to establish how much liquid was currently in the bottle in order to account

for evaporative losses Then the tubes were centrifuged at 1200 rpm for 15 minutes

diluted 120 and analyzed via ICP-AES at the Clemson Agricultural Service Laboratory

Results were corrected for dilution and normalized by the amount of soil originally placed

in solution so that results could be presented in terms of mgconstituentkgsoil

Model Fitting and Regression Analysis

Regression analyses were carried out using linear and multilinear regression tools

in Microsoft Excel 2003 Microcal Origin Version 60 and MiniTab 15 The nonlinear

regression tool in Origin was used to fit isotherm equations to results from the adsorption

experiments (Chapters 5 and 7) Isotherm parameters determined in this process were then

compared to soil characteristics using Excel and MiniTab as necessary (Chapter 7)

Similar regressions were carried out to relate soil characteristics to one another (Chapter 6)

Variablesrsquo significance was defined by p-value as is typical in the literature

models and parameters were considered significant at 95 certainty (p lt 005) although

some additional fitting parameters were considered significant at 90 certainty (p lt 010)

In general the coefficient of determination (R2) defined as the percentage of variability in

a data set that is described by the regression model was used to determine goodness of fit

For multilinear regressions the adjusted R2 (R2Adj) was used for this purpose in order to

appropriately account for additional variables and allow for comparison between

regressions R2Adj is defined by equation 4-2 where n is the number of measurements and p

is the number of fitting parameters

11)1(1 22

minusminusminus

minusminus=pn

nRR Adj (4-2)

In addition the dot plot and histogram graphing features in MiniTab were used to

group and analyze data Dot plots are similar to histograms in graphically representing

measurement frequency but they allow for higher resolution and more-discrete binning

CHAPTER 5

RESULTS AND DISCUSSION ISOTHERM DEVELOPMENT

Beginning with the 29 soils previously discussed (15 topsoils 14 subsoils)

isotherms were developed for 23 soils (11 topsoils 12 subsoils) Isotherms were not

developed for the remaining six soils (Lakelands Topsoil from the Sandhill REC Varina

Topsoil Fuquay Topsoil and Subsoil and Blanton Topsoil and Subsoil all from the Edisto

REC) for which PO4 percent adsorption was so low that it was unmeasurable Original

experimental data for all soils are included in the Appendix A Prior to developing

isotherms for the remaining 23 soils three different approaches for determining

previously-adsorbed PO4 and maximum adsorption capacity from raw data (Figure 5-1)

were evaluated along with one-surface vs two-surface isotherm fitting techniques

Cecil Subsoil Simpson REC

0 10 20 30 40 50 60 70 80

C mg-PO4L

Figure 5-1 Sample Plot of Raw Isotherm Data

Determining Previously-Adsorbed Phosphate and Maximum Adsorption Capacity

It was immediately observed that a small amount of PO4 desorbed into the

background electrolyte solution for all soils at the 0 mgPO4 L-1 dose level Desorption can

be thought of as negative adsorption therefore it is important to account for this

previously-adsorbed PO4 (Q0) when determining maximum adsorption capacity (Qmax) and

because it was thought that Q0 was important in its own right Three different approaches

for determining Q0 were evaluated (1) Using the Southern Cooperative Series (SCS)

Method (Graetz and Nair 2009) which extrapolates Q0 from a regression line relating the

amount of phosphate in solution to the amount sorbed to soil particles at the two lowest

concentrations (2) Using Microcal Origin Version 60 to fit curves to the original data

using a modified 1-surface Langmuir isotherm and (3) using Origin to fit curves to the

original data using a modified 2-surface Langmuir isotherm In the first case Qmax would

be determined by adding the estimated value for Q0 back to the original data prior to fitting

with Origin in the second and third cases the values estimated by Origin for Q0 and Qmax

were estimated from the original data

The first approach was established by the Southern Cooperative Series (SCS)

(Graetz and Nair 2009) Microsoft Excelrsquos linear regression feature was used to establish

a best-fit line of the form

Q = mC - Q0 (5-1)

where m is the slope of the regression line relating Q to C and Q0 is the y-intercept

representative of the amount of PO4 adsorbed where C = 0 (Graetz and Nair 2009) The

value found for Q0 is then added back to the entire data set which is subsequently fit using

Origin In this approachrsquos favor are the facts that (1) it is simple and logical (2) it has the

support of cooperative services in the southeast (3) it is derived from the portion of the

data which would intuitively likely have the most impact on Q0 and (4) it allows for quick

and easy analysis of Q0 as it is derived from the data by linear regression (Figure 5-2)

allowing statistics to be calculated to describe the validity of the regression

y = 41565x - 87139R2 = 07342

0 005 01 015 02 025 03

C mg-PO4L

Figure 5-2 Using the SCS Method to Determine Previously-Adsorbed Phosphate (Q0)

However the SCS procedure is based on the assumption that the two lowest

concentration points (shown in duplicate in Figure 5-2) have the most impact on Q0 while

reasonable the whole system collapses if this assumption is incorrect Equation 2-2

demonstrates that the SCS is only valid when C is much less than kl that is when the

Langmuir equation asymptotically approaches a straight line Another potential

disadvantage involves adjusting the rest of the data by Q0 prior to fitting with Origin

(Figure 5-3) This could result in over-estimating Qmax

The second approach to be evaluated used the non-linear curve fitting function of

Microcal Origin Version 60 to fit the data to a 1-surface Langmuir equation modified to

include Q0 always defined as a positive number (Equation 5-2) This method is referred to

in this thesis as the 3-Parameter method and it offers several potential benefits (1) It uses

the entire dataset of values for C and Q to establish Q0 and (2) Statistics for all variables are

0 10 20 30 40 50 60 70 80 90

C mg-PO4L

Adjusted Data Isotherm Model

Figure 5-3 Isotherm Developed Using the SCS Method to Determine Previously-Adsorbed Phosphate (Q0))

calculated as part of the curve-fitting process For a particular soil sample this approach

also lends itself to easy calculation of EPC0 if so desired While showing the

low-concentration values as negative adsorption (Figure 5-4) might not always be ideal

this can easily be remedied by eliminating the Q0 parameter when plotting the curve as

Qmax and kl are unchanged

A 5-Parameter method was also developed and evaluated This method uses the

same approach as the 3-Parameter method but for a modified 2-surface Langmuir isotherm

In general at the lowest values for C and Q the 5-Parameter method yields values for Q0

that are greater than Q0 estimated by the SCS and 3-Parameter methods While the

coefficient of determination (R2) is improved for this approach standard errors associated

with each of the five variables are generally very high and parameter values do not always

converge While it may provide a good approach to estimating Q0 its utility for

determining the other variables is thus quite limited

0 20 40 60 80 100

C mg-PO4L

Figure 5-4 3-Parameter Fit

0max 1

l minus⎥⎦

⎤⎢⎣

= (5-2)

Isotherm Development 1-Surface vs 2-Surface Langmuir Isotherm Equation

Using the SCS method for determining Q0 Microcal Origin was used to calculate

isotherm parameters and statistical information for the 23 soils which had demonstrated

experimentally-measurable adsorption Regressions for both the 1-Surface Langmuir

Equation and the 2-Surface Langmuir Equation were carried out Data for these

regressions including the derived isotherm parameters and statistical information are

presented in Appendix A Although statistical measures X2 and R2 were improved by

adding parameters for the 2-Surface model (Figure 5-5) standard error for individual

isotherm parameters was higher Because the purpose of this study is to find predictors of

isotherm behavior the increased standard error among the isotherm parameters was judged

more problematic than minor improvements to X2 and R2 were deemed beneficial

Therefore the 1-Surface Langmuir was selected as the equation of choice when fitting

isotherm models to the experimental data

0 10 20 30 40 50 60C mg-PO4L

SCS-Corrected Data SCS-1Surf SCS-2Surf

Figure 5-5 Comparison of 1- and 2-Surface Isotherms Developed for Vaucluse Topsoil Using the SCS Method

Isotherm Development Comparison of Methods for Estimating Q0 The effectiveness of the three approaches for estimating Q0 was evaluated using

two different techniques First three different soils one each with low intermediate and

high estimated values for kl were selected and graphed The three selected soils were the

Pelion Subsoil Dothan Topsoil and Vaucluse Topsoil (Figures 5-6 5-7 and 5-8) Raw

data for each soil were plotted along with isotherm curves shown only at the lowest

concentrations for C and Q Both the SCS and 5-Parameter methods are very effective at

fitting the lowest-concentration data points However the 5-parameter method seems to

introduce additional curvature and in the case of the Pelion and Dothan Topsoils it seems

to overestimate Q0

0 02 04 06 08 1C mg-PO4L

Raw Data 3 Parameter 5 Parameter Linear (Raw Data)

Figure 5-6 Comparison of Methods for Estimating Q0 Pelion Topsoil

-30-20

0 02 04 06 08 1C mg-PO4L

Figure 5-7 Comparison of Methods for Estimating Q0 Dothan

Topsoil

0 02 04 06 08 1C mg-PO4L

Figure 5-8 Comparison of Methods for Estimating Q0 Vaucluse Topsoil

In order to further compare the three methods presented here for determining Q0 10

soils were selected from the adsorption experiment data using Microsoft Excelrsquos random

number generator function Each of the 23 soils which had demonstrated

experimentally-detectable phosphate adsorption were assigned a number The random

number generator was then used to select one soil from each of the five sample locations

along with five additional soils selected from the remaining soils Then each of these

datasets were fit using the SCS 3-Parameter and 5-Parameter techniques

In general the 3-Parameter method provided the lowest estimates of Q0 for the

modeled soils the 5-Parameter method provided the highest estimates and the SCS

method provided intermediate estimates (Table 5-1) Regression analyses to compare the

methods revealed that the 3-Parameter method is not significantly related at the 95

confidence interval (p lt 005) to the SCS method or to the 5-Parameter method but that the

SCS method is significantly related to the 5-Parameter method (Figure 5-9 and Table 5-2)

This is not surprising based on Figures 5-6 5-7 and 5-8

Table 5-1 Comparison of Q0 Estimates for a Subset of Soils

3-Param Q(0) 5-Param Q(0) SCS Q(0) Appling Top 8774 17663 11421 Johnston Sub 5499 6759 4111 Dothan Top 1672 9152 4000 Norfolk Top 3330 5022 4082

Wadmalaw Top 2514 5088 3519 Coxville Sub 5930 12095 8759 Rembert Top 2220 2422 3499 Johnston Top 1608 18508 7357 Yonges Top -1298 20357 6914 Pelion Sub 8969 17503 4070

R2 = 04243

0 50 100 150 200 250

5 Parameter Q(0) mg-PO4kg-Soil

Figure 5-9 Regression Plot Comparing 5-Parameter Method to SCS Method Estimates for Q0

Table 5-2 Regression Statistics Relating Methods for Estimating Q0 to One Another

3-Param Q(0) 5-Param Q(0) SCS Q(0) 3-Param Q(0) Coefficient plusmn SE Intercept plusmn SE R2

5-Param Q(0) Coefficient plusmn SE Intercept plusmn SE R2

0063 plusmn 0181

3196 plusmn 22871 0016

SCS Q(0) Coefficient plusmn SE Intercept plusmn SE R2

025 plusmn 0281

4793 plusmn 1391 0092

027 plusmn 011

2711 plusmn 14381 042

1 p gt 005

Final Isotherms

Based on this analysis the 1-Surface Langmuir is the preferred approach for fitting

adsorption data and seeking predictive relationships based on soil characteristics due to the

fact that standard errors are reduced for the fitted parameters Regarding

previously-adsorbed PO4 it appears that the 5-Parameter method tends to overestimate Q0

leaving the 3-Parameter and SCS methods as the two preferred approaches with the SCS

method being probably superior Unfortunately estimates developed with these two

methods are not well-correlated with one another However overall the 3-Parameter

method is preferred because Q0 is the isotherm parameter of least interest to this study In

addition because the 3-Parameter method calculates Q0 directly it (1) is less

time-consuming and (2) does not involve adjusting all other data to account for Q0

introducing error into the data and fit based on the least-certain and least-important

isotherm parameter Thus final isotherm development in this study was based on the

3-Parameter method These isotherms sorted by sample location are included in Appendix

A (Figures A-41-6) along with a table including isotherm parameter and statistical

information (Table A-41)

CHAPTER 6

RESULTS AND DISCUSSION SOIL CHARACTERIZATION

Soil characteristics were analyzed and evaluated with the goal of finding

readily-available information or easily-measurable characteristics which could be related

to the isotherm parameters calculated as described in the previous chapter Primarily of

interest were soil texture SSA OM Mehlich-1 (standard soil test) data Fe Al and

previously-adsorbed PO4 Soil characteristics were related to data from the literature and

to one another by linear and multilinear least squares regressions using Microsoft Excel

2007 and MiniTab 15 set to a 95 confidence interval a characteristicrsquos significance was

indicated by p-values (p) lt 005

Soil Texture and Specific Surface Area

Soil texture is related to SSA (surface area per unit mass equation 6-1) as

demonstrated by the equations for calculating the surface area (SA) volume and mass of a

sphere of a given diameter D and density ρ

SMSASSA = (6-1)

2 DSA π= (6-2)

6 3DVolume π

= (6-3)

ρπρ 6

3DVolumeMass == (6-4)

Because specific surface area equals surface area divided by mass we can derive the

following equation for a simplified conceptual model

ρDSSA 6

= (6-5)

Thus we see that for a sphere SSA increases as D decreases The same holds true

for bulk soils those whose compositions include a greater percentage of smaller particles

have a greater specific surface area Surface area is critically important to soil adsorption

as discussed in the literature review because if all other factors are equal increased surface

area should result in a greater number of potential binding sites

Soil Texture

The individual soils evaluated in this study had already been well-characterized

with respect to soil texture by Price (1994) who conducted a hydrometer study to

determine percent sand silt and clay In addition the South Carolina Land Resources

Commission (SCLRC) had developed textural data for use in controlling stormwater and

associated sediment from developing sites Finally the county-wide soil surveys

developed by the Natural Resources Conservation Service (NRCS) or its predecessor the

Soil Conservation Service (SCS) provided additional data on soil texture (Herren 1979

Rogers 1977 Miller 1971 Morton 1996 Lawrence 1978)

Due to the fact that an extensive literature exists providing textural information on

many though not all soils it was hoped that this information could be related to soil

isotherm parameters However it also seemed prudent to compare Pricersquos (1994) data to

the data available in literature reviews This was carried out primarily with the SCLRC

data (Hayes and Price 1995) which provide low and high percentage figures for soil

fractions passing a 200 sieve (generally thought to contain the clay and silt fractions) and a

400 sieve (generally thought to contain the clay fraction) at various depths of each soil

Because the soil depths from which the SCLRC data were created do not precisely

correlate with Pricersquos definition of topsoil and subsoil overall arithmetic (xm) and

geometric (xg) means for each soil type were also created and compared Attempts at

correlation with the Price (1994) data were based on the low and high percentage figures as

well as arithmetic and geometric means In addition the NRCS County soil surveys

provide data on the percent of soil passing a 200 sieve for various depths These were also

compared to the Price data both specific to depth and with overall soil type arithmetic and

geometric means Unfortunately the correlations between top- and subsoil-specific values

for clay content from the literature and similar site-specific data were quite weak (Table

6-1) raw data are included in Appendix B It is noteworthy that there were some

correlations between the SCLRC and NRCS 200 Sieve Data perhaps indicating a common

origin

Poor correlations between the hydrometer data for the individual sampled soils

used in this study and the textural data from the literature are disappointing because it calls

into question the ability of readily-available data to accurately define soil texture This

indicates that natural variability within soil types is such that representative data may not

be available in the literature This would preclude the use of such data as a surrogate for a

hydrometer or specific surface area analysis

Table 6-1 Coefficients of Determination (R2) Between Hydrometer Data (Price 1994) and Soil Texture Data from the Literature

NRCS 200 Sieve Data ()2 Price ()3 Price Clay (Overall )4

Price Silt (Overall )3

Price Sand (Overall )3

Lower Higher xm xg Clay Silt (Clay

+ Silt)

xm xg xm xg xm xg

xm - - - - 031 - - - - - - - - SCLRC Clay ()2 xg - - - - 029 - - - - - - - -

xm - - - - - 0004 - - - - - - - SCLRC Silt ()2 xg - - - - - 61E-05 - - - - - - -

Lower 037 031 035 036 - - 020 - - - - - - Higher 079 055 058 058 - - 052 - - - - - -

xm 052 048 053 053 - - 0096 - - - - - -

SCLRC 200 Sieve Data ()2

xg 047 042 047 047 - - 0081 - - - - - - xm 0066 010 010 010 - - - 0020 - - - - - Clay xg 0080 012 012 012 - - - - 0058 - - - - xm 010 019 016 016 - - - - - 012 - - - Silt xg 022 037 034 034 - - - - - - 021 - - xm 012 020 018 018 - - - - - - - 016 - SC

Sand xg 013 023 02 020 - - - - - - - - 018 Lower - - - - 0262 0142 0412 0343 0293 0273 0273 0473 0483 Higher - - - - 0292 0142 0442 0593 0673 0193 0183 0553 0553

xm - - - - 0292 0152 0452 0563 0593 0273 0263 0603 0613

NRCS 200 Sieve Data ()

xg - - - - 0292 0152 0452 0533 0543 0273 0263 0603 0603

2 NRCS data are presented at a range of the percentage of soil passing a given sieve size ldquolowerrdquo and ldquohigherrdquo designations represent the ends

of this range 3 All three datasets have different ways of describing topsoil vs subsoil Price does so qualitatively while the SCLRC and NRCS do so with

various soil depths Data represented by these coefficients of determination are comparisons of soil texture data from similar soil strata although they are not defined identically between the three datasets

4 Arithmetic (xm) and geometric (xg) means were created for each soil type from various data for ldquotopsoilrdquo and ldquosubsoilrdquo or texture according to depth so that the three datasets could be compared

Soil Specific Surface Area

Soil specific surface area (SSA) should be directly related to soil texture Previous

studies (Johnson 1995) have found a strong correlation between SSA and clay content In

the current study a weaker correlation was found (Figure 6-1) Additional regressions

were conducted taking into account the silt fraction resulting in still-weaker correlations

Finally a multilinear regression was carried out which included the organic matter content

A multilinear equation including clay content and organic matter provided improved

ability to predict specific surface area considerably (Figure 6-2) using the equation

524202750 minus+= OMClaySSA (6-6)

where clay content is expressed as a percentage OM is percent organic matter expressed as

a percentage and SSA is specific surface area in m2 g-1 OMrsquos role in predicting SSA was

not unexpected as other researchers have noted positive correlations between the two

parameters (Kaiser and Guggenberger 2003) A table presenting regression statistics

(Table 6-2) is provided below and clay OM and SSA data are included in Appendix B

y = 09341x - 30278R2 = 0734

0 5 10 15 20 25 30 35 40 45

Clay Content ()

Figure 6-1 Clay Content vs Specific Surface Area

R2 = 08454

0 5 10 15 20 25 30 35 40 45

Predicted Specific Surface Area(m^2g)

Figure 6-2 Predicted Specific Surface Area Using Clay and Organic Matter Content vs Measured Specific Surface Area

Table 6-2 Regression Statistics Relating Specific Surface Area (SSA) to Soil Texture and Organic Matter (OM)

Coefficient(s) plusmn Standard Error (SE) y-intercept plusmn SE R2 Adj R2

Clay () 091 plusmn 013 -216 plusmn 2491 071 069 Silt () -0033plusmn 0371 1317 plusmn 6581 0024 000 Clay + Silt ()

078 plusmn 014 -1285 plusmn 483 063 058

OM () 379 plusmn 099 411 plusmn 2951 047 039 Clay () OM ()

075 plusmn 011 220 plusmn 060 -452 plusmn 209 085 081

Clay + Silt () OM()

1 p gt 005

Soil Organic Matter

As has previously been described the Clemson Agricultural Service Laboratory

carried out two different measurements relating to soil organic matter One measured the

percent carbon (Cb) of soil samples the other measured the percent organic matter (OM) of

the soil samples results for both analyses are presented in Appendix B

It would be expected that Cb and OM would be closely correlated but this was not

the case However a multilinear regression between Cb and DCB-released iron content

(FeDCB measured as mgFe kgSoil-1) and OM produced a very strong correlation (Figure 6-3)

which allows for a confident prediction of OM using the formula

160000130361 ++= DCBb FeCOM (6-7)

where OM and Cb are expressed as percentages This was not unexpected because of the

high iron content of many of the sample soils and because of ironrsquos presence in many

organic compounds and surface coatings (Parfitt 1978) The fit was slightly improved

further when DCB-released aluminum content (AlDCB measured as mgAl kgSoil-1) was also

included

2900000570000130371 +minus+= DCBDCBb AlFeCOM (6-8)

No such correlations were found for similar regressions using Mehlich-1 extractable iron

or aluminum (Table 6-3)

R2 = 09505

0 1 2 3 4 5 6 7 8 9

Predicted OM

Figure 6-3 Predicted OM Using FeDCB Content and Cb vs Measured OM

Table 6-3 Regression Statistics Relating Organic Matter (OM) to Other Soil Characteristics

Coefficient(s) plusmn Standard Error

y-intercept plusmn SE R2 Adj R2

Carbon (Cb ) 137E0 plusmn 0741 122 plusmn 0521 011 0078 FeDCB (mg kgsoil

-1) 128E-4 plusmn 107E-5 082 plusmn 017 084 083 FeMe-1 (mg kgsoil

-1) 366E-04 plusmn 199E-31 236 plusmn 075 00016 00000 AlDCB (mg kgsoil

-1) 703E-4 plusmn 120E-4 045 plusmn 0351 056 055 AlMe-1 (mg kgsoil

-1) 447E-4 plusmn 678E-41 172 plusmn 0911 0020 0000 Cb () FeDCB (mg kgsoil

-1) 137E0 plusmn 019

126E-4 plusmn 641E-06 016 plusmn 0161 095 095

Cb () AlDCB (mg kgsoil

-1) 122E0 plusmn 057

691E-4 plusmn 131E-4 -017 plusmn 0541 065 058

Cb () FeDCB (mg kgsoil

-1) AlDCB (mg kgsoil

138E0 plusmn 018 139E-4 plusmn 110E-5

-110E-4 plusmn 768E-51 029 plusmn 0181 095 095

1 p gt 005

A standard Mehlich-1 soil test was performed to determine whether or not standard

soil analyses as commonly performed by extension service laboratories nationwide could

provide useful information for predicting isotherm parameters Common analytes are pH

phosphorus potassium calcium magnesium zinc manganese copper boron sodium

cation exchange capacity acidity and base saturation (both total and with respect to

calcium magnesium potassium and sodium) In addition for this work the Clemson

Agricultural Service Laboratory analyzed Fe and As released by the Mehlich-1 extractant

using the ICP-AES instrument because Fe and Al have been previously identified as

predictors of PO4 adsorption Results from these tests are included in Appendix B

Mehlich-1 extractable iron (FeMe-1) and aluminum (AlMe-1) are compared to DCB-released

iron (FeDCB) and aluminum (AlDCB) in the next section and Mehlich-1 extractable

phosphorus (PMe-1) is compared to other measures of previously-adsorbed phosphorus in a

section which follows Regression statistics for isotherm parameters and all Mehlich-1

analytes are presented in Chapter 7 regarding prediction of isotherm parameters

correlation was quite weak for all Mehlich-1 measures and parameters

DCB Iron and Aluminum

The DCB procedure is much more aggressive than the Mehlich-1 extraction As a

result concentrations of iron and aluminum released by this procedure are much greater it

seems that the DCB procedure provides an estimate of total iron and aluminum that would

be more akin to a total digestion than to an extraction Results for FeDCB and AlDCB are

included in Appendix B and correlations between FeDCB and AlDCB and isotherm

parameters are presented in Chapter 7 regarding prediction of isotherm parameters

However because DCB analysis is difficult and uncommon it was worthwhile to explore

any relationships between FeDCB and AlDCB and FeMe-1 and AlMe-1 Unfortunately none

were evident (Table 6-4)

Table 6-4 Regression Statistics Comparing Mehlich-1 to DCB-Released Analyses of Iron and Aluminum Content VariableVariable FeDCB (mg kgsoil

-1) AlDCB (mg kgsoil-1)

FeMe-1 (mg kgsoil-1)

Coefficient plusmn Standard Error (SE) Intercept plusmn SE R2

-1365 plusmn 12121

1262397 plusmn 426320 0044

AlMe-1 (mg kgsoil-1)

Coefficient plusmn SE Intercept plusmn SE R2

093 plusmn 062 1

109867 plusmn 783771 0073

1 p gt 005

Previously Adsorbed Phosphorus

Previously adsorbed P is important both as an isotherm parameter and because this

soil-associated P has the potential to impact the environment even if a given soil particle

does not come into contact with additional P either while undisturbed or while in transport

as sediment Three different types of previously adsorbed P were measured as part of this

project (1) PO4 desorbed in water (PO4H2O-Desorbed) (2) DCB-released PO4 (PO4 DCB) and

(3) Mehlich-1 PO4 (PO4Me-1) These data are included in Appendix B and statistical

information regarding correlation with isotherm parameters is included in the final chapter

regarding prediction of isotherm parameters

Phosphorus Occurrence as Phosphate in the Environment

It is typical to refer to phosphorus (P) as an environmental contaminant yet to

measure or report it as phosphate (PO4) In this project PO4 was measured as part of

isotherm experiments because that was the chemical form in which the P had been

administered However to ensure that this was appropriate a brief study was performed to

ensure that the bulk of measured P was indeed occurring as PO4 Using Mehlich-1extract

solution leftover from the Clemson Agricultural Service Laboratoryrsquos ICP analysis of

standard soil analytes an IC measurement of PO4 was performed to ensure that the

mathematical conversion of ICP-measured P (PICP) to PO4 was defensible As expected

the experiment resulted in a strong nearly one-to-one correlation between the two

measures (Figure 6-4 Table 6-5) Therefore the reporting of PO4 rather than P was deemed

appropriate in all cases because approximately 81 of previously-adsorbed P consists of

PO4 and concentrations were quite low relative to the amounts of PO4 added in the

isotherm experiments When the Blanton Topsoil outlier was eliminated over 93 of

measured P was found to be present as PO4

R2 = 09895

0123456789

0 1 2 3 4 5 6 7 8 9 10

ICP mmols PL

Figure 6-4 ICP-Measured P vs IC-Measured PO4 With 11 Line Table 6-5 Relationship of PICP to PIC (mmolsP kgsoil

-1) Coefficient plusmn Standard

Error (SE) y-intercept plusmn SE R2

Overall PICP (mmolsP kgsoil

-1) 081 plusmn 002 023 plusmn 0051 099

Lower (mmolsP kgsoil-1) 093 plusmn 003 013 plusmn 004 098

Previously Adsorbed Phosphate Measures and Applicability PO4 desorbed in water (PO4H2O-Desorbed ) had been previously measured as part of

the original isotherm experiments it was the amount of PO4 measured in an equilibrated

solution of soil and water Although this is a very weak extraction it provides some

indication of the amount of PO4 likely to desorb from these particular soil samples into

water PO4Me-1 provides an estimate of the amount of PO4 available to plants so it is also a

useful indicator of these particular soil samplesrsquo ability to cause an adverse environmental

impact if in solution or of soil fertility if undisturbed PO4 DCB is akin to measurements of

total soil PO4 so its applicability in the environment would be limited to reduced

conditions which occasionally occur in the sediments of reservoirs and which could result

in the release of all Fe- and Al-associated PO4 None of these measurements would be

thought to provide a strong measure of previously-adsorbed PO4 across regions or even soil

types as this figure is dependent upon a particular soilrsquos history of fertilization land use

etc In addition none of these measures correlate well with one another (Table 6-6) there

are only very weak correlations between PO4H2O-Desorbed and PO4 DCB and between

PO4H2O-Desorbed and PO4 Me-1 It is noteworthy that PO4H2O-Desorbed is generally greater than

PO4 Me-1 This differs from the findings of Herrera (2003) and is likely due to the longer

equilibration time for the adsorption experiments which produced PO4H2O-Desorbed in

Table 6-6 Regression Statistics Relating PO4DCB PO4 Me-1 and PO4 H2O Desorbed VariableVariable PO4DCB

(mg kgsoil-1)

PO4 Me-1

(mg kgsoil-1)

PO4 H2O

Desorbed

(mg kgsoil-1)

PO4DCB (mg kgsoil-1)

PO4 Me-1 (mg kgsoil-1)

084 plusmn 058 1

55766 plusmn 111991 0073

PO4 H2O Desorbed (mg kgsoil-1)

1021 plusmn 331

19167 plusmn 169541 033

024 plusmn 0121 3210 plusmn 760

1 p gt 005

addition the Herrera soils contained higher initial concentrations of PO4 However that

study also found that Mehlich-1 and Mehlich-3 analyses were not reliable indicators of

water soluble phosphorus (WSP)

CHAPTER 7

RESULTS AND DISCUSSION PREDICTING ISOTHERM PARAMETERS

The ultimate goal of this project was to identify predictors of isotherm parameters

so that phosphate adsorption could be modeled using either readily-available information

in the literature or economical and commonly-available soil tests Several different

approaches for achieving this goal were attempted using the 3-parameter isotherm model

Figure 7-1 Coverage Area of Sampled Soils

General Observations

PO4 adsorption capacity (Qmax) and the affinity constant (kl) were both generally

greater for subsoils than for topsoils This may be due to clay and Fe leaching through the

soil column as data generally indicated varying levels of enrichment in subsoils relative to

topsoils There were two exceptions (1) The Johnston topsoil had a greater Qmax than the

Johnston subsoil and (2) The Wadmalaw topsoil had a greater kl than the Wadmalaw

subsoil There were 8 soils which had measurable phosphate adsorption for both top- and

subsoils Subsoil enrichment ratios were developed for these soils (Table 7-1) and

compared to isotherm parameters only organic matter enrichment was related to Qmax

enrichment and then only at a 92 confidence level although clay content and FeDCB

content have been strongly related to one another (Table 7-2)

Table 7-1 Subsoil Enrichment Ratios for Soil Characteristics and Isotherm Parameters

Soil Type OM Ratio

FeDCB Ratio

AlDCB Ratio

SSA Ratio

Clay Ratio

Qmax Ratio

kL Ratio

Qmaxkl Ratio

Madison 307 270 131 462 248 529 222 1174 Pelion 037 196 068 438 299 117 1647 1929

Johnston 020 076 067 081 181 033 743 248 Rembert 091 663 200 560 478 277 367 1015 Dothan 218 364 305 526 305 178 1188 2112

Coxville 167 018 104 198 278 173 120 207 Norfolk 050 996 505 5025 452 238 802 1911

Wadmalaw 041 125 124 425 354 289 010 027

Geography-Related Groupings

A GIS analysis using raster data from the USDA-NRCS (2009) indicates that the

soil series analyzed in this study cover approximately 30 of South Carolina (Figure 7-1)

This indicates that the sampled soils provide good coverage that should be typical of other

states along the south Atlantic coast However plotting the final isotherms according to

their REC of origin demonstrates that even for soils gathered in close proximity to one

another and sharing a common geological and land use morphology isotherm parameters

Table 7-2 Regression Statistics1 Between Subsoil Enrichment Ratios Qmax Ratio kL Ratio Clay Ratio OM Ratio

Clay Ratio Coefficient plusmn Standard Error (SE) Intercept plusmn SE R2

031plusmn059

128plusmn199 0045

-050plusmn231

800plusmn780

OM Ratio Coefficient plusmn SE Intercept plusmn SE R2

093plusmn0443 121plusmn066

-127plusmn218 785plusmn3303

FeDCB Ratio Coefficient plusmn SE Intercept plusmn SE R2

268plusmn082

-530plusmn274 065

0011 AlDCB Ratio Coefficient plusmn SE Intercept plusmn SE R2

-095plusmn047 -120plusmn160

0010plusmn028 114plusmn066 000022

SSA Ratio Coefficient plusmn SE Intercept plusmn SE R2

940plusmn552 -2086plusmn1863

005 1 Regression parameters are insignificant at the 95 confidence level (p gt 005)

unless otherwise noted 2 Significant at the 95 confidence level (p lt 005) 3 Significant at the 90 confidence level (p lt 010) can vary widely (Appendix A Figures A-41-6)) These parameters vary widely both

between and among top- and subsoils so even for soils gathered at the same location it

would be difficult to choose a particular Qmax or kl which would be representative

While no real trends were apparent regarding soil collection points (at each

individual location) additional analyses were performed regarding physiographic regions

major land resource areas and ecoregions Physiogeographic regions are based primarily

upon geology and terrain South Carolina has four physiographic regions the Southern

Blue Ridge Piedmont Uplands and the Sea Island section of the Coastal Plain The

Simpson Experiment Station (ES) is located in the Piedmont Uplands while the RECs

from which soils for this study were collected came from the Coastal Plain (USGS 2003)

In addition South Carolina has been divided into six major land resource areas

(MLRAs) the Blue Ridge Southern Piedmont Sand Hills Southern Coastal Plain

Atlantic Coast Flatwoods and Tidewater MLRAs are regions classified by relief

hydrologic units relief resource uses resource concerns and soil type Following this

classification scheme the Simpson ES would be alone in the Southern Piedmont MLRA

the Sandhill REC would be alone in the Sand Hills MLRA the Edisto and Pee Dee RECs

would be in the Southern Coastal Plain MLRA and the Coastal REC would be alone in the

Tidewater MLRA (USDA-NRCS 2006)

A similar spatial classification scheme is the delineation of ecoregions Ecoregions

are areas which are ecologically similar They are based upon both biotic and abiotic

parameters including geology physiography soils climate hydrology plant and animal

biology and land use There are four levels of ecoregions Levels I through IV in order of

increasing resolution South Carolina has been divided into five large Level III ecoregions

Blue Ridge (66) Piedmont (45) Southeastern Plains (65) Middle Atlantic Coastal Plain

(63) and Southern Coastal Plan (75) Level IV ecoregions are smaller geographically than

the Level III ecoregions they are denoted by an alphabetic suffix following their numeric

Level III identifiers South Carolina RECs are located in Level IV ecoregions as follows

Simpson ES is located in the Southern Outer Piedmont (45b) the Sandhill REC is in the

The Sand Hills (65c) the Edisto and Pee Dee RECs are in the Atlantic Southern Loam

Plains (65l) and the Coastal REC is in the Sea IslandsCoastal Marsh It should be noted

that in terms of Level III ecoregions Simpson ES is located in the Piedmont Sandhill

Edisto and Pee Dee RECs are in the Southeastern Plains and the Coastal REC is in the

Southern Coastal Plain (Griffith et al 2002)

Isotherms and isotherm parameters do not appear to be well-modeled

geographically (Tables 7-3 through 7-7 below) for the soils whose adsorption

characteristics were detectable While this is disappointing it should probably not be

surprising as MLRAs proved to be a poor indicator of soil texture for these same sampled

soils in a previous study (Price 1994) While other researchers (Van Der Zee et al 1988)

found less variability among adsorption isotherm parameters their work focused on

smaller areas and included more samples

Regardless of grouping technique a few observations may be made

1) The Simpson ES soils have more PO4 adsorption capacity than any other soils

analyzed Any geography-based isotherm approach would need to take this into

account

2) The Edisto and Pee Dee REC topsoils are essentially the same with respect to

adsorption capacity

3) The greatest difference regarding adsorption capacity between the Sandhill REC

soils and the Edisto and Pee Dee REC soils are found in the subsurface where the

Sandhill REC soils had a lower capacity

Table 7-3 Isotherm Parameters and Error Associated with Location Groupings of Soils Qmax (mgPO4 kgsoil

-1) plusmn Standard Error (SE)

kl (L mgPO4-1)

plusmn SE R2

Simpson ES All soils (5) Topsoils only (2) Subsoils only (3)

112121 plusmn 22298 42377 plusmn 4613

163477 plusmn 21446

020 plusmn 018 017 plusmn 0084 037 plusmn 024

033 082 064

Sandhill REC All soils (6) Topsoils only (3) Subsoils only (3)

Does Not Converge (DNC)

39223 plusmn 7707 22739 plusmn 4635

022 plusmn 019 178 plusmn 137

DNC 049 056

Edisto REC All soils (5) Topsoils only (2) Subsoils only (3)

020 076 091

Pee Dee REC All soils (4) Topsoils only (2) Subsoils only (2)

059 089 068

Coastal REC All soils (3) Topsoils only (2) Subsoils only (1)

65183 plusmn 8336 52156 plusmn 6613

101007 plusmn 15693

076 080 094

Table 7-4 Isotherm Parameters and Error Associated with Physiographic Groupings of Soils Qmax (mgPO4 L-1) plusmn

Standard Error (SE) kl (L mgPO4

-1) plusmn SE R2

Piedmont Uplands All soils (5) Topsoils only (2) Subsoils only (3)

112121 plusmn 22298 42377 plusmn 4613

163478 plusmn 21446

020plusmn 018

033 082 064

Coastal Plain All soils (18) Topsoils only (9) Subsoils only (9)

42706 plusmn 4020 63977 plusmn 8640

DNC 062 036

Table 7-5 Isotherm Parameters and Error Associated with MLRA Groupings of Soils Qmax (mgPO4 kgsoil

kl (L mgPO4-1) plusmn

Southern Piedmont All soils (5) Topsoils only (2) Subsoils only (3)

112121 plusmn 22298 42377 plusmn 4613

163477 plusmn 21446

033 082 064

Sand Hills All soils (6) Topsoils only (3) Subsoils only (3)

39223 plusmn 7707 22739 plusmn 4635

DNC 049 056

Southern Coastal Plain All soils (9) Topsoils only (4) Subsoils only (5)

50732 plusmn 9673 28912 plusmn 2397

83304 plusmn 13190

023 076 051

Tidewater All soils (3) Topsoils only (2) Subsoils only (1)

65183 plusmn 8336 52156 plusmn 6613

101007 plusmn 15693

076 080 094

Table 7-6 Isotherm Parameters and Error Associated with Level III Ecoregion Groupings of Soils Qmax (mgPO4 kgsoil

Piedmont All soils (5) Topsoils only (2) Subsoils only (3)

112121 plusmn 22298 42377 plusmn 4613

163478 plusmn 21446

033 082 064

Southeastern Plains All soils (15) Topsoils only (7) Subsoils only (8)

60697 plusmn 11735 35434 plusmn 3746

DNC 027 058

65183 plusmn 8336 52156 plusmn 6613

101007 plusmn 15693

076 080 094

Table 7-7 Isotherm Parameters and Error Associated with Level IV Ecoregion Groupings of Soils Qmax (mgPO4 kgsoil

kl (L mgPO4

-1) plusmn SE

Southern Outer Piedmont All soils (5) Topsoils only (2) Subsoils only (3)

112121 plusmn 22298 42377 plusmn 4613

163478 plusmn 21446

033 082 064

Does Not Converge

(DNC) 39223 plusmn 7707 22739 plusmn 4635

DNC 049 056

Atlantic Southern Loam Plains All soils (9) Topsoils only (4) Subsoils only (5)

50732 plusmn 9673 28912 plusmn 2397

83304 plusmn 13190

023 076 051

Sea IslandsCoastal Marsh All soils (3) Topsoils only (2) Subsoils only (1)

65183 plusmn 8336 52156 plusmn 6613

101007 plusmn 15693

076 080 094

4) Regarding affinity constant the Sandhill and Pee Dee REC soils are similar with

lower constants than the Edisto REC soils

5) All soils whose adsorption characteristics were so weak as to be undetectable came

from the Sandhill and Edisto RECs (the Lakelands Topsoil from the Sandhill REC

and the Varina Topsoil Fuquay Topsoil and Subsoil and Blanton Topsoil and

Subsoil all of the Edisto REC) so these regions appear to have the

weakest-adsorbing soils

6) Coastal REC soils demonstrated a greater adsorption capacity than the soils from

the Sandhill Edisto or Pee Dee RECs while affinity constants were low

In addition it should be noted that while error is high for geographic groupings of

isotherm parameters in general especially for the affinity constant it is not dramatically

worse than error for isotherm parameters of individual soils (Appendix A Table A-41)

This is encouraging Least squares fitting of the grouped data regardless of grouping is

not as strong as would be desired but it is not dramatically worse for the various groupings

than among soils taken from the same location This indicates that with the exception of

soils from the Piedmont variability and isotherm parameters among other soils in the state

are similar perhaps existing on something approaching a continuum so long as different

isotherms are used for topsoils versus subsoils

Making engineering estimates from these groupings is a different question

however While the Level IV ecoregion and MLRA groupings might provide a reasonable

approach to predicting isotherm parameters this study did not include soils from every

ecoregion or MLRA so it would be difficult to base an estimate upon them While the data

do not indicate a strong geographic basis for phosphate adsorption in the absence of

location-specific data it would not be unreasonable for an engineer to select average

isotherm parameters as set forth above for the Piedmont (Simpson ES data) versus the rest

of the state based upon location and proximity to the non-Piedmont sample locations

presented here

Predicting Isotherm Parameters Based on Soil Characteristics

Experimentally-determined isotherm parameters were related to soil characteristics

both experimentally determined and those taken from the literature by linear and

multilinear least squares regressions using Microsoft Excel 2007 and MiniTab15 The

confidence interval was set to 95 a characteristicrsquos significance was indicated by

p lt 005

Predicting Qmax

Given previously-documented correlations between Qmax and soil SSA texture

OM content and Fe and Al content each measure was investigated as part of this project

Characteristics measured included SSA clay content OM content Cb content FeDCB and

FeMe-1 and AlDCB and AlMe-1 A table providing regression statistics is presented below

(Table 7-8) for Qmax Given the obvious importance of FeDCB and the weak importance of

the commonly-available FeMe-1 these factors point to a potentially-important finding

indicated by the relationship between OM Cb and FeDCB (Figure 6-3 and Table 6-3)

while FeDCB is a fairly unusual measurement OM and Cb content are inexpensive analyses

($5 and $10 respectively for in-state samples at the CU ASL 2009) and readily available

allowing for the approximation of FeDCB This relationship is defined by the equation

Estimated 632103927526 minusminus= bDCB COMFe (7-1)

where FeDCB is presented in mgPO4 kgSoil

-1 and OM and Cb are expressed as percentages A

correlation is also presented for this estimated FeDCB concentration and Qmax Finally

given the strong correlations demonstrated for the SSA clay OM and FeDCB parameters

sum and product terms were also evaluated

Correlation results indicate that SSA is the best indicator of Qmax (Table 7-8)

Other significant predictors of Qmax were FeDCB and AlDCB confirming the observations of

Wiriyakitnateekul et al (2005) and OM Following a step-wise regression strategy fit

improves most when OM or FeDCB (Figure 7-2) are also included with little difference

between the two indicating the importance of OM-associated Fe regarding PO4 adsorption

Various surrogates were also evaluated including Cb in lieu of OM and clay content in lieu

of surface area Cb is not a strong indicator of Qmax indicating again that the OM fraction

most important for predicting Qmax is OM-associated Fe Clay content is an effective

although slightly less-robust surrogate for SSA as a predictor of Qmax Estimated FeDCB is

an effective surrogate for measured FeDCB although the need for either parameter is

questionable given the strong relationships regarding surface area or texture and organic

matter (which is predominantly composed of Fe as previously discussed) as predictors of

y = 09997x + 00687R2 = 08789

0 500 1000 1500 2000 2500

Predicted Qmax (mg-PO4kg-Soil)

Figure 7-2 Predicted Qmax Using SSA and FeDCB Content vs Measured Qmax

Table 7-8 Relationship of Phosphate Adsorption Capacity (Qmax) to Other Soil Characteristics Regression

Significance Coefficient(s) plusmn Standard Error

(SE) y-intercept plusmn SE R2 Adj R2

SSA (m2 g-1) 228E-09 4398 plusmn 444 13536 plusmn 7642 082 081 Clay () 248E-04 3637 plusmn 826 10106 plusmn 160931 048 045 Organic Matter (OM ) 224E-05 21811 plusmn 4025 20057 plusmn 120681 058 056 Carbon Content (Cb ) 853E-011 -4775 plusmn 25400181 71842 plusmn 19184 0002 000 FeMe-1 (mg kgSoil

-1) 908E-02 -094 plusmn 0531 98159 plusmn 20051 013 0089 FeDCB (mg kgSoil

-1) 181E-06 0032 plusmn 00048 36654 plusmn 8669 067 065 Estimated FeDCB (mg kgSoil

-1) 301E-06 0031 plusmn 00049 33965 plusmn 9165 065 064 AlMe-1 (mg kgSoil

-1) 738E-011 0066 plusmn 0201 61206 plusmn 26295 00050 00000 AlDCB (mg kgSoil

-1) 307E-05 020 plusmn 0039 18682 plusmn 125151 057 055 Soil pH (SU) 409E-011 21358 plusmn 253661 -35663 plusmn 1249971 003 000 SSA OM 679E-10 3442 plusmn 492

8760 plusmn 29031 5917 plusmn 69651 088 087

SSA FeDCB 680E-10 3207 plusmn 546

0013 plusmn 00043 15113 plusmn 6513 088 087

SSA OM FeDCB

474E-09 3241 plusmn 552

4720 plusmn 56611 00071 plusmn 000851

10280 plusmn 87551 088 086

SSA OM FeDCB AlDCB

284E-08

3157 plusmn 572 5221 plusmn 57801

00037 plusmn 000981 0028 plusmn 00391

6868 plusmn 100911 088 086

SSA Cb 126E-08 4499 plusmn 443

14028 plusmn 106531 4150 plusmn 103551 084 082 1 p gt 005

Table 7-8 (Continued) Relationship of Phosphate Adsorption Capacity (Qmax) to Other Soil Characteristics

Regression Significance

SSA Cb FeDCB

317E-09 3337 plusmn 549

11386 plusmn 91251 0013 plusmn 0004

7431 plusmn 88981 089 087

SSA-FeDCB Product 214E-08 000087 plusmn 000010 46837 plusmn 6313 078 077 SSA-Estimated FeDCB 482E-08 000080 plusmn 0000097 47192 plusmn 6554 077 075 SSA-OM Product 129E-08 595 plusmn 066 43328 plusmn 6325 079 078 Clay OM 451E-07 2450 plusmn 613

16634 plusmn 3338 -8036 plusmn 116001 077 074

Clay FeDCB 289E-07 1991 plusmn 638

0024 plusmn 00047 11852 plusmn 107771 078 076

Clay OM FeDCB

130E-06 2113 plusmn 653

7249 plusmn 77631 0015 plusmn 00111

3268 plusmn 141911 079 075

Clay- FeDCB Product 341E-08 000104 plusmn 0000124 43404 plusmn 6636 077 076 Clay-Estimated FeDCB Product 190E-08 00010 plusmn 000012

41984 plusmn 6520

078 077

Clay-OM Sum 187E-05 3717 plusmn 676 444 plusmn 14786 059 057 Clay-OM Product 520E-09 757 plusmn 080 35208 plusmn 6481 081 080

1 p gt 005

Initial Qmax distribution was nearly log-normal (Figures 7-3 and 7-4) However

normalizing by experimentally-determined values for SSA and FeDCB induced a

nearly-equal result for normalized Qmax values indicating the effectiveness of this

approach Outliers are the Coxville subsoil and Norfolk topsoil (Figures 7-5 and 7-6)

Applying the predictive equation based on the SSA and FeDCB regression produces a

log-normal distribution (Figures 7-7 and 7-8) lending further support to this approach

Normalizing experimental data for Qmax by clay content (Figures 7-9 and 7-10) and by clay

and OM content (Figures 7-11 and 7-12) yields a similar log-normal distribution Plots of

isotherms developed using these alternate normalizations are included in Appendix A

(Figures A-51-37)

Figure 7-3 Dot Plot of Measured Qmax

280024002000160012008004000

Qmax (mg-PO4kg-Soil)

Figure 7-4 Histogram of Measured Qmax

Figure 7-5 Dot Plot of Measured Qmax Normalized by SSA and FeDCB

0002000015000100000500000

Qmax (mg-PO4kg-Soilm^2mg-Fe)

Figure 7-6 Histogram of Measured Qmax Normalized by SSA and FeDCB

Figure 7-7 Dot Plot of Qmax as Predicted by Regression Equation for SSA and FeDCB

25002000150010005000

Qmax-Predicted (mg-PO4kg-Soil)

Figure 7-8 Histogram of Qmax as Predicted by Regression Equation for SSA and FeDCB

Figure 7-9 Dot Plot of Measured Qmax Normalized by Clay

120009000600030000

Qmax (mg-PO4kg-Clay)

Figure 7-10 Histogram of Measured Qmax Normalized by Clay

Figure 7-11 Dot Plot of Measured Qmax Normalized by Clay and OM

15000120009000600030000

Qmax (mg-PO4kg-Claykg-OM)

Figure 7-12 Histogram of Measured Qmax Normalized by Clay and OM

Predicting kl

Soil characteristics were analyzed to determine their predictive value for the

isotherm parameter kl (Table 7-9) The same characteristics that showed promise for

predicting Qmax were evaluated Unfortunately none of them provided a strong prediction

for kl only clay content (Figure 7-13) was significant at the 95 confidence level

Table 7-9 Relationship of the Affinity Constant (kl) to Other Soil Characteristics Regression

Significance Coefficient(s) plusmn

Standard Error (SE) y-intercept plusmn SE R2 Adj R2

SSA (m2 g-1) 414E-011 113E-02 plusmn 136E-021 569E-01 plusmn 234E-01 0032 000 Clay () 322E-02 3072E-02 plusmn 136E-02 214E-01 plusmn 234E-011 020 016 Organic Matter (OM ) 545E-011 -497E-02 plusmn 808E-021 824E-01 plusmn 242E-01 0018 000 Carbon Content (Cb ) 228E-011 -398E-01 plusmn 321E-011 942E-01 plusmn 242E-01 0068 0024 FeMe-1 (mg kgSoil

-1) 249E-011 -854E-04 plusmn 721E-041 976E-01 plusmn 272E-01 0063 0018 FeDCB (mg kgSoil

-1) 728E-011 -386E-06 plusmn 110E-051 752E-01 plusmn 197E-01 00059 000 Estimated FeDCB (mg kgSoil

-1) 829E-011 -237E-06 plusmn 108E-051 739E-01 plusmn 204E-01 00023 000

AlMe-1 (mg kgSoil-1) 577E-011 -144E-04 plusmn 254E-041 -144E-04 plusmn 2570E-04 0015 000

AlDCB (mg kgSoil-1) 526E-011 -4911E-05 plusmn 762E-051 833E-01 plusmn 247E-01 0019 000

Soil pH (SU) 737E-011 -115E-01 plusmn 336E-011 127E00 plusmn 166E001 001 000 SSA OM 229E-011 286E-02 plusmn 172E-021

-1581E-01 plusmn 1013E-11 706E-01 plusmn 243E-01 014 0051

SSA FeDCB 276E-011 311E-02 plusmn 192E-021

SSA OM FeDCB

406E-011 302E-02 plusmn 196E-021

126 E-01 plusmn 201 E-01 1 -568E-06 plusmn 300E-051

671E-01plusmn 311E-01 014 00026

SSA OM FeDCB AlDCB

403E-011

347E-02 plusmn 199E-02 1 -152E-01 plusmn 2015E-01 1

123E-05 plusmn 342E-051 -147E-04 plusmn 135E-041

853E-01 plusmn 352E-01 019 0012

SSA Cb 404E-011 871E-03 plusmn 137E-021

-362 E-01 plusmn 331 E-01 1 811E-01 plusmn 321E-01 0087 000 1 p gt 005

Table 7-9 (Continued) Relationship of the Affinity Constant (kl) to Other Soil Characteristics Regression

SSA C FeDCB

325E-011 274E-02 plusmn 196E-021 -319E-01 plusmn 326E-011 -2025E-05 plusmn 154E-051

758E-01 plusmn 318E-01 016 0031

SSA-FeDCB Product 911E-011 -319E-08 plusmn 282E-071 720E-01 plusmn 177E-01 00006 000

SSA-Estimated FeDCB Product 973E-011 894E-09 plusmn 262E-071 710E-01 plusmn 177E-01 00001 000

SSA-OM Product 969E-011 760E-05 plusmn 191E-031 709E-01 plusmn 182E-01 00001 000

Clay OM 240E-02 403E-02 plusmn 138E-02

-135E-01 plusmn 752E-01 1 361E-01 plusmn 261E-011 031 024

Clay FeDCB 212E-02 443E-02 plusmn 146E-02

Clay Estimated FeDCB 292E-02 430E-02 plusmn 148E-02

Clay OM FeDCB

253E-01 plusmn 332E-011 034 021

Clay- FeDCB Product 987E-011 551E-09 plusmn339E-071 711E-01 plusmn 182E-01 000 000

Clay- Estimated FeDCB Product 855E-011 605E-08 plusmn 327E-071 696E-01 plusmn 184E-01 00016 000

Clay-OM Sum 657E-021 247E-02 plusmn 127E-02 1 256E-01 plusmn 278E-011 015 000

Clay-OM Product 849E-011 463E-04 plusmn 240E-031 691E-01 plusmn 194E-01 00018 000 1 p gt 005

y = 09999x - 2E-05R2 = 02003

0 02 04 06 08 1 12 14 16Predicted kl (Lmg)

Figure 7-13 Predicted kl Using Clay Content vs Measured kl

While none of the soil characteristics provided a strong correlation with kl it is

interesting to note that in this case clay was a better predictor of kl than SSA This

indicates that regarding kl clay is more than a surrogate for SSA and that clay-associated

characteristics other than surface area drive kl Multilinear regressions for clay and OM

and for clay and FeDCB (Figure 7-14) improved the fit marginally indicating that clay in

association with OM and FeDCB drives kl but regression equations developed for these

parameters indicated that the additional coefficients were not significant at the 95

confidence level (however they were significant at the 90 confidence level) Given the

fact that organically-associated iron measured as FeDCB seems to make up the predominant

fraction of OM (Figure 6-3 and Table 6-3) it seems likely that OM is serving as a surrogate

for FeDCB in this case While neither multilinear regressions with OM nor with FeDCB

provide a particularly robust model for kl it is perhaps noteworthy that the economical and

readily-available OM measurement is almost equally effective in predicting kl

Further investigation demonstrated that kl is not normally distributed but is instead

collected into three primary groups (Figures 7-15 and 7-16) with two high outliers (Pelion

and Rembert subsoils) This called into question the regression approach just described so

an investigation into common characteristics for soils in the three groups was carried out

Ultimately the best approach seemed to be to break the soils into top- and subsoil groups

(Figures 7-17 through 7-20) This reduced the grouping considerably especially among

subsoils

y = 10005x + 4E-05R2 = 03198

0 05 1 15 2 25

Predicted kl (Lmg)

Figure 7-14 Predicted kl Using Clay Content and FeDCB vs Measured kl

Figure 7-15 Dot Plot of Measured kl For All Soils

3530252015100500

kL (Lmg-PO4)

Figure 7-16 Histogram of Measured kl For All Soils

Figure 7-17 Dot Plot of Measured kl For Topsoils

0806040200

Figure 7-18 Histogram of Measured kl For Topsoils

Figure 7-19 Dot Plot of Measured kl for Subsoils

3530252015100500

Figure 7-20 Histogram of Measured kl for Subsoils

Both top- and subsoils are nearer a log-normal distribution after treating them

separately however there is still some noticeable grouping among topsoils Unfortunately

the data describing soil characteristics do not have any obvious breakpoints and soil

taxonomy (Appendix B) also offers no obvious explanation Further regression analysis of

topsoil groups reveals that the lower kl group is more strongly correlated with OM than the

higher kl group which is more strongly correlated with FeDCB content However the cause

of the topsoil groupings is not clear based on the data Since FeDCB has been shown to be a

major component of OM the FeDCB fraction of OM was also determined and evaluated for

the presence of breakpoints which might explain the kl grouping none were evident

Tables 7-10 through 7-13 contain regression statistics describing this effort but generally

the confidence levels associated with these regressions are less than 95

Table 7-10 kl Regression Statistics All Topsoils

Signif Coefficient plusmn

Standard Error (SE)

Intercept plusmn SE R2 Adj R2

SSA (m2 g-1) 0491 138E-2plusmn191E-11 407E-1plusmn134E-1 0055 0000 Clay () 0581 208E-2plusmn362E-21 157E-1plusmn304E-11 0035 0000 FeDCB (mg kg-1) 0601 -567E-6plusmn106E-51 359E-1 plusmn 975E-2 0031 0000 OM () 0901 -926E-3plusmn745E-21 344E-1plusmn168E-11 0002 0000 SSA-FeDCB Product 0461 -778E-7plusmn102E-61 365E-1plusmn896E-2 0061 0000

Clay-OM Product 0881 122E-3plusmn777E-31 305E-1plusmn150E-11 0003 0000

Clay FeDCB 0721 249E-2plusmn381E-21

-693E-6plusmn14E-51 164E-1plusmn315E-11 0080 0000

Clay OM 0851 218E-2plusmn387E-21

-155E-2plusmn782E-21 179E-1plusmn341E-11 0040 0000 1 p gt 005 Table 7-11 Regression Statistics Low kl Topsoils

Standard Error (SE)

SSA (m2 g-1) 0591 -171E0plusmn295E01 481E1plusmn237E11 0063 0000 Clay () 0771 -188E0plusmn607E01 521E1plusmn510E11 0019 0000 FeDCB (mg kg-1) 0181 523E-6plusmn333E-61 124E-1plusmn379E-2 033 020 OM () 004 593E-2plusmn216E-2 -317E-2plusmn531E-21 060 052 SSA-FeDCB Product 0241 467E-7plusmn348E-71 132E-1plusmn382E-21 026 012

441E-6plusmn335E-61 246E-2plusmn983E-21 048 023

Clay OM 004 -273E0plusmn455E01

238E1plusmn107E1 513E0plusmn436E11 056 034 1 p gt 005

Table 7-12 Regression Statistics High kl Topsoils

Standard Error (SE)

SSA (m2 g-1) 0931 534E-3plusmn526E-21 581E-1plusmn256E-11 0005 0000 Clay () 0711 174E-2plusmn408E-21 462E-1plusmn343E-11 0083 0000 FeDCB (mg kg-1) 0131 650E-5plusmn266E-51 440E-1plusmn758E-2 075 062

OM () 0251 754E-2plusmn465E-21 487E-1plusmn852E-2 057 035 SSA-FeDCB Product 0071 738E-5plusmn371E-61 148E-1plusmn486E-2 087 081

Clay OM 0661 -166E-4plusmn430E-21

755E-2plusmn713E-21 336E-1plusmn239E-11 057 000 1 p gt 005

Table 7-13 kl Regression Statistics Subsoils

Standard Error (SE)

SSA (m2 g-1) 0701 -840E-3plusmn208E-21 122E0plusmn0476E-1 0016 000 Clay () 0651 124E-2plusmn263E-21 773E-1plusmn677E-11 0022 000 FeDCB (mg kg-1) 0391 -813E-2plusmn104E-11 125E0plusmn332E-1 0074 000

OM () 0451 -813E-2plusmn104E-11 127E0plusmn368E-1 0057 000 SSA-FeDCB Product 0451 -280E0plusmn356E01 119E0plusmn308E-1 0058 000

Clay-OM Product 0521 210E-3plusmn314E-31 121E0plusmn346E-1 0043 000

-135E-1plusmn118E-11 733E-1plusmn668E-11 015 000 1 p gt 005

Given the difficulties in predicting kl using soil characteristics another approach is

to use arithmetic means (xm) medians (m) trimmed means (xmacute) and medians (macute) and

interquartile ranges (IQR) for estimating kl Because the top- and subsoils are so obviously

different they are treated separately (Table 7-14)

Table 7-14 Descriptive Statistics for kl xm plusmn Standard

Deviation (SD) xmacute plusmn SD m macute IQR

Topsoil 033 plusmn 024 - 020 - 017-053

Subsoil 107 plusmn 090 091 plusmn 037 067 067 062-139

Because topsoil kl values fell into two groups only a median and IQR are provided

here Three data points were lower than the 25th percentile but they seemed to exist on a

continuum with the rest of the data and so were not eliminated More significantly all data

in the higher kl group were higher than the 75th percentile value so none of them were

dropped By contrast the subsoil group was near log-normal with two low and two high

outliers each of which were far outside the IQR These four outliers were discarded to

calculate trimmed means and medians but values were not changed dramatically Given

these data the arithmetic mean of kl = 033 would be preferred for use with topsoils and

the trimmed mean of kl = 091 would be preferred for use with subsoils

A comparison between the three methods described for predicting kl is presented in

Table 7-15 Values for kl predicted from regression equations for clay and for a multilinear

regression for clay and FeDCB were compared to actual values of kl as predicted by the

3-Parameter method Also compared were the topsoil mean and subsoil trimmed mean

The clay- and FeDCB-predicted values for kl best fit the data but the topsoil and subsoil

estimates fit the data almost as well Values for kl derived from clay and estimated-FeDCB

derived from Cb and OM averaged only 3 difference from values based upon

experimental values of FeDCB

Table 7-15 Comparison of Predicted Values for kl

Highlighted boxes show which value for predicted kl was nearest the actual value

TopsoilSubsoil-Predicted kl Clay-Predicted kl Clay- and FeDCB-Predicted kl Soil Type Fit

kl Pred kl

Actual Real Variation

Pred kl

Appling Top 020 033 16377 6377 037 18391 8391 003 1295 8705 Madison Top 029 033 11454 1454 059 20647 10647 032 11156 1156 Madison Sub 064 091 14239 4239 116 18152 8152 041 6416 3584 Hiwassee Sub 069 091 13167 3167 095 13719 3719 071 10214 214 Cecil Sub 063 091 14451 4451 133 21057 11057 088 13975 3975 Lakeland Sub 014 091 63949 53949 048 33382 23382 054 37803 27803 Pelion Top 020 033 16484 6484 046 23113 13113 052 25923 15923 Pelion Sub 330 091 2760 7240 096 2904 7096 119 3621 6379 Johnston Top 019 033 17780 7780 052 28242 18242 061 32781 22781 Johnston Sub 138 091 6601 3399 078 5630 4370 098 7116 2884 Vaucluse Top 060 033 5485 4515 053 8764 1236 059 9871 129 Varina Sub 141 091 6456 3544 036 2520 7480 035 2492 7508 Rembert Top 055 033 5948 4052 048 8728 1272 057 10291 291 Rembert Sub 204 091 4469 5531 151 7402 2598 194 9527 473 Dothan Top 005 033 61567 51567 044 82319 72319 045 83856 73856 Dothan Sub 064 091 14292 4292 091 14264 4264 092 14388 4388 Coxville Top 051 033 6481 3519 040 7760 2240 042 8250 1750 Coxville Sub 061 091 14930 4930 072 11775 1775 092 15080 5080 Norfolk Top 016 033 21140 11140 040 25511 15511 046 29301 19301 Norfolk Sub 125 091 7268 2732 105 8357 1643 132 10517 517 Wadmalaw Top 076 033 4354 5646 046 6105 3895 047 6213 3787 Wadmalaw Sub 007 091 126336 116336 110 152098 142098 136 189064 179064 Yonges Top 007 033 44655 34655 044 60122 50122 042 56220 46220 AVERAGE 071 063 21767 15261 071 25259 18027 071 25886 18946

Predicting Q0

Previously-adsorbed phosphate (Q0) itself is not likely to be important for most

modeling applications but depending on the site Q0 might actually be the most

environmentally-significant parameter as it is possible that an eroded soil particle might

not encounter any additional P during transport With this in mind the different techniques

for measuring or estimating Q0 are further considered here This study has previously

reported (Table 6-6) that various measures for Q0 are not well-correlated with one another

with PO4DCB greatly exceeding PO4Me-1 and PO4H2O Desorbed Here correlations are

presented between these three measures and Q0 estimated using the 3-parameter isotherm

technique (Table 7-16)

Table 7-16 Relationship of the Previously-Adsorbed Phosphate (Q0) Found By the 3-Parameter Isotherm Technique to Other Measures of Soil Phosphorus

y-intercept plusmn SE R2

PO4DCB (mg kgSoil

-1) 519E-04 0072 plusmn 0018 263 plusmn1781 044

PO4Me-1 (mg kgSoil

-1) 158E-011 037 plusmn 0261 3373 plusmn 560 0093

PO4H2O Desorbed (mg kgSoil

-1) 535E-021 081 plusmn 0401 1363 plusmn 0241 017

1 p gt 005

Admittedly Q0 has been estimated by extrapolation but it is perhaps noteworthy

that there are stronger correlations between it PO4DCB and PO4H2O Desorbed than among any

of the three experimentally-determined values If PO4DCB is thought of as the released PO4

which had previously been adsorbed to the soil particle as both the result of fast and slow

adsorption reactions as described previously it is reasonable that Q0 would be less

because Q0 is extrapolated from data developed in a fairly short-term experiment which

would likely only involve the outer surfaces of a soil particle as a result of fast adsorption

reactions This observation lends credence to the concept of Q0 extrapolated from

experimental adsorption data as part of the 3-parameter isotherm technique at the very

least it supports the idea that this approach to deriving Q0 is reasonable However in

general it seems that the most important observation here is that PO4DCB provides a good

measure of the amount of phosphate which could be released from PO4-laden sediment

under reducing conditions

Alternate Normalizations

Given the relationship between SSA clay OM and FeDCB additional analyses

focused on normalizing PO4 adsorption by these parameters rather than simply by soil in

the hope that controlling one of these parameters might collapse the wide-ranging data

spread of the statersquos soils into a narrower ldquobandrdquo Note that this exercise does not change kl

These isotherms are presented in Appendix A (Figures A-51-24)

Values for soil-normalized Qmax across the state were separated by a factor of about

14 Normalizations based on SSA and FeDCB resulted in greater spread with values for

Qmax spread by factors of about 29 and 18 respectively Normalizations based on clay and

OM resulted in less spread with values for Qmax spread by factors of about 9 and 10

respectively It is interesting to note that SSA and FeDCB were strongly correlated with an

individual soilrsquos Qmax yet that neither of them result in tightening of the data when these

normalizations are pursued across the state This seems to indicate that a parametersrsquo

significance in predicting Qmax varies across the state but that the surrogate parameters

clay and OM whose significance is derived from a combination of both SSA and FeDCB

content account for these regional variations rather well However neither parameter

results in significantly-greater improvements on a statewide basis so the attempt to

develop a single statewide isotherm whether normalized by soil or another parameter is

futile

While these alternate normalizations do not result in a significantly narrower

spread on a statewide basis some of them do result in improved spreads when soils are

analyzed with respect to collection location In particular it seems that these

normalizations result in improvements between topsoils and subsoils as it takes into

account enrichment ratios of clay and FeDCB of otherwise identical soils due to vertical

leaching through the soil column (Table 7-17 and Appendix A Figures A-57-18) While

kl does not change with the alternate normalizations a similar table showing kl variation

among the soils at the various locations is provided (Table 7-18) it is disappointing that

there is not more similarity with respect to kl even among soils at the same basic location

However according to this approach it seems that measurements of soil texture SSA and

clay content are most significant for predicting kl This is in contrast to the findings in the

previous section which indicated that OM and FeDCB seemed to be the most important

measurements for kl among topsoils only this indicates that kl among subsoils is largely

dependent upon soil texture

Another similar approach involved fitting all adsorption data from a given location

at once for a variety of normalizations Data derived from this approach are provided in

Table 7-19 for Qmax and Table 7-20 for kl This approach was generally more satisfying

but the result is basically the same SSA and clay content are the most-significant but not

the only factors in driving PO4 adsorption

Table 7-17 Qmax Variation Based on Location and Alternate Normalizations

Soil-Normalized (mgPO4 kgsoil

-1) SSA-Normalized

(mgPO4 m -2) Clay-Normalized

(mgPO4 kgclay-1)

FeDCB-Normalized (mgPO4 g FeDCB

-1) OM-Normalized (mgPO4 kgOM

-1) Statewide (23) Average Standard Deviation MaxMin Ratio

6908365 5795240 139204

01023 01666

292362

47239743 26339440

2122975 2923030 182166

432813645 305008509

104260 Simpson ES (5) Average Standard Deviation MaxMin Ratio

12025025 9373473 68248

00506 00080 15466

55171775 20124377

308938 111975 23568

207335918 89412290

32527 Sandhill REC (6) Average Standard Deviation MaxMin Ratio

3138355 1924539 39182

00963 00500 39547

28006554 21307052

1486587 1080448 49355

329733738 173442908

43253 Edisto REC (5) Average Standard Deviation MaxMin Ratio

7768883 4975063 52744

006813 005646 57377

58805050 29439252

1997150 1250971 41909

440329169 243586385

40420 Pee Dee REC (4) Average Standard Deviation MaxMin Ratio

4750009 2363103 29112

02530 03951

210806

40539490 13377041

6091098 5523087 96534

672821765 376646557

67874 Coastal REC (3) Average Standard Deviation MaxMin Ratio

7280896 3407230 28899

00567 00116 15095

62144223 40746542

1338023 507435 22600

682232976 482735286

78843 Highlighted boxes show adsorption normalizations which provided the best fit for a given REC

Table 7-18 kl Variation Based on Location Statewide Simpson Sandhill Edisto Pee Dee Coastal Average Standard Deviation MaxMin

07120 07577 615075

04899 02270 34298

09675 12337 231680

09382 07823 379869

06317 04570 80211

03013 03955 105234

Table 7-19 Qmax Regression Based on Location and Alternate Normalizations Soil-Normalized

(mgPO4 kgsoil -1)

SSA-Normalized (mgPO4 m -2)

Clay-Normalized (mgPO4 kgclay

-1) FeDCB-Normalized (mgPO4 kg FeDCB

-1) Statewide (23) R2 Qmax Standard Error

8307397 1024031

762687 97552

47158328 3041768

1813041124 342136497

346936330 33846950

Simpson ES (5) R2 Qmax Standard Error

11212101 2229846

480451 36385

50936814 4850656

289659878 31841167

195451505 23582865

Sandhill REC (6) R2 Qmax Standard Error

Does Not

Converge

1183646 127918

51981534 13940524

1887587339 391509054

275513445 43206610

Edisto REC (5) R2 Qmax Standard Error

5395111 1465128

452512 57585

43220092 5581714

1451350582 366515856

232031738 52104937

Pee Dee REC (4) R2 Qmax Standard Error

16129920 8180493

1588063 526368

35019815 2259859

5856020183 1354799083

780034549 132351757

Coastal REC (3) R2 Qmax Standard Error

6518327 833561

517508 63723

56970390 9851811

1011935510 296059587

648190378 148138015

Highlighted boxes show adsorption normalizations which provided the best fit for a given REC

Table 7-20 kl Regression Based on Location and Alternate Normalizations

-1) SSA-Normalized

(mgPO4 kgclay-1)

FeDCB-Normalized (mgPO4 kg FeDCB

-1) Statewide (23) R2 kl Standard Error

02516 01316 00433

01967 07410 04442

05766 01669 00378

01165 10285 8539

02886 06252 02893

Simpson ES (5) R2 kl Standard Error

03325 01962 01768

07605 03023 01105

06722 02493 01117

06013 02976 01576

05583 02682 01539

Sandhill REC (6) R2 kl Standard Error

Does Not

Converge

07584 00972 00312

05295 00512 00314

04390 01162 00743

04938 12578 13723

Edisto REC (5) R2 kl Standard Error

02019 12689 17095

05625 05663 03273

06017 04107 02202

02302 04434 04579

01283 02257 01330

Pee Dee REC (4) R2 kl Standard Error

05917 00238 00188

01877 11594 18220

08530 04814 01427

03236 10004 12024

05793 15258 08817

Coastal REC (3) R2 kl Standard Error

07598 01286 00605

06749 02159 00995

06103 01487 00274

03986 01082 00915

05282 01053 00689

CHAPTER 8

CONCLUSIONS AND RECOMMENDATIONS

Phosphate adsorption isotherms were developed for 23 of the 29 soils evaluated in

this study Best fits were established using a novel non-linear regression fitting technique

and isotherm parameters Qmax kl and Q0 were developed accordingly Isotherm

parameters were not strongly related to geography as analyzed by REC physiographic

region MLRA or Level III and IV ecoregions While the data do not indicate a strong

geographic basis for phosphate adsorption in the absence of location-specific data it would

not be unreasonable for an engineer to select average isotherm parameters as set forth

above for the Piedmont (Simpson ES data) versus the rest of the state based upon location

and proximity to the non-Piedmont sample locations presented here

Isotherm parameter Qmax was strongly-related to SSA OM content and FeDCB

content Fits improved for various multilinear regressions involving these parameters and

clay which proved to be an effective surrogate for SSA OM was an effective surrogate for

FeDCB The best fit for predicting Qmax involved the product term of SSA and FeDCB but as

measurements of the surrogates clay and OM are more economical and are readily

available it is recommended that they be measured from site-specific samples as a means

of estimating Qmax

Isotherm parameter kl was only weakly predicted by clay content Multilinear

regressions including OM and FeDCB improved the fit but below the 95 confidence level

This indicates that clay in association with OM and FeDCB drives kl While sufficient

uncertainty persists even with these correlations they remain better indicators of kl than

geographic area

While isotherm parameter Q0 is dependent upon a particular sitersquos history it can be

predicted using the DCB method or the water-desorbed method in conjunction with

analysis by ICP-AES if so desired This parameter is not necessary for the purposes of

predicting isotherm behavior because it is included in the Qmax term for which previous

regressions were developed however should this parameter be of interest for another

application it is worth noting that the Mehlich-1 soil test did not prove effective A better

method for determining Q0 if necessary would be to use a total soil digestion

Alternate normalizations were not effective in producing an isotherm

representative of the entire state however there was some improvement in relating topsoils

and subsoils of the same soil type at a given location This was to be expected due to

enrichment of adsorption-related soil characteristics in the subsurface due to vertical

leaching and does not indicate that this approach was effective thus there were some

similarities between top- and subsoils across geographic areas Further the exercise

supported the conclusions of the regression analyses in general adsorption is driven by

soil texture relating to SSA although other soil characteristics help in curve fitting

Qmax may be calculated using SSA and FeDCB content given the difficulty in

obtaining these measurements a calculation using clay and OM content is a viable

alternative (Figure 8-1) Although significant uncertainties exist when predicting kl this

study indicated that the best method for predicting kl would involve site-specific

measurements of clay and FeDCB content The following equations based on linear and

multilinear regressions between isotherm parameters and soil characteristics clay and OM

expressed as percentages and FeDCB expressed as mgFe kgSoil-1 are recommended

08352)(577max += OMClayQ (8-1) 200000201004430 +minus= DCBl FeClayk (8-2)

Site-specific measurements of clay OM and Cb content are further commended by

the low costs associated with each analysis an OM analysis costs $5 and Cb analysis costs

$10 (CU ASL 2009) while a sieve and hydrometer test to determine clay content costs

approximately $140 (G Tedder Soil Consultants Inc personal communication

December 8 2009) This compares to approximate material and analysis costs of $350 per

soil for isotherm determination plus approximately 12 hours of labor from a laboratory

technician and approximately 30 days of laboratory overhead As Table 6-1 indicated soil

texture values from the literature are not a reliable indicator of site-specific texture or clay

content so a soil sample should be taken for both analyses While FeDCB content might not

be a practical parameter to determine experimentally it can easily be estimated using

equation 7-1 and known values for OM and Cb In this case the following equation should

be used to predict kl (Figure 8-2) with clay content expressed as a percentage of total soil

mass and FeDCB expressed as mgFe kgSoil-1

21900000178004300 +minus= DCBl eEstimatedFClayk (8-3) Due to uncertainty surrounding other predictions of kl using the values kl = 033 for

topsoils and kl = 091 for subsoils is a viable alternative for the sampled soils

R2 = 08095

0 500 1000 1500 2000 2500 3000

Figure 8-1 Predicted Qmax Based on the Clay-OM Product vs Measured Qmax

R2 = 02971

0 02 04 06 08 1 12 14 16 18 2Predicted kl (Lmg)

Figure 8-2 Predicted kl Based on Clay and Estimated FeDCB vs Measured kl

Extrapolating beyond the range of values found in this study is not advisable for

equations 8-1 through 8-3 or for the other regressions presented in this study Detection

limits for the laboratory analyses presented in this study and a range of values for which

these regressions were developed are presented below in Table 8-1

Table 8-1 Study Detection Limits and Data Range

Parameter Detection Limit Data Range Al 005 ppm 499 ndash 7655 ppm Cb 80 ppm 670 ndash 20900 ppm Fe 001 ppm 368 ndash 57433 ppm

OM 5 ppm 3000 ndash 89000 ppm P 1 ppm 049 ndash 865 ppm

Clay 2 34 - 421 SSA 1 m2 g-1 04 - 439 m2 g-1

Isotherms predicted using equations 8-1 and 8-3 are presented in Appendix A

Figures A-61-23 with 95 confidence levels and isotherms based on fits to raw data

while not always good predictors the predicted isotherms seldom underestimate Q

especially at low concentrations for C In the absence of site-specific adsorption data such

estimates may be useful especially as worst-case screening tools

Engineering judgments of isotherm parameters based on geography involve a great

deal of uncertainty and should only be pursued as a last resort in this case it is

recommended that the Simpson ES values be used as representative of the Piedmont and

that the rest of the state rely on data from the nearest REC

Final Recommendations

Site-specific measurements of adsorption isotherms will be superior to predicted

isotherms However in the absence of such data isotherms may be estimated based upon

site-specific measurements of clay OM and Cb content Recommendations for making

such estimates for South Carolina soils are as follows

bull To determine Qmax use equation 8-1 along with site-specific measurements of clay

and OM content

bull To determine kl use equation 8-3 along with site-specific measurement of clay

content and an estimated value for Fe content Fe content may be estimated using

equation 7-1 this requires measurement of OM and Cb

bull A viable alternative for estimating kl is to use kl = 033 for topsoils and kl = 091 for

subsoils

CHAPTER 9

RECOMMENDATIONS FOR FURTHER RESEARCH

A great deal of research remains to be done before a complete understanding of the

role of soil and sediment in trapping and releasing P is achieved Further research should

focus on actual sediments Such study will involve isotherms developed for appropriate

timescales for varying applications shorter-term experiments for BMP modeling and

longer-term for transport through a watershed If possible parallel experiments could then

track the effects of subsequent dilution with low-P water in order to evaluate desorption

over time scales appropriate to BMPs and watersheds Because eroded particles not parent

soils are the vehicles by which P moves through the watershed better methods of

predicting eroded particle size from parent soils will be the key link for making analysis of

parent soils a good indication of eroded soilsrsquo behavior Therefore this area of research

should also be pursued and strengthened Finally adsorption experiments based on

varying particle sizes will provide the link for evaluating the effects of BMPs on

P-adsorbing and transporting capabilities of sediments

A final recommendation involves evaluation of the utility of applying isotherm

techniques to fertilizer application Soil test P as determined using the Mehlich-1

technique is considered optimized for soil fertility at 20-25 mgP kgsoil-1 (Sims 2009)

Because isotherms relate P in solution to adsorbed P an isotherm could be used to better

estimate the amount of P which must be added to achieve a desired soil concentration (Psoil)

Thus isotherms could provide an advance over simple mass-based techniques for

determining fertilizer recommendations Low-concentration adsorption experiments could

be used to develop isotherm equations for a given soil The first derivative of this equation

at any desired point would provide an estimate of the ldquoefficiencyrdquo of fertilizer application

at that point up to the point of optimum Psoil (Q using the terminology in this study) After

initial development of the isotherm future fertilizer recommendations would require only a

mass-based soil test to determine the current Psoil and the isotherm could be used to

determine more-exactly the amount of P necessary to reach optimum soil concentrations

Application of isotherm techniques to soil testing and fertilizer recommendations could

potentially prevent over-application of P providing a tool to protect the environment and

to aid farmers and soil scientists in avoiding unnecessary costs associated with

over-fertilization

APPENDICES

Appendix A

Isotherm Data

Containing

1 Adsorption Experiment Results 2 Data Comparing 1- and 2-Surface Isotherm Models 3 3-Parameter Isotherms 4 Alternate Normalizations Based on the 3-Parameter Isotherm 5 Predicted vs Fit Isotherms

A-1 Adsorption Experiment Results

Table A-11 Appling Topsoil Phosphate Adsorption

C Q Adsorbed mg L-1 mg kg-1

036 -7296 040 -7944 110 -1241 103 102 049 464 9611 938 460 9316 919 961 14881 719 968 12081 588

2964 18943 310 2836 25073 423 7785 5513 0351 7813 26795 169

1 Stray data points displaying less than 2

adsorption were discarded for isotherm fitting

Table A-12 Madison Topsoil Phosphate Adsorption

027 -5416 025 -5027 066 7487 3619 072 6147 2987 415 19014 1864 414 17368 1734 906 25718 1242 894 -48369 -37061

2825 45962 752 2868 35077 576 7737 36746 232 7691 37046 235

Table A-13 Madison Subsoil Phosphate Adsorption

025 -5106 018 -3674 021 16492 7997 024 15926 7691 078 88623 8505 070 86319 8608 354 137802 6602 358 135351 6534

2019 205371 3365 2046 195252 3224 6759 244047 1526 6740 237941 1497

Table A-14 Hiwassee Subsoil

Phosphate Adsorption C Q Adsorbed

mg L-1 mg kg-1 024 -4703 026 -5158 030 14740 7129 031 14180 6937 278 45583 4503 285 45064 4415 741 57479 2794 733 58327 2847

2151 67179 1350 2128 69463 1403 3661 73471 912 3677 69605 865

Table A-15 Cecil Subsoil Phosphate Adsorption

021 -4249 020 -3990 023 16088 7786 025 15592 7538 112 76981 7740 115 77064 7706 464 113813 5508 466 112368 5464

2283 155980 2546 2267 156902 2571 7011 178738 1130 6991 187998 1185

Table A-16 Lakeland Subsoil Phosphate Adsorption

022 -4302 024 -4796 111 -820 110 -429 489 2615 261 495 1935 192

1459 8316 278 1464 7478 250 3484 7869 1121 3484 7191 1021 4532 6758 0741 4383 11690 1321

adsorption were discarded for isotherm fitting Table A-17 Pelion Topsoil Phosphate Adsorption

026 -5159 031 -6290 108 157 073 109 -643 -303 484 4541 448 472 6494 644

1431 14971 498 1435 10738 361 3434 18887 268 3432 19382 275 4489 12785 141 4525 12180 1331

Table A-18 Pelion Subsoil Phosphate Adsorption

010 -1910 010 -1910 058 9846 4573 044 12561 5867 120 14292 3732 121 13474 3568 406 19389 1927 405 20058 1984 704 19779 1231 687 20162 1279 923 23694 1137 934 19165 930

Table A-19 Johnston Topsoil Phosphate Adsorption

017 -3405 014 -2768 061 8653 7144 055 9710 8843 356 27375 3843 361 26460 3664 827 38681 2339 835 37936 2271

2205 51122 1159 2177 58213 1337 3688 62909 853 3707 56706 765

Table A-110 Johnston Subsoil Phosphate Adsorption

010 -1906 014 -2895 070 6930 3301 075 6372 2984 142 9475 2510 141 9769 2578 433 14991 1477 430 14411 1437 726 13319 842 716 13651 871 953 17497 842 945 18440 890

adsorption were discarded for isotherm fitting Table A-111 Vaucluse Topsoil Phosphate Adsorption

005 -988 007 -1342 070 7227 3404 070 7581 3523 441 12354 1228 440 12794 1269 953 14272 697 950 16717 809

2366 21890 442 2379 19343 391 3987 11230 139 3890 19421 244

Table A-112 Varina Subsoil Phosphate Adsorption

040 -8077 035 -6944 110 -1298 -624 108 -937 -455 469 7200 713 479 6131 601 999 8540 410 987 9275 449

2427 6963 141 2457 -2878 -0591 3925 8839 111 3992 7425 092

Table A-113 Rembert Topsoil Phosphate Adsorption

007 -1412 010 -1908 058 9626 4520 059 9523 4478 214 17813 2939 215 17753 2921 468 25002 2108 482 22694 1907 737 31400 1756 748 27913 1573

1047 31994 1326 1051 31145 1291

Table A-114 Rembert Subsoil Phosphate Adsorption

007 -1483 LOST LOST 001 20912 9901 003 20370 9731 025 55779 9174 021 56176 9298 163 85417 7241 159 87031 7326 375 103421 5797 368 104386 5864 633 114590 4752 643 114940 4722

adsorption were discarded for isotherm fitting Table A-115 Dothan Topsoil Phosphate Adsorption

012 -2417 016 -3101 087 3669 1747 086 4066 1907 262 7698 1280 260 8566 1414 526 14583 1218 519 15274 1282 787 21303 1192 786 20137 1136

1077 26742 1104 1069 28247 1166

Table A-116 Dothan Subsoil Phosphate Adsorption

018 -3644 011 -2209 017 17644 8423 012 18725 8887 159 68853 6836 153 69847 6956 520 98596 4870 529 99317 4846 943 115623 3803 958 106533 3575

1324 130537 3305 1332 123500 3169

Table A-117 Coxville Topsoil Phosphate Adsorption

010 -2075 014 -2827 071 6770 3221 073 6094 2936 248 11264 1850 246 11437 1888 512 14842 1265 511 16873 1416 824 12475 704 808 17882 996

1102 21677 895 1092 22222 924

Table A-118 Coxville Subsoil Phosphate Adsorption

C Q Adsorption mg L-1 mg kg-1

023 -4676 012 -2418 056 10260 4786 059 9634 4511 382 24386 2420 380 25482 2511 732 33049 1842 733 33307 1851

1022 35440 1477 1023 34549 1444 1318 36646 1220 1313 39934 1320

adsorption were discarded for isotherm fitting Table A-119 Norfolk Topsoil Phosphate Adsorption

017 -3374 014 -2895 097 1908 894 098 1787 836 179 3766 952 173 5074 1280 448 11193 1111 457 9194 914 722 13354 847 716 13694 873 942 17569 853 940 17979 873

Table A-120 Norfolk Subsoil Phosphate Adsorption

007 -1482 006 -1209 017 17381 8394 016 17289 8415 236 54879 5376 227 53268 5393 679 69585 3389 687 68420 3323

2070 74470 1525 2062 82062 1660 3615 75804 949 3674 73903 914

Table A-121 Wadmalaw Topsoil Phosphate Adsorption

012 -2302 005 -1005 056 9179 4524 051 10196 5010 400 21854 2144 397 22017 2168 895 27677 1338 889 26617 1301

2292 35292 714 2333 28741 580 3824 33464 419 3882 29512 366

Table A-122 Wadmalaw Subsoil Phosphate Adsorption

012 -2485 015 -2963 047 11773 5563 040 13522 6296 340 32761 3249 341 32388 3222 841 37943 1841 816 41268 2018

2172 56286 1147 2164 60802 1232 3618 79252 987 3612 87108 1076

adsorption were discarded for isotherm fitting Table A-123 Yonges Topsoil Phosphate Adsorption

013 -2551 017 -3302 056 10148 4778 061 8842 4219 392 22990 2268 392 21148 2128 868 30852 1510 857 32569 1599

2209 55554 1118 2218 45506 931 3671 61069 769 3669 65232 817

Table A-21 Isotherm Parameters and Statistics as Determined Using a 1-Surface Langmuir Model and the SCS Method to Correct for Q0

Location Soil Type Qmax (mg kg-1)

Qmax Std Error

kl (L mg-1)

kl Std Error X2 R2

Simpson Appling Top 37483 1861 2755 05206 59542 96313

Simpson Madison Top 51082 2809 5411 149 259188 92546

Simpson Madison Sub 253599 10404 7711 14287 3617744 96237

Simpson Hiwassee Sub 1420518 2855 47009 5619 533618 94426

Simpson Cecil Sub 246858 8419 225922 37897 3516751 93756

Sandhill Lakeland Top1 - - - - - -

Sandhill Lakeland Sub 16345 1887 2597 0863 1399619 9495

Sandhill Pelion Top 26725 1940 2493 07309 62124 9336

Sandhill Pelion Sub 26339 1089 20503 42036 3978511 94912

Sandhill Johnston Top 71871 3478 2682 052 189091 9697

Sandhill Johnston Sub 22239 990 11496 21771 2197167 95839

Sandhill Vaucluse Top 21534 1178 76313 22522 40241 94189

Edisto Varina Top1 BDL BDL BDL BDL BDL BDL

Edisto Varina Sub 211 892 7554 1408 2027 9598

Edisto Rembert Top 38939 1761 6486 1118 37953 9767

Edisto Rembert Sub 110193 5045 4 40929 118093 8672376 94073

Edisto Fuquay Top1 - - - - - -

Edisto Fuquay Sub1 - - - - - - 1 Below Detection Limit

Data C

omparing 1- and 2-Surface Isotherm

Models

Table A-21 (Continued) Isotherm Parameters and Statistics as Determined Using a 1-Surface Langmuir Model and the SCS Method to Correct for Q0

REC Soil Type Qmax (mg kg-1)

Qmax Std Error

kl (L mg-1)

kl Std Error X2 R2

Edisto Dothan Top 5746142 899805 10333 02864 3186987 97447

Edisto Dothan Sub 1344583 6142545 6405 12835 5403517 97758

Edisto Blanton Top1 - - - - - -

Edisto Blanton Sub1 - - - - - -

Pee Dee Coxville Top 2790452 1600465 62191 13037 2870887 96505

Pee Dee Coxville Sub 4896565 1812305 82138 15187 7697144 97174

Pee Dee Norfolk Top 3203273 2794911 19407 0367 1135312 98137

Pee Dee Norfolk Sub 88604 2180 261442 44285 267416 97571

Coastal Wadmalaw Top 3568212 127021 85499 18129 60962 96694

Coastal Wadmalaw Sub 9402642 9848372 19004 0705 1046154 89132

Coastal Yonges Top 8001957 7215222 13484 03788 3816474 93883 1 Below Detection Limit

Data C

Models

Location Soil Type Qmax1

(mg kg-1)

Qmax1 Std

kl1 (L mg-1)

kl1 Std

Qmax2 (mg kg-1)

Qmax2 Std Error

kl2 (L mg-1)

kl2 Std

Error X2 R2

Simpson Appling Top 2049 765 0036 0063 2373 1151 0618 0407 4463 098

Simpson Madison Top 3339 3275 015 0292 2138 3594 229 555 22554 095

Simpson Madison Sub 13150 8570 0019 0038 19745 3383 128 0415 20945 098

Simpson Hiwassee Sub 3473 954 0087 0159 12263 1549 693 232 25248 098

Simpson Cecil Sub 10346 990 0045 0020 19262 831 410 0434 27212 100

Sandhill Lakeland Top1 - - - - - - - - - -

Sandhill Lakeland Sub 64E5 11E10 756E-06 0126 669 702 125 186 583 099 Sandhill Pelion Top 624 2834 138 747 2203 2486 0133 0288 7317 094

Sandhill Pelion Sub 2605 988 210 141 99178 15E9 40E-5 523 4969 095

Sandhill Johnston Top 2706 1480 160 127 5692 835 00707 0061 7954 099

Sandhill Johnston Sub 1992 1100 140 120 625E5 11E11 38E-6 0659 2623 096

Sandhill Vaucluse Top 1250 653 258 276 1423 400 0072 0119 2197 098

Edisto Varina Top1 - - - - - - - - - -

Edisto Varina Sub 1555 Did Not

Converge (DNC)

076 DNC 555 DNC 0756 DNC 2703 096

Edisto Rembert Top 3750 3349 0063 0155 2166 1157 175 131 2287 099

Edisto Rembert Sub 8379 963 0383 0181 5543 752 302 112 16000 099

Edisto Fuquay Top1 - - - - - - - - - -

Edisto Fuquay Sub1 - - - - - - - - - -

Edisto Dothan Top 56E6 35E10 388E-6 0024 822 356 192 157 830 099 1 Below Detection Limit

Data C

Models

Table A-22 (Continued) Isotherm Parameters and Statistics as Determined Using a 2-Surface Langmuir Model

and the SCS Method to Correct for Q0

REC Soil Type Q1 (mg kg-1)

Q1 Std

kl1 (L mg-1)

kl1 Std

Q2 (mg kg-1)

Q2 Std Error

kl2 (L mg-1)

kl2 Std

Error X2 R2

Edisto Dothan Sub 32E6 64E10 716E-6 0141 10084 4159 114 0780 48314 098

Edisto Blanton Top1 - - - - - - - - - -

Edisto Blanton Sub1 - - - - - - - - - -

Pee Dee Coxville Top 1740 588 155 0854 85E5 76E9 000001 00901 1293 099

Pee Dee Coxville Sub 3599 1645 139 0860 5429 26068 00236 0179 5894 098

Pee Dee Norfolk Top 10338 1105 081 105 718E5 34E9 000002 00864 680 099

Pee Dee Norfolk Sub 3682 336 8 034 0555 5776 3624 520 413 13396 099 Coastal Wadmalaw

Top 1488 2599 015 0504 2343 2949 171 256 5807 097

Coastal Wadmalaw Sub 3972 510 221 0729 126E7 101E11 113E-06 000903 12765 099

Coastal Yonges Top 2475 694 206 125 11667 6211 00177 00184 10410 099 1 Below Detection Limit

Data C

Models

Table A-31 3-Parameter-Derived Isotherm Parameters and Statistics

Sample Location Soil Type

Qmax (fit) (mg kg-1)

Qmax (fit) Std Error

kl (L mg-1)

kl Std

Error Q0

(mg kg-1) Q0

Std Error X2 R2 1 Simpson Appling Topsoil 36255 2328 020 006 8774 1940 56849 097 2 Simpson Madison Topsoil 46744 4154 029 012 5399 3813 230482 094 3 Simpson Madison Subsoil 247430 17542 064 023 17732 17605 3956215 096 4 Simpson Hiwassee Subsoil 82006 5567 069 026 8667 6296 350523 097 5 Simpson Cecil Subsoil 191316 11077 063 017 15639 10965 1601321 097 6 Sandhill Lakeland Topsoil1 - - - - - - - - 7 Sandhill Lakeland Subsoil 17775 3233 014 008 1608 2299 13079 096 8 Sandhill Pelion Topsoil 26079 2465 020 009 6314 2039 66688 094 9 Sandhill Pelion Subsoil 30549 3132 330 114 8969 6439 34427 096 10 Sandhill Johnston Topsoil 69644 3634 019 004 1608 2299 130803 098 11 Sandhill Johnston Subsoil 23277 1693 138 041 5499 1815 22797 096 12 Sandhill Vaucluse Topsoil 20977 1812 060 025 916 1633 43885 094 13 Edisto Varina Topsoil1 - - - - - - - - 14 Edisto Varina Subsoil 26310 3662 141 054 16667 3998 13959 098 15 Edisto Rembert Topsoil 38590 2109 055 014 2220 1614 39877 098 16 Edisto Rembert Subsoil 106720 10135 204 106 -10998 9095 1440678 093 17 Edisto Fuquay Topsoil1 - - - - - - - - 18 Edisto Fuquay Subsoil1 - - - - - - - - 19 Edisto Dothan Topsoil 78053 22344 005 002 1672 929 22164 098 20 Edisto Dothan Subsoil 138771 8149 064 020 3453 6645 838594 097 21 Edisto Blanton Topsoil1 - - - - - - - - 22 Edisto Blanton Subsoil1 - - - - - - - - 23 Pee Dee Coxville Topsoil 27688 1972 051 017 3099 1532 30784 097 24 Pee Dee Coxville Subsoil 47870 2646 061 022 5930 3061 81149 097 25 Pee Dee Norfolk Topsoil 33837 4151 016 005 3330 817 11707 098 26 Pee Dee Norfolk Subsoil 80605 3935 125 039 2737 4160 229307 098 27 Coastal Wadmalaw Topsoil 34968 2204 076 025 2514 2211 66518 097 28 Coastal Wadmalaw Subsoil 101055 15719 007 003 -4581 4286 627987 094 29 Coastal Yonges Topsoil 82404 8708 007 002 -1298 2582 213544 097

1 Below Detection Limits Isotherm Not Calculated

3-Parameter Isotherm

A-3 3-Parameter Isotherms

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Appling TopMadison TopMadison SubHiwassee SubCecil SubLakeland SubPelion TopPelion SubJohnston TopJohnston SubVaucluse TopVarina SubRembert TopRembert SubDothan TopDothan SubCoxville TopCoxville SubNorfolk TopNorfolk SubWadmalaw TopWadmalaw SubYonges TopLower Bound 95Upper Bound 9550th Percentile

Figure A-31 Isotherms for All Sampled Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Appling Top

Madison Top

Madison Sub

Hiwassee Sub

Cecil Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-32 Isotherms for Simpson ES Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Lakeland Sub

Pelion Top

Pelion Sub

Johnston Top

Johnston Sub

Vaucluse Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-33 Isotherms for Sandhill REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Varina Sub

Rembert Top

Rembert Sub

Dothan Top

Dothan Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-34 Isotherms for Edisto REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Coxville Top

Coxville Sub

Norfolk Top

Norfolk Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-35 Isotherms for Pee Dee REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Wadmalaw Top

Wadmalaw Sub

Yonges Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-36 Isotherms for Coastal REC Soils

A-4 Alternate Normalizations Based on the 3-Parameter Isotherm

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

2)Appling TopMadison TopMadison SubHiwassee SubCecil SubLakeland SubPelion TopPelion SubJohnston TopJohnston SubVaucluse TopVarina SubRembert TopRembert SubDothan TopDothan SubCoxville TopCoxville SubNorfolk TopNorfolk SubWadmalaw TopWadmalaw SubYonges TopLower Bound 95Upper Bound 9550th Percentile

Figure A-41 SSA-Normalized Isotherms for All Sampled Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Appling Top

Madison Top

Madison Sub

Hiwassee Sub

Cecil Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-42 SSA-Normalized Isotherms for Simpson ES Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Lakeland Sub

Pelion Top

Pelion Sub

Johnston Top

Johnston Sub

Vaucluse Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-43 SSA-Normalized Isotherms for Sandhill REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Varina Sub

Rembert Top

Rembert Sub

Dothan Top

Dothan Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-44 SSA-Normalized Isotherms for Edisto REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Coxville Top

Coxville Sub

Norfolk Top

Norfolk Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-45 SSA-Normalized Isotherms for Pee Dee REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Wadmalaw Top

Wadmalaw Sub

Yonges Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-46 SSA-Normalized Isotherms for Coastal REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Figure A-47 Clay-Normalized Isotherms for All Sampled Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Appling Top

Madison Top

Madison Sub

Hiwassee Sub

Cecil Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-48 Clay-Normalized Isotherms for Simpson ES Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Lakeland Sub

Pelion Top

Pelion Sub

Johnston Top

Johnston Sub

Vaucluse Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-49 Clay-Normalized Isotherms for Sandhill REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Varina Sub

Rembert Top

Rembert Sub

Dothan Top

Dothan Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-410 Clay-Normalized Isotherms for Edisto REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Coxville Top

Coxville Sub

Norfolk Top

Norfolk Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-411 Clay-Normalized Isotherms for Pee Dee REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Wadmalaw Top

Wadmalaw Sub

Yonges Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-412 Clay-Normalized Isotherms for Coastal REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

)Appling TopMadison TopMadison SubHiwassee SubCecil SubLakeland SubPelion TopPelion SubJohnston TopJohnston SubVaucluse TopVarina SubRembert TopRembert SubDothan TopDothan SubCoxville TopCoxville SubNorfolk TopNorfolk SubWadmalaw TopWadmalaw SubYonges TopLower Bound 95Upper Bound 9550th Percentile

Figure A-413 FeDCB-Normalized Isotherms for All Sampled Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Appling Top

Madison Top

Madison Sub

Hiwassee Sub

Cecil Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-414 FeDCB -Normalized Isotherms for Simpson ES Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Lakeland Sub

Pelion Top

Pelion Sub

Johnston Top

Johnston Sub

Vaucluse Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-415 FeDCB -Normalized Isotherms for Sandhill REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Varina Sub

Rembert Top

Rembert Sub

Dothan Top

Dothan Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-416 FeDCB -Normalized Isotherms for Edisto REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Coxville Top

Coxville Sub

Norfolk Top

Norfolk Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-417 FeDCB -Normalized Isotherms for Pee Dee REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Wadmalaw Top

Wadmalaw Sub

Yonges Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-418 FeDCB -Normalized Isotherms for Coastal REC Soils

100000

120000

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Figure A-419 OM-Normalized Isotherms for All Sampled Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Appling Top

Madison Top

Madison Sub

Hiwassee Sub

Cecil Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-420 OM-Normalized Isotherms for Simpson ES Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Lakeland Sub

Pelion Top

Pelion Sub

Johnston Top

Johnston Sub

Vaucluse Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-421 OM-Normalized Isotherms for Sandhill REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Varina Sub

Rembert Top

Rembert Sub

Dothan Top

Dothan Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-422 OM-Normalized Isotherms for Edisto REC Soils

100000

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Coxville Top

Coxville Sub

Norfolk Top

Norfolk Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-423 OM-Normalized Isotherms for Pee Dee REC Soils

100000

120000

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Wadmalaw Top

Wadmalaw Sub

Yonges Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-424 OM-Normalized Isotherms for Coastal REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

e)Appling TopMadison TopMadison SubHiwassee SubCecil SubLakeland SubPelion TopPelion SubJohnston TopJohnston SubVaucluse TopVarina SubRembert TopRembert SubDothan TopDothan SubCoxville TopCoxville SubNorfolk TopNorfolk SubWadmalaw TopWadmalaw SubYonges TopLower Bound 95Upper Bound 9550th Percentile

Figure A-425 SSA- and FeDCB-Normalized Isotherms for All Sampled Soils

000001

000002

000003

000004

000005

000006

000007

000008

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Figure A-426 SSA- and FeDCB-Normalized Isotherms for All Sampled Soils Scaled

00000005

0000001

00000015

0000002

00000025

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Appling Top

Madison Top

Madison Sub

Hiwassee Sub

Cecil Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-427 SSA- and FeDCB-Normalized Isotherms for Simpson ES Soils

000001

000002

000003

000004

000005

000006

000007

000008

000009

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Lakeland Sub

Pelion Top

Pelion Sub

Johnston Top

Johnston Sub

Vaucluse Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-428 SSA- and FeDCB-Normalized Isotherms for Sandhill REC Soils

000001

000002

000003

000004

000005

000006

000007

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Varina Sub

Rembert Top

Rembert Sub

Dothan Top

Dothan Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-429 SSA- and FeDCB-Normalized Isotherms for Edisto REC Soils

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Coxville Top

Coxville Sub

Norfolk Top

Norfolk Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-430 SSA- and FeDCB-Normalized Isotherms for Pee Dee REC Soils

0000002

0000004

0000006

0000008

000001

0000012

0000014

0000016

0000018

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Wadmalaw Top

Wadmalaw Sub

Yonges Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-431 SSA- and FeDCB-Normalized Isotherms for Coastal REC Soils

200000

400000

600000

800000

1000000

1200000

1400000

1600000

1800000

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Figure A-432 Clay- and OM-Normalized Isotherms for All Sampled Soils

100000

200000

300000

400000

500000

600000

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Figure A-433 Clay- and OM-Normalized Isotherms for All Sampled Soils Scaled

100000

120000

140000

160000

180000

200000

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Appling Top

Madison Top

Madison Sub

Hiwassee Sub

Cecil Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-434 Clay- and OM-Normalized Isotherms for Simpson ES Soils

100000

200000

300000

400000

500000

600000

700000

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Lakeland Sub

Pelion Top

Pelion Sub

Johnston Top

Johnston Sub

Vaucluse Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-435 Clay- and OM-Normalized Isotherms for Sandhill REC Soils

100000

200000

300000

400000

500000

600000

700000

800000

900000

1000000

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Varina Sub

Rembert Top

Rembert Sub

Dothan Top

Dothan Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-436 Clay- and OM-Normalized Isotherms for Edisto REC Soils

200000

400000

600000

800000

1000000

1200000

1400000

1600000

1800000

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Coxville Top

Coxville Sub

Norfolk Top

Norfolk Sub

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-437 Clay- and OM-Normalized Isotherms for Pee Dee REC Soils

200000

400000

600000

800000

1000000

1200000

1400000

0 10 20 30 40 50 60 70 80 90

C (mg-PO4L)

Wadmalaw Top

Wadmalaw Sub

Yonges Top

Lower Bound 95

Higher Bound 95

50th Percentile

Figure A-438 Clay- and OM-Normalized Isotherms for Coastal REC Soils

A-5 Predicted vs Fit Isotherms

Figure A-51 Predicted vs Fit Isotherms Appling Topsoil

Figure A-52 Predicted vs Fit Isotherms Madison Topsoil

Figure A-53 Predicted vs Fit Isotherms Madison Subsoil

Figure A-54 Predicted vs Fit Isotherms Hiwassee Subsoil

Figure A-55 Predicted vs Fit Isotherms Cecil Subsoil

Figure A-56 Predicted vs Fit Isotherms Lakeland Subsoil

Figure A-57 Predicted vs Fit Isotherms Pelion Topsoil

Figure A-58 Predicted vs Fit Isotherms Pelion Subsoil

Figure A-59 Predicted vs Fit Isotherms Johnston Topsoil

Figure A-510 Predicted vs Fit Isotherms Johnston Subsoil

Figure A-511 Predicted vs Fit Isotherms Vaucluse Topsoil

Figure A-512 Predicted vs Fit Isotherms Varina Subsoil

Figure A-513 Predicted vs Fit Isotherms Rembert Topsoil

Figure A-514 Predicted vs Fit Isotherms Rembert Subsoil

Figure A-515 Predicted vs Fit Isotherms Dothan Topsoil

Figure A-516 Predicted vs Fit Isotherms Dothan Subsoil

Figure A-517 Predicted vs Fit Isotherms Coxville Topsoil

Figure A-518 Predicted vs Fit Isotherms Coxville Subsoil

Figure A-519 Predicted vs Fit Isotherms Norfolk Topsoil

Figure A-520 Predicted vs Fit Isotherms Norfolk Subsoil

Figure A-521 Predicted vs Fit Isotherms Wadmalaw Topsoil

Figure A-522 Predicted vs Fit Isotherms Wadmalaw Subsoil

Figure A-523 Predicted vs Fit Isotherms Yonges Topsoil

Appendix B

Soil Characterization Data

Containing

1 General Soil Information

2 Soil Texture Data from the Literature

3 Experimental Soil Texture Data

4 Experimental Specific Surface Area Data

5 Experimental Soil Chemistry Data

6 Soil Photographs

7 Standard Soil Test Data

Table B-11 General Soil Information Availability of Detailed Particle Size Info Textural Description and Geologic Parent Material

na Information not available

USDANRCS County Soil Surveys (Anderson Barnwell Charleston Darlington Richland) Soil Type

SCS Detailed Particle Size Info

Topsoil Description

Likely Subsoil Description Geologic Parent Material

Appling na Sandy-Loam Sandy-Loam Granite Gneiss-Schist Madison na Sandy-Loam Clay Quartz-Mica Schist Hiwassee na Sandy-Loam Clay Dark colored rocks Cecil Bulletin 267 Sandy-Loam Clay Granite gneiss schist

Lakeland Bulletin 262 Sand Sand Thick beds of sandy marine or eolian sediment

Pelion na Loamy Sand Loamy Sand Loamy marine sediment Johnston na Loam ldquoMuckyrdquo Loam Loamy fluvial amp marine sediment Vaucluse na Loamy Sand Loamy Sand Loamy marine sediment Varina na Loamy Sand Sandy Clay Clayey coastal plain sediment Rembert na Loam Clay Loamy coastal plain sediment Fuquay na Sand Sand Loamy coastal plain sediment Dothan Bulletin 263 Loamy Sand Sand Loamy coastal plain sediment Blanton na Sand Sand Loamy coastal plain sediment Coxville Bulletin 263 Sandy-Loam Sandy Clay Beds of unconsolidated sand and clay Norfolk na Light Sandy-Loam Light Sandy-Loam Beds of unconsolidated sand and clay Wadmalaw na Fine Sandy Loam Fine Sandy Loam na Yonges na Loamy Fine Sand Loamy Fine Sand na

General Soil Inform

Table B-12 General Soil Information Reaction pH Permeability Hydrologic Soil Group and Erosion Factors USDANRCS County Soil Surveys (Anderson Barnwell Charleston Darlington Richland)

Soil Type Soil Reaction (pH) Permeability (inhr)

Hydrologic Soil Group

Erosion Factor K Erosion Factor T

Appling 45-55 20-60 B (B1) 2 4 Madison 45-60 20-60 B ( B1) 28 4 Hiwassee 45-65 6-20 B (B1) 28 5 Cecil 45-60 20-60 B (B1) 32 4 Lakeland 45-60 gt20 A (A1) 17 5 Pelion 45-65 20-60 BD (BD1) 24 3 Johnston 45-55 20-60 D (D1) 20 4 Vaucluse 45-55 60-20 C 17 3 Varina 50-65

45-55 20-60 6-20

C1 na na

Rembert 45-55 6-20 06-20

D1 na na

Fuquay 45-55 60-20 B1 na na Dothan 45-55 2-60 B1 na na Blanton 45-55 60-20 A1 na na Coxville 51-55 05-2 D1 na na Norfolk 56-60 25-50 B1 na na Wadmalaw 45-50 63-20 D1 na na Yonges 56-60 20-63 D1 na na

1 Denotes values taken from SC DHEC Erosion Related Information for South Carolina Soils na Information not available

General Soil Inform

Table B-13 Soil Classification Information (USDA-NRCS 2010a) Soil Type Location Order Suborder Great Group Sub Group Taxonomic Class Appling Simpson Ultisol Udult Kanhapludult Typic Fine kaolinitic thermic Madison Simpson Ultisol Udult Kanhapludult Typic Fine kaolinitic thermic

Hiwassee Simpson Ultisol Udult Kanhapludult Rhodic Very-fine kaolinitic thermic

Cecil Simpson Ultisol Udult Kanhapludult Typic Fine kaolinitic thermic Lakeland Sandhill Entisol Psamment Quartzipsamments Typic Thermic coated

Pelion Sandhill Ultisol Udult Kanhapludult Franiaquic Fine-loamy kaolinitic thermic

Johnston Sandhill Inceptisol Aquept Humaquept Cumulic Corse-loamy siliceous active

Vaucluse Sandhill Ultisol Udult Kanhapludult Fragic Fine-loamy kaolinitic thermic

Varina Edisto Ultisol Udult Paleudult Plinthic Fine kaolinitic thermic Rembert Edisto Ultisol Aquults Endoaqult Typic Fine kaolinitic thermic Fuquay Edisto Ultisol Udult Kandiudult Arenic Plenthic Loamy kaolinitic thermic

Dothan Edisto Ultisol Udult Kandiudult Plinthic Fine-loamy kaolinitic thermic

Blanton Edisto Ultisol Udult Paleudult Grossarenic Loamy siliceous semiactive thermic

Coxville Pee Dee Ultisol Aquults Paleaquult Typic Fine kaolinitic thermic

Norfolk Pee Dee Ultisol Udult Kandiudult Typic Fine-loamy kaolinitic thermic

Wadmalaw Coastal Alfisol Aqualf Endoaqualf Umbric Fine-loamy mixed semiactive

Yonges Coastal Alfisol Aqualf Endoaqualf Typic Fine-loamy mixed active

General Soil Inform

B-2 Soil Texture Data from the Literature

Table B-21 Soil Texture Data from NRCS County Soil Surveys

1 Values given from soil surveys are according to NRCS method of measuring soil texture An AASHO method for measuring soil texture produces different data which are sometimes included in soil surveys as well

2 Two sets of numbers denotes differences with depth through soil profile between sub- and topsoil

From USDA-NRCS County Soil Surveys (Anderson Barnwell Charleston Darlington Richland)

Percentage Passing Sieve Number (Parent Material)1 2

Soil Type

4 (47 mm) 10 (20mm) 40 (42 mm) 200 (074mm) Appling 86-100 80-100 55-75 15-35 Madison 85-100

90-100 80-100 85-100

60-90 75-97

26-49 57-85

Hiwassee 95-100 95-100

90-100 95-100

70-95 80-100

30-50 60-95

Cecil 84-100 97-100

80-100 92-100

67-90 72-99

26-42 55-95

Lakeland 90-100 90-100 60-100 5-12 Pelion 95-100 90-100 50-90 13-40 Johnston 100 100 60-95 30-75 Vaucluse 98-100 90-100 51-70 8-30 Varina na 100

100 80-90 85-95

15-35 45-70

Rembert na 100 100

70-90 85-95

45-70 65-80

Fuquay na 100 50-80 5-20 Dothan na 95-100 65-90 14-35 Blanton na 100 70-90 5-20 Coxville na 100 na 45-55 Norfolk na 100 na 35-45 Wadmalaw na 100 85-100 25-70 Yonges na 100 90-100 35-70

Table B-22 SCLRC Soil Texture Data (Hayes and Price 1995)

Passing Location Soil Type

Horizon Depth

(in) 200 Sieve (0075 mm)

400 Sieve (0038 mm)

0-9 15-35 5-20 0-9 15-40 5-30 0-9 40-70 20-35 9-35 51-80 35-60

Simpson Appling

35-46 40-75 20-50 0-6 26-55 5-20 0-6 30-49 5-15 0-6 48-80 25-35 6-30 57-85 30-50

30-35 50-80 25-35

Simpson Madison

35-66 26-55 5-20 0-7 30-50 7-20 0-7 50-90 10-25 0-7 50-85 10-35 7-61 51-95 35-60

Simpson Hiwassee

61-70 36-70 7-35 0-7 26-42 5-20 0-7 13-30 5-20 0-7 38-81 20-35 7-11 38-81 20-35

Simpson Cecil

11-50 55-95 35-70 0-43 5-12 2-8 Sandhill Lakeland 43-80 1-12 1-6 0-10 13-40 5-15 0-10 8-30 2-10

10-22 25-55 18-35 22-39 25-60 18-50

Sandhill Pelion

39-65 18-60 10-40 0-30 51-75 7-18 0-30 18-65 5-18

30-34 5-30 2-12 Sandhill Johnston

34-60 25-49 5-20 0-15 8-30 2-10 0-15 20-45 10-20 0-15 40-60 20-28

15-29 25-50 18-35 29-58 20-50 18-45

Sandhill Vaucluse

58-72 15-50 5-30

Table B-22 (Continued) SCLRC Soil Texture Data (Hayes and Price 1995)

Passing REC Soil Type

Horizon Depth

(in) 200 Sieve

(0075 mm) 400 Sieve

(0038 mm) 0-14 10-35 3-10 0-14 20-49 8-18

14-38 36-65 35-60 Edisto Varina

38-80 36-68 30-55 0-5 51-80 10-35 0-5 36-50 5-18 5-33 55-85 35-60

33-54 30-60 22-45 Edisto Rembert

54-65 20-50 8-25 0-34 5-35 2-10 0-34 5-20 1-7

34-45 23-45 10-35 Edisto Fuquay

45-96 28-49 10-35 0-13 13-30 5-15 0-13 20-40 10-18 0-13 10-30 1-10

13-33 23-49 18-35 Edisto Dothan

33-60 30-53 18-40 0-58 5-20 1-7 0-58 13-25 5-13 0-58 5-10 1-7

58-62 13-30 10-18 Edisto Blanton

62-80 25-50 12-40 0-11 46-75 5-27 Pee Dee Coxville 11-72 50-85 35-60 0-18 13-30 2-8 Pee Dee Norfolk 18-44 30-55 18-35 0-13 15-35 2-10 0-13 30-60 5-20

13-33 40-75 18-35 Coastal Wadmalaw

33-83 51-90 20-45 0-14 25-55 5-18 0-14 40-55 10-20 Coastal Yonges

14-42 40-70 18-40

B-3 Experimental Soil Texture Data

Table B-31 Experimental Site-Specific Soil Texture Data

(Price 1994) Location Soil Type CLAY

() SILT ()

SAND ()

Simpson Appling Topsoil 51 258 691 Simpson Appling Subsoil 238 177 585 Simpson Madison Topsoil 124 229 647 Simpson Madison Subsoil 308 163 528 Simpson Hiwassee Topsoil 137 83 78 Simpson Hiwassee Subsoil 239 16 602 Simpson Cecil Topsoil 94 222 684 Simpson Cecil Subsoil 362 117 521 Sandhill Lakeland Topsoil 53 59 889 Sandhill Lakeland Subsoil 85 47 868 Sandhill Pelion Topsoil 81 63 856 Sandhill Pelion Subsoil 242 12 638 Sandhill Johnston Topsoil 101 9 81 Sandhill Johnston Subsoil 183 156 661 Sandhill Vaucluse Topsoil 102 53 844 Sandhill Vaucluse Subsoil 179 55 765 Edisto Varina Topsoil 82 173 746 Edisto Varina Subsoil 46 273 681 Edisto Rembert Topsoil 88 163 75 Edisto Rembert Subsoil 421 115 464 Edisto Fuquay Topsoil 62 43 895 Edisto Fuquay Subsoil 141 84 774 Edisto Dothan Topsoil 74 124 802 Edisto Dothan Subsoil 226 15 624 Edisto Blanton Topsoil 34 66 90 Edisto Blanton Subsoil 84 82 834 Pee Dee Coxville Topsoil 59 233 708 Pee Dee Coxville Subsoil 164 225 611 Pee Dee Norfolk Topsoil 6 194 746 Pee Dee Norfolk Subsoil 271 124 606 Coastal Wadmalaw Topsoil 81 29 628 Coastal Wadmalaw Subsoil 287 242 471 Coastal Yonges Topsoil 75 156 769 Coastal Yonges Subsoil 217 147 636

B-4 Experimental Specific Surface Area Data

Table B-41 Experimental Specific Surface Area Data

Location Soil Type SSA (m2 g-1)

Simpson Appling Topsoil 95

Simpson Madison Topsoil 95

Simpson Madison Subsoil 439

Simpson Hiwassee Subsoil 162

Simpson Cecil Subsoil 324

Sandhill Lakeland Topsoil 04

Sandhill Lakeland Subsoil 15

Sandhill Pelion Topsoil 16

Sandhill Pelion Subsoil 7

Sandhill Johnston Topsoil 57

Sandhill Johnston Subsoil 46

Sandhill Vaucluse Topsoil 31

Edisto Varina Topsoil 19

Edisto Varina Subsoil 91

Edisto Rembert Topsoil 65

Edisto Rembert Subsoil 364

Edisto Fuquay Topsoil 18

Edisto Fuquay Subsoil 56

Edisto Dothan Topsoil 47

Edisto Dothan Subsoil 247

Edisto Blanton Topsoil 14

Edisto Blanton Subsoil 16

Pee Dee Coxville Topsoil 41

Pee Dee Coxville Subsoil 81

Pee Dee Norfolk Topsoil 04

Pee Dee Norfolk Subsoil 201

Coastal Wadmalaw Topsoil 51

Coastal Wadmalaw Subsoil 217

Coastal Yonges Topsoil 146

Table B-51 Experimental Site-Specific Soil Chemistry Data Location Soil Type OM

() C b ()

PO4Me-1 (mg kgSoil

-1) FeMe-1

(mg kgSoil-1)

AlMe-1 (mg kgSoil

-1) PO4DCB

(mg kgSoil-1)

FeDCB (mg kgSoil

-1) AlDCB

(mg kgSoil-1)

PO4Water-Desorbed (mg kgSoil

-1) Simpson Appling Top 370 008 083 5826 24128 94800 142065 1989231 443630 15240 Simpson Madison Top 290 004 067 2147 51560 137960 137551 2128866 546298 10443 Simpson Madison Sub 890 005 051 153 16624 101360 264647 5743252 714258 4390 Simpson Hiwassee Sub 450 003 047 307 16884 111560 76535 2750751 380124 4930 Simpson Cecil Sub 600 004 044 153 13664 96800 127058 4595251 571104 4119 Sandhill Lakeland Top 030 001 017 18859 35168 61760 65168 140261 57741 BDL1 Sandhill Lakeland Sub 030 001 007 7820 48400 52960 48443 192112 68799 4549 Sandhill Pelion Top 190 003 041 14106 25392 192600 76167 198279 136880 5725 Sandhill Pelion Sub 070 001 016 307 23076 55560 25648 389363 92637 1910 Sandhill Johnston Top 300 008 209 1073 95720 265240 26836 194108 123038 3086 Sandhill Johnston Sub 060 001 029 460 35472 107160 17368 147712 82108 2400 Sandhill Vaucluse Top 110 001 033 767 37576 82880 28913 288537 115824 1165 Edisto Varina Top 120 002 042 31585 18420 81480 90722 231367 64683 BDL1 Edisto Varina Sub 140 002 039 21312 9960 120440 80313 261263 98067 7510 Edisto Rembert Top 220 008 127 1993 40640 212200 36269 94059 100033 1660 Edisto Rembert Sub 200 005 044 153 11232 116600 21316 623506 200213 1483 Edisto Fuquay Top 060 002 039 12419 11772 64040 61135 181785 62982 BDL1 Edisto Fuquay Sub 100 001 020 3680 9284 90640 45979 467141 174713 BDL1 Edisto Dothan Top 110 004 052 8586 40040 126680 58084 389815 250598 2759 Edisto Dothan Sub 240 004 057 767 10064 245400 50686 1418428 765486 2926 Edisto Blanton Top 130 004 082 88776 9160 118040 130152 717214 65888 BDL1 Edisto Blanton Sub 030 001 010 24072 15120 61200 57698 161673 58566 BDL1 Pee Dee Coxville Top 030 001 009 613 20332 42840 22288 205432 97687 2451 Pee Dee Coxville Sub 050 002 026 613 20536 123000 10585 36773 101401 3547 Pee Dee Norfolk Top 240 003 114 2607 59240 149640 19404 41850 49978 3135 Pee Dee Norfolk Sub 120 003 020 307 25192 178720 14274 416686 252290 1345 Coastal Wadmalaw Top 270 014 155 3680 68840 57000 55224 437083 149952 1653 Coastal Wadmalaw Sub 110 004 035 307 6488 24500 17378 545169 185379 2724 Coastal Yonges Top 080 004 024 1533 8976 43120 23818 581003 160309 2926

1 Below Detection Limit

Experimental Soil C

hemistry D

B-6 Soil Photographs

Figure B-61 Appling Topsoil

Figure B-62 Madison Topsoil

Figure B-63 Madison Subsoil

Figure B-64 Hiwassee Subsoil

Figure B-65 Cecil Subsoil

Figure B-66 Lakeland Topsoil

Figure B-67 Lakeland

Subsoil

Figure B-68 Pelion Topsoil

Figure B-69 Pelion Subsoil

Figure B-610 Johnston Topsoil

Figure B-611 Johnston Subsoil

Figure B-612 Vaucluse Topsoil

Figure B-613 Varina Topsoil

Figure B-614 Varina Subsoil

Figure B-615 Rembert Topsoil

Figure B-616 Rembert Subsoil

Figure B-617 Fuquay Topsoil

Figure B-618 Fuquay

Subsoil

Figure B-619 Dothan Topsoil

Figure B-620 Dothan Subsoil

Figure B-621 Blanton Topsoil

Figure B-622 Blanton Subsoil

Figure B-623 Coxville Topsoil

Figure B-624 Coxville

Subsoil

Figure B-625 Norfolk Topsoil

Figure B-626 Norfolk Subsoil

Figure B-627 Wadmalaw Topsoil

Figure B-628 Wadmalaw Subsoil

Figure B-629 Yonges Topsoil

Soil pH

Buffer pH

P lbsA

K lbsA

Ca lbsA

Mg lbsA

Zn lbsA

Mn lbsA

Cu lbsA

B lbsA

Na lbsA

Appling Top 45 76 38 150 826 103 15 76 23 03 8

Madison Top 53 755 14 166 250 147 34 169 14 03 8

Madison Sub 52 745 1 234 100 311 1 20 16 04 6

Hiwassee Sub 5 755 2 75 108 71 09 51 17 03 7

Cecil Sub 5 75 1 128 432 139 07 31 14 05 8

Lakeland Top 52 78 123 33 296 45 25 14 21 0 5

Lakeland Sub 53 79 51 38 137 33 08 11 12 0 5

Pelion Top 5 76 92 92 472 53 27 56 09 02 6

Pelion Sub 5 78 2 42 337 45 15 2 19 02 10

Johnston Top 48 735 7 54 239 93 16 6 13 0 36

Johnston Sub 45 775 3 20 89 33 06 1 1 01 21

Vaucluse Top 48 77 5 20 156 31 11 4 09 01 10

Varina Top 58 785 206 115 1038 95 4 93 31 03 15 Varina Sub 57 765 139 80 1152 126 09 54 14 02 10

Rembert Top 44 74 13 31 137 26 13 4 11 02 13

Rembert Sub 47 73 1 15 194 71 07 1 1 04 19

Fuquay Top 62 79 81 64 620 134 18 22 08 02 6

Fuquay Sub 52 775 24 67 306 91 09 18 08 03 8

Dothan Top 46 765 56 173 669 93 48 81 11 01 8

Dothan Sub 51 745 5 216 853 204 05 12 1 04 11

Blanton Top 68 795 579 161 7591 268 3669 124 123 4 22

Blanton Sub 71 795 157 76 1098 52 424 40 29 05 7

Coxville Top 52 785 4 56 413 107 05 2 07 01 6

Coxville Sub 44 755 4 16 72 27 08 2 08 01 16

Norfolk Top 42 74 17 28 179 34 38 7 1 01 12

Norfolk Sub 44 725 2 27 122 64 08 1 16 03 22

Wadmalaw Top 43 73 24 37 1648 287 4 11 09 02 70

Wadmalaw Sub 51 76 2 27 2295 311 04 1 04 06 728

Yonges Top 63 78 10 93 2190 212 06 2 08 07 33

Standard Soil Test Data

Table B-71 Standard Soil Test Data

Soil Type CEC (meq100g)

Acidity (meq100g)

Base Saturation Ca ()

Base Saturation Mg ()

Base Saturation K

Base Saturation Na ()

Base Saturation Total ()

Appling Top 59 32 35 7 3 0 46

Madison Top 51 36 12 12 4 0 29

Madison Sub 63 44 4 21 5 0 29

Hiwassee Sub 43 36 6 7 2 0 16

Cecil Sub 58 4 19 10 3 0 32

Lakeland Top 26 16 28 7 2 0 38

Lakeland Sub 13 08 26 11 4 1 41

Pelion Top 47 32 25 5 3 0 33

Pelion Sub 27 16 31 7 2 1 41

Johnston Top 63 52 9 6 1 1 18

Johnston Sub 24 2 9 6 1 2 18 Vaucluse Top 3 24 13 4 1 1 19

Varina Top 44 12 59 9 3 1 72

Varina Sub 63 28 46 8 2 0 56

Rembert Top 53 48 6 2 1 1 10

Rembert Sub 64 56 8 5 0 1 13

Fuquay Top 3 08 52 19 3 0 73

Fuquay Sub 32 2 24 12 3 1 39

Dothan Top 51 28 33 8 4 0 45

Dothan Sub 77 44 28 11 4 0 43

Blanton Top 207 04 92 5 1 0 98

Blanton Sub 35 04 78 6 3 0 88

Coxville Top 28 12 37 16 3 0 56

Coxville Sub 39 36 5 3 1 1 9

Norfolk Top 55 48 8 3 1 0 12

Norfolk Sub 67 6 5 4 1 1 10

Wadmalaw Top 111 56 37 11 0 1 50

Wadmalaw Sub 119 32 48 11 0 13 73

Yonges Top 81 16 68 11 1 1 81

Table B-71 (Continued) Standard Soil Test Data

Appendix C

Additional Scatter Plots

Containing

1 Plots Relating Soil Characteristics to One Another 2 Plots Relating Isotherm Parameters to One Another 3 Plots Relating Soil Characteristics to Isotherm Parameters

C-1 Plots Relating Soil Characteristics to One Another

R2 = 03091

0 5 10 15 20 25 30 35 40 45 50

Arithmetic Mean SCLRC Clay

Figure C-11 SCLRC Clay vs Price (1994) Site-Specific Clay

R2 = 02944

0 10 20 30 40 50 60 70 80 90

Arithmetic Mean NRCS Clay

Figure C-12 NRCS Arithmetic Mean of Percent Passing 200 Sieve vs Price (1994) Site-Specific Clay

R2 = 05234

0 10 20 30 40 50 60 70 80 90 100

SCLRC Higher Bound Passing 200 Sieve

Figure C-13 SCLRC Higher Bound of Percent Passing 200 Sieve vs Price (1994) Site-Specific Percent (Clay + Silt)

R2 = 04504

0 10 20 30 40 50 60 70 80 90

NRCS Arithmetic Mean Passing 200 Sieve

Figure C-14 NRCS Arithmetic Mean of Percent Passing 200 Sieve vs Price (1994) Site-Specific Percent (Clay + Silt)

R2 = 06744

0 10 20 30 40 50 60 70 80 90 100

NRCS Overall Higher Bound Passing 200 Sieve

Figure C-15 NRCS Overall Higher Bound of Percent Passing 200 Sieve vs Geo-

metric Mean of Price (1994) Clay for Top- and Subsoil

R2 = 05574

0 10 20 30 40 50 60 70

NRCS Overall Arithmetic Mean Passing 200 Sieve

Figure C-16 NRCS Overall Arithmetic Mean of Percent Passing 200 Sieve vs

Arithmetic Mean of Price (1994) Clay for Top-and Subsoil

R2 = 00239

0 5 10 15 20 25 30 35

Price 1994 Silt

Figure C-17 Price (1994) Silt vs SSA

R2 = 06298

0 10 20 30 40 50 60

Price 1994 (Clay+Silt)

Figure C-18 Price (1994) Percent(Clay + Silt) vs SSA

R2 = 04656

000 100 200 300 400 500 600 700 800 900 1000

Figure C-19 OM vs SSA

R2 = 07477

-10 -5 0 5 10 15 20 25 30 35 40

Price 1994 (Clay+Silt) and OM-Predicted SSA (m^2g)

Figure C-110 Predicted SSA Based on Percent(Clay+Silt) (Price 1994) and OM vs Measured SSA

R2 = 08405

000 1000000 2000000 3000000 4000000 5000000 6000000 7000000

Fe(DCB) (mg-Fekg-Soil)

Figure C-111 FeDCB vs OM

R2 = 05615

000 100000 200000 300000 400000 500000 600000 700000 800000 900000

Al(DCB) (mg-Alkg-Soil)

Figure C-112 AlDCB vs OM

R2 = 06539

0 1 2 3 4 5 6 7

Al(DCB) and C-Predicted OM

Figure C-113 Percent Cb and AlDCB-Predicted OM vs Measured OM

R2 = 00437

-1000000

1000000

2000000

3000000

4000000

5000000

6000000

7000000

000 20000 40000 60000 80000 100000 120000

Fe(Me-1) (mg-Fekg-Soil)

Figure C-114 FeMe-1 vs FeDCB

R2 = 00759

100000

200000

300000

400000

500000

600000

700000

800000

900000

000 50000 100000 150000 200000 250000 300000

Al(Me-1) (mg-Alkg-Soil)

Figure C-115 AlMe-1 vs AlDCB

R2 = 00725

100000

150000

200000

250000

300000

000 10000 20000 30000 40000 50000 60000 70000 80000 90000 100000

PO4(Me-1) (mg-PO4kg-Soil)

Figure C-116 PO4Me-1 vs PO4DCB

R2 = 03282

100000

150000

200000

250000

300000

000 500 1000 1500 2000 2500 3000 3500

PO4(WaterDesorbed) (mg-PO4kg-Soil)

Figure C-117 PO4H2O Desorbed vs PO4DCB

R2 = 01517

000 2000 4000 6000 8000 10000 12000 14000 16000 18000

Water-Desorbed PO4 (mg-PO4kg-Soil)

Figure C-118 PO4Me-1 vs PO4H2O Desorbed

R2 = 06452

0 2 4 6 8 10 12

FeDCB Subsoil Enrichment Ratio

Figure C-119 Subsoil Enrichment Ratios FeDCB vs Clay

R2 = 04012

0 1 2 3 4 5 6

AlDCB Subsoil Enrichment Ratio

Figure C-120 Subsoil Enrichment Ratios AlDCB vs Clay

R2 = 03262

0 10 20 30 40 50 60

SSA Subsoil Enrichment Ratio

Figure C-121 Subsoil Enrichment Ratios SSA vs Clay

C-2 Plots Relating Isotherm Parameters to One Another

R2 = 00161

-20 0 20 40 60 80 100

3-Parameter Q(0) (mg-PO4kg-Soil)

Figure C-21 3-Parameter Q0 vs 5-Parameter Q0

R2 = 00923

-20 0 20 40 60 80 100

Figure C-22 3-Parameter Q0 vs SCS Q0

R2 = 00028

000 50000 100000 150000 200000 250000 300000

Figure C-23 Qmax vs kl Fit Using 3-Parameter Method

C-3 Plots Relating Soil Characteristics to Isotherm Parameters

R2 = 04316

0 05 1 15 2 25 3 35

OM Subsoil Enrichment Ratio

Figure C-31 Subsoil Enrichment Ratios OM vs Qmax

R2 = 00539

1012141618

0 05 1 15 2 25 3 35

Figure C-32 Subsoil Enrichment Ratios OM vs kl

R2 = 08237

0 5 10 15 20 25 30 35 40 45 50

SSA (m^2g)

Figure C-33 SSA vs Qmax

R2 = 048

0 5 10 15 20 25 30 35 40 45

Figure C-34 Clay vs Qmax

R2 = 0583

000 100 200 300 400 500 600 700 800 900 1000

Figure C-35 OM vs Qmax

R2 = 067

000 1000000 2000000 3000000 4000000 5000000 6000000 7000000

FeDCB (mg-Fekg-Soil)

Figure C-36 FeDCB vs Qmax

R2 = 0654

0 10000 20000 30000 40000 50000 60000 70000

Predicted FeDCB (mg-Fekg-Soil)

Figure C-37 Estimated FeDCB vs Qmax

R2 = 05708

000 100000 200000 300000 400000 500000 600000 700000 800000 900000

AlDCB (mg-Alkg-Soil)

Figure C-38 AlDCB vs Qmax

R2 = 08789

0 500 1000 1500 2000 2500

SSA amp OM -Predicted Qmax (mg-PO4kg-Soil)

Figure C-39 SSA and OM-Predicted Qmax vs Qmax

R2 = 08789

0 500 1000 1500 2000 2500

SSA amp FeDCB -Predicted Qmax (mg-PO4kg-Soil)

Figure C-310 SSA and FeDCB-Predicted Qmax vs Qmax

R2 = 08832

000 50000 100000 150000 200000 250000

SSA OM and FeDCB -Predicted Qmax (mg-PO4kg-Soil)

Figure C-311 SSA OM and FeDCB-Predicted Qmax vs Qmax

R2 = 08863

000 50000 100000 150000 200000 250000

SSA OM FeDCB amp AlDCB -Predicted Qmax (mg-PO4kg-Soil)

Figure C-312 SSA OM FeDCB and AlDCB-Predicted Qmax vs Qmax

R2 = 08378

000 50000 100000 150000 200000 250000

SSA amp C -Predicted Qmax (mg-PO4kg-Soil)

Figure C-314 SSA and Cb-Predicted Qmax vs Qmax

R2 = 0888

000 50000 100000 150000 200000 250000

SSA C amp FeDCB -Predicted Qmax (mg-PO4kg-Soil)

Figure C-315 SSA Cb and FeDCB-Predicted Qmax vs Qmax

R2 = 07823

000 50000 100000 150000 200000 250000 300000

SSAFeDCB-Predicted Qmax (mg-PO4kg-Soil)

Figure C-316 SSA-FeDCB Product-Predicted Qmax vs Qmax

R2 = 07651

000 50000 100000 150000 200000 250000 300000

SSAPredictedFeDCB-Predicted Qmax (mg-PO4kg-Soil)

Figure C-317 SSA-Estimated FeDCB Product-Predicted Qmax vs Qmax

R2 = 0768

000 50000 100000 150000 200000 250000

Clay amp OM-Predicted Qmax (mg-PO4kg-Soil)

Figure C-318 SSA-OM Product-Predicted Qmax vs Qmax

R2 = 07781

000 50000 100000 150000 200000 250000

Clay amp FeDCB-Predicted Qmax (mg-PO4kg-Soil)

Figure C-319 Clay and FeDCB -Predicted Qmax vs Qmax

R2 = 07879

0 500 1000 1500 2000 2500

Clay OM amp FeDCB-Predicted Qmax (mg-PO4kg-Soil)

Figure C-320 Clay OM and FeDCB -Predicted Qmax vs Qmax

R2 = 07726

0 500 1000 1500 2000 2500

ClayFeDCB-Predicted Qmax (mg-PO4kg-Soil)

Figure C-321 Clay-FeDCB Product-Predicted Qmax vs Qmax

R2 = 07848

000 50000 100000 150000 200000 250000

ClayPredictedFeDCB-Predicted Qmax (mg-PO4kg-Soil)

Figure C-322 Clay-Estimated FeDCB Product-Predicted Qmax vs Qmax

R2 = 059

000 20000 40000 60000 80000 100000 120000 140000 160000 180000

Clay+OM-Predicted Qmax (mg-PO4kg-Soil)

Figure C-323 Clay-OM Sum-Predicted Qmax vs Qmax

R2 = 08095

0 500 1000 1500 2000 2500

ClayOM-Predicted Qmax (mg-PO4kg-Soil)

Figure C-324 Clay-OM Product-Predicted Qmax vs Qmax

Figure C-325 Clay and OM-Predicted kl vs kl

Figure C-326 Clay OM and FeDCB-Predicted kl vs kl

Figure C-327 FeDCB-Predicted kl vs kl Low-kl Topsoils

Figure C-328 OM-Predicted kl vs kl Low-kl Topsoils

Figure C-329 SSA-FeDCB Product-Predicted kl vs kl Low-kl Topsoils

Figure C-330 Clay-OM Product-Predicted kl vs kl Low-kl Topsoils

Figure C-331 Clay and FeDCB-Predicted kl vs kl Low-kl Topsoils

Figure C-332 Clay and OM-Predicted kl vs kl Low-kl Topsoils

Figure C-333 FeDCB-Predicted kl vs kl High-kl Topsoils

Figure C-334 OM-Predicted kl vs kl High-kl Topsoils

Figure C-335 SSA-FeDCB Product-Predicted kl vs kl High-kl Topsoils

Figure C-336 Clay-OM Product-Predicted kl vs kl High-kl Topsoils

Figure C-337 Clay and FeDCB-Predicted kl vs kl High-kl Topsoils

Figure C-338 Clay and OM-Predicted kl vs kl High-kl Topsoils

Figure C-339 Clay and FeDCB-Predicted kl vs kl Subsoils

Figure C-340 Clay and OM-Predicted kl vs kl Subsoils

Figure C-341 PO4DCB vs Q0 Fit by 3-Parameter Method

Figure C-342 PO4Me-1 vs Q0 Fit by 3-Parameter Method

Figure C-343 PO4H20Desorbed vs Q0 Fit by 3-Parameter Meth

Appendix D

Sediments and Eroded Soil Particle Size Distributions

Containing

Introduction Methods and Materials Results and Discussion Conclusions

Introduction

Sediments are environmental pollutants due to both physical characteristics and

their ability to transport chemical pollutants Sediment alone has been identified as a

leading cause of impairment to 303(d) sites (Kuhnle et al 2001) in the US The EPA has

also historically identified sediment and sediment-related impairments such as increased

turbidity as a leading cause of general water quality impairment in rivers and lakes in its

National Water Quality Inventory (305(b) listed sites) ranging from 2000-2004 (Figure

2000 2002 2004

Figure D1 Contribution of Sediment to Surface Water Impairments in the US (US EPA 2002 2007 2009)

D Sediments and Eroded Soil Particle Size Distributions

Sediment loss can be a costly problem It has been estimated that streams in the

eastern US deposit 129 million tons of sediment in the Atlantic Ocean annually (Curtis et

al 1973) En route sediments can cause much damage Economic losses as a result of

sediment-bound chemical pollution have been estimated at $288 trillion per year

Additionally sediments cause $860 billion per year in biological damages (Osterkamp et

al 1998)

States have varying approaches in assessing water quality and impairment The

State of South Carolina does not directly measure sediment therefore it does not report any

water bodies as being sediment-impaired However South Carolina does declare waters

impaired based on measures directly tied to sediment transport and deposition These

measures of water quality include turbidity and impaired macroinvertebrate populations

They also include a host of pollutants that may be sediment-associated including fecal

coliform counts total P PCBs and various metals

Current sediment control regulations in South Carolina require the lesser of (1)

80 trapping efficiency of disturbed sediments or (2) 05 mLL peak settleable

concentration for site effluent (SC DHEC 2003) These goals are usually pursued through

the use of structural best management practices (BMPs) such as sediment ponds and traps

However these structures depend upon soil particlesrsquo settling velocities to work

According to Stokesrsquo Law then settling velocity becomes a function of eroded particle

size Thus many sediment control structures are only effective at removing the largest

particles which have the most mass In addition eroded particle size distributions the

bases for BMP design have not been well-quantified for the majority of South Carolina

soils nor are they predicted reliably using the Revised CREAMS equations (Price 1994)

This too calls current design practices into question

While removing most of the larger soil particles helps to keep streams from

becoming choked with sediment it does little to protect animals living in the stream In

fact many freshwater fish are quite tolerant of high suspended solids concentration

(measured as mass per volume) (Richards 1992) Meanwhile it is apparent that a better

means of predicting biological impairment is percentage of fine sediments in a water

(Chapman and McLeod 1987) This implies that the eroded particles least likely to be

trapped by structural BMPs are the particles most likely to cause problems for aquatic

organisms

There are similar implications relating to chemistry Smaller particles have greater

specific surface areas Thus small particles can adsorb a larger amount of pollutant per unit

mass by offering more adsorption sites per unit mass This makes fine particles an

important mode of pollutant transport both from disturbed sites and within streams

themselves This implies (1) that pollutant transport in these situations will be difficult to

prevent and (2) that particles leaving a BMP might well have a greater amount of

pollutant-per-particle than particles entering the BMP

Eroded soil particle size distributions are developed by sieve analysis and by

measuring settling velocities with pipette analysis Settling velocity is important because it

controls the effectiveness of BMPs in removing particles of a given size Pipette analysis is

used to measure settling velocity for assumed smooth spherical particles of equal density

in dilute suspension according to the Stokes equation

( )⎥⎦

⎤⎢⎣

⎡minus= 1

gDVs (D1)

where Vs is settling velocity D is particle diameter g is the acceleration due to gravity v is

the kinematic viscosity of the fluid and SG is the specific gravity of the particle (Johns

1998) In order to develop an eroded size distribution the settling velocity is measured and

used to solve for particle diameter for the development of a mass-based percent-finer

Current regulations governing sediment control are based on eroded size

distributions developed from the CREAMS and Revised CREAMS equations These

equations were derived from sieve and pipette analyses of Midwestern soils The

equations note the importance of clay in aggregation and assume that small eroded

aggregates have the same siltclay ratio as the dispersed parent soil in developing a

predictive model that relates parent soil texture to the eroded particle size distribution

(Foster et al 1985)

Unfortunately the Revised CREAMS equations do not appear to be effective in

predicting eroded size distributions for South Carolina soils probably due to regional

variations between soils of the Midwest and soils of the Southeast Two separate studies

using sieve and pipette analyses have demonstrated that the Revised CREAMS equations

are unable to reliably predict eroded soil particle size distributions for the soils in the study

(Price 1994 Johns 1998) However one researcher did find that grouping parent soils

according to clay content provided a strong indicator of a soilrsquos eroded size distribution

(Johns 1998)

Due to the importance of sediment control both in its own right and for the purposes

of containing phosphorus the Revised CREAMS approach itself was studied prior to an

attempt to apply it to South Carolina soils in the hope of producing a South

Carolina-specific CREAMS model in addition uncertainty associated with the Revised

CREAMS approach was evaluated

Methods and Materials

Revised CREAMS Approach

Foster et al (1985) describe the Revised CREAMS approach in great detail 28

soils were evaluated for sand silt and clay content both as dispersed matrix (parent) soil

and undispersed sediment Of these 28 soils 4 were from otherwise unpublished sources

and 24 were from published sources All published data was located and entered into a

Microsoft Excel spreadsheet for analysis they are included here as Table D1 Then with

the data available the Revised CREAMS approach was followed as described with the

goal of recreating the model However because the CREAMS researchers apparently used

different data at various stages of their model it was not possible to precisely recreate it

South Carolina Soil Modeling

Eroded size distributions and parent soil textures from a previous study (Price

1994) were evaluated for potential predictive relationships for southeastern soils The

Revised CREAMS approach was followed (Foster et al 1985) Soil characteristics of

interest include sand silt and clay content siltclay ratio and sediment enrichment ratios

Results and Discussion

Revised CREAMS ApproachD1

Noting that sediment is composed of aggregated and non-aggregated or primary

particles Foster et al (1985) proceed to state that undispersed sediments resulting from

agricultural soils often have bimodal eroded size distributions One peak typically occurs

from 002 to 006 mm and the other typically occurs from 02 to 10 mm To explain this

the authors identify five classes of soil particles a very fine particle class existing below

both peaks is defined as ldquoprimary clayrdquo while the smaller-diameter range is composed of

classes defined as ldquoprimary siltrdquo and ldquosmall aggregatesrdquo and the larger-diameter range is

composed of ldquoprimary sandrdquo and ldquolarge aggregatesrdquo

Young (1980) noted that most clay was eroded in the form of aggregated particles

rather than as primary clay Therefore diameters of each of the two aggregate classes were

estimated with equations selected based upon the clay content of the parent soil with

higher-clay soils having larger aggregates No data and limited justification were

D1 Note To accommodate sieve sizes the Revised CREAMS researchers defined clay as particles with diameters le0004 mm silt as particles with diameters ranging from gt0004 to 0063 mm and sand as particles with diameters greater than 0063 mm This is different from USDA soil texture definitions for clay (le0002 mm) silt (gt0002-005 mm) and sand (gt005 mm)

Table D1 Revised CREAMS Data (Foster et al 1985) Dispersed Soil Matrix Undispersed Sediment Dispersed Sediment

Soil Type Sand ()

Silt ()

Clay ()

Sand ()

Silt ()

Clay ()

Sand ()

Silt ()

Clay ()

Source

Grenada 4 78 18 35 59 6 3 787 183 Cascilla 11 73 16 35 58 7 94 75 156 Sharkey 2 44 54 49 39 12 1 437 553 Bruin 13 71 16 43 54 3 137 682 181 Vicksburg 6 84 10 10 85 5 49 866 85 Lexington 11 64 25 33 58 9 244 567 227 Arkabutla 24 57 19 35 58 7 227 599 174 Grenada 8 7 78 15 16 79 5 54 809 137 Morganfield 8 80 12 11 85 4 43 85 107

Meyer et al 1980

Barnes 37 50 13 40 54 6 37 52 11 Commerce 17 63 20 32 64 4 16 59 25 Marshall 4 60 36 16 72 12 3 61 36 Sharkey 7 41 52 37 49 14 12 51 37

Young et al 1980

Blount cultiv 296 54 164 308 675 17 247 579 174 Blount grassed 243 603 154 35 627 23 269 566 165 Morley cultiv 323 521 156 328 66 22 233 587 18 Morley grassed 309 565 126 307 666 27 25 575 175 Hoytville cultiv 121 664 215 359 616 25 16 537 303 Hoytville grassed 224 553 223 205 768 27 143 579 278

Fertig et al 1982

Hagener 88 8 4 87 12 1 86 8 6 Ida 8 80 12 38 53 9 5 71 24 Marshall 2 62 36 27 67 6 2 57 41 Lutton 2 46 52 33 52 15 1 44 55

Gabriels and Moldenhauer 1978

Miami 19 55 26 51 45 4 Alberts et al 1980 Hoytville 21 46 33 44 51 5 Parr 22 61 17 62 34 4 Martinsville 30 51 19 41 54 5 Blount 33 47 20 55 41 4

Neibling (Unpublished)

Sediments and Eroded Soil Particle Size D

istributions

presented to support the diameter size equations so these were not evaluated further

The initial step in developing the Revised CREAMS equations was based on a

regression relating the primary clay content of sediment to the primary clay content of the

parent soil (Figure D2) forced through the origin because there can be no clay in eroded

sediment if there was not already clay in the parent soil A similar regression line was

found here as reported by Foster et al (1985) (Table D2) Foster et al (1985) appear to

have plotted data from only 22 soils not all 28 soils provided in their data since no

explanation was given all data were plotted in Figure D2 and a similar result was achieved

When an effort was made to base data selections on what appears in Foster et al (1985)

Figure 1 for 18 identifiable data points this study identified the same basic regression

y = 0225x + 06961R2 = 06063

y = 02485xR2 = 05975

0 10 20 30 40 50 60Ocl ()

Clay Not Forced through Origin Forced Through Origin

Figure D2 Primary Clay in Sediment as a Function of Primary Dispersed Clay in Parent Soil After Foster et al (1985)

The next step of the Revised CREAMS derivation involved an estimation of

primary silt and small aggregate content Sieve size dictated that all particles in this class

(le0063 mm) must be composed of only clay particles andor silt particles Using 16 soils

for which the particle composition of small aggregates was known the CREAMS

researchers proceeded by multiplying the clay composition of these particles by the overall

fraction of eroded soil of size le0063 mm thus determining the amount of sediment

composed of clay contained in this size class (each sediment fraction was expressed as a

percentage) Primary clay was subtracted from this total to provide an estimate of the

amount of sediment composed of small aggregate-associated clay Next the CREAMS

researchers apply the assumption that the siltclay ratio is the same within sediment small

aggregates as within corresponding dispersed parent soil by multiplying the small

aggregate-associated clay by the siltclay ratio to determine the small aggregate-associated

silt fraction In order to estimate the total small aggregate fraction small

aggregate-associated clay and silt are then summed In order to estimate primary silt

content the authors applied an additional assumption enrichment in the 0004- to

00063-mm class is due to primary silt that is to silt which is not associated with

aggregates

In order to predict small aggregate content of eroded sediment a regression

analysis was performed on data from the 16 soils just described and corresponding

dispersed clay content of the parent soils Foster et al (1985) assumed that clay was

necessary for aggregation and thus forced the regression through the origin due to scatter

they also forced the regression to run through the mean of the data The 16 soils were not

specified Further the figure in Foster et al (1985) showing the regression displays data

from only 10 soils The sourced material does not clarify which soils were used as only

Meyer et al (1980) used the 0004- to 0063-mm classification system employed by Foster

et al (1985) although 18 soils used similar binning based upon the standard USDA

textural definitions So regression analyses for the Meyer soils alone (generally identified

by evaluating Foster et al 1985 Figure 2) and for the 18 soils which reported compositions

of small aggregates were performed the small aggregate fraction was related to the

primary clay fraction from dispersed parent soils (Figures D3-D4 and Table D2) Similar

results were found for soils with primary clay fraction lt25

Soils with clay fractions greater than 50 were modeled using a rounded average

of the sediment small aggregateparent soil primary clay ratio While the numbers differed

slightly using the same approach yielded the same rounded average when all 18 soils were

considered The approach then assumes that the small aggregate fraction varies linearly

with respect to the parent soil primary clay fraction between 25-50 clay with only one

data point to support or refute the assumption

y = 27108x

0 5 10 15 20 25 30 35 40

Ocl ()

All Data lt25 Ocl Arithmetic Mean Linear (Arithmetic Mean)

Figure D3 Small Aggregates in Sediment as a Function of Primary Dispersed Clay in Parent Soil After Foster et al (1985)

y = 19558x

0 10 20 30 40 50 60Ocl ()

Meyers Data lt25 Ocl Arithmetic Mean Linear (Arithmetic Mean)

Figure D4 Small Aggregates in Sediment as a Function of Primary Dispersed Clay in Parent Soil After Foster et al (1985) Selected Soils

To estimate the primary silt fraction Foster et al assumed that the eroded silt-sized

fraction (0004 to 0063 mm) consisting of primary silt and small aggregates equals the

dispersed silt fraction of the parent soil This assumption was evaluated for 20 unidentified

soils but a plot for parent soil silt versus silt-sized particles for all 28 soils given by Foster

et al was provided (Figure D5)

Primary sand and large aggregate classes were also estimated Estimates were

based on the assumption that primary sand in the sand-sized undispersed sediment

composes the same fraction as it does in the matrix soil Thus any additional material in the

sand-sized class must be composed of some combination of clay and silt Based on this

assumption Foster et al (1985) developed an equation relating the primary sand fraction of

sediment directly to the dispersed clay content of parent soils using a calculated average

value of five as the exponent Finally the large aggregate fraction is determined by

difference

For the sake of clarity it should be noted that there are several different soil textural

classes of interest here Among the eroded soils are unaggregated sand silt and clay in

addition to aggregates in both the sand-sized (large aggregates) and silt-sized (small

aggregates) classes Together these five classes compose 100 of eroded sediment and

they may be compared to undispersed eroded size distributions by noting that both silt and

silt-sized aggregates compose the silt-sized class and that both sand and sand-sized

aggregates compose the sand-sized class The aggregated classes are composed of silt and

clay that can be dispersed in order to determine the make up of the eroded sediment with

respect to unaggregated particle size also summing to 100

y = 07079x + 16454R2 = 05002

y = 09703xR2 = 04267

0102030405060708090

0 20 40 60 80 100

Osi ()

Silt Average

Not Forced Through Origin Forced Through Origin

Figure D5 Silt-Sized Fraction of Undispersed Sediment vs Primary Silt Fraction of Parent Soil After Foster et al (1985)

Table D2 Foster et al (1985) Regressions vs Reproduced Regressions Foster et al 1985 Reproduced (2010)

Compared to Measured Data

Description

Classification Regression Regression R2 Std Er

Small Aggregate Diameter (Dsg)D2

Ocl lt 025 025 le Ocl le 060

Ocl gt 060

Dsg = 0030 Dsg = 02(Ocl ndash 025) + 0030

Dsg = 0100 - - -

Large Aggregate Diameter (Dlg) D2

015 le Ocl 015 gt Ocl

Dlg = 0300 Dlg = 2(Ocl)

Eroded Primary Clay Content (Fcl) vs Ocl

- Fcl = 026(Ocl) All Data Fcl = 025(Ocl)

Selected Data Fcl = 026 (Ocl) 087 087

493 493

Ocl lt 025 Fsg = 18(Ocl) All Data Fsg = 27(Ocl)

Meyers Data Fsg = 20(Ocl) - D3 - D3

- D3 - D3

Eroded Small Aggregate Fraction (Fsg) vs Ocl 025 le Ocl le 050

Ocl gt 050 Fsg = 045-06(Ocl ndash 025)

Fsg = 06(Ocl) Fsg = 045-06(Ocl ndash 025) All Data Fsg = 06(Ocl)

- D3 - D3

Eroded Primary Silt Fraction (Fsi) vs Dispersed Silt Content of Parent Soil (Osi) and the Small Aggregate Fraction

- Fsi = Osi - Fsg Fsi = Osi - Fsg - D3 - D3

Eroded Primary Sand Fraction vs Dispersed Sand Content of Parent Soil (Osa) and Ocl

- Fsa = Osa(1-Ocl)5 Fsa = Osa(1-Ocl)6 - D3 - D3

Large Aggregate - Flg = 1 ndash Fcl ndash Fsi ndash Fsg - Fsa Flg = 1 ndash Fcl ndash Fsi ndash Fsg - Fsa - D3 - D3

D2 Based upon dispersed clay content of parent soil (Ocl) D3 Aggregates silt and sand were not measured directly as each bin contained some combination of aggregates and primary particles

istributions

Because of the difficulties in differentiating between aggregated and unaggregated

fractions within the silt- and sand-sized classes a direct comparison between measured

data and estimates provided by the Revised CREAMS method is impossible even with the

data used to develop the approach Two techniques for indirectly evaluating the approach

are (1) to compare undispersed sediment fractions to Revised CREAMS-estimated

fractions by comparing data for silt-and sand-sized classes and (2) comparing dispersed

sediment fractions to estimates of dispersed sand silt and clay provided by Foster et al

(1985) in the following equations estimating the amount of clay and silt contained in

aggregates

Small Aggregate Clay = Ocl(Ocl + Osi) (D2)

Small Aggregate Silt = Osi(Ocl + Osi) (D3)

Large Aggregate Clay = [Ocl - Fcl ndash (FsgSmall Aggregate Clay)]Flg (D4)

Large Aggregate Silt = [Osi ndash Fsi ndash (FsgSmall Aggregate Silt)]Flg (D5)

Both techniques for evaluating uncertainty are presented here Data for approach 1

are plotted in Figures D6-D8 Data for approach 2 are plotted in Figures D9-D11 Finally

a chart providing standard errors for the regression lines for both approaches is provided in

Table D3

y = 08709x + 08084R2 = 06411

0 5 10 15 20

Revised CREAMS-Estimated Clay-Sized Class ()

Data 11 Line Linear (Data)

Figure D6 Revised CREAMS-Estimated Clay-Sized Class Fraction vs Measured Clay-Sized Class Fraction from Undispersed Sediment

y = 07049x + 16646R2 = 04988

0 20 40 60 80 100

Revised CREAMS-Estimated Silt-Sized Class ()

Figure D7 Revised CREAMS-Estimated Silt-Sized Class Fraction vs Measured Silt-Sized Class Fraction from Undispersed Sediment

y = 0756x + 93275R2 = 05345

0 20 40 60 80 100

Revised CREAMS-Estimated Sand-Sized Class ()

Figure D8 Revised CREAMS-Estimated Sand-Sized Class Fraction vs Measured Sand-Sized Class Fraction from Undispersed Sediment

y = 14423x + 28328R2 = 08616

0 10 20 30 40

Revised CREAMS-Estimated Dispersed Clay ()

Figure D9 Revised CREAMS-Estimated Dispersed Clay Fraction vs Measured Clay Fraction from Dispersed Sediment

y = 08097x + 17734R2 = 08631

0 20 40 60 80 100

Revised CREAMS-Estimated Dispersed Silt ()

Figure D10 Revised CREAMS-Estimated Dispersed Silt Fraction vs Measured Silt Fraction from Dispersed Sediment

y = 11691x + 65806R2 = 08921

0 20 40 60 80 100

Revised CREAMS-Estimated Dispersed Sand ()

Figure D11 Revised CREAMS-Estimated Dispersed Sand Fraction vs Measured Sand Fraction from Dispersed Sediment

Interestingly enough for the soils for which the Revised CREAMS equations were

developed the equations actually provide better estimates of dispersed soil fractions than

undispersed soil fractions This is interesting because the Revised CREAMS researchers

seemed to be primarily focused on aggregate formation The regressions conducted above

indicate that both dispersed and undispersed estimates could be improved by adjustment

however In addition while the Revised CREAMS approach is an improvement over a

direct regressions between dispersed parent soils and undispersed sediments a direct

regression is a superior approach for estimating dispersed sediments for the modeled soils

(Table D4)

Table D3 Standard Error Associated With Revised CREAMS Estimates of Soil Fractions

Approach Soil Fraction Standard Error () Clay 1096 Silt 1126 Undispersed

Sand 227 Clay 613 Silt 625 Dispersed

Sand 512

Table D4 Direct Regressions Between Revised CREAMS Parent Soils and Undispersed (n = 28) and Dispersed (n=23) Sediments

Regression Coefficient Intercept

Sign St

Error ()

Coeff ()

St Error ()

Intercept ()

St Error ()

Undispersed Clay 94E-7 237 023 004 0701 091 061

Undispersed Silt 26E-5 1125 071 014 16451 842 050

Undispersed Sand 12E-4 1204 060 013 2494 339 044

Dispersed Clay 81E-11 493 089 007 3621 197 087

Dispersed Silt 30E-12 518 094 007 3451 412 091

Dispersed Sand 19E-14 451 094 005 0061 129 094

1 p gt 005

The South Carolina soils analyzed by Price (1994) generally exhibited the bimodal

eroded size distributions described by Foster et al (1985) Because aggregates are

important for settling calculations an attempt was made to fit the Revised CREAMS

approach to the South Carolina soils as previous attempts (Price 1994 and Johns 1998) at

modeling had demonstrated that the Revised CREAMS equations had not adequately

modeled eroded size distributions Clay content had been directly measured by Price

(1994) silt and sand content were estimated via linear interpolation

Unfortunately from the very beginning the Revised CREAMS approach seems to

break down for the South Carolina soils Primary clay in sediment does not seem to be

related to clay in dispersed parent soils (Figure D12) Similarly poor results were found for

the silt and clay fractions as well even when soils were broken into top- and subsoil groups

or grouped by location (Figure D13)

y = 01724x

0 10 20 30 40 50

Clay in Dispersed Parent Soil

R2 = 000

Figure D12 Clay Fraction in Dispersed Parent Soil vs Clay Fraction in Undispersed Sediment South Carolina Soils This was disappointing because it indicated that there were great differences

between the soils analyzed by the Revised CREAMS researchers and the South Carolina

soils In addition Johns (1994) and Price (1998) had not used dispersant to analyze

aggregation choosing only to model undispersed sediment So while it would be possible

to make some of the same assumptions used by the Revised CREAMS researchers they

would be impossible to evaluate or confirm Also even without the assumptions applied

by Foster et al (1985) to develop the equations for aggregated sediments the Revised

CREAMS soils showed fairly strong correlations between parent soil and sediment for

each soil fraction while the South Carolina soils show no such correlation Another

difference is that the South Carolina soils do not show enrichment in the sand-sized class

indicating the absence of large aggregates and lack of primary sand displacement Only the

silt-sized class is enriched in the South Carolina soils indicating that silt is either

preferentially displaced or that clay-sized particles are primarily contributing to small

silt-sized aggregates in sediment

10121416

0 10 20 30 40 50

Simpson Sandhills Edisto Pee Dee Coastal

Figure D13 Clay Fraction in Dispersed Parent Soil vs Clay Fraction in Undispersed Sediment South Carolina Soils Grouped by REC

These factors are generally opposed to the observations and assumptions of the

Revised CREAMS researchers However the following assumptions were made for

South Carolina soils following the approach of Foster et al (1985)

bull Sand seems to be preferentially undisturbed by rainfall Thus sand displacement

into sediment will be the next component to be modeled via regression

bull Remaining sediment must be composed of clay and silt Small aggregation will be

estimated based on the assumption that neither clay nor silt are preferentially

disturbed by rainfall

It appears that the data for sand are more grouped than for clay (Figure D14) A

regression line was fit through the data and forced through the origin as there can be no

sand in the sediment without sand in the parent soil Given the assumption that neither clay

nor silt are preferentially disturbed by rainfall it follows that small aggregates are

composed of the same siltclay ratio as in the parent soil unfortunately this can not be

verified based on the absence of dispersed sediment data

y = 07993x

0 20 40 60 80 100

Sand in Dispersed Parent Soil

R2 = 000

Figure D14 Sand Fraction in Dispersed Parent Soil vs Sand Fraction in Undispersed Sediment

The average enrichment ratio in the silt-sized class was 244 Given the assumption

that silt is not preferentially disturbed it follows that the excess sediment in this class is

small aggregate Thus equations D6 through D11 were developed to describe

characteristics of undispersed sediment

Fcl = 017Ocl (D6)Fsa = 080Osa (D7)Fsg = 144Osi (D8)

Fsi = 1 ndash Fcl ndash Fsa ndash Fsg (D9)Small Aggregate Clay = Fsg[Ocl(Osi+Ocl)] (D10)Small Aggregate Silt = Fsg[Osi(Osi+Ocl)] (D11)

The accuracy of this approach was evaluated by comparing the experimental data

for undispersed sediment to estimates derived using equations D6-D11 Unfortunately

regressions were quite poor (Table D5) This indicates that the data do not support the

assumptions made in order to develop equations D6-D11 which was suspected based upon

the poor regressions between size fractions of eroded sediments and parent soils this is in

contrast to the Revised CREAMS soils for which data provided strong fits for simple

direct regressions In addition the absence of data on the dispersed size distribution of

eroded sediments forced the assumption that the siltclay ratio was the same in eroded

sediments as in parent soils

Table D5 Regressions Between Soil Fraction Estimates Developed Using Equations D6-D11 and Measured Data from Undispersed (n=34) Sediments

Sign St

Error ()

Coeff ()

St Error ()

Intercept ()

St Error ()

Clay 0631 322 0161 033 309 100 001 Silt 0241 1531 0381 032 20001 1325 004 Sand 0161 1702 0441 030 32411 1747 006

1 p gt 005

While previous researchers had proven that the Revised CREAMS equations do not

fit South Carolina soils well this work has demonstrated that the assumptions made by

Foster et al (1985) in developing the model also cannot be applied to South Carolina soils

as defined by existing experimental data Possible explanations include the fact that the

South Carolina soils have a lower clay content than the Revised CREAMS soils In

addition there was greater spread among clay contents for the South Carolina soils than for

the Revised CREAMS soils One of the key assumptions in the Revised CREAMS

approach is that clay plays an important role in aggregation so clay content of South

Carolina soils could be an important contributor to the failure of this approach In addition

the Revised CREAMS soils are generally dissimilar to SC soils in terms of soil taxonomy

(Table D6)

Conclusions

The Revised CREAMS equations effectively modeled the soils upon which they

were based However direct regressions would have modeled eroded particle size

distributions for the selected soils almost as well Based on the analyses of Price (1994)

and Johns (1998) the Revised CREAMS equations do not provide an effective model for

estimating eroded particle size distributions for South Carolina soils Using the raw data

upon which the previous analyses were based this study indicates that the assumptions

made in the development of the Revised CREAMS equations are not applicable to South

Carolina soils

Table D6 Revised CREAMS Soils (Foster et al 1985 USDA-NRCS 2010a) Soil Type Order Suborder Great group Subgroup

Modifier Particle Size Mineralogy Soil Temp States MLR

Grenada Alfisols Udalfs Fraglossudalfs Oxyaquic Fine-silty Mixed Thermic AR KY LA MS TN 134

Cascilla Inceptisols Udepts Dystrudepts Fluventic Fine-silty Mixed Thermic KY LA MS TN 134

Sharkey Vertisols Aquerts Epiaquerts Chromic Very-fine Smectitic Thermic AR KY LA MO MS TN

Bruin Inceptisols Udepts Eutrudepts Oxyaquic Coarse-silty Mixed Thermic LA MS 131

Vicksburg Entisols Fluvents Udifluvents Typic Coarse-silty Mixed Thermic KY LA MS TN

131 134

Lexington Alfisols Udalfs Hapludalfs Ultic Fine-silty Mixed Thermic AR KY LA MS TN

133A 134

Arkabutla Inceptisols Aquepts Endoaquepts Fluventic Fine-silty Mixed Thermic AR KY LA MS OK TN 134

Morganfield Entisols Fluvents Udifluvents Typic Coarse-silty Mixed Thermic AR KY LA MS TN 134

Barnes Mollisols Udolls Hapludolls Calcic Fine-loamy Mixed Frigid MN ND SD

102A 55A 55B

Commerce Inceptisols Aquepts Endoaquepts Fluvaquentic Fine-silty Mixed Thermic AR KY LA MO MS TN 131

Marshall Mollisols Udolls Hapludolls Typic Fine-silty Mixed Mesic IA KS MO NE

102B 106 107 109

Blount Alfisols Aqualfs Epiaqualfs Aeric Fine Illitic Mesic IL IN MI OH WI

108 110 111 95B

97 98 99

Morley Alfisols Udalfs Hapludalfs Oxyaquic Fine Illitic Mesic IL IN MI OH WI

108 110 111 95B

97 98 99

istributions

Table D6 (Continued) Revised CREAMS Soils (Foster et al 1985 USDA-NRCS 2010a) Soil Type Order Suborder Great group Subgroup

Modifier Particle Size Mineralogy Soil Temp States MLRAs

Hoytville Alfisols Aqualfs Epiaqualfs Mollic Fine Illitic Mesic IN MI OH

Hagener None Available

None Available None Available None Available None Available None

Available None

Available IL None Available

Ida Entisols Orthents Udorthents Typic Fine-silty Mixed Mesic IA MO NE 107

Lutton None Available

Available None

AvailableNone

Available None

Available

Miami Alfisols Udalfs Hapludalfs Oxyaquic Fine-loamy Mixed Mesic IL IN MI OH WI

108 110 111 113 114 115 95B 97

98 Parr

Mollisols Udolls Argiudolls Oxyaquic Fine-loamy Mixed Mesic IL IN OH

108 110 111 95B

Martinsville Alfisols Udalfs Hapludalfs Typic Fine-loamy Mixed Mesic IL IN MI OH

105 108 110 111 114 115 95B 97 98 99

istributions

Appendix E

BMP Study

Containing

Introduction

The goal of this thesis was based on the concept that sediment-related nutrient

pollution would be related to the adsorptive potential of parent soil material A case study

to develop and analyze adsorption isotherms from both the influent and the effluent of a

sediment pond was carried out to evaluate this hypothesis and to evaluate its application to

a common construction best management practice (BMP) Thus the pondrsquos effectiveness

in stopping not only total suspended solids but also phosphate and phosphate-adsorbing

potential could be evaluated

Permission was obtained to sample a sediment pond at a development in southern

Greenville County South Carolina The drainage area had an area of 705 acres and was

entirely composed of Cecil soil (Figures E1 and E2) approximately 5 acres were disturbed

at the time of sampling Runoff was collected and routed to the pond via storm drains

which had been installed along curbed and paved roadways The pond was in the shape of

a long rectangle approximately 15 feet wide by 150 feet long (Figure E3) and was

equipped with an 18-inch concrete inlet pipe and a 75-inch polyvinyl chloride (PVC)

outlet pipe installed on a 1 grade and discharging below the pond behind double silt

fences The pond discharge structure was located in the lower end of the pond it was

composed of a corrugated polyethylene standpipe of approximately 40rdquo in diameter

E BMP Study

surrounded by gravel The standpipe connected to the 75-inch PVC outlet below a row of

eight 5-inch holes (Figure E4)

Figure E1 Location Map Showing Drainage Area and 4-ft Topographical Contours (GVL CO SC 2010b)

E BMP Study

Figure E2 NRCS Soil Survey (USDA NRCS 2010)

Figure E3 Sediment Pond

E BMP Study

Figure E4 Sediment Pond Discharge Structure

The sampled storm took place over a one-hour time period in April 2006 The

storm resulted in approximately 04-inches of rain over that time period at the site The

pond was discharging a small amount of water that was not possible to sample prior to the

storm Four minutes after rainfall began runoff began discharging to the pond the outlet

began discharging eight minutes later Runoff ceased discharging to the pond about 2

hours after the storm had passed and the pond returned to its pre-storm discharge condition

about 40 minutes later

Over the course of the storm samples of both pond influent and effluent were taken

at timed ten-minute intervals with clean 500-mL mason jars Jars were swept across the

entire flow of the inlet pipe and the outlet standpipe structure Flow rate was established

E BMP Study

when samples were taken using a calibrated bucket and stopwatch Samples were then

composited according to a flow-weighted average

Total suspended solids and turbidity analyses were conducted as described in the

main body of this thesis This established a TSS concentration for both the influent and

effluent composite samples necessary for proper dosing with PO4 and for later

normalization by FeDCB and AlDCB It also provided a measure of quality assurance within

the isotherm experiment itself An adsorption experiment was then conducted as

previously described in the main body of this thesis and used to develop isotherms using

the 3-Parameter Method

Next a variation of the dithionite-citrate-bicarbonate (DCB) experiment was

conducted in order to establish FeDCB AlDCB and PO4DCB content of the eroded soil

material flowing into and out of the sediment pond In this case 25 mL of stirred

composite solution were withdrawn and sodium citrate and bicarbonate were added as dry

measured powders (22 g and 21 g respectively) rather than adding sodium citrate and

bicarbonate solutions to a measured amount of dry soil as before

Finally the composite samples were analyzed for particle size by sieve and pipette

analysis

Sieve Analysis

Sieve analysis was conducted by straining the water-sediment mixture through a

series of six stainless steel sieves with mesh openings of 20 10 05 025 0125 and

0063 mm The 20 10 and 05 mm sieves were placed in a 1325 L bucket and the

mixture strained through each sieve three times Then these sieves were replaced by 025

E BMP Study

0125 and 0063 mm sieves which were also used to strain the mixture three times What

was left in suspension was saved for pipette analysis The sieves were washed clean and the

sediment deposited into pre-weighed jars The jars were then dried to constant weight at

105degC and the mass of soil collected on each sieve was determined by the mass difference

of the jars (Johns 1998) When large amounts of material were left on the sieves between

each straining the underside was gently sprayed to loosen any fine material that may be

clinging to larger sediments otherwise data might have indicated a higher concentration

of large particles (Meyer and Scott 1983)

Pipette Analysis

Pipette analysis was used to establish the eroded particle size distribution and is

based on the settling velocities of suspended particles of varying size assuming spherical

shape and a specific gravity equal to 265 The water-sediment mixture was thoroughly

mixed and 12 liters were poured into a glass cylinder The test procedure is

temperature-dependent based on Stokesrsquo Law and the viscosity of water so the

temperature of the water-sediment solution was recorded The sample in the glass cylinder

was mixed by stirring and pipettes were used to draw off 25 mL of liquid at specified

depths and at specified times (Table E1)

Solution withdrawal with the pipette began 5 seconds before the designated

withdrawal time so that the pipette was halfway-filled at the specified time (Meyer and

Scott 1983) An aluminum dish pre-weighed to the nearest 01 milligram for each 25 mL

sample was placed in a drying oven set at 105degC for twelve hours or more to achieve

E BMP Study

constant weight Then weight differences were calculated to establish the mass of sediment

in each aluminum dish (Johns 1998)

Table E1 Time and depth requirements for pipette tests at 20degC (Haan et al 1994) Particle Diameter (mm)

0063 062 031 016 008 004 002

Withdrawal Depth (cm) 15 15 15 10 10 5 5

Time (hrminsec) 00000 00044 00252 00740 03040 10119 40400

The final step in establishing the eroded particle size distribution was to develop

cumulative particle size distribution curves that show the percentage of particles (by mass)

that are smaller than a given particle size First the total mass of suspended solids was

calculated For the sieved particles this required summing the mass of material caught by

each individual sieve Then mass of the suspended particles was calculated for the

pipette-analyzed portion To do this sediment mass for an unsettled (t=0) sample at initial

concentration was found and used to calculate the total mass of pipette-analyzed suspended

solids Total mass of suspended solids was found by adding the total pipette-analyzed

suspended solid mass to the total sieved mass Example calculations are given below for a

25-mL pipette

MSsample = MSsieve + MSpipette (E1)

MSsample = total sample mass (g) MSsieve = Σ(sieved masses for each size class) (g)

MSpipette = [Mt=025mL Vbucket(mL)] (g) Vbucket = Volume of Sieve Sample Bucket

E BMP Study

The mass of material contained in each sieve particle-size category was determined by

dry-weight differences between material captured on each sieve The mass of material in

each pipetted category was determined by the following subtraction function

MSpipette size = MSprevious pipette size ndash MSparticle size of interest (E2)

This data was then used to calculate percent-finer for each particle size of interest (20 10

050 025 0125 0063 0031 0016 0008 0004 0002 mm) (Johns 1998)

Flow measurements were complicated by the pondrsquos discharge structure and outfall

location The pond discharged into a hole from which it was impossible to sample or

obtain flow measurements Therefore flow measurements were taken from the holes

within the discharge structure standpipe Four of the eight holes were plugged so that little

or no flow was taking place through them samples and flow measurements were obtained

from the remaining holes which were assumed to provide equal flow However this

proved untrue as evidenced by the fact that several of the remaining holes ceased

discharging as runoff decreased at the eventrsquos conclusion Further proving the faultiness of

this assumption was the fact that summed flows for effluent using this method would have

resulted in an effluent volume (55789 L) nearly four times as large as the influent volume

(14673 L) This could not have been correct as a pond cannot discharge more water than

it receives therefore a normalization factor relating total influent flow to effluent flow was

developed by dividing the summed influent volume by the summed effluent volume The

E BMP Study

resulting factor of 026 was then applied to each discrete effluent flow measurement by

multiplication the resulting hydrographs are shown below in Figure E5

0 50 100 150 200 250

Minutes After Pond Began to Receive Runoff

Influent Effluent

Figure E5 Influent and Normalized Effluent Hydrographs

Sediments

Results indicated that the pond was trapping about 26 of the eroded soil which

entered This corresponded with a 4-5 drop in turbidity across the length of the pond

over the sampled period (Table E2) As expected the particle size distribution indicated

that the pond effluent contained a greater percentage of fine particles (Figure E6) this was

expected because sediment pond design results in preferential trapping of larger particles

Due to the associated increase in SSA this caused sediment-associated concentrations of

PO4DCB FeDCB and AlDCB to increase on a per-mass basis (Table E3) In addition this

resulted in increases in Q0 kl and Qmax as fit using the 3-Parameter Method (Tables E4-6

and Figures E7 and E8)

E BMP Study

Table E2 TSS and Turbidity in Composite Influent and Effluent Samples

TSS (g L-1)

Turbidity 30-s(NTU)

Turbidity 60-s (NTU)

Influent 111 1376 1363 Effluent 082 1319 1297

Figure E6 Eroded Size Distributions Sediment Pond Inflent and Effluent Table E3 Sediment-Associated Concentrations of PO4DCB FeDCB and AlDCB

PO4DCB (mgPO4 kgSoil

-1) FeDCB

(mgFe kgSoil-1)

AlDCB (mgAl kgSoil

-1) Influent 338325 3223675 556458 Effluent 512946 3819701 730243

E BMP Study

Table E4 Influent Sediment Ad- Table E5 Effluent Sediment Ad- sorption Data sorption Data

C Q Adsorbed mg L-1 mg kg-1 ()

015 -13281 020 -17674 043 52461 5751 040 52703 5933 130 158223 5732 163 129364 4664 421 170135 3085 403 189698 3420 685 199565 2435 665 221593 2688 937 192557 1849 933 200494 1918

C Q Adsorbedmg L-1 mg kg-1 ()

013 -15731 015 -18071 034 79060 6581 035 77450 6471 166 166919 4548 160 175295 4768 285 3912621 5324 427 224346 3034 692 260782 2382 691 263102 2401 812 187240 1606 834 237681 1912

1 Discarded as obvious outlier Table E6 3-Parameter Fit Isotherm Parameters

Qmax (mgPO4 kgSoil

-1) kl

(L mg-1) Q0

(mgPO4 kgSoil-1)

Influent 422979 227 102831 Effluent 503062 315 127500 Cecil Subsoil 191316 063 15639

Figure E7 Adsorption Data and 3-Parameter Fit Isotherm Pond Influent

E BMP Study

Figure E8 Adsorption Data and 3-Parameter Fit Isotherm Pond Effluent

Because the disturbed soils would likely have been defined as subsoils using the

definitions of Price (1994) the Cecil subsoil adsorption data and isotherm parameters

previously described should be representative of the parent soil type The greater kl and

Qmax determined in this BMP study are likely due to the greater surface area of eroded soils

relative to parent soils as smaller particles are more likely to be displaced by rainfall

Further evidence of this phenomenon is found by comparing increases in Qmax and kl as a

result of trapping of larger particles in the pond (Figure E8) Thus preferential trapping of

larger particles results in greater PO4-adsorption potential per unit mass among the smaller

particles which remain in solution

E BMP Study

Figure E9 Isotherm Models of Phosphate Adsorption for a Cecil Subsoil vs Pond Influent and Effluent

Estimates of the pondrsquos effectiveness in removing pollutants and PO4-adsorbing

potential from solution can be determined by calculating the mass of sediment trapped in

the pond and then by applying the values for PO4DCB and Qmax to the trapped sediment via

multiplication Since no runoff was apparently detained in the pond the influent volume

(14673 L) was approximately equal to the effluent volume This volume was multiplied

by the TSS concentrations determined previously to provide mass-based estimates of the

amount of sediment trapped by the pond Results are provided in Table E7

Table E7 Trapped Pollutants and Phosphate-Adsorbing Potential Sediment

(kg) PO4DCB

(g) PO4-Adsorbing Potential

(g) Influent 1629 - - Effluent 1203 - - Trapped 426 1441 1802

E BMP Study

Conclusions

At the time of the sampled storm this pond was not particularly effective in

removing sediment from solution or in detaining stormwater Clearly larger particles are

preferentially removed from this and similar ponds due to gravity settling The smaller

particles which remain in solution both contain greater amounts of PO4 and also are

capable of adsorbing greater amounts of PO4 per unit mass However the sediment that

was trapped resulted in removal of both previously-adsorbed PO4 (measured as PO4DCB)

and of PO4-Adsorbing Potential (identified by measurements of Qmax) from solution

REFERENCES

Atalay A (2001) Variation in phosphorus sorption with soil particle size Soil and Sediment Contamination 10(3) 317-335

Barrow NJ (1983) A mechanistic model for describing the sorption and desorption of phosphate by soil Journal of Soil Science 34 733-750

Barrow NJ (1984) Modeling the effects of pH on phosphate sorption by soils Journal

of Soil Science 35 283-297 Bennett EM Carpenter SR and Caraco NF (2001) Human impact on erodable

phosphorus and eutrophication A global perspective BioScience 51(3) 227- 234

Carpenter SR Caraco NF Correll DL Howarth RW Sharpley AN and

Smith VH (1998) Nonpoint pollution of surface waters with phosphorus and nitrogen Ecological Applications 8(3) 559-568

Chapman DW and McLeod KP (1987) Development of criteria for fine sediment in

the Northern Rockies ecoregion EPA9109-87-162 United States Environmental Protection Agency Seattle WA

Chen B Hulston J and Beckett R (2000) The effect of surface coatings on the

association of orthophosphate with natural colloids The Science of the Total Environment 263 23-35

[CU ASL] Clemson University Agricultural Service Laboratory (2000) Soil Analysis

Procedures Clemson University Clemson SC Accessed 14 September 2006 lthttpwwwclemsoneduagsrvlbprocedures2interesthtmgt

[CU ASL] Clemson University Agricultural Service Laboratory (2009) Compost

Analysis Procedures Clemson University Clemson SC Accessed 16 August 2009 lthttphttpwwwclemsonedupublicregulatoryag_svc_labcompost compost_procedurestotal_nitrogenhtmlgt

Curtis WF Culbertson JK and Chase EB (1973) Fluvial-sediment discharge to the

oceans from the conterminous United States 17 US Geological Survey Circular 670

Foster GR Young RA Neibling WH (1985) Sediment composition for nonpoint

source pollution analyses Transactions of the ASAE 28(1) 133-139

Frossard E Brossard M Hedley MJ and Metherell A (1995) Reactions controlling the cycling of P in soils pg 107-138 in Phosphorus in the Global Environment Transfers Cycles and Management ed H Tiessen John Wiley amp Sons New York

Graetz DA and Nair VD (2009) Phosphorus sorption isotherm determination pg

35-38 in Methods of Phosphorus Analysis for Soils Sediments Residuals and Waters Southern Cooperative Series Bulletin No 408 North Carolina State University Raleigh NC Accessed February 27 2010 lthttpwwwsera17 extvteduDocumentsP_Methods2ndEdition2009pdfgt

Greenbert AE Clesceri LS Eaton AD eds (1992) Standard Methods for the

Examination of Water and Wastewater pg 2-56 Washington DC American Public Health Association

[GVL CO SC] Greenville County South Carolina (2010a) IDEAL Integrated Design Evaluation and Assessment of Loadings Model Accessed April 15 2010 lthttp wwwgreenvillecountyorgland_developmentpdfStorm_Water_IDEAL_Model_ Section5pdfgt

[GVL CO SC] Greenville County South Carolina (2010b) Internet Mapping System

Accessed March 4 2010 lthttpwwwgcgisorgwebmappubgt Griffith GE Omernik JM Comstock JA Schafale MP McNab WH Lenat DR

MacPherson TF Glover JB and Shelburne VB (2002) Ecoregions of North Carolina and South Carolina (color poster with map descriptive text summary tables and photographs) Reston Virginia USGeological Survey Accessed 16 August 2009 ltftpftpepagovwedecoregionsnc_scncsc_frontpdfgt

Haan CT Barfield BJ and Hayes JC (1994) Design Hydrology and Sedimentology

for Small Catchments Academic Press San Diego

Hayes JC and Price JW (1995) Estimation of eroded particle sizes for sediment control in South Carolina Stormwater Management and Sediment Control Sedimentology Resource SC DHEC

Herren EC (1979) Soil Survey of Anderson County South Carolina United States Department of Agriculture Soil Conservation Service US Govrsquot Printing Office

Herrera DA (2003) Relationships Among Mehlich-1 Mehlich-3 and Water-Soluble Phosphorus Levels in Manure-Amended Inorganically Fertilized and Phosphatic Soils unpublished Masterrsquos thesis University of Florida Gainesville FL

Hesse PR (1973) Phosphorus in lake sediments pg 573-583 in Environmental Phosphorus Handbook ed EJ Griffith et al John Wiley amp Sons New York

Hiemstra T Antelo J Rahnemaie R van Riemsdijk WH (2010a) Nanoparticles in natural systems I The effective reactive surface area of the natural oxide fraction in field samples Geochimica et Cosmochimica Acta 74 41-58

Hiemstra T Antelo J van Rotterdam AMD van Riemsdijk WH (2010b)

Nanoparticles in natural systems II The natural oxide fraction at interaction with natural organic matter and phosphate Geochimica et Cosmochimica Acta 74 41-58

Hur J Schlautman MA Karanfil T Smink J Song H Klaine SJ Hayes JC (2007) Influence of drought and municipal sewage effluents on the baseflow water chemistry of an upper Piedmont river Environmental Monitoring and Assessment 132171-187 Jarvie HP Juumlrgens MD Williams RJ Neal C Davies JJL Barrett C and White

J (2005) Role of river bed sediments as sources and sinks of phosphorus across two major eutrophic UK river basins The Hampshire Avon and the Herefordshire Wye Journal of Hydrology 304 51-74

Johns JP 1998 Eroded Particle Size Distributions Using Rainfall Simulation of South

Carolina Soils unpublished Masterrsquos thesis Clemson University Clemson SC Johnson WH 1995 Sorption Models for U Cs and Cd on Upper Atlantic Coastal Plain

Soils unpublished Doctoral thesis Georgia Institute of Technology Atlanta GA

Kaiser K and Guggenberger G (2003) Mineral surfaces and soil organic matter European Journal of Soil Science 54 219-236

Kuhnle RA Simon A and Knight SS (2001) Developing linkages between sediment

load and biological impairment for clean sediment TMDLs Wetlands Engineering and RiverRestoration Conference Reno Nevada August 27-31 2001 ASCE

Lawrence CB (1978) Soil Survey of Richland County South Carolina United States

Department of Agriculture Soil Conservation Service US Govrsquot Printing Office Lovejoy SB Lee JG Randhir TO and Engel BA (1997) Research needs for water

quality management in the 21st century a spatial decision support system Journal of Soil and Water Conservation 50 383-388

McDowell RW and Sharpley AN (2001) Approximating phosphorus release from soils to surface runoff and subsurface drainage Journal of Environmental Quality 30 508-520

McGechan MB and Lewis DR (2002) Sorption of phosphorus by soil part 1

Principles equations and models Biosystems Engineering 82 1-24 Meyer LD and Scott SH (1983) Possible errors during field evaluations of sediment

size distributions Transactions of the ASAE 12(6)754-758762

Miller EN Jr (1971) Soil Survey of Charleston County South Carolina United States Department of Agriculture Soil Conservation Service US Govrsquot Printing Office

Morton R (1996) Soil Survey of Darlington County South Carolina United States Department of Agriculture Natural Resources Conservation Service US Govrsquot Printing Office

Mueller DK and Spahr NE (2006) Nutrients in streams and rivers across the nation ndash 1992-2001 United States Geologic Survey Scientific Investigations Report 2006-5107

Osterkamp WR Heilman P and Lane LJ (1998) Economic considerations of a

continental sediment-monitoring program International Journal of Sediment Research 13 12-24

Parfitt RL (1978) Anion adsorption by soils and soil minerals Advances in

Agronomy 30 1-42

Price JW (1994) Eroded Soil Particle Distributions in South Carolina unpublished Masterrsquos thesis Clemson University Clemson SC

Reid LK (2008) Stormwater Export of Nitrogen Phosphorus and Carbon From Developing Catchments in the Upper Piedmont Physiographic Province unpublished Masterrsquos thesis Clemson University Clemson SC

Richards C (1992) Ecological effects of fine sediments in stream ecosystems

Proceedings of the USEPA and USDA FS Technical Workshop on Sediments pg 113-118 Corvallis Oregon

Rogers VA (1977) Soil Survey of Barnwell County South Carolina Eastern Part United States Department of Agriculture Soil Conservation Service US Govrsquot Printing Office

Rongzhong Y Wright AL Inglett K Wang Y Ogram AV and Reddy KR (2009) Land-use effects on soil nutrient cycling and microbial community dynamics in the Everglades Agricultural Area Florida Communications in Soil Science and Plant Analysis 40 2725ndash2742

Sayin M Mermut AR and Tiessen H (1990) Phosphate sorption-desorption

characteristics by magnetically separated soil fractions Soil Science Society of America Journal 54 1298-1304

Schindler DW (1977) Evolution of phosphorus limitation in lakes Science 195260-

Sims JT (2009) Soil test phosphorus Mehlich 1 pg 15-16 in Methods of Phosphorus Analysis for Soils Sediments Residuals and Waters Southern Cooperative Series Bulletin No 408 North Carolina State University Raleigh NC Accessed February 27 2010 lthttpwwwsera17extvteduDocuments P_Methods 2ndEdition2009pdf gt

Sposito G (1984) The Surface Chemistry of Soils Oxford University Press New York [SCDHEC] South Carolina Department of Health and Environmental Control (2003)

Stormwater Management and Sediment Control Handbook for Land Disturbance Activities Accessed September 14 2006 lthttpwwwscdhecgoveqcwater pubsswmanualpdfgt

[USDA-NRCS] United States Department of Agriculture Natural Resources Conservation Service (2006) Land Resource Regions and Major Land Resource Areas of the United States the Caribbean and the Pacific Basin United States Department of Agriculture Washington DC Handbook 296 Accessed 15 August 2009 ltftp ftp-fcscegovusdagovNSSCAg_Handbook_296Handbook_296_lowpdfgt

[USDA-NRCS] United States Department of Agriculture Natural Resources Conservation

Service (2009) Soil Data Mart Washington DC Accessed August 17 2009 lthttpsoildatamartnrcsusdagovgt

[USDA-NRCS] United States Department of Agriculture Natural Resources Conservation Service (2010a) Official Soil Series Descriptions Washington DC Accessed March 11 2010 lthttpsoilsusdagovtechnicalclassificationosdindexhtmlgt

[USDA-NRCS] United States Department of Agriculture Natural Resources Conservation Service (2010b) Web Soil Survey Washington DC Accessed March 4 2010 lthttpwebsoilsurveynrcsusdagovappWebSoilSurveyaspxgt

[USEPA] United States Environmental Protection Agency (2002) National Water Quality Inventory 2000 Report to Congress Office of Water Washington DC EPA-841-R-02-001

[USEPA] United States Environmental Protection Agency (2007) National Water

Quality Inventory 2002 Report to Congress Office of Water Washington DC EPA-841-R-07-001

[USGS] United States Geologic Survey (1999) The quality of our nationrsquos waters nutrients and pesticides Circ 1225 Denver CO USGS Info Serv

[USGS] United States Geologic Survey (2003) A Tapestry of Time and Terrain Accessed 15 August 2009 lthttptapestryusgsgovDefaulthtmlgt

Van Der Zee SEATM MM Nederlof Van Riemsdijk WH and De Haan FAM

(1988) Spatial variability of phosphate adsorption parameters Journal of Environmental Quality 17(4) 682-688

Wang Z Zhang B Song K Liu D Ren C Zhang S Hu L Yang H and Liu Z

(2009) Landscape and land-use effects on the spatial variation of soil chemical properties Communications in Soil Science and Plant Analysis 40 2389-2412

Weld JL Parsons RL Beegle DB Sharpley AN Gburek WJ and Clouser WR

(2002) Evaluation of phosphorus-based nutrient management strategies in Pennsylvania Journal of Soil and Water Conservation 57(6) 448-454

Wiriyakitnateekul W Suddhiprakarn A Kheuruenromne I and Gilkes RJ (2005) Extractable iron and aluminum predict the P sorption capacity of Thai soils Australian Journal of Soil Research 43 757-766

Young RA (1980) Characteristics of eroded sediment Transactions of the ASAE 23

1139-1142

Clemson University

TigerPrints

5-2010
Modeling Phosphate Adsorption for South Carolina Soils
Jesse Cannon
Recommended Citation
Microsoft Word - JWCThesis_SAMampJWC_FINAL2doc