SPARROW Modeling in the Mississippi and Atchafalaya River Basins (MARB) Dale Robertson, Richard Alexander, and David Saad U.S. Geological Survey USGS/USEPA WEBEX Meeting June 29, 2007
Mar 29, 2015
SPARROW Modeling in the Mississippi and Atchafalaya River Basins (MARB)
Dale Robertson, Richard Alexander, and David Saad
U.S. Geological Survey
USGS/USEPAWEBEX Meeting
June 29, 2007
Outline• Overview of SPARROW
• Recent advances in SPARROW and applications to the Mississippi/Atchafalaya R. Basin
• Summary of nitrogen and phosphorus results for large regional basins
• Preliminary watershed rankings – nutrient delivery to the Gulf
• Future SPARROW modeling
SPARROW Water-Quality ModelSPAtially Referenced Regression on Watershed Attributes
http://water.usgs.gov/nawqa/sparrow; Smith et al. 1997
Hybrid statistical and mechanistic process structure; mass-balance constraints; data-driven, nonlinear estimation of parameters
Spatially explicit; separates land and water processes
Physically interpretable coefficients; model supports hypothesis testing and uncertainty estimation
Predictions of mean-annual flux reflect long-term, net effects of nutrient supply and loss processes in watersheds
Sources
Land-to-watertransport
SPARROW modeling approach:- Regress water-quality conditions (monitored load) on upstream sources and factors controlling transport- Incorporates instream decay of nutrients
Monitored load
Instreamtransport
Nonlinear Regression Model
SPARROWSPAtially Referenced Regressions On Watershed Attributes
For each watershed
Estimation of mean-annual nutrient load
at stream monitoring sites – Model Inputs
Log
Load
(k
g/da
y)
1992 Base YearActual LoadPredicted Load
Load Confidence Interval
1978
1982
1986 19
9019
9419
76 1980 19
8419
8819
92
Mean-annual TN load for 1992 base year (detrended; flow-adjusted 1975-2000)
SPARROW’s Reach-Scale Mass Balance
Reach network relates watershed datato monitored loads
Load leaving a reach =
Load generated within upstream reaches and
transported to the reach via the stream network
+
Load originating within the reach’s incremental watershed and delivered
to the reach segment
Earlier SPARROW ResultsEarlier SPARROW ResultsTotal Nitrogen Delivery to the Gulf of MexicoTotal Nitrogen Delivery to the Gulf of Mexico
1987 Base Year1987 Base YearAgriculture Municipal Wastewater
Atmosphere
Alexander et al. 2000, Nature
• Model structure: specification, flux-routing algorithms, stream monitoring loads, documentation
• Data infrastructure: climate, 1-km DEM, 30-m NLCD land use, cropping and drainage systems
• Result:• Added complexity• Model accuracy
improved by 20%
Recent Advances in SPARROW
ClimateWatersheds
Topography
Land Use
Water
Ice, snow
High intensity residential
Low intensity residential
Quarries, strip mines, gravel pits
Transitional
Bare rock, sand, clay
Commercial, industrial, transportation
Deciduous forest
Mixed forest
Evergreen forest Grasslands, herbaceous
Pasture, hay
Orchards, vineyards, other
Shrubland Row crops
Small grains
Urban, recreational grasses
Fallow
Emergent herbaceous wetlands
Woody wetlands
NLCD 1K
Artificial Drainage
SPARROW Sources andTransport Features
NUTRIENT SOURCES (1992)•Urban and population sources•Atmospheric N deposition•Farm fertilizer use allocated to
major crops:– County fertilizer sales and
expenditures; crop acreage– NLCD agricultural land use– State application rates (corn,
soybeans, cotton, wheat, other crops)
– Corn/soybean rotations•N2 fixation – cultivated lands•Animal manure:
– Non-recoverable on pasture/rangelands
– Recoverable on crops•Natural and residual sources
(lands in forest, barren, shrub)
AQUATIC ATTENUATION•Streams
– First-order decay ~ f(water travel time, flow and depth)
•Reservoirs– First-order decay ~ f(areal
hydraulic load—ratio of outflow to surface area)
LAND-TO-WATER DELIVERY•Climate (precipitation,
temperature)•Soils (permeability)•Topography/subsurface (slope,
specific catchment area)•Artificial drainage (tiles,
ditches)
SPARROW Delivery of Agricultural Nutrients to Streams
CROPNUTRIENTS
COMMERCIAL FERTILIZER
BIOLOGICALN2 FIXATION
ANIMAL MANURE(Non-recoverable)
UNCONFINED ANIMALS
RECOVERABLEMANURE
CONFINED ANIMALS
Model Source &Delivery Coefficients
Model Source &Delivery Coefficients
STREAMS & RESERVOIRS
N
P
Harvesting
The Improved SPARROW Nutrient Models
Observed vs. Predicted Yield
Stream and Reservoir Transport for 1992
SPARROW Rates of Aquatic Nutrient Loss
Nutrient removal rate declines in larger riversand more rapidly flushed reservoirs
Nitrogen literature rates from Howarth et al. 1996; Seitzinger et al. 2002; Bohlke et al. 2004; Mulholland et al. 2004)
STREAMS RESERVOIRS
Percentage of Stream Nutrients Delivered to the Gulf of Mexico
Total Nitrogen Total Phosphorus
Remove1 kg at Gulf outlet
Remove1.1 kg = 1/0.9
Remove4 kg = 1/0.25
Aquatic Removal of Nutrients in MARB Regional Watersheds - 1992
Regional Watersheds
Nutrient Source Contributions
to Stream Flux:Types and Regional
Geography
Sources Contributions to Stream Nutrient Flux - 1992
Mississippi River at St. Francisville, LA
Regional Contributions to the Stream Nutrient Flux to the Gulf of
Mexico - 1992Regional Watersheds
Changes in Stream Nutrient Flux, 1992 to
2002•Simulate changes in flow-adjusted stream nutrient flux
•Account for changes in population and agriculture (animal manure; crop fertilizer application, acreage, and production)
•Account for changes in harvested nutrients with changes in marginal rate of crop production
•Assume steady-state conditions with constant model coefficients over time
Simulated Changes in Flow-Adjusted Nutrient Flux, 1992 to
2002
•Changes in flux typically less than 5%
•Geography of 2002 source shares are generally unchanged from 1992
*statistically sig. (p<0.06)
MARB Watershed Rankings
Nutrient Delivery to the Gulf of Mexico for 1992
Preliminary watershed rankings based on nutrient delivery to the Gulf
Try to answer questions that have been popping up during past discussions.
Future SPARROW modeling- Additional Refinements to the SPARROW models
Definitions:Load – Total amount of a constituent transported – (kgs)
Incremental Yield – Amount of a constituent transported per unit area between two points – kg/km2
Delivered Incremental Yield – Amount of a constituent transported per unit area between two points that is delivered or transported to some specific point – kg/km2
Yield – Total amount of a constituent transported per unit area – kg/km2
Total Nitrogen Load – Compare with Monitoring Data
- Can be used to estimate instream concentrations
Top 4 %
1992 Nitrogen SPARROW Model Output
Total Nitrogen – Incremental Yield- Can be used to demonstrate the highest export areas
1992 Nitrogen SPARROW Model Output
Total Nitrogen – Incremental YieldTop 10 %
1992 Nitrogen SPARROW Model Output
Total Nitrogen – Delivered Incremental Yield to the Gulf
1992 Nitrogen SPARROW Model Output
Total Nitrogen – Delivered Incremental Yield to the Gulf
Top 10 %
1992 Nitrogen SPARROW Model Output
Total Nitrogen – Delivered Incremental Yield
HUC 8 Scale
1992 Nitrogen SPARROW Model Output
Total Nitrogen – Delivered Incremental Yield
Top 100
Top 100 represent 42% of the Total Load
Total Nitrogen – Delivered Incremental Yield
Top 100
1992 Nitrogen SPARROW Model Output
Ranked Top 100HUC 8’s
Total Nitrogen – Ranked based on total delivered incremental yield
Ranked Top 10HUC 8’s
Rank HUC HUC NAME STATES Area (km2)
Total Nitrogen Incremental Load (kgs)
Delivered Incremental Yield (kg/km2)
1 7120003 Chicago IL IN 1,256 6,561,711 5,224.32 5120205 Flatrock-Haw IN 1,101 3,162,907 2,872.83 8020201 New Madrid-St. Johns KY MO 1,340 3,828,405 2,857.04 5120113 Lower Wabash IL IN KY 3,303 9,075,317 2,747.65 5120107 Wildcat IN 1,898 5,088,816 2,681.16 5120206 Upper East Fork White IN 2,246 5,719,073 2,546.37 7120005 Upper Illinois IL 2,482 5,875,363 2,367.28 5140202 Highland-Pigeon IN KY 2,707 6,403,427 2,365.59 5120204 Driftwood IN 2,872 6,759,048 2,353.410 5120105 Middle Wabash-Deer IN 1,828 4,260,060 2,330.4
1992 Nitrogen SPARROW Model Output
1,500 km2
HUC8 MRB3 National
Types of Basins
101
102
103
104
105
106A
rea
(km
2)
SPARROW Calibration Basins
160 km2207 km2
5th Percentile
3,200 km2
SPARROW Median Basin Size
86 km2
10% <12 km2
Best to predict for Basins > 200 km2
0
200,000,000
400,000,000
600,000,000
800,000,000
1,000,000,000
1,200,000,000
1,400,000,000
1,600,000,000
0 100 200 300 400 500 600 700 800
Ranked HUCs
To
tal N
itro
ge
n L
oa
din
g, k
gs
0%
20%
40%
60%
80%
100%
Incremental Load
Accumulated Nitrogen Loading
1992 Nitrogen SPARROW Model Output
To Obtain a 30% Reduction in the Total Nitrogen Load
Number of Basins (HUC 8) Required to Aquire Specific Nitrogen Load Reductions to the Gulf
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 100 200 300 400 500 600 700 800Number of Basins Required to Achieve Reduction
Pe
rce
nt
Re
du
cti
on
to
th
e G
ulf
0
200,000,000
400,000,000
600,000,000
800,000,000
1,000,000,000
1,200,000,000
1,400,000,000
1,600,000,000
To
tal P
os
sib
le L
oa
d R
ed
uc
tio
n,
Kg
s
100% Removal in HUC90%80%70%60%50%40%30%20%10%
With a 100% Removal in TN Load, it would require the top 68 HUCs
With a 50% Removal in TN Load, it would require the top 171 HUCs
Total Phosphorus Loading
Top 4 %
1992 Phosphorus SPARROW Model Output
Total Phosphorus – Delivered Incremental Yield
Top 10 %
1992 Phosphorus SPARROW Model Output
Total Phosphorus – Delivered Incremental Yield
HUC 8 Scale
1992 Phosphorus SPARROW Model Output
Total Phosphorus – Delivered Incremental Yield
HUC 8 ScaleTop 100
Top 100 > 42% of the Total Load
Total Phosphorus – Delivered Incremental Yield
HUC 8 ScaleTop 100
1992 Phosphorus SPARROW Model Output
Ranked Top 100HUC 8’s
Total Phosphorus – Ranked based on total delivered incremental yield
Rank HUC HUC NAME STATES Area (km2)
Total P Incremental Load (kgs)
Delivered Incremental Yield (kg/km2)
1 7120003 Chicago IL IN 1,256 633,967 504.7502 8020201 New Madrid-St. Johns KY MO 1,340 403,649 301.2303 5120101 Upper Wabash IN OH 4,606 1,226,551 266.2944 8020204 Little River Ditches AR KY MO 6,336 1,617,929 255.3555 8030207 Big Sunflower AR MS 6,956 1,687,703 242.6266 8030209 Deer-Steele AR LA MS 1,863 434,846 233.4127 5100102 South Fork Licking KY 2,391 540,428 226.0268 8070100 Lower Mississippi-Baton Rouge LA 2,292 464,357 202.5999 8090100 Lower Mississippi-New Orleans LA 1,583 311,200 196.58910 5120206 Upper East Fork White IN 2,246 441,432 196.541
Ranked Top 10HUC 8’s
1992 Phosphorus SPARROW Model Output
0
20,000,000
40,000,000
60,000,000
80,000,000
100,000,000
120,000,000
140,000,000
0 100 200 300 400 500 600 700 800
Ranked HUCs
To
tal P
ho
sp
ho
rus
Lo
ad
ing
, kg
s
0%
20%
40%
60%
80%
100%
Incremental Load
Accumulated Phosphorus Loading
1992 Phosphorus SPARROW Model Output
Number of Basins (HUC 8) Required to Aquire Specific Phosphorus Load Reductions to the Gulf
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
110%
0 100 200 300 400 500 600 700 800Number of Basins Required to Achieve Reduction
Pe
rce
nt
Re
du
cti
on
to
th
e G
ulf
0
20,000,000
40,000,000
60,000,000
80,000,000
100,000,000
120,000,000
140,000,000
To
tal P
os
sib
le L
oa
d R
ed
uc
tio
n,
Kg
s
100% Removal in HUC90%80%70%60%50%40%30%20%10%
To Obtain a 30% Reduction in the Total Phosphorus Load
With a 100% Removal in TP Load, it would require the top 70 HUCs
With a 50% Removal in TP Load, it would require the top 162 HUCs
U.S. Geological Survey SPARROW modelsNational Model – Richard Alexander, G. Schwarz, and R. Smith
1992 and 2002 Models
Dale Robertson, WI
Richard Rebich, MS
Lori Sprague, CO
Major River BasinLead MRB Scientists
Anne Hoos, TN
Richard Moore
Mississippi River SPARROW Model
Dale Robertson, WI
Richard Rebich, MS
Lori Sprague, CO
Mississippi River SPARROW Coordinator: Dale Robertson
Richard Alexander, VA
Future Improvements from Regional SPARROW Models
1. Better spatial resolution – More sites and especially more smaller sites, should lead to more accurate predictions at smaller scales.
2. Further reductions in biases.
3. Better definition of source terms – better point-source data, more sites in unique areas, possible better local GIS inputs.
4. Better able to address more regional and local questions.
Approximately 475-600 sites used in National SPARROW Models(Number of sites used in models varies by constituent)
~1000 Potential Load Sites for MRB3 SPARROW Model
WQ Data: -NWIS -STORET -States (IN,IL, WI)
Flow Data: -NWIS
Potential load site
EXPLANATION
Additional Sites to be Added for Model Calibration
~1000 Potential Load Sites for MRB3 SPARROW ModelPotential Sites in the Missouri River Basin
~350 sites~750-1000 sites in the Lower Mississippi River Basin
Standardized Residuals Map for Total Phosphorus Further Reduction in Spatial Biases
Standardized Residuals
Refinement of Source Contributions to Stream Nutrient Flux
Mississippi River at St. Francisville, LA
Better able to address more regional and local questions
SPARROW Modeling in the Mississippi and Great Lakes Basins
Captain Jack Sparrow