subwatershed prioritization in arkansas - a modeling approach

Subwatershed Prioritizationin Arkansas

- A Modeling Approach

Presented by:Dharmendra Saraswat

Co-Authors: N. Pai, M. Daniels, and M. Leh

ANRC Task Force Meeting, September 21, 2010

Reviewed by: U of A Division of AgricultureEnvironmental Task Force

http://www.oregon.sierraclub.org/groups/rogue/photos/IllinoisRiver_G.%20Lyford.jpg

http://images.google.com/imgres?imgurl=http://www.farmersguardian.com/Pictures/web/d/x/t/poultry.jpg&imgrefurl=http://www.farmersguardian.com/story.asp?sectioncode=38&storycode=19553&usg=__12e-63tY8XFhcYZAqAQkzc02R7w=&h=348&w=370&sz=80&hl=en&start=69&sig2=sN9xiwPoyIktBeafyJPvcw&um=1&tbnid=ncnx6vRcYwf6eM:&tbnh=115&tbnw=122&prev=/images?q=poultry&ndsp=20&hl=en&rls=com.microsoft:en-us:IE-SearchBox&sa=N&start=60&um=1&ei=3hBMSp65K4_GlAfsxqHLDQ

http://images.google.com/imgres?imgurl=http://www.farmersguardian.com/Pictures/web/d/x/t/poultry.jpg&imgrefurl=http://www.farmersguardian.com/story.asp?sectioncode=38&storycode=19553&usg=__12e-63tY8XFhcYZAqAQkzc02R7w=&h=348&w=370&sz=80&hl=en&start=69&sig2=sN9xiwPoyIktBeafyJPvcw&um=1&tbnid=ncnx6vRcYwf6eM:&tbnh=115&tbnw=122&prev=/images?q=poultry&ndsp=20&hl=en&rls=com.microsoft:en-us:IE-SearchBox&sa=N&start=60&um=1&ei=3hBMSp65K4_GlAfsxqHLDQ

MODELING OBJECTIVETo assist in assessing the relative risk of sub-watersheds (12-digit hydrologic unit code) for sediment and nutrients contribution from selected 8-digit HUC priority watersheds using State-of-the Art modeling techniques


Priority Watersheds 12-digit HUC 12-digit HUC Priority

MODELING TASKSa) Collect and organize watershed specific input data for setting up the Soil and

Water Assessment Tool (SWAT) model

b) Calibrate and validate the SWAT model for flow, sediment, and nutrients (total phosphorus and nitrate-nitrogen)


MODEL PHILOSOPHY

Public Domain, supported by USDA-ARS, Worldwide

Users and developers base, 682 published articles


MODEL PROCESS- LAND PHASEFor smallest modeling scale, SWAT simulates spatially-variable

soil and water processes that influence movement of nutrients and sediment from land to water

(Source: Neitsch et al. 2005)


MODEL PROCESS- LAND PHASEMajor Simulation Elements

Weather Hydrology Plant growth Erosion Nutrients (N, P) Pesticides Management



MODEL PROCESS- IN-STREAM



MODELING INPUTS- SPATIAL DATA

SWAT 2009

Elevation(DEM-10/30m)

USGS Boundary(12 Digit HUC)

Soil Data(SSURGO)

Landuse(MULTIPLE

YEARS)

Streams(NHD)

NEXRAD RADAR/Weather

gages


MODELING INPUTS- TABULAR DATA

SWAT 2009

Non-Point SourceAgricultural Land-use

practices for each county

Point source discharge


Average daily loadings for:water, sediment, ammonia, Carbonaceous biochemical oxygen demand dissolved oxygen and phosphorus

MODEL EVALUATION PROTOCOL

Six components of model evaluation:

Model examination Algorithm examination Data evaluation: quality of data

Sensitivity analysis: highest influence

Validation: system performance

Uncertainty analysis (recent trends)

(Source: American Society for Testing and Materials, 1984)


WHY TAKE ALL THIS TROUBLE?

Reduce potential modeler bias Provide a road map to follow Allow other to repeat the study Improve acceptance of model

results


1. Arrange simulated annual pollutant concentration

2. Sort low to high

3. Calculate rank

4. Calculate percentile rank

Tied Rank functionality in Excel 2010

Low

To

High

PERCENTILE RANKING


CASE STUDY

Beaver Reservoir

Lake Conway Point Remove (LCPR)

Bayou Bartholomew

Illinois River Drainage Area in Arkansas (IRDAA)


CASE STUDY

ILLINOIS RIVER DRAINAGE AREA

IN ARKANSAS (IRDAA)


MODEL SETUP- COMPARISONModel: SWAT2005DEM: 10 mSoil: SSURGOLand Use: 2006 HRUs: Criteria(10% land use, 5% soil type, 0%

slope)Ponds: 9019 (3831 acres)Study Period:Calibration: 1996-2005(1992-

95)Validation: 2006-2007 Flow at three gages, sediment and nutrients at outlet

Nitrate-nitrogen yield from a few forestry dominated subwatersheds in southern part of IRDAA did not match with expected yield

Lack of all forms of nitrogen data from point sources could result in higher loading from landscape.


Model: SWAT2009DEM: 10 mSoil: SSURGOLand Use: 1992, 1993, 1999,

2001, 2004, and 2006 HRUs: 848Ponds: 9019 (3831 acres)Study Period:Calibration: 1996-2005(1992-

95)Validation: 2006-2008 Flow at total seven gages, sediment and nutrients at three

MODEL SETUP- COMPARISON

Few studies have used watershed models in spatially distributed fashion

Spatially distributed calibration and validation account for hydrologic patterns in subwatersheds*

* Qi and Grunwald, 2005


Weather

* NOAA -NWS

DATA USED

NEXT GENERATION RADAR (NEXRAD)


ETF Recommendation:

4%, 21%, and 75% for ammonia-, organic-, and nitrate-nitrogen- based on ETF recommendation

Point Source Loading

DATA USED


Poultry litter application adjusted

-Litter application rate adjusted for type of grasses and temporally as well*-Cool-season: 2 ton/ac Warm season: 3 ton/ac-January 2004 onwards, rate declined to 1 ton/ac

Management Data

DATA USED

(*Sharpley et al., 2009)


CALIBRATION/VALIDATION

Subwatershed#16- ~44%, (80%) #19- ~60% (~89%), #18- 24% (~65%)


Percentile Approach – Sediment

SUBWATERSHED IDENTIFICATION

Highest priority:

Subwatershed#

Urban%

Pasture%

Forest%

1 44.3 35.5 14.6

2 5.3 62.1 24.9

4 13.9 67.7 14.1

5 31.1 54.5 10.9

12 5.3 56.9 33.4

23 46.0 35.9 12.7

Soil under Hydrologic Group C & D- 73.4%)


HOC, MFL, LOC, SC,MC,SC

Subwatershed#

Urban%

Pasture%

Forest%

7 3.9 58.9 32.1

11 2.5 36.1 58.7

15 1.5 27.9 67.2

16 5.3 40.3 47.7

25 0.5 22.8 73.5



Lowest priority:

Slopes > 8%- 46.4% Soil under Hydrologic Group C & D- 67.6%)


UBC, CH, LW, UMF, UEC

Subwatershed#

Urban%

Pasture%

Forest%

2* 44.3 35.5 14.6

3 7.2 52.8 34.8

4 13.9 67.7 14.1

5* 31.1 54.5 10.9

18* 11.4 56.3 25.8

23* 46.0 35.9 12.7

Percentile Approach – Total Phosphorus


Highest priority:

*Point source affected *23- 303(d) listSoil -C & D- 74%


MFC, HFC, LOC,SC, GC, SC

Subwatershed#

Urban%

Pasture%

Forest%

6 4.7 35.6 55.6

11 2.5 36.1 58.7

15 1.5 27.9 67.2

16 5.3 40.3 47.7



Lowest priority:


LF, CH, LW, UMF

Subwatershed#

Urban%

Pasture%

Forest%

2 5.3 62.1 24.9

4 13.9 67.7 14.1

5* 31.1 54.5 10.9

6 4.7 35.6 55.6

7 3.9 58.9 32.1

Percentile Approach – Nitrate-Nitrogen


Highest priority:

*point source affected, 5 is in 303(d) for nitrate Slopes > 8%- 9.6%


MFC, LOC, SC, LF, UBC



Lowest priority:

Subwatershed#

Urban%

Pasture%

Forest%

9 2.0 38.1 55.9

21 33.7 29.1 29.4

22 40.5 28.7 21.5

26 3.2 36.6 54.6

27 2.2 39.6 54.3


CASE STUDY

BEAVER RESERVOIR WATERSHED

(BRW)


Model: SWAT2009Delineation: USGS 12-digit HUCDEM: 30 mSoil: SSURGOLand Use: 1992, 1999, 2001, 2004,

and 2006 Subwatersheds: 76HRUs: 2199(5% land use, 10% soil type, 0% slope)Ponds: 6100 (4857 acres)Study Period:Calibration: 1996-2005 (2003-2005)Validation: 2006-2008 Point Sources: 23

MODEL SETUP


Weather

* NOAA -NWS

DATA USED

NEXT GENERATION RADAR (NEXRAD)


MONITORING GAGES



SUBWATERSHED IDENTIFICATION Highest priority:

Subwatershed#

Forest%

Pasture%

Urban%

1* 49 38 58* 65 16 615 41 47 216 43 46 117 65 21 246 57 20 1647 49 16 2959 62 29 160 73 19 161 70 23 162 47 39 663 63 28 264* 51 31 1066 37 53 267 57 34 3


West Fork - 303(d) – sediment since 1998 , TMDL

* PS


SUBWATERSHED IDENTIFICATION Lowest priority:

Subwatershed#

Forest%

Pasture%

Urban%

2* 38 47 47 45 41 2

10 60 28 012 61 11 920 80 7 022 84 1 224 76 14 426 82 4 227 58 3 532 56 29 434 83 10 241 70 20 144 73 19 145 57 32 271 80 11 4

* flatter topography results in lower sediment loss


Percentile Approach – Total P


Subwatershed#

Forest%

Pasture%

Urban%

3* 24 57 64 21 61 26 26 52 3

8* 65 16 69 37 46 411 48 35 213 61 7 1415 41 47 216 43 46 117 65 21 245 57 32 263 63 28 264* 51 31 1065 43 45 466 37 53 2

* Point source affected


Percentile Approach – Total P

SUBWATERSHED IDENTIFICATION Lowest priority:

Subwatershed#

Forest%

Pasture%

Urban%

7 45 41 222 84 1 224 76 14 426 82 4 232 56 29 435 80 13 136 84 11 139 79 14 140 73 18 142 72 18 169 91 5 071 80 11 472 81 8 673 95 3 074 95 3 175 90 7 1


Percentile Approach – Nitrogen-N


Subwatershed#

Forest%

Pasture%

Urban%

1* 49 38 53 24 57 64 21 61 25 45 40 26 26 52 38* 65 16 69 37 46 410 60 28 011 48 35 216 43 46 117 65 33 245 57 32 2

64* 51 31 1065 43 45 466 37 53 2

* Point source affected



Subwatershed#

Forest%

Pasture%

Urban%

2 38 47 420 80 7 122 84 1 235 80 13 136 84 11 140 73 18 151 82 12 152 50 30 953 82 13 156 83 12 069 81 12 171 80 11 473 95 3 074 95 3 175 90 7 176 88 7 2

Percentile Approach – Nitrogen-N Lowest priority:


CASE STUDY

LAKE CONWAY POINT REMOVE

WATERSHED (LCPRW)


Model: SWAT2009Delineation: USGS 12-digit HUCDEM: 10 mSoil: SSURGOLand Use: 1999, 2004, and 2006 Subwatersheds: 29HRUs: 2668(0% land use, 5% soil type, 0% slope)Ponds: 6100 (4857 acres)Study Period:Calibration: 2001-2005(Flow)Validation: 2006-2008 (Flow)Point Sources: 34

MODEL SETUP


LEAST MONITORED, MOST RAINGAGES

Weather

DATA USED


Row Crops : Dominant Crop- Soybean

Management Data

DATA USED

Date Operation Rate

May 1 - May 15Tillage couple days before

planting

Disk 2X and

Float 1X

May 1 - May 15 Planting

May 15 - May 30 Fertilization with 0-0-60 100 lb/acre

May 15 - July 15Pesticide glyphosate once per

month 1 lb/acre/month

July 15 - Sep 15 Irrigation ~3 inches/month

Oct 1 - Oct 20 Harvest and kill


Arkansas River as Inlet

DATA USED

USGS gauge near Dardanelle (# 07258000)

Stage-Discharge relationship:Discharge (cfs) = 7408 x Stage (feet) - 30234; R2 = 0.77

+/- 23% error in estimating missing data should be expected (J.E. Funkhouser, USGS)


CALIBRATION/VALIDATION

Flow calibration: 2001-2005, Validation: 2006-2008 Land Use of covered subwatersheds: Forests: 61.7% - 98.4% Pasture: 1.2% - 35.2% Urban: 0.1% - 2.7%

Forest (~54%) and Pasture (~22%)

Flow missing: October 2004- September 2005 October 2007 – September 2008




Highest priority:

Subwatershed#

Forest%

Crop%

Pasture%

Urban%

14 20 44.6 20.4 5.518 58.8 - 16.9 3.925 39.3 - 39.1 1626 36.2 - 29.6 5.828 14.4 44.7 30.1 9.429 42.7 17.8 23.4 10.8


Subwatershed #18- Beaverdam Creek



Lowest priority:

Subwatershed#

Forest%

Crop%

Pasture%

Urban%

1 76.8 - 20.6 1.62 98.4 - 1.2 0.13 64.2 - 29.0 3.24 52.5 - 35.2 3.8

19 72.8 - 15.2 5.3




Highest priority:

Subwatershed#

Forest%

Crop%

Pasture%

Urban%

14 20 44.6 20.4 5.518* 58.8 - 16.9 3.925 39.3 - 39.1 16.026 36.2 - 29.6 5.828 14.4 44.7 30.1 9.429 42.7 17.8 23.4 10.8

Soils C & D *Low streamflow




Lowest priority:

Subwatershed#

Forest%

Crop%

Pasture%

Urban%

1 76.8 - 20.6 1.62 98.4 - 1.2 0.14 52.5 - 35.2 3.87 66.1 - 31.1 1.99 75.0 - 23.4 1.1




Highest priority:

Subwatershed#

Forest%

Crop%

Pasture%

Urban%

6 33.3 8.6 31.6 20.014 20.0 44.6 20.4 5.517 33.2 - 26.1 35.8

19* 72.8 - 15.2 5.325 39.3 - 39.1 16.026 36.2 - 29.6 5.8

#6- Whig Creek (10 mi)#17- City of Conway WWTP


Subwatershed#

Forest%

Crop%

Pasture%

Urban%

2 98.4 1.2 0.14 52.5 35.2 3.8

15 69.9 - 26.6 2.616 46.1 - 45.5 6.229 42.7 17.8 23.4 10.8



Lowest priority:


CASE STUDY

BAYOU BARTHOLOMEWWATERSHED (BBW)


Model: SWAT2009Delineation: USGS 12-digit HUCDEM: 10 mSoil: SSURGOLand Use: 1992, 1999, 2004, and

2006 Subwatersheds: 44HRUs: 3657(0% land use, 5% soil type, 0% slope)Study Period:Calibration: 1996-2005Validation: 2006-2008 (2004-

2008)-PortlandPoint Sources: 9

MODEL SETUP


DATA USEDSub

watershedWET_MXS

A (ha)WET_NS

A (ha)WET_MXVO

L (m3)WET_NVO

L (m3)WET_F

R1 286 86 1,432,403 429,721 0.042 2,550 765 12,751,107 3,825,332 0.273 716 215 3,579,180 1,073,754 0.094 1,078 323 5,391,309 1,617,393 0.166 1,206 362 6,028,520 1,808,556 0.167 802 241 4,009,672 1,202,902 0.139 1,024 307 5,119,612 1,535,884 0.15

10 767 230 3,837,069 1,151,121 0.1011 678 203 3,391,550 1,017,465 0.0612 902 271 4,510,018 1,353,005 0.1713 1,841 552 9,205,229 2,761,569 0.1714 136 41 681,884 204,565 0.0118 3,423 1,027 17,113,701 5,134,110 0.2319 2,368 710 11,838,890 3,551,667 0.2222 955 286 4,774,685 1,432,406 0.1023 1,724 517 8,621,501 2,586,450 0.1525 1,410 423 7,048,546 2,114,564 0.1426 847 254 4,236,976 1,271,093 0.0627 1,412 424 7,062,041 2,118,612 0.1530 1,260 378 6,298,187 1,889,456 0.1631 2,077 623 10,385,029 3,115,509 0.1837 1,080 324 5,398,067 1,619,420 0.1043 1,365 410 6,826,668 2,048,000 0.2444 2,395 718 11,972,565 3,591,770 0.35

2001 NLCD land use map 24 of the 44 subwatersheds in BBW had some wetland surface area


NEXRAD

Weather

* NOAA -NWS

DATA USED





Preliminary Results

OVERALL SUMMARY

State of the art modeling and latest data was used for calibrating and validating SWAT model Total number of gages utilized varied from one to seven in different selected watersheds NEXRAD data used for precipitation as per availabilityModel outputs such as, total flow, surface flow, base flow, sediment, nitrate-nitrogen, and total phosphorus were rigorously evaluated using latest performance criteria In most of the subwatersheds, generally land use, soils, and slope association seen with high/low concentrations for water quality constituents


OVERALL SUMMARY..CONTD.

Uncertainty estimation to assess risk associated with model outputs is gaining ground. Quantification of input data uncertainty is desirable to assess model performance The model output has been used to rank subwatersheds on a relative basis. Absolute conc. of water quality constituent may not be of concern Data gaps noticed in LCPR It is encouraged that modeling results are also compared with other local information for decision making purposes Monitoring and Modeling are complimentary and monitoring data is required for having confidence in modeling results


ACKNOWLEDGMENTS


Dr. Robert (Bob) Morgan, Beaver Water District

Dr. Srinivasan (Srini) Raghavan

Dr.(s) Jeff Arnold, Mike White and Nancy Sammons

Dr. Amy Apon and Mr Jeff Pummil,UA High Performance Computing Center

QUESTIONS?/ SUGGESTIONS

[email protected]


subwatershed prioritization in arkansas - a modeling approach

Documents

leh anrc task force

components of model

model evaluation protocol

model setup comparisonmodel

model philosophy public

water source

smallest modeling scale

digit huc priority watersheds