Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Maumee Basin SWAT Modeling Todd Redder Joseph V. DePinto Chelsie Boles
University of Michigan Water Center WBLE Scenario Workshop June 23, 2015
Presentation Outline • Summary of LimnoTech modeling activities in WBLE
• Overview of Maumee SWAT model
• Maumee SWAT calibration: – “Goodness of fit” calibration statistics (flow, pollutant loads) – Additional model-data comparisons (e.g., tile flow, crop yield)
• SWAT scenario evaluation: – Review SWAT scenario capabilities – Examples of past SWAT scenario applications & results
Relevant Modeling Activities in WBLE • Watershed Modeling:
– Blanchard River – AnnAGNPS (USACE, 2009-10) – Tiffin River – SWAT (USACE, 2011-12)
Ephemeral gully erosion represented Farmer reverse auctions – cover crops, filter strips, tile
drain management – Maumee SWAT modeling (2011-present)
Preliminary evaluation of “upscaled” BMPs (GLPF) - e.g., cover crops, conservation tillage
“4R” nutrient management evaluation (IPNI)
• “Western Lake Erie Ecosystem Model” (WLEEM) – Simulates nutrient transport & fate, harmful algal
bloom (HABs) – Various funding sources - recently applied for Annex 4
SWAT Model – Code Enhancements • Incorporated “tillage-induced ephemeral gully
erosion” sub-model • Enhanced model algorithms for representing
detention basins / wetlands: – “Potholes” (HRU level) – “Ponds” (subbasin level)
• Phosphorus in tile drains: – Fixed “disconnection” issue for soluble reactive P (SRP) – Added option for specifying fixed SRP concentration in
tiles
• Model calibration & diagnostic support: – Added many diagnostic variables (e.g., tile flow,
NO3/SRP concentrations) – Comprehensive post-processing & visualization tools
SWAT Custom Visualization Tools
Purpose/Role of Maumee SWAT Model Implementation of Nutrient
Management Practices • Type of practice(s) • Affected land area
Soil & Water Assessment Tool
(SWAT)
Western Lake Erie Ecosystem Model
(WLEEM)
e.g., changes to fertilizer timing, quantity
Flow, sediment, nutrient loading near mouth
Flow
, sed
imen
t, nu
trie
nt lo
adin
g Reduced Nutrient & Sediment Delivery
(@ tributary mouths)
Reduced Harmful Algal Production in
Western Lake Erie • Microsystis bloom intensity • Microsystis bloom duration
Ecological Endpoints
Model Linkage
Maumee Basin SWAT Model – Subbasin Delineation
Tiffin
Tiffin
Lower Maumee
Blanchard
St. Marys
St. Joseph
Upper Auglaize
Lower Auglaize
Upper Maumee
Flow Sediment Nutrients
Western Lake Erie
Basin Michigan Ohio
Subbasin Count: 203 HRU Count: 2,341
Maumee SWAT Model Lineage • Based on original EcoFore SWAT model (Bosch et al. 2011):
– Subbasin delineation roughly based on HUC-12 delineation – Simulation time period: 1995-2005
• Key LimnoTech refinements include: – SWAT 2012 framework adopted (updated from SWAT 2005) – Simulation period updated to 1995-2010 – Incorporation of detailed cropland management input data
from ARS Maumee CEAP SWAT model – Preliminary representation of soluble reactive P (SRP) transport
through tile drains
Hydrology Calibration Metrics
• Model originally calibrated to Waterville stations: USGS gage: 04193500 Heidelberg monitoring
(sediment, nutrients)
• Additional stations to monitor model fit for tributary branches and stations upstream of Waterville
• At least one station on each major tributary considered for flow calibration
Model Calibration Locations Tiffin River
@ Stryker
Maumee R. @ Waterville
Blanchard R. @ Findlay
Model Calibration: Metrics & Monthly Targets
Moriasi, D. N., J. G. Arnold, M. W. Van Liew, R. L. Bingner, R. D. Harmel, and T. L. Veith. 2007. “Model Evaluation Guidelines for Systematic Quantification of Accuracy in Watershed Simulations.” Transactions of the ASABE 50 (3): 885–900.
• All stations meet the standards confirming at least a “Satisfactory” calibration for monthly guidelines
• Most stations meet the “Very Good” guidelines for monthly NSE
• No Moriasi et al., 2007 standards given for R2, but statistic is greater than 0.50 for all locations
Note: Daily PBIAS values equal monthly PBIAS values.
Flow Calibration Statistics
Hydrograph Comparisons @ Waterville Water Year 2005
Water Year 2009
Water Budget for Entire Maumee Basin
Tile Drainage (as % of precipitation): • Target range: 10-31% (mean: ~21%)
• Model: ~19% on tile-drained land
Tile Flow by HUC-8 Watershed
Water Quality Calibration Metrics
Landscape Nutrient Yields
12.9
1.6
0.67
14.7
1.3
0.65
0
1
10
100
TN, lb/acre TP, lb/acre Sediment, tons/acre
Unit-
Area
Loa
d, lo
g sc
ale
Average Landscape Loading, Agriculture
Literature Average Maumee SWAT
Tile Drainage: Dissolved Nutrients
0
0.01
0.02
0.03
0.04
0.05
0.06
SRP
Tile
SRP
Conc
entra
tion,
mg/
L
SRP in Tile Drainage
0
5
10
15
20
25
NO3
Tile
NO3 C
once
ntra
tion,
mg/
L
NO3 in Tile Drainage
Observed Range: ~ 0.01 – 1.0 mg/L (K. King, OSU)
• Simulated NO3 loading of 21.15 kg/ha for tile-drained AGRR lands.
• Compares favorably with 23.17 kg/ha measured by Kladivko et al., 2004 for Indiana
Crop Yield Comparison (2000 – 2005)
020406080
100120140160180
2000 2001 2002 2003 2004 2005
YIEL
D (B
U/ac
re)
CORN YIELDS MAUMEE SWAT NASS
0
10
20
30
40
50
60
70
2000 2001 2002 2003 2004 2005
YIEL
D (B
U/ac
re)
SOYBEAN YIELDS MAUMEE SWAT NASS
Nutrient & Sediment Calibration Metrics • All stations meet the standards
confirming at least a “Satisfactory” calibration for monthly guidelines for nitrogen and phosphorus constituents.
• Most stations meet the “Very Good” guidelines for PBIAS (Abs(PBIAS) < 25)
• Monthly sediment loads at Waterville are just outside the “Satisfactory” range (±55)
Note: Instream sediment calibration has no influence on nutrients
All values reported are on a monthly time-step
Monthly TP Load @Waterville (1998-2010)
Monthly TN Load @Waterville (1998-2010)
“Spring” P Loads: Feeding Harmful Algal Bloom Production (Waterville)
1:1 Line
R2: 0.78 NSE: 0.62 (Good / Satisfactory) PBIAS: -10.6 (Very Good)
Maumee SWAT: Evaluating Nutrient Management Scenarios
SWAT General Management Capabilities • One or multiple rotation years (Maumee SWAT: 4 years) • Daily scheduling of crop land management practices:
– Crop planting – Crop harvest (or “kill”) – Tillage – Fertilizer application (timing, amount, degree of incorporation)
• ‘Direct’ implementation of some BMPs – e.g.: – Conservation tillage / ‘no till’ – Filter strips/buffers – Nutrient management – Conservation crop rotation – Wetland / detention basin treatment of ag runoff
Example SWAT BMP Evaluations • Little Cannon (MN) SWAT model:
– Simulate “detention pond” treatment of 30% of agricultural land (23% TP reduction, 3% TN reduction)
• Tiffin River SWAT model: – Support MSU reverse auction (Defiance, Fulton counties) – Ranking of farmer bids based on proposed cover crops, filter
strips, drain water mgmt. • Maumee Basin SWAT model:
– Preliminary demonstration of nutrient management (rate, incorporation)
Example Results: single HRU, various BMPs (previous version of Maumee Basin SWAT model)
General Considerations for BMP Implementation in SWAT
• Suitability/robustness of SWAT & supporting data:
– “Direct” vs. “indirect” representation – How critical is local topography to
representing a BMP realistically?
• Extent of implementation across watershed – e.g.:
– 25% of cropland acres – 40% of cropland acres
• Prioritization of implementation: – Randomly distributed implementation, or: – Targeting of higher sediment/nutrient yield
areas
Questions?
Contact Information: Todd Redder LimnoTech Ann Arbor, MI Email: [email protected] Phone: 734-332-1200
Funding Acknowledgements: • Great Lakes Protection Fund • International Plant Nutrition Institute (IPNI):
4R Nutrient Stewardship Research Fund