SWAT Model Analysis of Water Quality Effects from Increased Corn Biofuel Production
Client: U.S. EPA Office of Science and Technology, Washington D.C.
Description
Recent federal initiatives seek to increase alternative fuel production to 35 billion gallons annually by 2017. Corn?based ethanol production capacity is projected to increase over the next ten years to meet the growing demand for alternative fuels. With increased corn production, waterways are subject to higher nutrient loading and increased concentrations of sediment and pesticides. The objective of this project was to model the effects of increased corn production on surface waters in the Upper Mississippi River Basin (UMRB). The UMRB plays a central role in U.S. corn production and offered a broad opportunity to model the effects ethanol production has on surface water, including the hypoxia phenomenon observed in the Gulf of Mexico.
Participants
AQUA TERRA Consultants led the case study assessment of this project with support from the Texas A&M University Spatial Sciences Laboratory (TAMU-SSL) and the EPA. The Soil and Water Assessment Tool (SWAT) modeling application was employed as the primary analysis tool. To calibrate the SWAT model, the TAMU-SLL team developed parameter values, model procedures and conducted a baseline scenario analysis to compare the output of the model with historical results. Following the baseline analysis, AQUA TERRA was tasked with performing long term model runs and producing output summaries.
Model Development
The UMRB was modeled in SWAT using HUC 8-digit watersheds (Figure 1). The baseline condition was defined to represent UMRB conditions corresponding to the approximate time frame of 2003 to 2006. Flow and sediment, nitrogen and phosphorous loads were output at the HUC-8 level for all scenarios. Data sets used in the model include 2001 NLCD data for the land use coverage, USDA-NRCS STATSGO for soils data and AgCensus 2002/1997 for identifying cropping rotation and management practices. Based on the management information at the 8-digit level, each sub-watershed/HRU was assigned the same management information. The SWAT model used NRCS weather data from 1970-2002 available from climatic data center, which is a derivative of NOAA-NWS-NCDC datasets. The weather data were partially interpolated to assign one weather station per subbasin. No point sources were considered for the modeling, and only major reservoirs were included in the model.
With assistance from the EPA, correlations between economic factors
(e.g., corn prices, ethanol subsidies) and physical parameters were
established and incorporated into the model by adjusting physical inputs.
Application
The SWAT model was applied using the available input climate record of 1960 - 2001 and conditions for land use, cropping (e.g. increases corn production, conversion to switchgrass), management practices and nutrient inputs were projected for the years 2010, 2015, 2020 and 2022. The model was then further adjusted to compare with baseline loadings. For this study the year 2005 was selected as the mid-point of the target period for baseline conditions. The current national average for corn yield, 150 bushels per acre (bu/ac), was used to establish baseline yield levels. The baseline average yield for the UMRB was established at 140.7 bu/ac, which was considered acceptable due to the significant amount of crop area in Northern states where yield values are lower than the national average. Additionally, model nutrient outputs were assessed and related to nutrient reduction goals for the Gulf of Mexico. A summary of yield and ethanol production for each scenario is shown in Table 1.
Table 1: Summary of corn yield, corn area, and ethanol produced for UMRB scenarios
|
Scenario |
Average Yield, (bu/ac) |
Corn Area (Million Acre.) |
Ethanol Produced (BGY) |
Baseline |
140.7 |
23.65 |
- |
2010 |
149.6 |
33.35 |
5.1 |
2015 |
159.0 |
35.57 |
6.3 |
2020 |
169.0 |
34.69 |
6.3 |
2022 |
173.2 |
34.35 |
6.3 |
|
Results
In order to assess the potential impacts of increased corn production,
due to projected increased demands for ethanol, the SWAT model was applied
to the UMRB to first establish that the model can reasonably represent the
flow, sediment, and nutrient loads within and leaving this large regional
river basin. This baseline condition was defined to represent UMRB conditions
corresponding to the approximate time frame of 2003 to 2006. Then the model
was run with changed conditions to reflect alternative future scenarios, for
the ethanol demands expected for the time horizons of 2010, 2015, 2020, and
2022. The differences between the baseline and the other time horizons reflect
the impacts that might be expected due to the growing demands for ethanol as
a biofuel.
As shown in Table 2, average annual outflow of total nitrogen from the study
area increased by 5.5% from the baseline to the 2010 scenario. Scenario runs
beyond 2010 showed decreasing total nitrogen outflow with the 2022 scenario
showing less than a 2% increase over the baseline. A detailed analysis of the
scenario results included review and analysis of nitrogen application and plant
uptake for all scenarios. This showed that the decreasing of nitrogen loads can
be attributed to the increased yield production of future scenarios, resulting in
greater plant uptake of nitrogen. In addition, as noted above, less corn acreage
was needed for the 2020 and 2022 scenarios due to the increased yield values.
Table 2: Comparison of average annual total nitrogen loads.
|
Scenario |
Corn Area (Million acre.) |
Unit Load (lbs./acre.) |
Total Load (Million lbs.) |
Outflow (Million lbs.) |
Removed by Assimilation (%) |
Corn Area Increase (%) |
Outflow Change (%) |
Baseline |
23.6 |
15.61 |
1,897.00 |
1,433.50 |
24.4 |
- |
- |
2010 |
33.4 |
16.405 |
1,993.70 |
1,512.80 |
24.1 |
41 |
5.53 |
2015 |
35.6 |
16.262 |
1,976.30 |
1,500.30 |
24.1 |
50 |
4.66 |
2020 |
34.7 |
15.951 |
1,938.50 |
1,469.50 |
24.2 |
47 |
2.51 |
2022 |
34.4 |
15.851 |
1,926.40 |
1,459.60 |
24.2 |
45 |
1.82 |
|
As shown in Table 3, Average annual outflow of total phosphorous from the study
area increased by 2.8% from the baseline to the 2010 scenario. For scenario runs
beyond 2010, phosphorous loads followed a similar pattern to nitrogen.
Table 3: Comparison of average annual total phosphorous loads
|
Scenario |
Corn Area (Million acre.) |
Unit Load (lbs./acre.) |
Total Load (Million lbs.) |
Outflow (Million lbs.) |
Removed by Assimilation (%) |
Corn Area Increase (%) |
Outflow Change (%) |
Baseline |
23.6 |
1.453 |
176.6 |
132.4 |
25 |
- |
- |
2010 |
33.4 |
1.487 |
180.7 |
136.1 |
24.6 |
41 |
2.79 |
2015 |
35.6 |
1.470 |
178.6 |
134.7 |
24.6 |
50 |
1.74 |
2020 |
34.7 |
1.462 |
177.6 |
133.7 |
24.7 |
47 |
0.98 |
2022 |
34.4 |
1.459 |
177.3 |
133.4 |
24.7 |
45 |
0.76 |
|
Table 4 shows results for both sediment and flow, neither of which showed any
dramatic change from one scenario to the next. This is primarily due to the
corn being modeled as a well-managed crop in terms of sediment loss, primarily
due to the corn stover remaining on the fields following harvest.
Table 4: Comparison of average annual sediment outflows and streamflows
|
Scenario |
Corn Area (Million acre.) |
Sediment (Million tons) |
Flow (ft3/sec.) |
Sediment Change (%) |
Flow Change (%) |
Baseline |
23.6 |
6.339 |
122084 |
- |
- |
2010 |
33.4 |
6.373 |
122257 |
0.54 |
0.14 |
2015 |
35.6 |
6.355 |
122013 |
0.25 |
-0.06 |
2020 |
34.7 |
6.349 |
121819 |
0.16 |
-0.22 |
2022 |
34.4 |
6.348 |
121759 |
0.14 |
-0.27 |
|