Modeling Soil Organic Carbon in the Soil and Water Assessment Tool (SWAT)
Results
Things that you need to know:
1. The runoff and crop yield calibrations went smoothly with both obtaining a satisfactorily inefficiencies for the Nash-Sutcliffe (NSE) and PBIAS.
2. Calibrating biomass yield was the problem and neither a satisfactorily (NSE) or PBIAS were obtained.
3. The best theories to why the biomass yield was difficult to calibrate is land burning and land cover. At the present time SWAT is not able to factor in land burning and land burning does happen in the Konza. During calibration Big Blue Steam was selected and even though the main land cover at the Konza is Blue Big Steam it is not the only grass on the Prairie.
4. Even though the biomass yields could not be collected a uncalibrated test has been done to to see the trend that carbon levels take after harvest.
5. This study was a preliminary study that will be beneficial to more extensive research on soil organic carbon.
Figure 1. Shows the runoff from the McDowell creek and the model's predictions
Figure 2. Shows the grain sorghum yields and the yields that the model predicted for McDowell area.
Figure 3. Shows the actual or "observed" data with the predicted biomass yields from SWAT.
DFLT = Default
Nash-Sutcliffe:
Range ( -∞ -> 1)
Optimally = 1
Satisfactory > 0.5
Negative values = unacceptable
PBIAS:
Range (-∞ -> ∞)
Optimally = 0
Satisfactory varies for each parameter
Negative values = overestimation
Positive values = underestimation
Future Studies
- An uncalibrated SWAT model was run that modeled residue, manure, and soil carbon mass. This is where this research project would have gone if time had been available.
Statistical Efficiency
Figure 4. shows the levels of carbon residues and soil carbon levels during a full year from harvest to harvest.