Enhanced NAPRA WWW spatial decision support system (SDSS) with river water quality modeling capability
Abstract
The NAPRA WWW system was enhanced by enabling simulation of multiple pesticide applications for each crop in crop rotations. The NAPRA WWW and the LTHIA/NPS GIS system water quality results for watersheds were compared with measured nutrient data from IDEM and with measured pesticide data from USGS in the White River basin in Indiana to evaluate the predictive ability of the enhanced NAPRA WWW system. The R2 value (R2 = 0.72) for the comparison of short-term (1990s) measured and predicted nitrate concentrations is slightly higher than that (R2 = 0.60) for the comparison of long-term measured and predicted nitrate concentrations with two outliers removed. The QUAL2E model was used with the NAPRA WWW and the LTHIA/NPS GIS predicted nutrient results to further evaluate the potential for scaling NAPRA results to watersheds. Only dilution of nutrients with point source discharges along reaches was considered, since point source nutrient data was not available. The R2 values considering all watersheds when QUAL2E was used were slightly better than results without QUAL2E. The R2 value (R2 = 0.65) for the comparison of short-term measured nitrate data (1990s) with predicted nitrate data was a little higher than that (R2 = 0.53) for the comparison of long-term measured data (1970s ∼ 1990s) when QUAL2E was utilized with two outliers removed. When two outliers were removed, the R2 values without QUAL2E are slightly higher than those with QUAL2E. The relationship between the predicted phosphorus concentration values and the measured phosphorus concentration values is poor (R2 is 0.25 with two outliers removed after QUAL2E runs). The enhanced NAPRA WWW system with the LTHIA/NPS GIS system can be efficiently used to identify critical nitrate loss in runoff but not phosphorus loss areas. Eight pesticides applied to corn and two pesticides applied to soybeans were simulated for three crop rotations with the NAPRA WWW system with R 2 values for pesticide concentration ranging from 0.002 to 0.280. For most cases, measured pesticide values were higher than predicted values. The differences in the amounts of pesticides actually applied compared to the estimated pesticide application rates and the timing of pesticide applications are likely responsible for mismatches in pesticide simulations with observed results.
Degree
Ph.D.
Advisors
Engel, Purdue University.
Subject Area
Agricultural engineering|Environmental science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.