Revisiting the Fresno DBCP Case Study Simulations: The Effect of Upscaling

The Fresno case study (see Loagne et al., 1998b) was a regional‐scale assessment of nonpoint‐source DBCP (1‐2‐dibromo‐3‐chloropropane) leaching that employed upscaled near‐surface information. The effort reported here focuses on the effect that upscaling spatially variable soil and crop cover information had on the Fresno case study simulations. The DBCP water‐table loading concentrations estimated by Loague et al. were evaluated by statistical comparison with three new water‐table loading scenarios that were developed in this study from simulations that did not employ the same upscaling techniques as previously used. Four model performance statistics were used to assess the upscaling effects revealed by the comparison. The relative statistical results indicate that: (i) the area most sensitive to the upscaling was small relative to the total study area, (ii) the new concentration estimates were within the range of the DBCP concentrations reported by Loague et al., and (iii) the effects from the upscaling of soil‐order information far exceeded the effects from the upscaling of the crop covers. The results of the upscaling analysis suggest that the intensely variable concentrations observed for the Fresno study area would probably not have been predicted from new near‐surface simulations involving smaller elements (and less upscaling) than the 1‐km2 elements used by Loague et al. Inasmuch as this study demonstrates that a somewhat coarse grid may be suitable for certain locations, thoughtful planning for variations in the grid size, using the methods reported here, should help to target where upscaling is a potential problem for future regional‐scale nonpoint‐source simulation efforts.