Nonpoint‐Pollution Model Sensitivity to Grid‐Cell Size

Abstract

Nonpoint‐pollution models estimate loadings of chemicals, sediment, and nutrients that degrade water quality. Before controls can be implemented, location and severity of pollution must be identified in the watershed basin. Geographic information systems (GISs) are computer‐automated, data management systems simplifying the input, organization, analysis, and mapping of spatial information. Because nonpoint‐pollution models simulate distributed watershed basin processes, a heterogeneous and complex land surface must be divided into computational elements such as grid cells. Model parameters can be derived from each grid cell directly from maps using GIS. Cell size selection, if arbitrarily determined though, yields ambiguous if not erroneous results. This paper investigates the effects of cell size selection through a sensitivity analysis of input parameters for the nonpoint‐pollution model, Agricultural Nonpoint Source Pollution Model (AGNPS), using a GIS for a small research watershed. Model grid‐cell sizes were found to be the most important factor affecting sediment yield. As the grid‐cell sizes increase, stream meanders are short‐circuited. The shortened stream lengths cause sediment yield to increase by as much as 32%.