Studies have been done at watershed scale to determine the relationship of a stream's quality with land use, soil, topography, and population at watershed scale, with each variable having a single value per watershed. The assumption of this spatial aggregation is that everywhere in the watershed is homogeneous, which is not true in reality. This research aims to explore the full potential of GIS's capability in handling spatially disaggregated data in environmental models. Point-based variables will not be aggregated, with each observation used as a separate input to the model. Nonpoint source data will be aggregated into an upstream catchment area, thus eliminating the unwanted downstream impact.