Development of a Population Density and Land Use Based Regression Model to Calculate the Amount of Imperviousness

Nonpoint source pollution has been cited as the nation’s number one water quality problem and urban runoff is the pollutant of greatest concern. It has been noted that the amount of urban runoff and its impacts on stream conditions and water quality are correlated to anthropogenic watershed imperviousness. While much research has focused on quantifying this relationship, little work has been done to develop methods to measure impervious surface area at the watershed-scale. This paper provides details of research to develop an impervious surface estimation model for use with National Land Cover Data (NLCD) and Census Tract population density data. High-resolution planimetric data for buildings and transportation features served as the footprints for impervious surfaces from towns in Connecticut, Massachusetts, and New York and were used to calibrate the model. Census Population data most contemporaneous with the planimetric data were collected for the more than 100 tracts covering these towns. Percent imperviousness was calculated for each tract regressed against predictor variables of population density and NLCD class percentage per tract. A random sample of 80% of all the tracts was used to produce a linear regression model with an R-squared of 0.95, and RMSE less than 6%, with predicted imperviousness being within 1-2% for most tracts, when applied to independent validation data. A method has been developed to port this model to watersheds using interpolated population density and NLCD data. This has produced a regionally-applicable impervious surface estimation tool to be used by watershed managers. Research results are being distributed through the national NEMO Network and are being shared with cooperators at NOAA Coastal Services Center to refine the Impervious Surface Analysis Tool (ISAT).