A comparison of three methods for estimating impervious area in an urban watershed : Second Creek, Knoxville, Tennessee

In order to comply with federal water quality standards for urban stormwater runoff and target areas for watershed remediation, urban planners must perform baseline studies of stream runoff and pollutant loads. A necessary component of runoff and pollutant load models is being able to accurately estimate the percentage of impervious area. One technique for estimating impervious area is "averaging-by-land-use," or simply "land use," whereby an average percentage imperviousness is applied to the area of each of the land use types in the watershed. The result is a weighted mean for the whole watershed. This method is relatively simple and cost-effective, but, prior to this study, its accuracy for use in runoff load calculations had yet to be assessed. To evaluate the accuracy of the land use technique as it has been applied to the Second Creek watershed of Knoxville, Tennessee, I applied the technique to a subwatershed of Second Creek and compared the results to those obtained using Geographic Information System (GIS) analysis and aerial photograph analysis. Results showed that a critical factor of the averaging-by-land-use method is the accurate estimation of paved street right-of-way. I expected results from the aerial photograph and GIS techniques to be similar, since the GIS coverages were derived from the same aerial photographs that I used. This was not the case, and I believe that this can be explained by a combination of an unintended bias on my part in the air photo technique, and the generally poor quality of the GIS linework. However, imperviousness values obtained from the land use technique using the most likely values of paved right-of-way fell within a reasonable range between the other two methods. Based on the results of my evaluation, I conclude that the averaging-by-land-use technique is at least as accurate as the other two methods and, given that paved street right-of-way is accurately estimated, the most effective and suitable method of the three for use in pollutant load modeling.