Classifying Total and Effective Impervious Surfaces for Urban Hydrologic Modeling

Abstract

This paper describes an approach using maximum likelihood classification of high-resolution (< 4–m) multi-spectral data to determine the total impervious areas in a watershed in Houston, Texas, USA. The areas studied included a commercial district, an industrial site, and two residential areas, one with a very small amount of tree canopy coverage and another with significant canopy coverage. An unsupervised approach was tested along with several supervised approaches and the results suggested that the use of the high-resolution data provided reasonable approximations of total impervious area for the sample areas with negligible canopy coverage, and less accurate results in the area with significant canopy coverage. Currently we are addressing the issue of under the canopy impervious area and developing an approach to estimate effective impervious area by integrating the derived total impervious area surface with topography, drainage network, building morphology, and roadway data in a geographic information system. The techniques to estimate under the canopy impervious area and the effective impervious area are being based on defining flow paths and summing the pixels that drain to a drainage channel, drainage inlet, or some other entry point to the drainage system. The topography data being used is a highresolution (1-m) horizontal dataset derived from airborne LIDAR technology. An automated analysis tool within a geographic information system is being developed and will be described at the conference. Results from the application of the method for a large urban watershed in Houston will be presented at the conference.