High-resolution land cover datasets, composite curve numbers, and storm water retention in the Tampa Bay, FL region

Abstract

Policy makers need to understand how land cover change alters storm water regimes, yet existing methods do not fully utilize newly available datasets to quantify storm water changes at a landscape-scale. Here, we use high-resolution, remotely-sensed land cover, imperviousness, and tree canopy density data to calculate modified Soil Conservation Service (SCS) composite curve numbers. Policy makers can interpret composite curve numbers as a continuous, relative index of storm water mitigation ecosystem services provided by the landscape, allowing for better comprehension of the implications of land use decisions than current discrete methods. We also compare composite curve number calculations from regional land cover/land use data to calculations from national land cover data and show that they differ significantly in each of the watersheds that drain into Tampa Bay. The use of discrete urban classes to determine curve numbers is also explored. We show that, for the Tampa Bay region, assumed urban imperviousness values published in Technical Release 55 (a document that provides curve numbers for soil/land cover complexes) overestimate the actual imperviousness as reported by the national and regional datasets. This error in estimation caused composite curve numbers calculated using remotely-sensed data to be lower than they would have been if curve numbers had been assigned solely by discrete SCS classes. Furthermore, the average imperviousness for all urban classes increased from 1995 to 2005, reflecting region-wide increases in imperviousness reported by the regional dataset. Our comparison illustrates that using a constant imperviousness value for urban classes over time, a method used in numerous studies and models, leads to inaccurate estimation of temporal changes in storm water runoff. Average imperviousness for urban classes also differed significantly between the regional and national datasets, signifying that the two datasets cannot be used interchangeably.