The multivariate climatic and anthropogenic elasticity of streamflow in the Eastern United States

Abstract

Abstract Study region Eastern United States excluding Florida. Study focus We used elasticity to assess the sensitivity of mean and drought flows to changes in climate, land use and land cover (LULC), and water use across a broad region. Three multivariate regression analyses were used for elasticity estimation: nonparametric (NP), double-logarithm (DL), and variable transformation (VT). We demonstrate the importance of using multivariate analysis for elasticity estimation and show that the reliability of elasticity estimates depends critically upon model goodness-of-fit. VT analysis was found to provide more reliable estimates than other analyses across the Eastern U.S., except the Northeast where slight multicollinearity existed in the VT but was absent in the NP models. New hydrological insights for the region Changes in climate, LULC, and water use all significantly affected mean and drought flows. The interactions of these factors moderated the effect of precipitation on mean flow for regions where anthropogenic influences cannot be ignored. Human-induced land use changes were found to have a greater influence on drought flow than mean flow. Increased water use significantly reduced mean flow in the Northeast, where urbanization was more prevalent. Although the effect of water use on drought flow was found to be greater than its effect on mean flow, the variations of most water use effect estimates were too large to be concluded as significant.