Use of Multiple Linear Regression to Estimate Flow Regimes for All Rivers Across Illinois, Michigan, and Wisconsin

摘要

We estimated flow regimes for all river reaches in Illinois, Michigan, and Wisconsin, as a step in developing consistent ecological river mapping and assessment frameworks across this diverse climatic and physiographic region. Our objectives were to: 1) build and evaluate, for each state, multiple linear regression models that predict attributes of stream flow regimes, using catchment summaries of climate and landscape attributes as independent variables; 2) compare the performance of the state models with alternative full region and ecoregion models; and 3) predict stream flow regimes for all ungaged stream reaches. Recent stream discharge regimes were characterized using data from a set of 206 U.S. Geological Survey stream gages scattered across the three states. As independent variables we used data summarizing climate and landscape attributes for catchments of selected gage sites and also for every stream reach within the three states. We successfully built multiple linear regression models for a range of exceedance discharges representing several seasons, using gage data stratified by either state, ecoregion, or entire three-state region.