Accuracy of Hourly Water Temperatures in Rivers Calculated from Air Temperatures

Abstract

Water temperature is a critical variable for water quality control and management. The primary objective of this paper was to develop and compare simple methods to estimate hourly water temperatures in rivers. The wave function (WF) model, originally used to calculate hourly air temperature, was modified and applied to eight Alabama rivers. The results show significant improvement by using the modified WF model instead of direct linear and non-linear (polynomial and logistic) regression models with time lags (4–5 h). The average Nash–Sutcliffe coefficient (NS) used to evaluate model accuracy for the eight rivers improved from 0.71 for the linear model to 0.89 for the modified WF model with NS for most rivers exceeding 0.90. A lumped modified WF model was also developed by combining water temperature data for all eight rivers and can be applied for rivers in Alabama when no observed water temperatures are available to develop a site-specific WF model. The procedure to develop a modified WF model can be applied to other regions.