Evaluating seasonal and regional calibration of temperature-based methods for estimating potential evaporation in Ontario

Abstract

Development of appropriate water management strategies in the context of projected air temperature increases under a changing climate necessitates assessing how increases in atmospheric demand for water might constrain future water availability. This is most easily assessed using estimates of potential evaporation (PE). Most temperature-based methods for estimating PE have been calibrated in different landscapes in more southern locations using a singular calibration coefficient. This study assessed whether regionally calibrated temperature-based models improve estimates of PE in Ontario and whether further improvements could be made by adjusting coefficients monthly to address a marked seasonality in Ontario or spatially to address the province’s size and varied landscape. Monthly calibration coefficients were calculated for four temperature-based methods (Thornthwaite, Blaney–Criddle, Hamon and Hargreaves–Samani) using climate data and evaporation estimates from 18 pan evaporation stations in Ontario. Calibrated models were then evaluated against an additional 19 pan evaporation stations by comparing estimates of PE calculated using the adjusted and literature-derived coefficients with pan evaporation estimates. Finally, PE estimated using adjusted coefficients was compared to evaporation (E) estimated using a water balance approach for 64 watersheds across Ontario. As anticipated, the regional calibration process increased the accuracy of the PE estimates relative to those calculated using published model coefficients. Use of monthly model coefficients addressed the marked seasonality of evaporation in Ontario and resulted in improved PE estimates. Adjusted model coefficients varied spatially but the paucity of measurement stations precluded further regionalization. PE estimates improved using adjusted coefficients, varied less between models, and were highly correlated with water balance estimates of E. The Hargreaves–Samani model produced the best estimates of PE using unadjusted and adjusted coefficients and was most highly correlated with long-term water balance estimates of E.