Estimation of standardized reference evapotranspiration on the Canadian Prairies using simple models with limited weather data

摘要

Potential evapotranspiration (PET) can be accurately calculated with a Penman-Monteith model, however this model requires inputs that are not all measured at most weather stations. The agricultural region of the Canadian Prairies is a large geographicregion of dry continental weather (cold semi-arid to subhumid) and could benefit from simple, yet accurate empirical models for determination of evapotranspiration (ET). The purpose of our study was to develop simple (limited data) ET models to estimatestandardized reference evapotranspiration. We used daily weather data from 2003 and 2004 from ten stations across the agricultural region of the prairies to develop temperature, temperature-relative humidity, and temperature-relative humidity-wind speedmodels. Our models were developed by regression using the ASCE (2005) standardized reference model for a well-watered short crop surface (ET_(os)) as the 'observed' variable. We compared our developed models to other published simple models by Baier-Robertson, Hargreaves, and Linacre. To verify the accuracy of the models (relative to ET_(os)), we considered averages, standard errors, regression parameters (r~2, slope, and intercept), as well as the coefficient of efficiency. Of the temperature based models, our model and Hargreave's model were the most accurate. Our temperature-relative humidity model improved the estimate of evapotranspiration compared to temperature based models. The Linacre and the Baier-Robertson temperature and temperature-relative humidity models were not accurate and are not recommended for estimating evapotranspiration in the Canadian Prairies. Incorporation of temperature, relative humidity, and wind speed into our statistically developed model resulted in the most accurateand precise estimation of standardized reference evapotranspiration.