Modeling of reference evapotranspiration by multiple linear regression

Abstract

Evapotranspiration is an important parameter for many projects related to climate characterization, hydrological modeling and water resources. This work was established as the first study in Rio Branco, eastern Acre, in order to derive empirical relations to estimate the reference evapotranspiration in the annual range from meteorological data readily available using the multiple linear regression analysis. Meteorological data of mean temperature (maximum and minimum), wind speed and insolation were obtainded from the National Institute of Meteorology for the period 1980-2014, which can be considered representative of the local climate. To estimate the reference evapotranspiration was used the Penman-Monteith-FAO, and multiple regression analysis was used as a selection process of significant variables for the model fit. Generated values by the proposed evapotranspiration models were compared to observed values for validation. Results indicated that the model with three variables (mean temperature, wind speed and insolation) satisfactorily estimated reference evapotranspiration for Rio Branco, AC, with great performance for annual data. Models with one variable (insolation) and two variables (mean temperature and insolation) showed less accuracy. However, they have advantage due to simplicity, since they can estimate the reference evapotranspiration from a few climatic parameters. From a practical point of view, these models can be regarded as a method to estimate the reference evapotranspiration when the input weather variables are insufficient to other methods.