Geospatial interpolation of reference evapotranspiration (ETo) in areas with scarce data: case study in the South of Minas Gerais, Brazil

Abstract

The reference evapotranspiration is an important hydrometeorological variable; its measurement is scarce in large portions of the Brazilian territory, what demands the search for alternative methods and techniques for its quantification. In this sense, the present work investigated a method for the spatialization of the reference evapotranspiration using the geostatistical method of kriging, in regions with limited data and hydrometeorological stations. The monthly average reference evapotranspiration was calculated by the Penman-Monteith-FAO equation, based on data from three weather stations located in southern Minas Gerais (Itajubá, Lavras and Poços de Caldas), and subsequently interpolated by ordinary point kriging using the approach "calculate and interpolate." The meteorological data for a fourth station (Três Corações) located within the area of interpolation were used to validate the reference evapotranspiration interpolated spatially. Due to the reduced number of stations and the consequent impossibility of carrying variographic analyzes, correlation coefficient (r), index of agreement (d), medium bias error (MBE), root mean square error (RMSE) and t-test were used for comparison between the calculated and interpolated reference evapotranspiration for the Três Corações station. The results of this comparison indicated that the spatial kriging procedure, even using a few stations, allows to interpolate satisfactorily the reference evapotranspiration, therefore, it is an important tool for agricultural and hydrological applications in regions with lack of data.

Keywords

• hydrometeorology, Penman-Monteith-FAO, kriging
• spatialization