Eficiencia de las técnicas de interpolación, relleno y generación de datos de lluvia en función del tamaño de la muestra disponible

Abstract

In this work, six precipitation interpolation techniques were compared such as: Thiessen, IDW, polynomial, Gandín method, multicuadrática and Kriging, applied to the daily precipitation records of 48 meteorological stations, spatially distributed between the provinces of Loja and El Oro. Which have been delimited in homogeneous regions by characteristics of the study area and sample data statistics. In this sense, the performance of each interpolation technique was determined by comparing the resulting values of R² and RMSE, generated by estimating precipitation data when different record lengths are available compared to the actual sample data. As a result, it is shown that all the techniques used serve to estimate daily precipitation data. However, the Kriging interpolation turns out to be more efficient when having records with different data densities, the IDW and multi-quadratic techniques have good performance, when the reference stations are close and homogeneous to the station that we want to estimate the data. The polynomial interpolation by the Lagrange and Thiessen method gave higher RMSE values.