Abstract

Missing monthly rainfall and temperature data were estimated to complete a 34-year meteorological record from 92 weather stations distributed in the Sierra de Amula region in the state of Jalisco, Mexico, by applying the methods of arithmetic mean, simple linear regression and the technique employed by the U.S. National Weather Service. Estimation of data consistency and homogeneity was achieved by means of double mass analysis, and Student’s t- and Cramér’s t-tests processed by Geographic Information Systems. Missing data was filled for up to 36,036 rainfall and 9,730 temperature data points. Data consistency errors were found for rainfall data of 10 months, with the exception of July and August, showing R² = 0.99, (α = 0.05); for temperature, data exhibited R² = 1 with a confidence and prediction of 100%. Student’s t- and Cramér’s t-test values below 1.56 were found and indicates a high degree of data homogeneity and quality (R² = 0.99). The best method to interpolate rainfall data was polynomial interpolation, whereas for temperature, J-Bessel provides a reliability close to 100%.

Keywords: Geographic Information Systems; Cramér’s t-test; Student’s t-test