Space-time variation of rainfall in Mexico: an approach for assessing impacts

Abstract

This work discusses the development of a process to approach the study of spatial and temporal rainfall patterns in Mexico, considering historical information for at least 20 years from 2 165 weather stations belonging to the National Weather Service. Outlier data were eliminated and estimates were made using ClimGen, provided that the missing information did not exceed 25% of the total. Goodness-of-fit was tested with respect to the two-parameter gamma distribution for the annual period and the period from May to October, the latter of which registers 80% of the annual rainfall in Mexico. It was noted that 1 786 of the stations for the annual period and 1 727 stations for May to October adjusted to the gamma distribution. The following spatial interpolation techniques were assessed: kriging, co-kriging, inverse distance and thin-plate smoothing spline. The latter showed lower mean square error and was considered in the interpolation of the shape and scale parameters to generate precipitation maps for 70, 80 and 90% rainfall probability. Finally, we analyzed the frequency, duration and intensity of wet and dry periods in Mexico considering the classification by the Standardized Precipitation Index (SPI), which is interpolated to generate annual raster maps for the period 1961-2003 with a resolution of 900 meters. An increase was observed in the duration of droughts and a decrease in their intensity. This research aims to help improve the modeling of the hydrological cycle, the study of climate change and the delineation of vulnerable areas in Mexico.

Keywords

• precipitación, interpolación espacial, bondad de ajuste, spi