Análisis espacial de las precipitaciones extremas en la provincia de Chimborazo (1990-2018)

Abstract

In recent years, climate change has been a trigger for the stochasticity of the behavior of meteorological variables. The province of Chimborazo, having microclimates presents precipitations with specific localized characteristics. Recent research studies were interested in the behavior of rainfall in the city of Riobamba, initiating the use of extreme value theory from a univariate approach in said area, that is, using the records of a weather station. For the purpose of giving continuity to this research this thesis work proposes strategies by recurring analysis of periods of extreme rainfall in the province of Chimborazo. Records of rainfall from January 1, 1990 to December 31, 2018 provided by the GEAA-INHAMI of 23 meteorological stations were used. An exploratory analysis of each station was done, where, few measurements were found at the Palmira and Chambo station thus they were withdrawn. The Moran index indicates that there is no spatial autocorrelation at 95% reliability. Two types of thresholds were identified, one fixed for the entire area of study and a specific one for each meteorological station, for this purpose the techniques of the quantile and the mean residual life graph. With Bootstrap resampling the sampling distributions were found, to later estimate the parameters of scale and shape of the generalized Pareto distribution in confidence intervals. finally identified the extreme rainfall and its values at risk, which were represented through maps with IDW interpolation. The use of specific thresholds in each station allowed to characterize the behavior of extreme rainfall and its value at risk in a localized manner. for all the analysis process R tools were used.