Evaluation of space-time dynamics in extreme precipitation frequency using geostatistical cosimulation with elevation

摘要

This study evaluates local dynamics in extreme precipitation frequency from 1940 to 1999 in the South of Portugal. The analysis is based on a climate index defined by the annual count of days with precipitation above the 30 mm threshold (R30mm). The space-time scenarios of this index, and their uncertainty evaluation, were produced through direct sequential cosimulation (coDSS) with elevation. The methodology incorporates space-time models that follow the premises that elevation and precipitation extremes may interact differently both in time and space. The results indicate that the relationship between elevation and the R30mm index has decreased through time over the study region, especially in the southeast area. Furthermore, the spatial patterns of the extreme precipitation index have become more homogenous during the last decades of the twentieth century. The more frequent heavy rainfall events occur in the mountainous areas of the South, which are desertification prone areas at risk of water erosion and floods. As expected, the space-time scenarios have greater spatial variability at regions less densely sampled. However, the uncertainty in mountainous regions is noticeably small given that elevation was used as secondary exhaustive information. The coDSS proved to be a valuable tool to deepen the knowledge on the local dynamics of the extreme precipitation frequency.