Detecting precipitation trend using a multiscale approach based on quantile regression over a Mediterranean area

Abstract

One of the most relevant and debated topics related to the effects of the climate change is whether intense rainfall events have become more frequent over the last decades. It is a crucial aspect, since an increase in the magnitude and frequency of occurrence of heavy rainfall events could result in a dramatic growth of floods and, in turn, human lives losses and economic damages. Because of its central position in the Mediterranean area, Sicily has been often screened with the aim to capture some trends in precipitation, potentially related to climate change. While Mann-Kendall test has been largely used for the rainfall trend detection, in this work a different procedure is considered. Precipitation trends are here investigated by processing the whole rainfall time-series, provided by the regional agency SIAS at a 10-min resolution, through the quantile regression method by aggregating precipitation across a wide spectrum of durations and considering different quantiles. Results show that many rain gauges are characterized by an increasing trend in sub-hourly precipitation intensity, especially at the highest quantiles, thus suggesting that, from 2002 to 2019, sub-hourly events have become more intense in most of the island. Moreover, by analysing some spatial patterns, it has been revealed that the south and the east of Sicily are more interested in significant increasing rainfall trends, especially at the 10-min duration. Finally, the comparison between the two procedures revealed a stronger reliability of the quantile regression in the trend analysis detection, mainly due to the possibility of investigating the temporal variation of the tails of precipitation distribution.