On the distributions of annual and seasonal daily rainfall extremes in central Arizona and their spatial variability

JOURNAL OF HYDROLOGY

卷:559

Article

Mascaro, Giuseppe¹ 

[1] Arizona State Univ Tempe, Sch Sustainable Engn & Built Environm, Tempe, AZ 85281 USA

关键词: Rainfall extremes;    Peak-over-threshold analysis;    Generalized Pareto Distribution;    Seasonal extremes;    Annual extremes;    Spatial variability;   

DOI  :  10.1016/j.jhydrol.2018.02.011

来源: Elsevier

【 摘 要 】

This study uses daily rainfall records of a dense network of 240 gauges in central Arizona to gain insights on (i) the variability of the seasonal distributions of rainfall extremes; (ii) how the seasonal distributions affect the shape of the annual distribution; and (iii) the presence of spatial patterns and orographic control for these distributions. For this aim, recent methodological advancements in peak-over-threshold analysis and application of the Generalized Pareto Distribution (GPD) were used to assess the suitability of the GPD hypothesis and improve the estimation of its parameters, while limiting the effect of short sample sizes. The distribution of daily rainfall extremes was found to be heavy-tailed (i.e., GPD shape parameter xi > 0) during the summer season, dominated by convective monsoonal thunderstorms. The exponential distribution (a special case of GPD with xi = 0) was instead showed to be appropriate for modeling wintertime daily rainfall extremes, mainly caused by cold fronts transported by westerly flow. The annual distribution exhibited a mixed behavior, with lighter upper tails than those found in summer. A hybrid model mixing the two seasonal distributions was demonstrated capable of reproducing the annual distribution. Organized spatial patterns, mainly controlled by elevation, were observed for the GPD scale parameter, while xi did not show any clear control of location or orography. The quantiles returned by the GPD were found to be very similar to those provided by the National Oceanic and Atmospheric Administration (NOAA) Atlas 14, which used the Generalized Extreme Value (GEV) distribution. Results of this work are useful to improve statistical modeling of daily rainfall extremes at high spatial resolution and provide diagnostic tools for assessing the ability of climate models to simulate extreme events. (C) 2018 Elsevier B.V. All rights reserved.

【 授权许可】

Free