Temporal and Spatial Characteristics of Multidimensional Extreme Precipitation Indicators: A Case Study in the Loess Plateau, China

Chaoxing Sun,Guohe Huang,Yurui Fan

doi:10.3390/w12041217

Abstract

Extreme precipitation can seriously affect the ecological environment, agriculture, human safety, and property resilience. A full-scale and scientific assessment in extreme precipitation characteristics is necessary for water resources management and providing decision-making support to mitigate the potential losses brought by extreme precipitation. In the present study, a multidimensional risk assessment framework is developed to investigate the spatial–temporal changes in different extreme precipitation indicators. The Gaussian mixture model (GMM) is applied to fit the distribution for each indicator and carry out single index risk assessment. The joint probabilistic features of multiple extreme indicators can be explored through coupling the GMM distributions into copulas. In addition, the moving window approach and the Mann–Kendall test are integrated to examine non-stationary risks (evaluated by “AND”, “OR”, and Kendall return periods) of multidimensional indicators along with their changing trends and significance. The proposed assessment framework is applied to the Loess Plateau, China. Four extreme precipitation indicators are characterized: the amount (P95), the number of days (D95), the intensity (I95), and the proportion (R95) of extreme precipitation. The spatial–temporal changes of these indicators and their multidimensional combinations (including six two-dimensional and three three-dimensional combinations) are fully identified and quantitatively evaluated.

Highlights

In recent years, the increasingly catastrophic impact of hydrometeorological disasters has aroused public concern over extreme events due to climate change [1,2,3]
In order to reduce the uncertainty brought by differences in marginal distributions, the Gaussian mixture model (GMM) [22] is applied to fit the marginal distributions for all extreme precipitation indicators in this study
The probability density function (PDF) of GMM is expressed by a weighted sum of n-component Gaussian PDFs: p(x) =

Summary

Introduction

The increasingly catastrophic impact of hydrometeorological disasters has aroused public concern over extreme events due to climate change [1,2,3]. The drastic changes in the global water cycle may lead to increasing extremes such as extreme precipitation and drought in local areas [4,5]. Risk analysis for extreme precipitation is important for hydrologic designs and environmental systems management [9,10,11]. The characteristics of extreme precipitation have presented dynamic features [12], in which the changes of magnitude or/and frequency of extreme precipitation may have significant impacts on local flooding [10,13]. The commonly used risk assessment techniques with stationary assumption

Objectives

Methods

Results

Conclusion