Projected regional responses of precipitation extremes and their joint probabilistic behaviors to climate change in the upper and middle reaches of Huaihe River Basin, China

Highlights

• Significant decreases will occur in the intensity and amount of heavy precipitation for upcoming decades.

• Change rates of precipitation indices combinations illustrate obvious spatial heterogeneity.

• The frequency of simultaneous floods and droughts in a year and that of extreme heavy precipitation events will augment.

Abstract

The objective of this paper is to investigate the projected regional responses of univariate and bivariate behaviors of extreme precipitation to climate change over the upper-middle Huaihe River Basin. Based on twelve GCM outputs under historical, RCP4.5 and the observations at 32 rainfall stations, the equidistant cumulative distribution function matching method (EDCDFm) was utilized to bias correct daily precipitation during the historical (1961–2005) and future (2021–2080) periods. Four precipitation indices combinations were introduced based on eight precipitation indices to characterize the regional-scale changes of precipitation events, which designate the duration, intensity and amount of heavy and weak precipitation in a year. Their dependence structures were captured by Copulas. Kendall return period (KRP) were applied to discuss hazard scenarios and we quantified the spatial variability of KRPs under different marginal values. The results indicated that projected precipitation characteristics including the average intensity, the amount of annual precipitation, the intensity and amount of extreme precipitation together with annual extremes displayed increasing trends, while the changes of consecutive wet and dry days did not present pronounced trends. Decreased KRPs in the vast majority of the territory manifested that the frequency of simultaneous floods and droughts in a year as well as that of extreme heavy precipitation events would augment. Obvious spatial heterogeneity of the changes of KRP was partly attributed to the topography difference, especially the coastal areas along the main stream of the Huaihe River. Consequently, there will be a higher risk of water resources-related issues in this region for upcoming decades.

Introduction

Climate extremes are universally perceived as imperative triggers of meteorological and hydrological hazards (Kusangaya et al., 2014; Sisco et al., 2017). Furthermore, changes of extreme climate events are particularly relevant to socio-economic and ecosystems because of their potentially severe impacts on water resources-related issues, especially floods and droughts in terms of the intensification in magnitude, intensity and severity (Monirul Qader Mirza, 2002; Sillmann et al., 2013). In view of the consequences of these increasing catastrophes as well as the demand for anticipating and responding to climate extreme events, tremendous researches aiming to furnish robust and consistent predictions of possible changes in future climate extremes have emerged from various aspects.

One notable contribution comes from the CCI/WCRP/JCOMM Expert Team on Climate Change Detection and Indices (ETCCDI), which has constructed a set of temperature-based and precipitation-based climate extreme indices to characterize climate changes and variabilities (Sillmann et al., 2013; Zhang et al., 2011). These indices have been calculated for monitoring and detecting changes in observations (Alexander et al., 2006; Zhang et al., 2012) and investigating future changes of climate extremes as well (Li et al., 2015; Sillmann and Roeckner, 2008).

In the latter application of ETCCDI indices, another indispensable contribution towards long-term meteorological variables projections, the efforts of the fifth phase of the Coupled Model Intercomparison Project (CMIP5), should be highlighted. The Global Circulation Models (GCMs) participate in CMIP5 and their corresponding outputs under different IPCC scenarios are identified as one of the most available and fundamental tools for predicting future meteorological elements (Stocker et al., 2013). Nevertheless, it is inadvisable to use a single climate model to simulate future precipitation projections since there are a profusion of uncertainties stemming from a hierarchy of sources (Chen and Sun, 2009). To address this problem, the multi-model ensemble (MME) technique has been developed as an adequate tool for amending climate prediction on behalf of diminishing uncertainty. Additionally, numerous studies have been implemented to dissect different MME methods so far (Kug et al., 2008; Sun et al., 2015). Among them, the simple arithmetic mean method (AMM) is the most extensively used and has been proved to perform better compared with complex models (Kharin and Zwiers, 2002; Min et al., 2014).

In general, precipitation-based climate extreme indices are not independent. Different combinations of rainfall amount and rainfall intensity are likely to generate precipitation events presenting entirely different characteristics; the identification of extreme precipitations hinge on the joint features of rainfall peak, amount and duration. These paradigmatic instances illustrate the significance to link different precipitation extreme variables in order to render a comprehensive description that reflects the intrinsically multivariate features of precipitation events. Nowadays, copulas, a function that links joint distributions to their different margins (Salvadori and De Michele, 2004), have been extensively applied especially in hydrology and related fields, such as for rainfall, flood and drought frequency analysis (Grimaldi and Serinaldi, 2006; Kao and Govindaraju, 2010; Renard and Lang, 2007; Shiau, 2006; Zhang and Singh, 2007a, Zhang and Singh, 2007b), spatial dependence modelling (Ghizzoni et al., 2010; Suroso and Bárdossy, 2018), return period identification (Salvadori et al., 2011; Salvadori and De Michele, 2010), and multivariate probabilistic forecasting (Khedun et al., 2014; Liu et al., 2015; Madadgar and Hamid, 2013).

Focusing on precipitation extreme analysis, with copula methods maturing and advanced, besides investigating changing characteristics of statistical behaviors of precipitation extremes (Chen, 2013; Du et al., 2014; Fatichi and Caporali, 2009; Li et al., 2013), precipitation extremes combinations have also been designed to inquire as deeply and authentically as possible into bivariate responses to climate change. Zhang et al. (2012) studied the spatiotemporal variations of precipitation extremes in Xinjiang, China. Li et al. (2015) and Zhang et al. (2013) analyzed the coincidence of extreme heavy precipitation and drought, and the joint extreme precipitation events considering different extreme precipitation combinations of intensity, amount and duration across China. Goswami et al. (2018) developed seven extreme combinations to explore the designated copula-based joint behavior of precipitation extremes throughout the eastern Himalayan region. Jhong and Tung (2018) investigated the spatial variability of bivariate return periods of climate extremes to illustrate the occurrence possibility of future natural disasters in Shih-Men Reservoir Watershed. Among these previous studies, primary return period (“AND” case) (Salvadori, 2004) have been adopted to represent the joint behavior of bivariate variables, which is calculated based on the scenario that both variables exceed the prescribed thresholds. However, this definition method is not eligible to discriminate safe and dangerous regions of interest and eventually underestimate the potential risk of hazard-induced impacts. To overcome this deficiency, the concept of Kendall return period (KRP), also known as secondary return period, has been proposed by Salvadori et al. (2011), which guarantees the accuracy and rationality of identifying multivariate dangerous zone. Sufficient works associated with the applications of KRPs can be found in Salvadori et al., 2013a, Salvadori et al., 2013b, Graler et al. (2013). Thus, our work takes advantage of KRPs to discuss hazard scenarios.

Huaihe River Basin is immensely vulnerable to the changing climate, especially the upper-middle region, which has been often exposed to the threats of frequently floods (Shi et al., 2011; Zhang and You, 2014). The majority of the existing researches have been devoted to estimate the future possible trends of precipitation or focused on changes in monthly or quarterly average precipitation (Chen, 2013; Wu et al., 2018; Ye and Li, 2017), while limited studies focus on analyzing the changing characteristics of precipitation extremes in terms of precipitation indices. Moreover, joint probabilistic characteristic of precipitation events in the upper-middle region of Huaihe River Basin have not ever been specialized investigated. Given the gap in the current literature, we seek here to (1) evaluate the performance of selected climate models to simulate different precipitation statistics; (2) analyze changing characteristics of precipitation indices; (3) illustrate the projected regional responses of bivariate behaviors of precipitation extremes to climate change.