Global precipitation characteristics and flood situations are expected to undergo significant changes as a consequence of global climate change including increasing temperatures and varying precipitation patterns projected for the future climate (IPCC, 2014). The incessant increase in magnitude and frequency of weather extremes (Yang et al., 2019; Blöschl et al., 2019; Vormoor et al., 2015), and uncertainties under climatic change scenarios may cause extremely severe disasters (Stott, 2016; Chen et al., 2018; Swain et al., 2020; Hao and Singh, 2020). For instance, extreme hydrological events are becoming one of the most serious natural disasters in densely populated regions, such as the Yangtze River Delta, the most developed region of China (Han et al., 2015; Jiang et al., 2020; Wang et al., 2016). Accurate detection of the response in hydroclimatic extreme to changing environments has, therefore, become an urgent and difficult issue of study for disaster-causing process in critical watersheds as well as its flood protection (Garrote et al., 2016; Lin et al., 2021).
Climate change, characterized by global warming and precipitation extremes, as well as anthropogenic activities including rapid urban growth, are involved in the furious turbulence of hydroclimatic extremes (Clark et al., 2016; Nathan et al., 2019). Global warming caused by greenhouse gas emissions can change atmospheric humidity and disturb the condition of atmospheric circulation, posing a dramatic effect on the characteristics of precipitation and the redistribution equilibrium of water resources (Li et al., 2019; Nilawar and Waikar et al., 2019; Paparrizos at al., 2015; Zhang et al., 2016). Many studies have therefore explored the impacts of climate change on the occurrence of extreme events, and demonstrated that the climate was projected to be warmer and wetter with time and the magnitude of the changes in extreme hydrological events would experience vital changes (Niu et al., 2021; Zhu et al., 2021; Dong et al., 2020).
It is effective to utilize Global Circulation Model (GCM) projections to drive hydrological model, such as the SWAT, VIC, MIKE-SHE and so on, to explore how climate conditions work on the hydrological process (Pastén-Zapata et al., 2020; Li and Fang, 2021). GCMs that have good capability in explicitly simulating average characteristics of temperature, precipitation, and circulation factors on a global scale provide credible information regarding the past, now, and future (Zhang et al., 2016). Given that GCMs lack accuracy in spatial and temporal resolution, researchers use regionally more relevant scale climate models to downscale future climate scenarios (Chen et al., 2011; Fowler and Kilsby, 2007). Particularly, the Statistical Downscaling Method (SDSM) model possesses the competence to apply climate information with greater physical significance and more accurate simulation in GCMs output to the statistical model (Wilby et al., 2002; Shen at al., 2018; Wilby and Dawson, 2007). This could reduce the system error of GCMs and as a consequence own an easy operation with excellent simulation effect. In hydrological and climatological studies, future climate scenarios are a reasonable description of the time and space distribution of the future climate state, which is based on certain driving forces and scientific assumptions. Representative Concentration Pathways (RCPs), as the most widely used future climate scenario, are advantageous for understanding the risks connected with different emission scenarios (Bhatta et al., 2019).
Over the past decades, many studies, highlighting the projected changes to weather and climate extremes, have been conducted, and most of them mainly focused on individual extremes or variables (Tofiq and Guven, 2015; Meaurio et al., 2017; Bulti et al., 2020). Conventionally, each hydrological variable interacts intricately instead of existing independently. There are many limitations in practical applications when analyzing the frequency only considering the influence of a single factor which cannot accurately describe the relationship between the variables. To investigate the joint probabilistic behavior of extreme events, multivariate probabilistic framework comes into being. The copula function (Skarl, 1995), which without restriction on the specification of marginal distributions and their uniformity, is conducive to deduce the joint probability distribution of random variables (Favre et al., 2004;Joe, 2014). A number of univariate and copula-based multivariate approaches were developed for co-occurrence frequency analysis of extremes, such as drought (Ayantobo et al., 2018; Sun et al.,2019; Li et al.,2020; Xu et al., 2020), and flooding hazard (Muñoz et al., 2020; Balistrocchi et al., 2017; Yu et al., 2019; Yin et al., 2018; Liu et al., 2020; Vinnarasi and Dhanya, 2019), thereby providing a joint frequency perspective of the associative structure between the depicted factors. For instance, Sun et al(2019)utilized a multidimensional copula model to combine the two kinds of disaster in terms of investigating the potential two future scenarios impacts on the simultaneous occurrence and joint return period of droughts and hot extremes in the Loess Plateau of China, and found that the compounding occurrence of drought with long-term hot extremes will be more severe and frequent. Muñoz et al (2020) evaluated compounding effects of terrestrial and coastal flood drivers and wetland elevation accuracy on maximum floodwater height and velocity with a bivariate statistical analysis framework. Yin et al (2018) explored the variations of flood peak and volume in Ganjiang River basin under different climate scenarios by fitting univariate and copula-based bivariate distribution, which indicated that the impacts of climate change on the future bivariate flood quantiles are considerable.
Centered on the Yangtze River Delta region, the joint frequency analysis of the multivariate characteristics based on the coupled extreme precipitation–streamflow under changing climate is rarely inquired about in literature. Furthermore, flood is one of the most destructive and widespread natural disasters over China, and climate change has anticipated to exacerbate the frequency of extreme flood events in the future. Herein, this research aims to detect the joint responses of extreme precipitation and floods under the climate changing environment by establishing suitable marginal and two-dimensional Copula function of the extreme precipitation and streamflow, which expands on our previous work (Wang et al., 2020). The results would improve the acknowledge of the disaster-causing process in watershed flood; thereafter, actionable insights are provided for decision makers to formulate the planning scheme of flood control and disaster reduction.
Clearly, the overarching structure of the research is organized as follows. In section 2, the study area and data involved in this paper are displayed. Section 3 is devoted to the methodologies used in the paper, including the prediction technique of extreme events and the copula method. In Section 4, the best fitted marginal and joint distributions are established through Copula functions, in order to evaluate the impact of climate change on the frequency of univariate and bivariate joint design events. At last, conclusions are presented.