Projection of disaster-causing risk of extreme precipitation in the Yangtze River Basin based on CMIP6

The Sixth Assessment Report of the United Nations Intergovernmental Panel on Climate Change points out that since 1750, global warming has been increasingly influenced by human-driven factors and has led to a gradual increase in the frequency, intensity and duration of extreme climate events (IPCC 2022). As one of the processes of the water cycle, precipitation is the key link of water-vapor exchange. In the context of global climate change, the speed of the global water cycle and the spatio-temporal distribution of precipitation have significantly changed (Hu et al. 2021a). The frequency, intensity and impact scope of floods caused by extreme precipitation have also increased, which seriously affects social and the economic development, people's lives, the safety of property and the ecological environment. The study of extreme precipitation change has become the focus and also attracted the wide attention of governments, the public and scholars under the background of global warming.

Extreme precipitation is very sensitive to climate change (Easterling et al. 2000; Meehl et al. 2000; Jiang et al. 2009; Li et al. 2020b, 2021a). The increase in temperature leads to the increase of water vapor, which affects the occurrence of heavy and extreme precipitation events. In regions where the total precipitation increases, the heavy precipitation events are likely to increase in a larger proportion. In some basins, the amount and frequency of heavy or extreme precipitation also increased under the condition of total precipitation remaining unchanged or reduced (Karl & Trenberth 2003; Ding et al. 2008; Mishra et al. 2012; Moustakis et al. 2021), which has undoubtedly increased the risk of flooding (Allan & Soden 2008). The influence of climate change on extreme precipitation is also reflected in the sharp increase of regional short-duration rainstorm intensity and the number of extreme heavy precipitation days, that is, the extreme precipitation value and the average intensity of extreme precipitation have an increasing trend and the ratio of extreme precipitation to the total precipitation tends to increase (Zhai & Pan 2003; Ding et al. 2006; Allan & Soden 2008; Mladjic et al. 2011). For example, compared with the beginning of the 20th century, the number of days with daily rainfall exceeding 50.8 and 101.6 mm increased in the United States at the end of the 20th century and both the intensity and frequency of heavy precipitation increased. Moreover, the proportion of extreme precipitation to the total precipitation increased by about 5–10% in the Mississippi River Basin, the midwest and southwest of the United States and the Great Lakes region (Easterling et al. 2000). Goswami et al. (2006) obtained similar results based on the analysis of Indian data and found that the extreme precipitation events (daily precipitation greater than 100 mm) during 1981–2000 increased significantly compared with those during 1951–1970.

China is sensitive to the impact of global climate change and is one of the countries with the highest frequency, the greatest intensity and the most serious disasters caused by extreme weather events (Huang et al. 2019). Statistically, the direct economic losses caused by extreme precipitation and floods reached as much as 167.86 billion yuan and the number of people affected exceeds 100 million every year; more importantly, there is a tendency for extreme weather events to increase year by year (Fang et al. 2014; Li & Zhao 2022). Due to the influence of the East Asian monsoon, the Yangtze River Basin has always been a severely afflicted area and a frequent area of flood disasters. At the same time, the Yangtze River Basin is an important economic core area in China with a dense population and rapid urbanization and its urban security and sustainable development are facing great challenges (Qin 2015). Previous studies revealed that in the past 50 years, the beginning time of extreme precipitation events in the Yangtze River Basin was significantly advanced, while the ending time had a trend towards postponement. The spatial distribution of precipitation was more uneven and the contribution rate of heavy precipitation in the Yangtze River Basin has been as high as 30.75% since the 21st century (Shi et al. 2017; Sun et al. 2018; Lu et al. 2020). Gao et al. (2001) and Li et al. (2013) found that the increase in atmospheric CO₂ concentration can lead to an increase in extreme weather in southern China. Moreover, Qin et al. (2005) pointed out that against the background of climate warming, extreme precipitation events in China are becoming more and more frequent and intense, especially in the Yangtze River Basin (Jiang et al. 2008). However, most of the previous studies in the Yangtze River Basin focus on the spatio-temporal changes of climate elements under different scenarios in the future and analyze and predict the change trend of future precipitation in the Yangtze River Basin based on CMIP5 and CMIP6 data (Chu et al. 2015; Li et al. 2020a, 2022), but pay little attention to the quantitative assessment of disaster risk caused by extreme precipitation. It has become an important measure to effectively deal with climate change and prevent flood disasters to clarify the disaster risk of extreme precipitation and the temporal and spatial distribution characteristics of high-risk areas under different climate change scenarios in the future.

At present, the coupled model intercomparison project (CMIP) has been developed into the sixth phase (CMIP6). Numerous studies have shown that, compared with its previous generation CMIP5, the experimental design of CMIP6 is more perfect and the accuracy and spatial resolution of the data have been greatly improved (Lin & Chen 2020; Tang et al. 2022). Zamani et al. (2020) for the first time applied CMIP5 and CMIP6 data to the arid region and discussed the seasonal and annual precipitation performance in northeastern Iran from 1987 to 2005. They found that the model ensemble performance of CMIP6 was better than that of CMIP5 in most seasons and stations. Zhu et al. (2020) quantitatively evaluated the simulation of future climate change in China by 12 CMIP6 models and 30 CMIP5 models from the perspectives of annual variability and spatial distribution, which confirmed that the model diffusion of CMIP6 was significantly smaller than that of CMIP5 and the spatial deviation of annual precipitation also decreased from 127% of CMIP5 to 79% of CMIP6. Additionally, Pan et al. (2022) evaluated the performance of CMIP5 and CMIP6 in simulating precipitation in the Yangtze River Basin under the moderate emission scenario and found that CMIP6 had a lower relative deviation and better spatio-temporal simulation results for the middle and lower reaches of the Yangtze River. Therefore, as the latest generation model, CMIP6 provides strong data support for studying the spatio-temporal variation trend of future precipitation under different warming conditions.

The objectives of the study are: (1) to analyze and predict the spatial and temporal trends of extreme precipitation in the Yangtze River Basin, including the number of heavy rain days, heavy precipitation, 5-day maximum precipitation and annual precipitation, under different climate scenarios in the future based on seven CMIP6 models; and (2) to quantitatively assess the spatial and temporal patterns of disaster-causing risk of extreme precipitation under different climate scenarios in the Yangtze River Basin from 2021 to 2100.

CMIP6 is a project which has participated in the largest number of models, designed the most abundant numerical experiments and provided the largest amount of simulation data in the past 20 years (Xiang et al. 2021). Considering the computational efficiency and model performance, seven different models of CMIP6 (i.e., CanESM5, EC–Earth3–Veg, GFDL–ESM4, IPSL–CM6A–LR, MIROC6, MPI–ESM1-2–LR and MRI–ESM2-0) were selected in the study to get the daily surface precipitation data. These Global Climate Models (GCMs) were widely used and achieved good performance in many regions, especially in China or the Yangtze River Basin (Table 1). Moreover, five different climate change scenarios, including SSP1-1.9, SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5, were selected to represent the scenarios of ultra-low, low, medium, medium-high and high radiative forcings. These data are freely collected from the website of https://esgf-node.llnl.gov/search/cmip6/. In this study, 1984–2014 was selected as the historical period and the period for climate projection was 2021–2100. To facilitate intercomparison, the spatial resolution of all model outputs was interpolated into a common 0.5° × 0.5° grid using the bilinear interpolation method.

In order to evaluate the simulation accuracy of precipitation by different climate models in the Yangtze River Basin, the precipitation data in the dataset of the China Surface Climatological Data (V3.0) provided by the National Meteorological Center and the China Meteorological Data Network (http://data.cma.cn/) are selected as the true observed values. The quality and integrity of the dataset are significantly improved compared with the previous ground-based data products (Sun & Zhang 2017). This dataset contains the daily climate element data of 224 reference weather stations in the Yangtze River Basin since 1951, and the distribution of stations is shown in Figure 1. These data have been widely used in various hydrological studies (Li et al. 2014, 2015, 2016, 2017a, 2017b). Based on the inverse distance weighting method (IDW), the daily precipitation at gauges in the Yangtze River Basin was interpolated to obtain the spatial distribution of precipitation with a spatial resolution of 0.5° × 0.5°.

The elevation data of the Yangtze River Basin are derived from the spatial distribution dataset of elevation (DEM) of China provided by the Resources and Environmental Science and Data Center (https://www.resdc.cn/), with a spatial resolution of 1 km. This dataset was generated by resampling based on the latest shuttle radar topography mission (SRTM) V4.1 radar image data, which has the characteristics of strong reality, high precision and can accurately reflect the topography features of the basin.

Compared with a single model, the multi-model ensemble mean (Jiang et al. 2008) has generally better simulation ability and can effectively reduce the error caused by the model uncertainty of a single model (Srivastava et al. 2020). Therefore, the equal-weight multi-mode ensemble average method is used to reduce the simulation error of the CMIP6 model. Meanwhile, the simulation accuracy of seven selected CMIP6 models and multi-model ensemble mean (MME) for extreme precipitation in historical periods is evaluated by the Taylor diagram, including the standard deviation, root mean square error and coefficient of spatial correlation.

In this study, four extreme precipitation indices are selected as evaluation indicators, including heavy rain days (R50), heavy precipitation (R95p), 5-day maximum precipitation (RX5day) and annual precipitation (PRCPTOT), and the specific information of each index is shown in Table 2. The R50, R95p and Rx5day can reflect the extreme characteristics of precipitation and the PRCPTOT represents the spatial distribution characteristics of precipitation.

According to the values of ⁠, the disaster-causing risk of extreme precipitation in the Yangtze River Basin was divided into five risk levels, i.e., Level I (0–0.35), Level II (0.35–0.5), Level III (0.5–0.65), Level IV (0.65–0.85) and Level V (0.85–1).

The future period is divided into three: near-term (2021–2040), medium-term (2041–2070) and long-term (2071–2100) periods. The mean values of extreme precipitation indices in the Yangtze River Basin during different periods are shown in Table 3. Compared with the observed data, the average values of PRCPTOT and R95p increase in the future with the average increase rates of 30.54 and 51.09%, respectively. In the SSP5-8.5 scenario, the increase rate of PRCPTOT is the largest in the long term, increasing from 1,073 mm in the historical period to 1,504 mm, with a rate of 40.14%. The increase rates of R95p in three future periods are 44.11, 50.01 and 56.82%, respectively, showing an increasing trend with time. The main increase period of R95p in the Yangtze River Basin is from 2071 to 2100, while for R50 and RX5day, although they show the long-term upward trends, their mean values in the future are lower than that in the historical period (Table 3). Figure 4 and Table 3 suggest that the increase of precipitation in the Yangtze River Basin in the future is mainly due to the increase of heavy precipitation, but the duration of extreme precipitation will decrease.

In the future, the high values of R50 are primarily distributed in the Dongting and Poyang Lake basins. The average value of R50 increases from 1.71 days in the near term to 1.98 days in the medium term and 2.33 days in the long term, with the maximum value of 4.06 days in the northeast of Poyang Lake Basin. Under the scenario of SSP3-7.0 and SSP5-8.5, the area with R50 greater than 2 days in the Dongting and Poyang Lake Basins expands from the near term to the long term, accounting for 18.2 and 20.9% of the total basin area in the long term, respectively. Under the scenario of SSP1-1.9, the R50 of the main stream area from the Yibin to Hukou, the Jialing and Hanjiang River Basins will not exceed 1 day in the three future periods, with an average value of 0.95 days. In the other four scenarios, R50 reaches 1 day in the medium term, and reaches the maximum of 1.7 days in the long term under the scenario of SSP5-8.5.

The distribution pattern of R95p and RX5day shows high spatial consistency and the area with high values is widely distributed in the middle and lower reaches of the Yangtze River. In the three future periods, the area with high values will continue to expand, showing a northward expansion trend and the regional average will continue to increase. The averages of R95p in the middle and lower reaches of the Yangtze River are 474.92, 493.30 and 513.93 mm, respectively, in the three future periods and the maximum value of 630.26 mm appeared in the northeast of Poyang Lake Basin. Under the scenario of SSP1-1.9, the mean value of RX5day in the middle and lower reaches of the Yangtze River was 62.72 mm, while it increased to 68.33 mm in the case of the SSP5-8.5 scenario.

In a word, the Dongting, Poyang Lake Basins and the lower researches of the Yangtze River are the primary distribution areas with high values of extreme precipitation indices in the future. Especially in the northeast of Poyang Lake Basin and the downstream of the Yangtze, the frequency and intensity of extreme precipitation are likely to be high in the future.

In addition, the changes of disaster-causing risk of extreme precipitation in the Yangtze River Basin in different periods are mainly manifested in the decrease of low-risk areas (Levels I and II) and the increase of medium-risk areas (Levels III and IV) (Table 4). Under the scenarios of SSP1-1.9 and SSP1-2.6, the area of risk Level IV continues to expand in the western Sichuan Basin from the near term to the medium term and in the long term, the area of risk Level IV increases to 34.79% and 34.00%, respectively. Under the scenario of SSP2-4.5, compared with the near term, the area of risk Levels III and IV increased by 3.98% in the future long term and with the expansion of the medium-risk area, the area of risk Level II in the southwest of the Yangtze River Basin decreased from 15.92% in the near term to 13.40% in the medium term and even to 12.50% in the long term. It is worth noting that the area of risk Level V shows an obvious expansion trend under the scenarios of SSP3-7.0 and SSP5-8.5 and continues to expand to the Dongting Lake Basin, the Han River Basin and the middle and lower main streams of the Yangtze River. In the long term, the area of risk Level V increase to 11.36% and 12.02%, respectively, which is twice as much as that in the near term.

Overall, the Poyang Lake Basin is the area with the highest risk of disasters caused by extreme precipitation in the Yangtze River Basin in the future. To reduce the losses caused by extreme precipitation leading to disasters, it is necessary to focus on prevention and strengthen the regulation of water conservancy facilities. The risks in the Dongting Lake Basin and the middle and lower reaches of the Yangtze River follow that of the Poyang Lake Basin. The risk level of disaster caused by extreme precipitation in the southwest of the basin has a gradual increasing trend, due to the complex terrain and the influence of the southwest monsoon.

The continuous change of global climate promotes the water cycle process and intensifies the spatio-temporal variation of precipitation (Lu et al. 2022). In this study, it is found that under all scenarios, the future precipitation in the Yangtze River Basin showed an increasing trend. Especially under the scenario of SSP5-8.5, the increase rate of PRCPTOT reaches 29.42 mm/(10a). With the increase of episodic radiative forcing, this trend of precipitation increase becomes more significant. Zhu et al. (2021) used CMIP5 and CMIP6 data to simulate the future precipitation in the Yangtze River Basin, respectively, and found that with the increase of the level of episodic radiative forcing, the degree of wetness in the Yangtze River Basin increased. The study of Li et al. (2021b) also indicated that the precipitation from the source region to the lower reaches of the Yangtze River Basin showed a significant increasing trend and SSP5-8.5 was the scenario with the largest wetting trend in the selected future scenarios. Against the background of continuous strengthening of the greenhouse effect and rising temperature, the increase of water vapor content in the atmosphere and the enhancement of the heat exchange effect between land and sea lead to the stronger East Asian monsoon, which is the leading factor behind the increase of precipitation in the Yangtze River Basin (Ma et al. 2017; Chen et al. 2020). In the future, the middle and lower reaches of the Yangtze River will still be the primary precipitation centers in the basin and the precipitation will increase significantly. Under the five scenarios, the average annual precipitation in this region increases from 1,458 mm in the near term to 1,515 mm in the medium term and to 1,575 mm in the long term, which increased by 20% compared with that in the historical period. The results in this study are consistent with the analysis of the characteristics of extreme precipitation changes in the middle and lower reaches of the Yangtze River under different warming scenarios by Liu et al. (2017). Yue et al. (2021) also pointed out that the precipitation increase rate in the Yangtze River Basin would gradually increase from the northwest source area to the southeast in the future.

Assessment results showed that the high-risk areas caused by extreme precipitation in the Yangtze River Basin are mainly distributed in the middle and lower reaches and the high-risk area shows a continuous increasing trend under the scenarios of high radiation stress (SSP3-7.0 and SSP5-8.5). In this study, the assessment of disaster-causing risk of extreme precipitation was based on the disaster-causing factor and the disaster-inducing environment. As the main components of disaster-inducing environment, terrain and slope have a low probability of change in the foreseeable future and their impacts on the risk level can be ignored. This means that the frequency and intensity of disaster-causing risk of extreme precipitation in the future period are mainly controlled by the changes in extreme precipitation (Shao & Zheng 2018; Huang & Li 2022). Spatially, the areas with risk Levels IV and V are mainly distributed in the middle and lower reaches of the Yangtze River, while the area of risk Level I is concentrated in the source area. In addition, by comparing the area changes of different risk levels in different periods, it is found that the area with middle-high risk increases and the area with low risk decreases in the Yangtze River Basin in the future. These trends are also consistent with the change characteristics of extreme precipitation and the increase of extreme precipitation correspondingly leads to an increase of risk level.

Meanwhile, although the models and data used in the study have inherent advantages in reflecting the characteristics of extreme precipitation and its disaster-causing risk in the future in the Yangtze River Basin, some uncertainties still exist in this study. There are some errors in the simulation of precipitation in the Yangtze River Basin by CMIP6 models. The limitations of the model calculation and complexity of the climate system are the biggest sources of uncertainty in the simulation process when predicting the changes in extreme precipitation under future climate scenarios (Zhao et al. 2022b). In this study, the lack of bias correction for extreme precipitation in the Yangtze River Basin undoubtedly affected the simulation results (Wu et al. 2022; Zhao et al. 2022a). Such as, the predicted future precipitation in the Yangtze River Basin may be high overall, which resulted in an overestimation of the disaster-causing risk level. For this, we will correct the bias of the CMIP6 raw data through various bias correction methods to enhance the simulation accuracy of extreme precipitation (Su et al. 2020). Moreover, this study did not consider the impacts of changes in human social development, population distribution and disaster tolerance. With the development of the social economy, the impacts of human activities are increasing and the projection of disaster-causing risk of extreme precipitation needs to take human activities into account. The next study will consider the changes in land cover type, population density, economic development level and other factors to further reduce the uncertainties in the future (Xu et al. 2014).

This study analyzes and predicts the spatial and temporal trends of extreme precipitation in the Yangtze River Basin and quantitatively assesses the spatial and temporal patterns of disaster-causing risk of extreme precipitation under different climate scenarios in the Yangtze River Basin from 2021 to 2100. The main results are as follows:

There are some biases in the simulation of extreme precipitation in the Yangtze River Basin by CMIP6 and there are obvious overestimates in the eastern part of the Qinghai–Tibet Plateau. The precipitation distribution in other regions has a good spatial consistency with that of the observed data.

In time, the extreme precipitation in the Yangtze River Basin shows an increasing trend and all indices of extreme precipitation show a fluctuating upward trend in the future. Under the five scenarios, future changes in PRCPTOT and R95p show significant increases, but R50 and RX5day decrease compared with the historical periods. In space, the high values of PRCPTOT, R50, R95p and RX5day of the Yangtze River Basin in the future period are mainly distributed in the eastern part of the Qinghai–Tibet Plateau, the Dongting Lake Basin, the Poyang Lake Basin and the downstream main stream area. Meanwhile, from the near term to the medium term and the long term, these peak areas gradually move northward.

The distribution of disaster-causing risk of extreme precipitation in the Yangtze River Basin will be mainly in Levels III and IV, accounting for 57.23 ∼ 65.99% of the total basin area. The area of risk Level V is mainly distributed in the Poyang, the Dongting Lake Basins and some parts of the lower main stream. Under all scenarios, the changes of disaster-causing risk of extreme precipitation in the Yangtze River Basin at different periods were mainly represented by the decrease of the area with low risk (Levels I and II) and the increase of the area with medium (Levels III and IV). In the future, extreme precipitation in the Poyang Lake Basin may lead to the highest risk of disaster, which needs to be focused on prevention.

This work was jointly funded by the National Key Research and Development Project of China (Grant No. 2018YFE0206400), the National Natural Science Foundation of China (Grant No. 41871093 and 42071028) and the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDA23040202).

All relevant data are available from an online repository or repositories (https://esgf-node.llnl.gov/search/cmip6/, http://data.cma.cn/, and https://www.resdc.cn/).

The authors declare there is no conflict.