Projected changes in heat, extreme precipitation, and their spatially compound events over China’s coastal lands and seas through a high-resolution climate models ensemble

China’s coastal lands and seas are highly susceptible to the changing environment due to their dense population and frequent economic activities. These areas experience more significant impacts from climate change-induced extreme events than elsewhere. The most noticeable effects of climate change are extreme high temperatures and extreme precipitation. We employ an ensemble of RCMs (Regional Climate Models) to investigate and project changes in temperature, precipitation, and Compound Heat-Precipitation Extreme events (CHPEs) over selected China’s coastal lands and seas for both historical (1985–2004) and future periods (2080–2099). The multi-model ensemble projects that daily temperature extremes will increase by 2.9 °C to 5.4 °C across China’s coastal lands and seas, with land areas showing a higher temperature increase than marine areas. Extreme precipitation shows a high geographical heterogeneity with a 2.8–3.9 mm d−1 reduction over the 15–25°N marine areas while a 2.2–5.4 mm d−1 increment over the 25°N-35°N land areas. We use the Clausius–Clapeyron relationship to reveal that the peak of daily extreme precipitation will increase by 2–7 mm d−1 and the temperature at which extreme precipitation peaks will increase by 2 °C to 6 °C by warming. The land area of 25–30°N has the highest peak precipitation increase of 9.87 mm d−1 and a peak temperature increase of 6 °C. As precipitation extremes intensify with daily temperature extremes increase, CHPEs are projected to occur more frequently over both land and marine areas. Compared with the historical period, the frequency of CHPEs will increase by 40.9%-161.2% over marine areas, and by 36.2%-163.6% over land areas in the future. The 15–20°N area has the highest frequency increase of CHPE events, and the 25–30°N area has the largest difference in frequency increase under two different scenarios. It indicated that the 25–30°N area will be more easily affected by climate change.


Introduction
In recent years, climate change has brought more frequent and intense extreme events to the world [1-3], posing a significant threat to human health [4][5][6], economic stability [7][8][9], and environmental safety on a regional and global scale [10][11][12].Coastal areas are influenced by interactions between the ocean and the atmosphere, leading to the development of weather patterns, sea breezes, heatwaves, as well as coastal storms [13].These interactions play a crucial role in shaping local climate conditions.In particular, the contrast in heating between land and sea creates pressure differences, influencing the formation of weather systems.For instance, during a heat wave, the land heats up faster than the sea, causing low-pressure areas over land [10,13,14].This contrast can lead to the development of local or regional weather events, affecting precipitation patterns.Therefore, understanding these interactions is crucial for assessing the impact of climate change on coastal areas.
China is affected by climate change, with the frequency of extremely high temperatures and extreme precipitation events increasing [15][16][17][18].Due to the rapid warming and frequent extreme events, compound extreme events in China are receiving more attention from researchers and policymakers [19][20][21].Compared with inland China, the coastal lands and seas of China are typically vulnerable, with dynamic and diverse natural systems on the Earth's surface and intense human activity [22][23][24][25].The Chinese coastal lands and seas are influenced by monsoons and sea-land interactions, and are also one of the fastest-growing and most dynamic regions in the world, bearing a high degree of vulnerability [18,22,23].Therefore, the Chinese coastal zone is more susceptible to compound events caused by climate change [14,16,19,26].We promote the development of the marine economy in offshore areas, including aquaculture, fishing, offshore oil, gas, and wind power resource development, marine tourism, etc [14,24,25], compound extreme events occurring in coastal marine areas should also be given attention like those occurring in land areas [26,27].
Climate change has led to a noticeable increase in extreme events, most notably the rise in extreme temperatures and alterations in extreme precipitation patterns [1, 9,25,27].These changes often result in Compound Heat-Precipitation Extreme events (CHPEs), which are spatially compound extremes that occur successively within a specific temporal and spatial framework [15,16,26,27].The weather or climatic conditions of heat events are conducive to the occurrence of heavy precipitation [15,[28][29][30].The elevated surface temperature and sensible heat flux characteristic of these events can amplify the effective potential energy of convection, thereby providing favorable conditions for the development and sustenance of heavy precipitation [18,27,31,32].Recent studies have drawn attention to the prevalence of CHPEs in China.You & Wang [28] discovered that between 1981 and 2005, 22% of China's land experienced CHPEs, with shorter, more intense heatwaves being more likely to be followed by heavy precipitation.Wu et al [29] found that in the Guangdong region, 28.25% of extreme precipitation events occurred within 3 days of an extremely high-temperature event.
Ning et al [32] reported that approximately a quarter of extreme precipitation events in western China occurred within three days of an extremely high-temperature event during the summer.In terms of generational changes, the frequency of CHPEs has significantly increased in most parts of China [30][31][32][33], with a national average increase of 2.51% every decade between 1961 and 2016 [33].Projections indicate that CHPEs in China will become more abrupt and frequent in the future, leading to an escalation in secondary risks, [27,[31][32][33].
Climate model ensemble data is a tool for studying past and future climate change through computer simulation [15,[28][29][30].In research, we can use the results of climate model runs to study past or future climate change [3,5,30].In terms of compound event research, some scholars have used climate models to study compound extreme events.For example, Bevacqua et al [29] used multiple climate models' Single Model Initialcondition Large Ensemble (SMILE) to complete climate data for hundreds or even thousands of years in the past for assessment and construction of stable model predictions; Kim et al [31] discussed 4 types of compound extreme events composed of 8 different types of disasters and conducted multivariate bias correction for regional climate models (RCMs), making climate models more accurately describe compound events.
While climate model ensemble data offers several advantages, its application is not without limitations.These limitations stem from the uncertainty in the model's spatial and temporal distribution, differences in dynamic downscaling approaches, and the complexity of the climate over coastal lands and seas [3,27,32].For instance, Yang et al [33] discovered that the precipitation regression correction schemes in southern China (east of 95°E, south of 35°N) significantly differ from those in northern and western China when correcting multimodel ensemble precipitation data under the CMIP5 RCMs in China.This discrepancy may be attributed to variations in the statistical properties of the precipitation series.Chokkavarapu et al [34] noted that differences could arise due to varying dynamic downscaling methods between climate models and reference data.Furthermore, the dynamic and diverse climate, sea-land interactions, and intense human activities over coastal lands and seas make it challenging for climate model ensembles to accurately capture temperature and precipitation changes.This results in deviations between model data and reference data, both spatially and temporally.Despite the progress made in the application of climate model ensembles, further research and improvements are necessary to address these limitations and challenges [3,27].
The objective of this study is to validate and analyze heat, extreme precipitation, and their spatially compound events through a high-resolution climate model ensemble.The adopted climate models ensemble's skill in reproducing the historical climate over China's coastal lands and seas will be gauged by validating its historical simulation with the reanalysis dataset in terms of temperatures and precipitation.Following validation, the climate models ensemble will be used to project independent climate variables and their spatially compound extreme events under different greenhouse gas emission scenarios.Particularly we focus on the spatial distribution differences of heat, extreme precipitation, and Compound Heat-Precipitation events (CHPEs) between marine and land parts within the research area.Eventually, this will enable a comprehensive assessment of the impacts of climate change on heat, extreme precipitation, and CHPEs across China's coastal lands and seas.

Models, experimental design, and data
In this study, we employed the ensemble mean of four Regional Circulation Models (RCMs) under the CMIP5 framework: CCLM, HCLM, MCLM, and PREC (see table 1).Each of the four CMIP5 models used different GCMs as driving conditions.CCLM, HCLM, and MCLM were all driven by the same RCM, which has been thoroughly validated for simulating the Chinese region [5,15].
To investigate the relationship between temperature and precipitation in China's coastal lands and seas, considering the vast latitudinal span of these areas, which results in varying natural resource endowments and regional economic development conditions [22,23].We divided China's coastal lands and seas from south to north into intervals of 5°latitude.As depicted in figure 1, this division resulted in 6 coastal lands and seas divisions: 15°N-20°N, 20°N-25°N, 25°N-30°N, 30°N-35°N, 35°N-40°N, and 40°N-45°N.These represent 6 regions: Hainan Island Area (HN), Subtropical Coastal Area (ST), Southeastern Coastal Area (SC), Eastern Coastal Area (EC), Yellow and Bohai Sea Area (YB), and Liaodong Bay Area (LD).Each sub-region is further divided into two areas by the coastline: the marine part and the land part.In each sub-region, we calculated the change sequences of temperature and precipitation, two indicators obtained by regional averages in each subregion, for subsequent research.
In order to validate the skills of chosen climate models to reproduce historical climate, we require observational data or climate reanalysis data as references [4].The results of the climate model ensemble mean are compared with these references.For validation, we utilized the ERA-5 Hourly Data on Single Levels from 1940 to the Present reanalysis dataset provided by the European Centre for Medium-Range Weather Forecasts  (ECMWF) [35].This dataset includes hourly meteorological variables such as temperature, precipitation, and wind speed near the surface of the globe from 1940 to the present, with a spatial resolution of 0.25°×0.25°[35].ERA-5 is ECMWF's fifth-generation reanalysis dataset for global climate and weather over the past 80 years [35].
It combines model data with observational data from around the world into a comprehensive and consistent global dataset using the law of physics, providing datasets spanning decades [35].The climate model ensemble mean in this study will be tested for spatio-temporal consistency with the ERA-5 dataset [35].
Prior to the experiment, we standardized the selected climate model data into a daily scale data.We used percentile values (90%, 95%) to represent the spatial distribution characteristics of extreme temperature and extreme precipitation in the research zone.In this study, two percentile indicators (T & R) represent extremely high temperature and extreme precipitation data.Based on these indicators (T90p, T95P, R90p, R95p), we studied the spatial distribution of temperature and precipitation in the research zone.
Over a 20-year time scale, we computed daily temperature and precipitation in each sub-region over the research area using the climate model ensemble data for the historical simulation and future projection.The daily maximum temperature indicator was selected as extreme temperature, and the interval for this study was chosen to be 0 to 35 °C [36].We then calculated the average value above the 90th percentile as daily extreme precipitation in the bins of 1 °C daily extreme temperature.This study looks at daily extreme temperatures and extreme precipitation over China's coastal lands and seas [36][37][38].
The rate of change of extreme precipitation is calculated based on the Clausius-Clapeyron (C-C) equation as follows [36][37][38]: The formula (1) and (2) can be derived as follows [11,12]: Precipitation-Temperature curves were constructed based on the (3) formula, and the least squares regression fitting was used to calculate the value of parameter c.According to the C-C equation, parameter c can be interpreted as the C-C scaling rate of the area [4,37].In areas with lower latitudes and sufficient water vapor supply, the relationship between temperature and precipitation is more likely to conform to the super C-C scaling [12].On the other hand, in areas with higher latitudes and insufficient water vapor supply, the relationship is more likely to conform to the sub-C-C scaling [11,12].
In the study of CHPEs, extremely high temperatures and extreme precipitation are chosen as key variables.An event is considered a high temperature or extreme precipitation event when the variable threshold exceeds the corresponding percentile of the historical stage [39][40][41].A CHPE is determined to have occurred when the time interval between high temperature and precipitation extremes does not exceed 3 days [16,19,40,42].In this study, we establish two types of CHPEs.The CE1-type, uses the 90th percentile as the threshold, while the CE2-type, uses the 95th percentile as the threshold.With two types of CHPEs, we can investigate the spatial distribution of CHPEs in the research zone.

Simulations of historical temperature and precipitation
In order to validate the historical model, we selected two indicators: the monthly average temperature and total monthly precipitation.These were used to verify the model's spatial distribution of time correlation.The outcomes are depicted in figure 2. Figure 2(a) illustrates the spatial distribution of time correlation.The ensemble model effectively simulates the temperature within the research zone, with a covariance value exceeding 0.9 across the entire region.The model's temperature fitting is superior in land areas compared to marine areas within each sub-region of the research zone.Specifically, the ensemble model exhibits optimal fitting in the land areas of the LD Area (40-45°N) and the YB Area (35-40°N), while it performs slightly less effectively in the marine areas of the ST Area (20-25°N) and the HN Area (15-20°N).
As for precipitation time correlation indicators, as shown in figure 2(b), the ensemble model provides a relatively ideal fit for most areas within the research zone, achieving above 0.5.The areas with the highest time correlation are concentrated in the ST Area (20-25°N), the YB Area (35-40°N), and the LD Area (40-45°N), where time correlation coefficients can reach 0.6 and above.However, the correlation fit is less than ideal in the SC Area (25-30°N) and the EC Area To gauge the skills of the ensemble model in reproducing the mean temperatures (Tas) and the mean precipitation (Pr) over the lands and seas over the research zone, the Taylor Diagram is introduced to summarize how closely the patterns from the ensemble model match the reference model [4].The Taylor model exhibits the correlation coefficient (CORs) and the normalized standardized deviations (SDN).On the plot, the pattern will have a closer distance to the reference point marked 'REF' on the x-axis.Figure 4 shows that, in China's coastal lands and seas, the ensemble model shows merits in both land and sea areas.For the Tas index, the ensemble model  data results in a great correlation over the research areas.Results from the marine areas have CORs greater than 0.9 and SDN between 0.8 and 1, the Land areas also have high CORs.For the Pr index, the ensemble data has a good performance in the areas between 30°N and 45°N.Overall, the ensemble model has sufficient skill to reproduce the spatial distribution of mean temperatures and mean precipitation over China's coastal lands and seas.
Figure 5 presents the spatial distribution of T90p and T95p in China's coastal lands and seas, derived from the climate models ensemble and ERA-5 data spanning from 1985 to 2004.As depicted in Figures 5(b) and (d), the ERA-5 data indicates that the highest temperature in China's coastal lands and seas is in South China.Within the research area, the temperature gradually decreases from south to north, with the Liaodong Peninsula and the Bohai Sea area recording the lowest temperatures.Apart from the sea area of the HN Area (15-20°N), the temperature over the marine areas is lower than that on land areas with the same latitude.In comparison, as shown in Figures 5(a      urban agglomeration, and the Beijing-Tianjin-Hebei urban agglomeration.Southern Guangxi experiences higher temperatures due to its proximity to the tropics and its extensive land area.The three urban agglomerations of the Pearl River Delta, Yangtze River Delta, and Beijing-Tianjin-Hebei exhibit higher extreme temperatures due to their large urban built-up areas and significant urban heat island effects. In terms of warming range, the research zone overall displays a trend of higher warming in the north and lower warming in the south.The warming range over the sea is smaller than that over land at the same latitude, with all marine areas (except for the Bohai Sea area) showing a lower level of warming range.Under the RCP4.5 scenario, for both 90% and 95% percentile temperature indicators, the warming range in the research zone is between 1.7 °C-3.4 °C.The marine area from 15°N to 30°N has the lowest warming range in the sea area, with an average warming of less than 2.0 °C; the HN Area (15-20°N) has the smallest warming range on land, with a warming range of about 2.4 °C.The marine area with the highest warming range is in the Bohai Sea area, with a warming range of about 3.0 °C; on land, it is in the LD Area (40-45°N) and YB Area (30-35°N), with a warming range of more than 2.8 °C, with an extreme warming value of 3.4 °C appearing in the central part of the LD area (40-45°N).
Under the RCP8.5 scenario, the warming range in the research zone is between 2.9 °C-5.4 °C.The sea area south of 30°N has a lower warming range of 2.9 °C-3.6 °C; the HN Area (15-20°N) on land has a minimum warming range of 3.9 °C-4.4 °C.The marine area with the highest warming range is in the YB area (35-40°N) with a warming range of 4.6 °C-5.0 °C; on land, it is in the LD Area (40-45°N) where it reaches up to 5.4 °C.Additionally, for the SC Area (25-30°N), the EC Area (30-35°N), and the YB Area (35-40°N) on land, when considering 95th percentile temperature, they have higher warming ranges than when considering 90th percentile temperature.
In the context of climate change, extremely high temperatures can lead to increased atmospheric instability [10,16,43].This results in an accelerated exchange of water vapor between the sea, land surfaces, and the atmosphere, thereby increasing the rate of water vapor accumulation in the atmosphere and intensifying extreme precipitation events.Over the marine areas, high temperatures increase the evaporation rate of sea surface water vapor, facilitating the formation of strong convective weather systems along the coast.Concurrently, the rise in seawater temperature and sea surface temperature makes it easier for marine heat waves to form in nearshore areas, negatively impacting marine ecology and seawater aquaculture [7, 10, 44].Over the land areas, high temperatures are more likely to evaporate surface moisture, transforming latent heat flux from the surface into sensible heat flux.This can trigger compound extreme events characterized by high temperatures and drought [2,19,42].On the other hand, an increase in temperature leads to a rise in the water vapor content in the atmosphere.This can trigger heavy precipitation more easily, potentially leading to floods.Such climatic changes pose significant challenges to public health and infrastructure [15,45].
Figure 10 depicts the spatial distribution of future precipitation at the 90% and 95% percentiles under the RCP4.5 and RCP8.5 scenarios from 2080 to 2099 in the research zone, in comparison with the historical precipitation spatial distribution.Historically, precipitation was primarily concentrated in areas from 15°N to 30°N.Under the RCP4.5 and RCP8.5 scenarios, precipitation is mainly concentrated in the SC Area (25-30°N), particularly at the coastline of Fujian and Zhejiang.In terms of changes in extreme precipitation compared to the historical period, under both RCP4.5 and RCP8.5 scenarios, there is a trend of a decrease in areas from 15°N to 25°N and an increase in areas from 25°N to 45°N for both marine and land areas.The increase in extreme precipitation is concentrated over both marine and land areas of 25-35N°.
Under the RCP4.5 scenario, the area with the largest decrease in extreme precipitation is the west coast of Hainan Island, reaching 2.8-3.9 mm d −1 , while the area with the largest increase is at the coastline of Fujian and Zhejiang, with an increase in extreme precipitation of 2.2-5.4 mm d −1 .Under the RCP8.5 scenario, whether it is an increase or a decrease, the change in extreme precipitation is smaller than under the RCP4.5 scenario.The area with decreased precipitation is mainly concentrated in the ST Area (20-25°N), with a decrease in extreme precipitation of 2.1-3.2mm d −1 .The area with increased extreme precipitation is mainly concentrated in the SC Area (25-30°N) and EC Area (30-35°N), with an increase in extreme precipitation of 1.0-2.8mm d −1 .
In the context of climate change, a decrease in precipitation can lead to a reduction in surface soil moisture [44,46].This transformation of latent heat into sensible heat can facilitate the onset of high temperatures, potentially triggering Compound Heatwave-Drought Extremes [16,28,42].These events can have detrimental impacts on various sectors.Conversely, heavy rainfall events can disrupt local atmospheric stability in a short timeframe, increasing the likelihood of extremely high temperatures [28].Additionally, the surge in humidity from heavy rainfall, when combined with high temperatures, can result in Compound heat waves and Humid events.Such events can adversely affect the health of residents [15,47].As depicted in figure 11, the P-T curve obtained through the temperature bin in the research zone is presented.The general trend in the research zone is a gradual decrease in extreme precipitation as latitude increases.The overall pattern shown in figure 11 indicates that the increase in precipitation in marine areas is typically slightly larger than that in land areas at the same latitude.Concurrently, with climate change progression, both the peak value of extreme precipitation on the P-T curve in each area and the corresponding peak temperature are increasing.This is illustrated by the rightward shift of the RCP4.5 (Black curves) and RCP8.5 (Red curves) curves compared to the historical period curve (Blue curve).
Among the 12 sub-regions curves, the P-T curves of the marine parts of the ST Area are all open sea areas with no significant land areas influencing local circulation.Water vapor can be smoothly replenished from the sea surface to the atmosphere, thus displaying a hook-shaped curve.Previous research suggests that limited water supply under high temperatures may result in peak-shaped curves in the P-T relationship in this region.For the other three marine sub-regions, due to Hainan Island, Shandong Peninsula, and Liaodong Peninsula lying across the sea area, water vapor circulation in these areas is greatly affected by land surface, and sea-land circulation also affects water vapor replenishment.Therefore, these three marine subregions, along with other land sub-regions, show peak-shaped curves.As latitude increases, P-T curves fluctuate to varying degrees due to annual changes in temperature and precipitation.In sub-regions from 35°N to 45°N, the influence of annual changes in temperature and precipitation on P-T curves results in larger curve fluctuations.Compared with the historical period, under both RCPs, P-T curves in each sub-region within the research zone have different degrees of upward and rightward shifts compared to historical periods.Our research shows that the study of P-T curves under RCP4.5 and RCP8.5 scenarios, reveals a rightward and upward shift compared to the historical period, representing the tendency of warming and more precipitation in the future.Due to rising temperatures, peaks of daily precipitation increase in our research zone, with the largest increase (9.87 mm d −1 ) observed in the land part of the SC Area (25-30°N).Daily peak temperatures which precipitation peaks are between 21 °C and 29 °C, showing 2 °C-6 °C of increase compared to historical periods.This is consistent with the previously mentioned trend of increasing spatial extent of extreme precipitation in the research zone.The peak daily precipitation in most of the research zone will rise by 2-7 mm d −1 , and the temperature at which the daily temperature peaks will rise by 2 °C-6 °C, which is similar to previous studies [12,36].The land part of the SC Area (25-30°N) shows the highest peak daily precipitation increase by 9.87 mm d −1 and the highest peak daily temperature increase of 6 °C among the sub-regions.
As presented in table 2, we calculated the C-C scaling rates in each area under historical, RCP4.5, and RCP8.5 scenarios based on the P-T curves shown in figure 8. Overall, the C-C scaling rates in the marine parts of the HN Area (15-20°N) and the ST Area (20-25°N) exceed 10%, conforming to super C-C scaling in the area.In contrast, inland parts from 30°N to 45°N, C-C values are less than 6%, conforming to sub-C-C scaling.The C-C  scaling rates in other areas conform to C-C scaling.The general trend in the research zone is a decrease in C-C scaling rates on both marine and land parts as latitude increases.
With the increase of radiative forcing level, the C-C scaling rates in each area exhibit varying levels of change.In the marine part of the HN Area (15-20°N), the C-C scaling rate increases with radiative force rise, as water vapor circulation is promoted with rising temperature in this area.On land, due to land surface constraints, water vapor replenishment is limited, and the parameter value decreases with temperature rise.The C-C scaling rate in the ST Area (20-25°N) shows a trend of first decreasing and then increasing with the increase of radiative forcing.The C-C scaling rate under the RCP4.5 scenario is the lowest among the three scenarios, and under the RCP8.5 scenario, it's smaller than that during the historical period.This is because while the temperature rises in this area, precipitation decreases, resulting in a calculated C-C scaling rate smaller than that during the historical period.Both marine and land parts in the SC Area (25-30°N) show a decrease in scaling values with temperature rise.In the EC Area (30-35°N), the marine parts' C-C scaling rate decreases while the land part increases.In the YB Area (35-40°N) and the LD Area (40-45°N), marine parts' C-C scaling rate increases with the increase of radiative forcing while land parts' show a decreasing trend; with climate change in this area, water vapor circulation over the sea is promoted while over land is suppressed.

Frequency changes of compound heat-precipitation extreme events
While global warming, the Clausius-Clapeyron (C-C) relationship suggests that as near-surface air temperature rises, the water vapor content in the atmosphere will increase according to the C-C scaling rate.As discussed earlier, under future scenarios, temperature will rise with increased radiative forcing, and the C-C scaling rate in some areas will also increase correspondingly in China's coastal lands and seas.Therefore, extreme precipitation values and frequencies should increase correspondingly.With the increase in frequency of high temperature and precipitation extremes, the probability of their co-occurrence will greatly increase.Hence, in this section, we set two types of CHPEs (CE1 and CE2) to analyze the frequency and changes in our research zone.
As illustrated in figure 12, under the CE1-type, historical CHPEs mostly occur in the northern land area of the YB Area (35-40°N) and land area of the LD Area (40-45°N), reaching up to 740 times in a 20-year time scale.The marine area of the ST Area (20-25°N) also experiences a high frequency of CHPEs, while areas with lower frequency are the SC Area (25-30°N) and the EC Area (30-35°N).Under RCP4.5 and RCP8.5 scenarios, the occurrence rate of CHPEs in all areas has greatly increased.CHPEs are frequent in marine and land areas of the HN Area (15-20°N), the ST Area (20-25°N), and the land part of the LD Area (40-45°N).It is noteworthy that with increased future radiative forcing levels, the frequency of CHPEs in land part of the HN Area (15-20°N) and the ST Area (20-25°N) will greatly increase, which will more easily affect coastal cities and related sea areas.Simultaneously, North China Plain also experiences a high frequency of CHPEs.The research indicates that in areas with a high level of urbanization, such as the Pearl River Delta urban agglomeration, the Yangtze River Delta urban agglomeration, and the Beijing-Tianjin-Hebei urban agglomeration, the frequency of CHPEs will increase as radiation levels rise.This is consistent with existing research [16,41,47].We also made a frequency distribution chart for each sub-region.As shown in figure 13 over the research zone, CE1-type CHPEs show regional differences, with more occurrences in the north and south parts and fewer in the middle.The EC Area (30-35°N) has the lowest frequency of CHPEs in the research zone, and the growth rate of CHPEs in this area is the smallest under RCP4.5 and RCP8.5 scenarios.As can be seen from figure 13(a), during the historical period, the areas most prone to CHPEs are the LD Area (40-45°N) and the ST Area (20-25°N ), on a future scale, the HN Area (15-20°N) and the ST Area (20-25°N) have faster growth rates and higher frequencies of CHPEs than others.As shown in figure 13(b), there is also a similar frequency distribution trend in land areas as in marine areas.
Compared to the historical period, each sub-region in the research zone under the future RCP4.5 scenario has an increase of 36.2%-85% in CE1-type CHPEs, and under the RCP8.5 scenario, the increase is 36.2%-86.8%.In marine areas, the area with the highest growth rate is the HN Area (15-25°N), where the frequency of CHPEs increases by 73.2% and 75.6% under RCP4.5 and RCP8.5 scenarios respectively.The area with the smallest growth rate is the LD Area (40-45°N), with growth rates of 40.9% and 51% respectively.In land areas, the frequency of CHPEs also increases most in the HN Area (15-20°N), with growth rates reaching 85.0% (RCP4.5)and 86.8% (RCP8.5)respectively.The area with the smallest increase in the frequency of CHPEs is also the LD Area (40-45°N), with an increase of 36.2% for both scenarios.Under different RCPs, the area with the largest difference in frequency growth of CHPEs compared to the historical period is the marine part of the SC Area (25-30°N).Over the land part of the, under the RCP4.5 scenario, the frequency of CHPEs increases by 52.0%, while under the RCP8.5 scenario, it increases by 76.3%, a difference of 24.3%, indicating that the frequency of CHPEs in this area is more easily affected by climate change.
In summary, the overall increase in frequency of CHPEs shows a decreasing trend from south to north under CE1-type.Comparing the marine and the land areas, CE1-type CHPEs are more likely to occur in the marine areas.The HN Area (15-20°N) and the ST Area (20-25°N) will have higher frequencies and larger growth rates of CHPEs in the future.Comparing RCP4.5 and RCP8.5 scenarios, it can be seen that the marine part of the SC Area (25-30°N) has the most obvious increase in the frequency of CHPEs with increasing radiative forcing.
The CE2-type of CHPEs shows spatial distribution similar to the CE1-type.As presented in figure 14, during the historical period, they are more likely to occur in the land part of the LD Area (40-45°N), the marine part of the ST Area (20-25°N), and the HN Area (15-20°N), and less likely to occur in the EC Area (30-35°N).Under RCP4.5 and RCP8.5 scenarios, the areas with frequent future CHPEs have shifted to areas south of 30°N, especially frequent in near-sea and coastal zones.The most frequent areas are the west coast of Hainan Island and along the Taiwan Strait.In terms of latitudinal distribution, from 15°N to 25°N, the frequency of CHPEs in marine areas is higher than in land areas.From 25°N to 45°N, the frequency of the CHPEs between marine areas and land areas is similar.It showed the same phenomenon with CE1-type CHPEs in urbanized areas experiencing a higher frequency of CHPEs.
The frequency of CE2-type CHPEs in different sub-regions is shown in figure 15.Similar to the CE1-type CHPEs, CE2-type CHPEs are more likely to occur in marine areas as well.Overall, under the RCP4.5 scenario, the frequency in each sub-region has increased by 69.8%-141.4% compared to the historical period, while under the RCP8.5 scenario, the increase in frequency is 100.1%-163.6%.
As shown in figure 15(a), the area most prone to CHPEs in the marine part is the ST Area (20-25°N), and the area with the lowest frequency is the EC Area (30-35°N).In the RCP4.5 scenario, the highest frequency of CHPEs over the marine areas is in the HN Area (15-20°N), and under the RCP8.5 scenario, it is in the SC Area (25-30°N).Compared with the historical period, under the RCP4.5 scenario, the frequency of CHPEs in the HN Area (15-20°N) increased by 141.4%, which is the highest among the six marine sub-regions, and the lowest is in the LD Area (40-45°N), which is 95.5%.Under the RCP8.5 scenario, the highest increase in composite event frequency occurs in the SC Area (25-30°N), which is 161.2%, and the lowest value is also in the LD Area (40-45°N ), which is 127.2%.Under both radiation forcing paths, the area with the highest difference in CHPEs frequency is in the SC Area (25-30°N).
As shown in figure 15 ) and the ST Area (20-25°N) in the RCP4.5 scenario, while in the RCP8.5 scenario, they occur most frequently on the HN Area(15-20°N).The lowest occurrence of CHPEs is in the EC Area (30-35°N).Compared with the historical period, among six sub-regions, in the RCP4.5 scenario, the highest increase in frequency occurs in the ST Area (20-25°N), with an increase of 130.4% for CHPEs, and the lowest value is shown in the EC Area (30-35°N ), which is 55.5%.The highest increase occurs in the HN Area (15-20°N), which is 163.6%, and the lowest value occurs in the EC Area (30-35°N), which is 100.1% while the RCP8.5 scenario.At the radiation forcing paths, the highest difference in CHPEs frequency occurs in the SC Area (25-30°N), which is the same as in marine areas.
With different RCPs, the area with the largest difference in frequency growth of CHPEs compared to the historical period is also the SC Area (25-30°N), which is similar to the CE1-type CHPEs.Over the marine part of the SC Area (25-30°N), under the RCP4.5 scenario, the frequency of CHPEs increases by 110.7%, while under the RCP8.5 scenario, it increases by 161.2%, showing a difference of 50.5%; over the land part, the increase of CE2-type CHPEs' frequency is 89.3% (RCP4.5)and 136.4% (RCP8.5),showing a difference of 47.1%.The result above indicates that the frequency of CE2-type CHPEs in this area is more easily affected by climate change.
Compared with CE1-type, and CE2-type CHPEs, there will be a greater increase in frequency.At the same time, with our result in the part of future precipitation prediction part and the P-T curve part, the SC Area (25-30°N) has the future precipitation center, highest peak temperature increase, highest peak temperature increases, and the largest difference in CHPEs frequency increase under different RCPs, indicates the zone will be greatly affected by climate change and CHPEs in the research zone.It also predicts that future compound event frequency of CE2-type CHPEs will be higher and more severe along China's coastal lands and seas.It may pose more serious challenges to people's health and economic production within urban agglomerations and on the near-sea along coastlines.High-temperature processes in CHPEs can adversely affect infrastructure such as power.Heavy rainfall processes in CHPEs may wash organic matter into river channels utilizing urban surface currents.While bringing waterlogging to the city, it also brings the hidden danger of eutrophication to the river water body.When these nutrient-rich water bodies enter the ocean with surface runoff, they are easy to cause red tide and green tide in offshore waters, which has adverse effects on offshore economic activities.However, the limitation of our method is that it is not possible to study the interval, duration, and seasonal distribution between high temperature, precipitation, and CHPEs in more detail.The uncertainty of different land-sea interactions for the occurrence of compound events between them in the future is also worth discussing.Compared with the CMIP5 RCMs used in this research, we also expect that high-resolution GCMs and RCMs under the framework of CMIP6 will simulate the extreme temperature and precipitation events over China's coastal lands and seas more accurately.

Conclusion
This study employed a high-resolution climate models ensemble under the CMIP5 framework to study the response of extreme temperature and extreme precipitation, as well as their spatially Compound Heat-Precipitation Extreme Events (CHPEs) under RCP4.5 and RCP8.5 over China's coastal lands and seas.First, this study examines the performance of ensemble models in simulating historical climate from 1985 to 2004.The ensemble model is capable of simulating temperature and precipitation and reproducing their spatial distribution in China's coastal lands and seas.It has high time correlations, field correlations, and similar standard deviations compared with reanalysis model data ERA-5.
The ensemble model predicts that by the end of the 21st century.The daily extreme temperature in China's coastal lands and seas will rise by 2.9 °C to 5.4 °C.The smallest warming range is over the marine areas from 15°N to 30°N, with a warming range of 3.0 °C, and the largest warming range is in the Bohai Sea area, with a warming range of 5.4 °C; The area with the highest warming rate on land is in Beijing-Tianjin-Hebei urban agglomeration, with a warming range reaching 5.4 °C.In terms of precipitation, overall changes show a trend of increasing in the north part of the research zone and decreasing in the south with 25°N as the boundary.For daily precipitation, we found precipitation decreases in the HN Area (15-20°N) marine part are between 2.8-3.9 mm d −1 while increases in the SC Area (25-30°N) are between 2.2-5.4 mm d −1 .
In our research of Precipitation-Temperature (P-T) curves and Clausius-Clapeyron (C-C) scaling values, we found that P-T curves in our research area mainly show peak curves due to sufficient water vapor replenishment.Sub-regions with insufficient water vapor supply, the P-T curves exhibit a peak-shaped curve, while sub-regions with sufficient water vapor supply show a hook-shaped curve.Under the RCPs, the peak of daily extreme precipitation and temperature at which precipitation peaks will increase while warming.The peak daily precipitation in most of the research zone will rise by 2-7 mm d −1 , and the temperature at which the daily temperature peaks will rise by 2 °C-6 °C.The land part of the SC Area (25-30°N) shows a peak daily precipitation increase of 9.87 mm d −1 and a peak daily temperature increase of 6 °C.When investigating C-C scaling values, we found that daily extreme precipitation and temperature generally show a C-C relationship.The area between 15-25°N shows a super C-C distribution relationship.
Overall, two types of CHPEs are more likely to occur in marine areas, with urban areas on land experiencing a higher frequency of such events.Within the research zone, the frequency of CHPEs tends to be higher in the south and north, and less in the middle.Under both RCP4.5 and RCP8.5 scenarios, there is a significant increasing trend in the frequency of CHPEs in each sub-region within the research area.CHPEs the HN Area

Figure 1 .
Figure 1.Study Area Overview and Sub-region Divisions.
(30-35°N), with low time correlation observed on the northern coast of Fujian.In conclusion, in terms of time correlation indicators, the model has demonstrated effective fitting results for both temperature and precipitation.This validates that the ensemble model effectively simulates historical climate.As field correlation, we compared the field correlation indicators of each model with the ensemble model.The findings are presented in figure 3.Figure 3(a) reveals that the four independent climate models effectively simulate historical temperatures in the research zone, with CCLM demonstrating the most effective simulation.The ensemble model process enhances the simulation of historical temperature.As depicted in figure 3(b), for precipitation, the median of the field correlation indicators for historical precipitation across the four climate models is within the range of 0.2-0.4.The median value of the historical precipitation field correlation surpasses 0.4, indicating a significant improvement in the simulation by the ensemble model of historical precipitation.In conclusion, the ensemble model can enhance the simulation level of temperature and precipitation data for this region's actual climate in our research zone.

Figure 2 .
Figure 2. Regional Distribution Map of Temperature (a) and Precipitation (b) Time Correlation.

Figure 3 .
Figure 3. Field Correlation Graph of the Temperature (a) and Precipitation (b).
) and (c), the ensemble model captures the high-temperature center in the subtropical region of Guangdong and Guangxi, as well as the low-temperature center in the Liaodong Bay area.Overall, it appears that the ensemble model data underestimates the temperature over the marine part of the HN Area (15-20°N) while overestimating it over the land part in the ST Area (20-25°N).

Figure 6
presents the 90th and 95th percentile (R90p, R95p) precipitation data derived from the model data and ERA-5 data.According to the ERA-5 data, in terms of the R90p and R95p indicators, the ST Area (20-25°N), the SC Area (25-30°N), and the EC Area (30-35°N) record higher precipitation values on land.Conversely, there is less precipitation on land in the YB Area (35-40°N) and the LD Area (40-45°N).In comparison to ERA-5, the ensemble model data underestimates the precipitation on land in the ST Area (20-25°N) and the SC Area (25-30°N), while overestimating it over the marine part of the ST Area (20-25°N) and on land in the YB Area (35-40°N).The ensemble model identifies the precipitation center over the South China Sea, whereas the ERA-5 reanalysis locates it in the Pearl River Delta region and at the coastline of Zhejiang and Fujian provinces.Despite some bias in fitting validation data in terms of precipitation, the ensemble model still effectively simulates historical climate.The ensemble model has high time correlations, field correlations, and similar standard deviations compared with validation data ERA-5.It is capable of simulating temperature and precipitation and reproducing their spatial distribution in China's coastal lands and seas.Figures 7 and 8 depict the spatial distribution of two types of Compound Heat-Precipitation Extreme events (CHPEs) across China's coastal lands and seas.These distributions are derived from both the ensemble of climate models and the ERA-5 data.Generally, the ensemble mean data effectively captures the spatial distribution of both types of CHPEs.As illustrated in figure 7, for the CE1 type of CHPEs, the land part of the YB Area (35-40°N) and the LD Area (40-45°N) record higher values of CHPEs.Conversely, lower values are recorded in the marine part of the EC Area (30-35°N).Compared with the validation data, the ensemble mean model tends to underestimate values in the land part of the YB Area (35-40°N), while overestimating values in the marine part of the ST Area (20-25°N).As depicted in figure 8, for the CE2 type of Compound Heat-Precipitation Extreme events (CHPEs), the land part of the YB Area (35-40°N) and the LD Area (40-45°N) record higher values of CHPEs.Conversely,

Figure 9
Figure9illustrates the spatial distribution of the estimated 90% and 95% percentile temperature indicators relative to the historical period, as derived from the ensemble model under RCP4.5 and RCP8.5 scenarios for 2080-2099.The temperature spatial distribution in the research zone generally exhibits a pattern of higher temperatures in the southern region and lower temperatures in the northern region.High-temperature centers are primarily located in southern Guangxi, the Pearl River Delta urban agglomeration, the Yangtze River Delta

Figure 7 .
Figure 7. Spatial distribution of CE1 CHPEs in the research zone from 1985 to 2004, Ensemble Model (a) and Validation Dataset ERA-5 (b).

Figure 11 .
Figure 11.The Precipitation-Temperature (P-T) curve for each sub-region.(The blue curve -the historical stage scenario; the black curve -the RCP4.5 scenario; the red curve -the RCP8.5 scenario.).

Figure 13 .
Figure 13.Frequency Distribution Chart of each Sub-region over the Marine Areas (a) and the Land Areas (b) of the Research Zone under CE1-type CHPEs.
(b), for land areas, CHPEs occur most frequently in the HN Area (15-20°N) and the LD Area (40-45°N), and least frequently in the SC area (25-30°N).The frequency of CHPEs on land conforms to the same distribution frequency in the CE1-type.CHPEs occur most frequently on land in the HN Area (15-20°N

Figure 15 .
Figure 15.Frequency Distribution Chart of each Sub-region over the Marine Areas (a) and the Land Areas (b) of the Research Zone under CE2-type CHPEs.

Table 1 .
Information of Climate Models Used in the Study.

Table 2 .
The Clausius-Clapeyron (C-C) Scaling Rate in Each Subregion of Every Scenario.