Nested MPI-CWRF dynamical downscaling and future projections of summer extreme precipitation changes in China

Abstract

This study develops a nested MPI-CWRF dynamical downscaling system by using the regional Climate-Weather Research and Forecasting Model (CWRF, ~30 km) driven by the MPI-ESM1-2-HR global climate model (GCM, ~100 km), aiming to improve the future projections of summer extreme precipitation over China. This system is implemented for both present climate simulations (1980−2014) and future projections (2016−2050) under the highest emission scenario of SSP5-8.5. Comparative analyses with high-resolution MPI-ESM1-2-XR (~50 km) demonstrate that dynamical downscaling achieves superior improvements in simulating summer precipitation extremes than mere GCM resolution enhancement. Future projections indicate divergent trends and interannual variability across eastern China: summer precipitation averages (PRA), daily precipitation extremes (P95), and intensity of persistent precipitation events (Rx5day) are projected to have greater increases and enhanced interannual variability under MPI-CWRF compared to GCM, while the number of rainy days (NRD) is projected to decline with reduced interannual variability in south-eastern regions. Specifically, Central China exemplifies this pattern with NRD decreasing by 4.0%, but PRA, P95, and Rx5day increasing by 4.9%, 8.9%, and 11.8%, respectively. These projected changes correlate with key atmospheric shifts: the south-eastward expansion of the South Asian High, the southward-displaced East Asian jet, and anomalous low-level cyclonic circulations over the Yellow-Bohai Seas, as well as strengthened moisture convergence and convective available potential energy through intensified southerly monsoon flows. The physical coherence among these key circulation changes ensures the reliability of extreme precipitation projections made by the MPI-CWRF dynamical downscaling system.

Graphical abstract

Keywords

regional climate model / dynamical downscaling / summer extreme precipitation / future projection

1 Introduction

Under global warming, China has experienced increased frequency and intensity of extreme precipitation since the 1960s, yet with pronounced regional differences. For example, the strongest increasing trends of the intensity and frequency of summer extreme precipitation are observed in the middle-lower Yangtze River Basin, coastal south China, and arid northwest China. Among them, east China exhibits the largest increase in extreme precipitation amount, while northwest China shows the fastest increase rate in the frequency (Lu et al., 2020). Sun and Zhang (2017) also identified a triple pattern with significant increasing extremes in southeast and northwest China, contrasting with decreasing trends in north China. Despite overall decreasing trends in southwest, north, and northeast China, recent years have witnessed high-impact extreme precipitation events with high interannual variability, such as the 2022 Beichuan mountain flood disaster (southwest China), the 2022 Songliao River basin dam-break floods (northeast China), and the 2023 Typhoon Doksuri-induced mega-rainfall in Beijing-Tianjin-Hebei region (north China). Many climate models have conducted simulations on the distribution and variation of historical summer mean and/or extreme precipitation over China, but systematic biases and challenges still exist (He and Zhai, 2018; Lu et al., 2020; Wang et al., 2021).

For instance, global models such as Coupled Model Intercomparison Project Phase 6 (CMIP6) still exhibit systematic biases in regional extreme precipitation simulations (Wang et al., 2021; Huang et al., 2024). These include underestimation of extreme precipitation across southern China and persistent extreme rainfall intensity in north China, contrasted with overestimation of extreme precipitation intensity and duration in eastern Inner Mongolia and northeast China. Especially in northwest and Xizang, the models falsely simulate heavy rainfall occurring on the southern Xizang Plateau (Wang et al., 2021). All such biases fundamentally compromise the reliability of CMIP6-based projections for future extreme precipitation changes.

Regional climate models (RCMs), through dynamical downscaling with refined model resolution and physics, offer a feasible pathway to improve regional simulations. Weather Research and Forecasting model (WRF) and Regional Climate Model systems (RegCM) have been widely used for downscaling the global reanalysis or global climate models (GCMs) to reduce regional biases in China. Notably, the RegCM4.6 model shows more significant improvements than the WRFv3.6 model in simulating the spatiotemporal distribution of extreme temperatures in China (Kong et al., 2019), while WRFv3 model exhibits higher fidelity over RegCM4 in simulating extreme precipitation in eastern China (Wang et al., 2016). However, these RCMs still have limitations in simulating summer mean and extreme precipitation. For instance, dynamical downscaling using WRF driven by IPSL-CM5A-LR effectively reduced but did not eliminate precipitation underestimation in south-eastern China, despite mitigating systematic biases in east China’s daily mean and extreme precipitation intensities (Wei et al., 2019). Wang et al. (2024) showed that as the horizontal resolution increased from 50 km to 25 km and further to 10 km, RegCM better improves the summer precipitation simulation over northern Xizang than WRF, yet both persistently produce overestimation bias over southern Plateau.

Recently, the new generation of regional Climate-Weather Research and Forecasting Model (CWRF) has been developed upon the WRF model with optimized configuration for domain and lateral boundary conditions, as well as improved representations of key physical processes, including convection-microphysics, cloud-aerosol-radiation, and land-sea-atmosphere interactions (Liu et al., 2008; Liang et al., 2012; Liu and Liang, 2017). In long-term simulation experiments driven by European Centre for the Medium-Range Weather Forecasts (ECMWF) reanalysis, the CWRF downscaling significantly reduces the precipitation biases in the ECMWF reanalysis, particularly for extreme precipitation in the Yangtze River Basin and south China (Yuan and Liang, 2011; Liang et al., 2019; Jiang et al., 2021; Xu et al., 2024; Zhang et al., 2024). Given the CWRF’s demonstrated capability, it is scientifically imperative to employ this model for dynamical downscaling studies and regional climate change projections of summer extreme precipitation over China.

Driving field has a significant impact on the downscaling performance of regional climate models. The CMIP6 models are generally used as driving fields for regional downscaling simulations, and a large number of studies have been conducted to evaluate the performance of multiple CMIP6 models. For example, Xu et al. (2024) evaluated the precipitation simulation capabilities of 23 CMIP6 models over China and found that MPI-ESM1-2-HR (MPI-HR) is the best-performing model for southern China. Yang et al. (2024) evaluated the capability of 30 CMIP6 models in simulating summer monsoon precipitation in northern China and found that MPI-HR is also one of the top four models that best simulates the summer rainfall spatial distribution. Therefore, MPI-HR is selected as the driving field for downscaling simulations and future projections in this study. Regarding the selection of future emission scenarios, previous studies have shown that under different emission scenarios, the trends in future changes of precipitation intensity in most regions of China are basically consistent, but the signals of future extreme changes are more pronounced under higher emission scenarios (Zhou et al., 2021). Hence, this study selects the highest emission scenario of Shared Socioeconomic Pathways (SSP5-8.5) (Zhang et al., 2019) to better demonstrate the changing trend of future extreme precipitation.

Therefore, this study constructs a nested MPI-CWRF dynamical downscaling system based on the advantageous global climate model of MPI-HR and regional climate model of CWRF to (i) quantify improvements of the dynamical downscaling model over the global climate model in simulating summer extreme precipitation characteristics in different regions of China; (ii) demonstrate the superiority of dynamical downscaling over merely resolution enhancement of GCM by comparing with MPI-ESM1-2-XR (MPI-XR), the high-resolution version of MPI-HR. The main objective is to employ this nested MPI-CWRF model to make reliable projections of summer extreme precipitation over China. Furthermore, the possible physical mechanisms driving the future increase in summer extreme precipitation in the eastern China are also elucidated through the impact of key atmospheric circulation systems, moisture convergence, and convective available potential energy (CAPE).

The paper is structured as follows. Section 2 describes the regional climate model CWRF, the global climate model of coarse-resolution MPI-HR as driving field, and the high-resolution MPI-XR. It also includes observa-tional data, extreme precipitation indices, and statistical methods used to evaluate the spatial distribution and interannual variability simulations. Section 3 evaluates the performance of the nested MPI-CWRF dynamical downscaling model (following abbreviated as CWRF) in simulating key features of extreme precipitation for the present climate (1980−2014) in comparison with MPI-HR and MPI-XR models. Section 4 presents the projected spatial distribution and interannual variability of summer extreme precipitation indices for the future climate (2016−2050). Section 5 explores the causes of future changes in summer extreme precipitation, and Section 6 summarizes the key findings.

2 Model, data, and methods

Fig.1 shows the CWRF computational domain, using Lambert projection with the central latitude and longitude of 35.18°N and 110.0°E. The domain consists of 232 × 172 grid points with a horizontal resolution of 30 km, a buffer zone of 14 grids, and 36 vertical layers, with the top layer located at 50 hPa. The physical parameterization schemes used in CWRF include the Ensemble Cumulus Parameterization (ECP) for deep convection (Qiao and Liang, 2016, 2017), the University of Washington (UW) shallow convection scheme (Park and Bretherton, 2009), the GSFCGCE microphysics scheme (Tao et al., 2003), the CAM boundary layer scheme (Holtslag and Boville, 1993), the CSSP land surface process (Choi et al., 2007), and the GSFCLXZ radiation scheme (Chou and Suarez, 1999; Chou et al., 2001). This set of schemes has been designated as the configuration for the CWRF control experiments, demonstrating considerable advantages in simulating seasonal averages, interannual variability, and particularly summer extreme precipitation in the eastern China when driven by the ECMWF reanalysis in long-term experiments (Yuan and Liang, 2011; Jiang et al., 2021).

In this study, the driving field is MPI-ESM1-2-HR (Müller et al., 2018), released by the Max Planck Institute, with a horizontal resolution of approximately 100 km × 100 km. The institute has also developed a higher-resolution version of this model, MPI-ESM1-2-XR (MPI-XR, Gutjahr et al., 2019) under the CMIP6 HighResMIP protocol (Haarsma et al., 2016). MPI-XR has a horizontal resolution of approximately 50 km × 50 km, which is closer to the resolution of CWRF (~30 km) and will serve as a comparator to evaluate the added value of nested MPI-CWRF dynamical downscaling against conventional GCM resolution enhancement approaches.

For model evaluation, daily precipitation observations from 1980 to 2014 are obtained from the CN05.1 data set with a horizontal resolution of 0.25° × 0.25°, which is constructed from over 2400 meteorological stations across China by Wu and Gao (2013). Besides, the latest fifth-generation reanalysis dataset from ECMWF (ERA5), with a horizontal resolution of 0.25° × 0.25°, is adopted as the reference proxy for circulation features (Hersbach et al., 2020). To compare the results, all datasets are interpolated to the 30-km horizontal grids used by CWRF for consistency. The study region is divided into eight subregions (Fig.1), with the division method following Jiang et al. (2021) and Liu et al. (2021).

According to Liang et al. (2019) and Jiang et al. (2021), four key precipitation indices from the Expert Team on Climate Change Detection and Indices (ETCCDI) (Karl et al., 1999) are adopted to characterize extreme precipitation features: daily average precipitation (PRA), the 95^th percentile (P95), maximum consecutive 5-day precipitation (Rxday5), and the number of rainy days (NRD). The specific definitions of these indices are provided in Tab.1.

The statistics used in this study include spatial correlation coefficient (CORR), root mean square error (RMSE), and interannual variability skill score (IVS) (Gleckler et al., 2008). IVS compares the standard deviation of year-by-year data between model simulations and observations, which can more objectively measure the similarity between the modeled and observed interannual variability. It has been widely used for evaluating the predictive skills of climate models (Scherrer, 2011; Yang et al., 2021), which is calculated as

(1)IVS={\left(\frac{{STD}_{\mathrm{m}}}{{STD}_{\mathrm{o}}}-\frac{{STD}_{\mathrm{o}}}{{STD}_{\mathrm{m}}}\right)}^2,

where STD_m and STD_o are the standard deviation of model and observation, respectively. A smaller IVS value indicates a closer match between the model and observation.

3 Present climate simulations

Fig.2 shows the 1980−2014 summer spatial distributions of four extreme precipitation indices (PRA, P95, Rx5day, NRD) in China, alongside the bias distributions simulated by MPI-HR, MPI-XR, and CWRF, with regional mean biases quantitatively compared in Fig.3. The 35-year historical observations (Fig.2(a)) reveal distinct spatial patterns: precipitation amount (PRA) peaks in eastern and southern China, aligning with main rainfall belts in SC, CC, SW, southern NE, and eastern NC. The spatial distributions of P95 and Rx5day are similar, showing maxima in eastern CC and along the SC coast, whereas NRD reaches its highest values over the southern TB, reflecting its near-continuous summer rainfall. Regions like SC, SW, CC, and NE experience over 45 rainy days per summer.

The MPI-HR model shows relatively small biases in capturing extreme precipitation indices over the four key regions of northern China (NC, NE, IM, NW), except for a pronounced NRD underestimation in the NW region. However, systematic biases dominate south-western regions (SW, TB): PRA, P95, and Rx5day are overestimated, while the NRD shows opposite errors (overestimation in SW, underestimation in TB) In south-eastern China (CC, SC), the MPI-HR model consistently underestimates PRA, P95, and Rx5day, but overestimates NRD.

However, the increased resolution of the MPI-HR does not effectively resolve the biases in simulating extreme precipitation in China. Specifically, the MPI-XR model, as the high-resolution version of MPI-HR, only yields limited benefits, marginally reducing PRA overestimation south of the TP region, but exacerbating wet biases in SW and intensifying the errors in SC (underestimated P95 and Rx5day) and CC (overestimated NRD).

CWRF downscaled from MPI-HR demonstrates marked improvements over both MPI-HR and MPI-XR in simulating all four extreme precipitation indices. It effectively mitigates the overestimation of PRA, P95, and Rx5day in the SW and TB regions, while correcting the systematic underestimation of PRA, P95, and Rx5day in the CC and SC regions, and the severe overestimation of NRD in the CC region. However, CWRF exacerbates the underestimation of NRD in the TB region, and over-corrects the underestimation biases of MPI-HR in the SC region, resulting in PRA and Rx5day overestimations.

Tab.2 quantifies the nationwide RMSE and CORR statistics for the four extreme precipitation indices simulated by these three models. Overall, CWRF achieves a 34% reduction in PRA bias (from 0.41 to 0.27 mm/day) and a 43% decrease in the P95 bias (from −0.91 to −0.52 mm/day) compared to MPI-HR, though it with slight increases in Rx5day and NRD biases (Fig.2). However, CWRF outperforms both GCMs across most metrics, exhibiting the lowest RMSE and highest CORR for P95 and Rx5day. While the RMSE of PRA slightly exceeds that of MPI-HR and NRD metrics show minor degradations, CWRF demonstrates superior spatial distributions for extreme precipitation overall. In contrast, MPI-XR ranks lowest in spatial performance among the three models.

Fig.4 compares the IVS for four extreme precipitation indices simulated by each model during the summer of 1980−2014 over China and its eight subregions. Overall, the MPI-HR model exhibits substantial IVS biases, while the MPI-XR model, with a resolution of 50 km, achieves considerable improvements for P95, Rx5day, and NRD but shows negligible progress for PRA. However, the CWRF model notably reduces the nationwide IVS biases compared to MPI-HR.

For the four key regions of northern China (NC, NE, IM, NW), the CWRF model systematically overestimates the IVS of the four indices in NW compared to the MPI-HR model. However, it shows great improvements in the other three regions: PRA and NRD in the NC region, P95 and Rx5day in the NE region, as well as PRA and P95 in the IM region. For the two key regions of south-western China (SW, TB), the CWRF model primarily corrects the IVS biases of NRD in the SW region and PRA in the TB region compared to the MPI-HR model.

For the two key regions of southeastern China (CC, SC), CWRF significantly reduces the bias of P95 and Rx5day, but largely overestimates the interannual variability of PRA, particularly in the CC region. This different performance arises from distinct precipitation drivers for these monsoon-dominated areas: large-scale circulation and localized convective processes. For summer mean precipitation, it is strongly influenced by the model’s response to large-scale circulations. The overestimated interannual variability may indicate that CWRF is overly sensitive to the interannual fluctuations in the driving fields. In contrast, summer extreme precipitation is predominantly governed by local convective processes, which can be better captured by CWRF with refined resolution and improved physics, resulting in smaller biases in the interannual variability of extreme precipitations. This finding is also supported by Liang et al. (2019), who validated CWRF’s ability to replicate both mean and extreme precipitation variability in central China when forced by ERA5’s accurate circulation fields.

Fig.5 provides an overall evaluation of model performance across eight sub-regions by ranking the regional mean biases, spatial distribution statistics (CORR, RMSE), and IVS for four extreme precipitation indices. Each metric’s top-performing model earns one point, yielding a maximum of 16 points per region. The final scores are normalized for cross-region comparability, with higher values indicating better simulation performance. The CWRF model shows clear superiority in simulating the spatiotemporal characteristics of summer extreme precipitation over China. Nationally, the CWRF model performs the best, scoring above half of the maximum points in the SW, SC, CC, and TB regions, with the highest score in the CC region, where extreme precipitation is most frequent during the summer. However, it is noted that the CWRF model underperforms in the IM region, scoring lower than MPI-HR and MPI-XR. This anomaly arises from the uniformly high model agreement in IM, where all the three models exhibit consistently high CORR (>0.9) and substantially lower RMSE and Bias values, leaving limited room for CWRF improvement despite its slight statistical lag.

The results indicate that the CWRF model substantially improves the ability of the MPI-HR model in capturing summer extreme precipitation events in China. Therefore, the next section primarily focuses on the projection of future extreme precipitation changes based on the CWRF model.

4 Future changes

Fig.6 presents the projected changes in summer extreme precipitation indices (2016−2050 vs. 1980−2014) along with the corresponding interannual variability changes for the driving MPI-HR model and the nested CWRF model. Regional mean changes of four indices are further quantified in Fig.7. Both the MPI-HR and CWRF models project intensified summer extreme precipitation indices (PRA, P95, Rx5day) across eastern China (NE, NC, CC, SC), along with reduced NRD in the western and southern regions (TB, SW, SC) in future.

However, the CWRF model predicts stronger P95 and Rx5day increases in the two key regions of south-eastern China (CC, SC) compared to the MPI-HR model. Notably, these two models are different in projected trends of NRD in the CC region. The MPI-HR model projects increased NRD in eastern CC, whereas the CWRF model forecasts a significant decrease across most of CC, as shown in Fig.6(a) and Fig.6(b). Consistent with the future changes, the CWRF model exhibits greater interannual variability for PRA, P95, and Rx5day in the eastern regions (NE, NC, CC, SC) compared to the MPI-HR model. For NRD, both models display similar distributions, with higher values concentrated primarily in parts of the northern regions (NE, IM, NC) and eastern TB region, but with decreasing interannual variability in the CC and SC regions, and the reduction is less pronounced in the CWRF model (Fig.6(c) and Fig.6(d)). The combination of the intensified precipitation and stronger interannual variability suggests escalating extremes in eastern China.

Given CWRF’s demonstrated skill in effectively mitigating GCMs’ NRD overestimations and minimizing biases of extreme precipitation indices (P95, Rx5day) in the CC region during the present-climate simulations, its future projections carry high credibility. Thus, it could be inferred that under the future high-emission scenario, CC, the key region of southern China, is expected to experience a 4.0% reduction in NRD, and substantial intensification of PRA, P95, and Rx5day, with average increases of 4.9%, 8.9%, and 11.8%, respectively.

To investigate the stability of bias signal propagation in the nested model, Fig.8 presents the correlation of regional mean differences of the four indices between the CWRF model and the driving MPI-HR model for both present climate and future projections. The results show that the differences between the CWRF and MPI-HR, whether in extreme precipitation intensity (PRA, P95, Rx5day) or precipitation days (NRD), exhibit a highly significant linear relationship between the present and future. This suggests that the bias correction signals from the CWRF model for extreme precipitation simulation in the present climate can be linearly propagated into future projections. Therefore, the CWRF’s excellent performance in present climate simulations ensures its credibility for future climate projections. The following section will further explore the physical mechanisms behind the significant increase of summer mean precipitation (PRA) and extreme precipitation intensity (P95, Rx5day) over the eastern China.

5 Understanding of precipitation projections

Previous studies have explored the critical role of large-scale circulation systems in modulating the interannual variability of summer mean precipitation in eastern China, particularly over the Yangtze River Basin, highlighting the importance of South Asian High (SAH), East Asian Westerly Jet (EAJ), and low-level southerly jet (Liang and Wang, 1998; Jiang et al., 2011; Wei et al., 2015; Li et al., 2021; Zhao et al., 2024). For instance, a recent study by Zhao et al. (2024) identified a teleconnection between the summer long-term mean precipitation simulation biases over the YRB and the large-scale circulation biases in CWRF simulations. The teleconnection pattern closely matches to the observed correlation between interannual variations in precipitation and circulation. Moreover, Zhang et al. (2023) found that the interannual variability of summer extreme precipitation in the Yangtze River Basin, is significantly associated with regional moisture convergence and CAPE. Based on these findings, the following analysis investigates the potential physical mechanisms underlying the projected increase in future summer extreme precipitation indices (PRA, P95, Rx5day) over most of eastern China (NC, CC, SC), focusing on changes in the key atmospheric circulation systems and regional factors including moisture convergence and CAPE.

Fig.9(a) compares present-climate (1980−2014) and future (SSP5-8.5) summer mean upper- and lower-level circulations from ERA5 reanalysis and CWRF simulations. Key circulation features include the SAH and its associated EAJ at 200 hPa, as well as the East Asian Monsoon and the low-level jet at 850 hPa. In the observed present climate shown in Fig.9(a1), the core of the EAJ is centered around 40°N, with the CC region located beneath its right exit-region upper-level convergence zone. The SAH covers over most of southern China (SW, SC, CC, TB), extending eastward to ~120°E. The CWRF model (Fig.9(a2)−Fig.9(a3)) captures the location of the EAJ core well, but slightly underestimates the jet intensity and shifts its exit westward. At the same time, the SAH is displaced eastward. Both the EAJ and SAH biases lead to a slight underestimation of precipitation in eastern CC. However, the CWRF model more realistically simulates the southerly summer monsoon south of the Yangtze River Basin, but the simulated monsoon trough over the Indian Ocean is slightly stronger, causing the rain belt to shift south-westward.

In the future climate, as shown in Fig.9(a4), the SAH amplifies and expands eastward into the north-western Pacific, while the EAJ exit region shifts southward, placing the CC and NC regions beneath its left exit’s upper-level divergence zone, which favors strong uplifting. At 850 hPa, wind anomalies over the SC and CC regions show enhanced westerly and southerly flows, intensifying the oceanic moisture advection into northern regions, which results in a significant increase in precipitation over northern SC. Concurrently, a strengthened low-level cyclonic circulation occurs over the Yellow-Bohai Seas. This dynamic coupling between upper-level divergence (EAJ shift) and low-level convergence (cyclonic anomaly) further amplifies the ascending motion, contributing to pronounced summer mean precipitation increases in the northern CC and southern NC regions.

Fig.9(b) and Fig.9(c) analyze the projected change in column-integrated moisture transport flux and its divergence, as well as the CAPE distribution corresponding to Fig.9(a). For the present climate, CWRF successfully captures the overall spatial patterns of moisture transport flux and CAPE, except in regions with complex terrain (NW, TB). In eastern CC, the dry bias in extreme precipitation correlates with moisture flux divergence and a lower-than-observed CAPE value. Conversely, for the SC region, the pronounced wet bias aligns with excessive moisture flux convergence and elevated CAPE values.

In the future climate, the regions in northern CC, southern NC, and northern SC, where the CWRF model projects intensified extremes (Fig.6(b)) exhibit two synergistic changes, including enhanced moisture flux convergence and intensified convective instability. These thermodynamic changes provide favorable conditions for extreme rainfall development. Therefore, based on the CWRF projections, the future summer precipitation extreme increases over these regions are associated with several key features, including the south-eastward expansion of the SAH, the southward EAJ displacement, and low-level cyclonic anomalies over the Yellow-Bohai Seas, as well as the regional thermodynamic enhancements.

6 Conclusions

This study establishes a nested MPI-CWRF dynamical downscaling system to simulate summer extreme precipitation in China under present (1980−2014) and future SSP5-8.5 (2016−2050) climates. The MPI-HR (~100 km) global climate model is selected as the driving field due to its superiority in simulating precipitation over China among the CMIP6 models, while the regional climate model CWRF with a finer resolution (~30 km) and optimized physics representations is used for dynamical downscaling. Through comparative analyses against MPI-HR and its high-resolution counterpart (MPI-XR), the study first demonstrates the advantages of the nested MPI-CWRF downscaling model in simulating summer extreme precipitation across China under the present climate, and then identifies the future projections and associated physical mechanisms. The main conclusions are as follows.

1) For the present climate simulations, the nested MPI-CWRF downscaling model shows marked improvements over MPI-HR and MPI-XR in simulating spatial distribution and interannual variability of summer extreme precipitation over China, particularly over the CC region. Specifically, it effectively mitigates the MPI-HR’s overestimation biases of extreme precipitation intensity (PRA, P95, Rx5day) in the south-western China (SW, TB) and rainy days (NRD) in the CC region, while correcting the systematic underestimation biases of PRA, P95, and Rx5day in the south-eastern China (CC, SC).

2) For the future (2016−2050) under the SSP5-8.5 emission scenario, the nested MPI-CWRF model projects amplified increases in extreme precipitation intensity (P95, Rx5day) and interannual variability for two key regions in south-eastern China (CC, SC) compared to MPI-HR. Specifically, over the CC region, NRD is projected to decrease by 4.0%, contrasting with increases in the intensity of PRA, P95, and Rx5day by 4.9%, 8.9%, and 11.8%, respectively. Notably, CWRF’s bias correction for present-climate MPI-HR simulations can be linearly propagated into future projections, and thus enhances confidence in its downscaled projections.

3) The projected future changes in summer mean/extreme precipitation under the nested MPI-CWRF model domonstrate robust physical consistency with key atmospheric circulation, moisture flux convergence, and CAPE. Specifically, the future increase of summer mean precipitation (PRA) in the eastern regions (CC, NC, SC) is associated with the south-eastward expansion of the SAH, the southward shift of the EAJ, and low-level cyclonic anomalies over the Yellow-Bohai Seas. These factors place the eastern China within upper-level divergence zone (left region of jet exit), in conjunction with the low-level convergence, resulting in intensified updrafts and thus stronger precipitation. Future enhancement of the southerly summer monsoon, accompanied by substantial increases in moisture convergence and CAPE, leads to a significant increase in both daily precipitation amounts and the intensity of persistent precipitation events in northern CC, southern NC, and northern SC regions. Hence, consistent model representations of circulation changes, moisture convergence, and convective activity reinforce the credibility of MPI-CWRF projections.

However, there are still some deficiencies in the nested MPI-CWRF downscaling simulations. For instance, model tends to underestimate NRD in the TB region, and overestimates the PRA/Rx5days in the SC region, suggesting deficient rainfall occurrence in TB but excessive convective rainfall over SC. These issues require further improvements in model’s resolution and precipitation physics representations such as shallow and deep convection schemes. As suggested by Liu et al. (2024), shallow cumulus dominates TB in the summer and the representation of shallow convection has a significant impact on the TB’s summer rainfall simulations. Their study shows that using the UW shallow convection scheme (also implemented in our configuration) is prone to reduce and delay the CAPE accumulation, and tends to increase the incoming radiation and decrease the soil moisture, ultimately warming/drying the boundary layer and suppressing precipitation. For SC, the overestimated convective rainfall aligns with the findings of Liang et al. (2019) regarding deep convection scheme deficiencies, particularly in closure assumptions and trigger functions. Therefore, addressing these limitations require further improvements in model resolution and physics parameterizations, with priority given to developing shallow/deep convection schemes.

References

[1] Choi H I, Kumar P, Liang X Z (2007). Three-dimensional volume-averaged soil moisture transport model with a scalable parameterization of subgrid topographic variability.Water Resour Res, 43(4): w04414

[2] ChouMD, SuarezM J (1999). A Solar Radiation Parameterization for Atmospheric Studies. NASA/TM-1999–104606/VOL15

[3] ChouMD, SuarezM J, LiangXZ, Yan M MH (2001). A Thermal Infrared Radiation Parameterization for Atmospheric Studies. NASA/TM-2001–104606/VOL19

[4] Gleckler P J, Taylor K E, Doutriaux C (2008). Performance metrics for climate models.J Geophys Res D Atmospheres, 113: D06104

[5] Gutjahr O, Putrasahan D, Lohmann K, Jungclaus J H, von Storch J S, Brüggemann N, Haak H, Stössel A (2019). Max Planck Institute Earth System Model (MPI-ESM1.2) for the High-Resolution Model Intercomparison Project (HighResMIP).Geosci Model Dev, 12(7): 3241–3281

[6] Haarsma R J, Roberts M J, Vidale P L, Senior C A, Bellucci A, Bao Q, Chang P, Corti S, Fuckar N S, Guemas V, von Hardenberg J, Hazeleger W, Kodama C, Koenigk T, Leung L R, Lu J, Luo J J, Mao J F, Mizielinski M S, Mizuta R, Nobre P, Satoh M, Scoccimarro E, Semmler T, Small J, von Storch J S (2016). High Resolution Model Intercomparison Project (HighResMIP v1.0) for CMIP6.Geosci Model Dev, 9(11): 4185–4208

[7] He B R, Zhai P M (2018). Changes in persistent and non-persistent extreme precipitation in China from 1961 to 2016.Adv Clim Chang Res, 9(3): 177–184

[8] Hersbach H, Bell B, Berrisford P, Hirahara S, Horányi A, Muñoz-Sabater J, Nicolas J, Peubey C, Radu R, Schepers D, Simmons A, Soci C, Abdalla S, Abellan X, Balsamo G, Bechtold P, Biavati G, Bidlot J, Bonavita M, De Chiara G, Dahlgren P, Dee D, Diamantakis M, Dragani R, Flemming J, Forbes R, Fuentes M, Geer A, Haimberger L, Healy S, Hogan R J, Hólm E, Janisková M, Keeley S, Laloyaux P, Lopez P, Lupu C, Radnoti G, de Rosnay P, Rozum I, Vamborg F, Villaume S, Thépaut J N (2020). The ERA5 global reanalysis.Q J R Meteorol Soc, 146(730): 1999–2049

[9] Holtslag A A M, Boville B A (1993). Local versus nonlocal boundary-layer diffusion in a global climate model.J Clim, 6(10): 1825–1842

[10] Huang X, Wang Y H, Ma X F (2024). Simulation of extreme precipitation changes in Central Asia using CMIP6 under different climate scenarios.Theor Appl Climatol, 155(4): 3203–3219

[11] Jiang R S, Sun L, Sun C, Liang X Z (2021). CWRF downscaling and understanding of China precipitation projections.Clim Dyn, 57(3−4): 1079–1096

[12] Jiang X W, Li Y Q, Yang S, Wu R G (2011). Interannual and interdecadal variations of the South Asian and western Pacific subtropical highs and their relationships with Asian-Pacific summer climate.Meteorol Atmos Phys, 113: 171–180

[13] Karl T R, Nicholls N, Ghazi A (1999). CLIVAR/GCOS/WMO Workshop on indices and indicators for climate extremes - Workshop summary.Clim Change, 42(1): 3–7

[14] Kong X H, Wang A H, Bi X Q, Wang D (2019). Assessment of temperature extremes in China using RegCM4 and WRF.Adv Atmos Sci, 36(4): 363–377

[15] Li Q Q, Wang T, Wang F, Liang X Z, Zhao C B, Dong L L, Zhao C Y, Xie B (2021). Dynamical downscaling simulation of the East Asian summer monsoon in a regional climate-weather research and forecasting model.Int J Climatol, 41(S1): E1700–E1716

[16] Liang X Z, Sun C, Zheng X H, Dai Y J, Xu M, Choi H I, Ling T J, Qiao F X, Kong X H, Bi X Q, Song L C, Wang F (2019). CWRF performance at downscaling China climate characteristics.Clim Dyn, 52(3−4): 2159–2184

[17] Liang X Z, Wang W C (1998). Associations between China monsoon rainfall and tropospheric jets.Q J R Meteorol Soc, 124(552): 2597–2623

[18] Liang X Z, Xu M, Yuan X, Ling T J, Choi H I, Zhang F, Chen L G, Liu S Y, Su S J, Qiao F X, He Y X, Wang J X L, Kunkel K E, Gao W, Joseph E E, Morris V, Yu T W, Dudhia J, Michalakes J (2012). Regional climate-weather research and forecasting model.Bull Am Meteorol Soc, 93(9): 1363–1387

[19] Liu G, Liang X (2017). Progress of the climate extension of weather research and forecast (CWRF) model application in China.Adv Earth Sci, 32: 781–787

[20] Liu J, Yang K, Wang J, Zhou X, Jiang Y, Shao C, Lu H, Yao X, Sun J, Shi J (2024). Impacts of a shallow convection scheme on kilometer-scale atmospheric simulations over the Tibetan Plateau.Clim Dyn, 62(8): 8019–8034

[21] Liu Q, Qiao F, Zhu Y, Liang X, Liu Y, Zhang H, Wang R (2021). Evaluation of the spatio-temporal variations of extreme temperature simulations in China based on the regional climate-weather research and forecasting model.Climat Environ Res, 26: 333–350

[22] Liu S Y, Liang X Z, Gao W, Zhang H (2008). Application of climate - weather research and forecasting model (CWRF) in China: domain optimization.Chin J Atmos Sci, 32: 457–468

[23] Lu S, Hu Z, Wang B, Qin P, Wang L (2020). Spatio-temporal patterns of extreme precipitation events over China in recent 56 Years.Plateau Meteorol, 39: 683–693

[24] Müller W A, Jungclaus J H, Mauritsen T, Baehr J, Bittner M, Budich R, Bunzel F, Esch M, Ghosh R, Haak H, Ilyina T, Kleine T, Kornblueh L, Li H, Modali K, Notz D, Pohlmann H, Roeckner E, Stemmler I, Tian F, Marotzke J (2018). A higher-resolution version of the Max Planck Institute Earth System Model (MPI-ESM1.2-HR).J Adv Model Earth Syst, 10(7): 1383–1413

[25] Park S, Bretherton C S (2009). The University of Washington shallow convection and moist turbulence schemes and their impact on climate simulations with the community atmosphere model.J Clim, 22(12): 3449–3469

[26] Qiao F X, Liang X Z (2016). Effects of cumulus parameterization closures on simulations of summer precipitation over the United States coastal oceans.J Adv Model Earth Syst, 8(2): 764–785

[27] Qiao F X, Liang X Z (2017). Effects of cumulus parameterization closures on simulations of summer precipitation over the continental United States.Clim Dyn, 49(1−2): 225–247

[28] Scherrer S C (2011). Present-day interannual variability of surface climate in CMIP3 models and its relation to future warming.Int J Climatol, 31(10): 1518–1529

[29] Sun J, Zhang F (2017). Daily extreme precipitation and trends over China.Sci China Earth Sci, 60(12): 2190–2203

[30] Tao W K, Starr D, Hou A, Newman P, Sud Y (2003). A cumulus parameterization workshop.Bull Am Meteorol Soc, 84: 1055–1062

[31] Wang X Y, Tang J P, Niu X R, Wang S Y (2016). An assessment of precipitation and surface air temperature over China by regional climate models.Front Earth Sci, 10(4): 644–661

[32] Wang X, Gu H, Dai B, Zhang H, Yu Z (2024). Simulation of climate characteristics in the Qinghai-Tibet Plateau by regional climate models at different horizontal resolutions.Arid Zone Res, 41: 363–374

[33] Wang Y, Li H, Wang H, Sun B, Chen H (2021). Evaluation of CMIP6 model simulations of extreme precipitation in China and comparison with CMIP5.Acta Meteorol Sin, 79: 369–386

[34] Wei P, Dong G, Shi J, Zhang B (2019). Dynamical downscaling simulation and projection of extreme precipitation over East China.Climat Environ Res, 24: 86–104

[35] Wei W, Zhang R H, Wen M, Kim B J, Nam J C (2015). Interannual variation of the South Asian High and its relation with Indian and East Asian summer monsoon rainfall.J Clim, 28(7): 2623–2634

[36] Wu J, Gao X J (2013). A gridded daily observation dataset over China region and comparison with the other datasets.Chin J Geophys, 56: 1102–1111

[37] Xu H R, Liang X Z, Xue Y K (2024). Regional climate modeling to understand Tibetan heating remote impacts on East China precipitation.Clim Dyn, 62(4): 2683–2701

[38] Yang L, Zhao J, Mao M, Li Y, Liu G (2024). Evaluation of the simulation capability of CMIP6 model for summer precipitation in the northern monsoon region of China.Climat Environ Res, 29: 549–562

[39] Yang X L, Zhou B T, Xu Y, Han Z Y (2021). CMIP6 evaluation and projection of temperature and precipitation over China.Adv Atmos Sci, 38(5): 817–830

[40] Yuan X, Liang X Z (2011). Improving cold season precipitation prediction by the nested CWRF-CFS system.Geophys Res Lett, 38(2): L02706

[41] Zhang H, Liang X Z, Dai Y J, Song L C, Li Q Q, Wang F, Zhang S L (2024). CWRF downscaling with improved land surface initialization enhances spring-summer seasonal climate prediction skill in China.J Clim, 37(17): 4437–4459

[42] Zhang L, Chen X, Xin X (2019). Short commentary on CMIP6 scenario model intercomparison project (ScenarioMIP).Climat Change Res, 15: 519–525

[43] Zhang S Y, Wang M H, Wang L N, Liang X Z, Sun C, Li Q Q (2023). Sensitivity of the simulation of extreme precipitation events in China to different cumulus parameterization schemes and the underlying mechanisms.Atmos Res, 285: 106636

[44] Zhao Y, Qiao F X, Liang X Z, Yu J H (2024). Understanding and improving Yangtze River Basin summer precipitation prediction using an optimal multi-Physics ensemble.Front Earth Sci, 18(1): 256–277

[45] Zhou T, Chen Z, Chen X, Zuo M, Jiang J, Hu S (2021). Interpreting IPCC AR6: future global climate based on projection under scenarios and on near-term information.Climat Change Res, 17: 652–663

RIGHTS & PERMISSIONS

Higher Education Press