Global climate model (GCM) outputs from Coupled Model Inter-comparison Project Phase 5 (CMIP5) were widely used to investigate climate change impacts last 10 years. It is important to know whether Coupled Model Inter-comparison Project Phase 6 (CMIP6) is more reliable than CMIP5. Number of studies compared GCMs from two CMIPs in simulating climate variables, but they are not in the field of hydrology for large quantities of watersheds. The objective of this study is to compare CMIP5 and CMIP6 climate model projections in projecting hydrological changes between future (2071–2100) and historical (1976–2005) periods and inter-model variability of hydrological impacts for the future period over 343 catchments in China's mainland. The results show that the GCMs in CMIP6 show more increase in daily mean temperature and mean annual precipitation. However, GCMs in CMIP6 and CMIP5 show similar increases in mean and peak streamflow. Moreover, GCMs in CMIP6 show less inter-model variability for streamflow in southern and northeastern China, but more in central China, which is consistent to that for precipitation after bias correction. Overall, this comparison provides the consistency of future change and uncertainty in predicted streamflow with climate simulations, which bring confidence for hydrological impact studies using CMIP6.

  • GCMs in CMIP6 under SSP5-85 show more increase in temperature and precipitation than those in CMIP5 under RCP8.5 in China.

  • GCMs in CMIP6 and CMIP5 show similar increases in mean and peak streamflow.

  • GCMs in CMIP6 show less inter-model variability for mean and peak flow in southern and northeastern China, but more in central China.

Graphical Abstract

Graphical Abstract
Graphical Abstract

The Fifth and Sixth Assessment Report (AR5 and AR6) of the Intergovernmental Panel on Climate Change (IPCC) (Stocker et al. 2013; Pörtner et al. 2022) both stated that the global temperature has a significant increase during the last century and the global warming increases the likelihood of severe, pervasive, and irreversible impacts, especially for the terrestrial hydrology. This is because the rising of air temperature increases the water holding capacity in the atmosphere and changes the spatial distribution of precipitation on the earth. In other words, the global water cycle process would be affected by the increase of air temperature (Syukuro 2019), which may result in more hydrological extremes such as waterlog and drought (Greene et al. 2009; Hirabayashi et al. 2013; AghaKouchak et al. 2014; Huang et al. 2017; Lin et al. 2020).

The quantification of climate change impacts on hydrology is usually achieved by driving hydrological models using global climate model (GCM) simulations with downscaling or bias correction techniques (Velázquez et al. 2013; Papadimitriou et al. 2016; Paulo et al. 2020). Various GCMs have been developed by different climate centers in many countries. However, due to different parameterizations of some unknown physical processes (Choi et al. 2017), large variability has been found related to climate projections in hydrological impact studies, especially for specific regions (Jiang et al. 2007; Her et al. 2019; Lehner et al. 2019; Zhou et al. 2020). A total of 41 GCMs in Coupled Model Comparison Project Phase 6 were evaluated in simulating monthly precipitation and found that the selection of appropriate models was a critical factor for climate change impact studies (Li et al. 2022). The multiple climate model ensemble (MME) is usually used to represent the variability related to climate models (Rasmussen et al. 2016; Ishizaki et al. 2017; Fathalli et al. 2019; Christiansen 2020).

Since the AR5 of IPCC, GCMs from Coupled Model Inter-comparison Project Phase 5 (CMIP5) were commonly used to provide climate projections for hydrological impact studies. For example, Papadimitriou et al. (2016) used five GCM simulations from CMIP5 to study the high-end climate change impact on European runoff and low flows and found an intensification of the water cycle at higher levels of warming. Asadieh & Krakauer (2017) investigated the impact of 21st century anthropogenic forcing under Representative Concentration Pathway 2.6 (RCP2.6) and RCP8.5 on streamflow extremes and found increased flow in the Arctic Ocean and reduced flow in subtropical arid areas under both forcing scenarios. Pokhrel et al. (2021) used an ensemble of hydrological simulations driven by four GCM simulations from CMIP5 and found that climate change could reduce terrestrial water storage in many regions globally. Yin et al. (2021) predicted the streamflow with 6 GCM simulations under RCP8.5 scenario for 151 catchments over the mainland of China and found that earlier flood timing and larger flood quantiles is likely to occur for most catchments in the future period than in the historical period.

More recently, the GCMs with higher spatial resolution and improvements in physical processes participating in CMIP6 have been available, which use the new shared socioeconomic pathway (SSP)/RCP-based emission scenarios for future simulations (Eyring et al. 2016; O'Neill et al. 2016). The variables of precipitation and temperature from these GCMs were used instead of those in CMIP5 archive for hydrological impact studies. For example, Cook et al. (2020) investigated changes in drought with the variables of precipitation, soil moisture, and runoff projected by 13 GCM simulations under 4 SSPs in CMIP6 and found that the drying in the mean state and risk of the historically most extreme events would increase. Bian et al. (2021) investigated variations in precipitation, runoff, and flood by using four GCMs in CMIP6 and found that the probability of floods would increase in the future over the upper Huai River basin. The evaluation of simulated runoff from 12 GCMs from CMIP6, 6 global hydrological models from the inter-sectoral impact model inter-comparison project and 3 land surface models from the global land data assimilation system show that projections of runoff changes based on these global outputs contain great uncertainties and should be interpreted with caution (Hou et al. 2022).

Since the climate models in CMIP6 provides a new opportunity to obtain a more credible understanding of climate change, it is necessary to compare the performance of GCM in CMIP6 and CMIP5 in simulating climate variables, especially for precipitation and temperature which are the most important variables for hydrological impact studies. Previous studies have focused on the performance of climate simulations, which compared the difference in climate simulations between CMIP5 and CMIP6 GCMs. For example, Lehner et al. (2020) found that simulated temperature and precipitation in CMIP6 showed greater model response variability than those in CMIP5, using 28 GCMs under RCP2.6, RCP4.5, and RCP8.5 from CMIP5 and 21 GCMs under SSP1-26, SSP2-45, SSP3-70, and SSP5-85 from CMIP6. Moreover, Zhang (2021) compared the uncertainties in projections of precipitation and temperature extremes between CMIP5 and CMIP6 archives at the global scale using 24 GCMs from each archive and found that more GCMs are needed to ensure the robustness of climate projections in CMIP6 than CMIP5. However, Fan et al. (2020) found that some individual CMIP6 models performed better than CMIP5 model in simulating summer days, tropical nights, cold spell duration, and diurnal temperature range in spatial pattern scores with 24 GCMs from CMIP6 and 28 GCMs from CMIP5.

GCMs in both CMIP5 and CMIP6 have been used to project climate change and its impact on hydrology in China. For example, the temporal distribution of precipitation was projected to become more uneven in the future than the historical period when using 20 Atmosphere-Ocean General Circulation Models (AOGCMs) in CMIP6 under four SSPs (Zhu et al. 2021). A larger flood is likely to occur in the future period than in the historical period in the mainland of China under the simulations of hydrological models when using GCMs from CMIP5 or CMIP6 (Bian et al. 2021; Yin et al. 2021). In addition, the duration and intensity of drought were projected to increase in some regions of China when using eight GCMs in CMIP6 under three SSPs (Yin et al. 2022).

Even though there were many studies comparing the performance of GCMs between CMIP6 and CMIP5 in simulating climate variables, the comparison of GCM simulations between CMIP6 and CMIP5 in future hydrological impacts is not found in the literatures, especially for the region of China. Given the transformation from climate to hydrology is highly nonlinear, this comparison is necessary. Thus, the objective of this study is to compare GCMs simulations between CMIP5 and CMIP6 in projecting future streamflow changes and their variability related to the choice of climate models. Specifically, changes and inter-model variability of mean flow and annual peak streamflow simulated by 2 hydrological models using bias corrected climate simulations (2 bias correction methods) from CMIP5 and CMIP6 as inputs are compared over 343 catchments in the mainland of China. The rest of the paper is organized as follows. Section 2 presents a brief introduction to the study area and datasets, and Section 3 presents the methodology, including bias correction methods, hydrological models, and data analysis methods. Section 4 presents the results, followed by the discussion and conclusion in Section 5.

Observations

This study used a gridded meteorological dataset (0.5° × 0.5°) over China for the period of 1961–2016 to represent observed data. This dataset contains three climate variables including daily precipitation, and daily minimum and maximum temperatures, which were downloaded from the China Meteorological Data Sharing Service System (http://www.cma.gov.cn/). The gridded dataset was generated from 2472 in situ observation gauge stations across China by the thin plate spline interpolation method and GTOPO30 (Global 30 Arc-Second Elevation) data sampling, which has been commonly used in many hydro-climatological studies in China (Xiao et al. 2013; Zhang et al. 2015; Fan et al. 2019; Zheng et al. 2020).

The daily streamflow series over 343 catchments in China were also used (Figure 1). These catchments with a wide range of climatic conditions and hydrological regimes span over all the 9 major river basins in China. The size of the catchment ranges from 584 to 1,462,621 km2. The streamflow dataset covers the 1961–2016 period with the length between 22 and 52 years. The missing values were not included when calibrating hydrological models.
Figure 1

The selected catchments and their areas in km2 (The points in figure represent the watershed outlets).

Figure 1

The selected catchments and their areas in km2 (The points in figure represent the watershed outlets).

Close modal

Climate model

For the climate simulations, daily precipitation and temperatures (maximum and minimum temperatures) derived from the CMIP5 and CMIP6 archives (21 GCMs from CMIP5 and 21 GCMs from CMIP6) were selected (Table 1). The CMIP5 and CMIP6 outputs were downloaded from the Earth System Grid Federation. Even though they are not exactly from the same climate centers, most of them are from the same climate centers with advanced versions in CMIP6. These 42 GCMs provide the climate variables under the historical forcing for the reference period (1976–2005), RCP8.5 (CMIP5) and SSP5-85 (CMIP6) forcing for the 2071–2100 period. The use of a large number of GCM simulations is to adequately cover the inter-model variability.

Table 1

Information of the used CMIP5 and CMIP6 climate models in this study

No.CMIP5
CMIP6
GCMsResolution (lon. × lat.)Institution/Country (Region)GCMsResolution (lon. × lat.)Institution/Country (Region)
BCC-CSM1.1 2.8° × 2.8° BCC/CN BCC-CSM2-MR 1.125° × 1.1213° BCC/CN 
BNU-ESM 2.8125° × 2.8125° BNU/CN FGOALS-g3 2° × 2.25° CAS/CN 
CanESM2 2.8° × 2.8° CCCma/CAN CanESM5 2.8125° × 2.7893° CCCma/CAN 
CNRM-CM5 1.4° × 1.4° CNRM-CERFACS/FR CNRM-CM6-1 1.4063° × 1.4004° CNRM-CERFACS/FR 
CSIRO-Mk3.6.0 1.8° × 1.8° CSIRO-QCCCE/AUS CNRM-ESM2-1 
EC-EARTH 1.125° × 1.125° EC-Earth-Consortium/EU ACCESS-ESM1-5 1.8750° × 1.25° CSIRO/AUS 
IPSL-CM5A-LR 3.75° × 1.8° IPSL/FR ACCESS-CM2 CSIRO-ARCCSS/AUS 
IPSL-CM5A-MR 2.5° × 1.25° EC-Earth3 0.7031° × 0.7017° EC-Earth-Consortium/EU 
FGOALS-g2 2.8125° × 3° CAS/CN EC-Earth3-Veg 
10 MIROC-ESM 2.8° × 2.8° MIROC/JPN INM-CM4-8 2° × 1.5° INM/R.F 
11 MIROC-ESM-CHEM  INM-CM5-0 
12 MPI-ESM-LR 1.8750° × 1.8750° MPI-M/DE IPSL-CM6A-LR 2.5° × 1.2676° IPSL/FR 
13 MPI-ESM-MR MIROC6 1.4063° × 1.4004° MIROC/JPN 
14 MRI-CGCM3 1.1° × 1.1° MRI/JPN MIROC-ES2L 2.8125° × 2.7893° 
15 CCSM4 1.25° × 0.9375° NCAR/U.S.A HadGEM3-GC31-LL 1.8750° × 1.25° MOHC/UK 
16 NorESM1-M 2.5° × 1.8750° NCC/NOR UKESM1-0-LL 
17 HadGEM2-AO 1.875° × 1.2414° NIMR-KMA/KOR MPI-ESM1-2-HR 0.9375° × 0.9349° MPI-M/DE 
18 GFDL-CM3 2.5° × 2.0° NOAA-GFDL/U.S.SA MPI-ESM1-2-LR 1.875° × 1.8647° 
19 GFDL-ESM2G MRI-ESM2-0 1.1250° × 1.1213° MRI/JPN 
20 GFDL-ESM2M NorESM2-MM 1.25° × 0.9424° NCC/NOR 
21 CESM1-CAM5 1.25° × 0.9375° NSF-DOE-NCAR/U.S.A GFDL-ESM4 1.25° × 1° NOAA-GFDL/U.S.A 
No.CMIP5
CMIP6
GCMsResolution (lon. × lat.)Institution/Country (Region)GCMsResolution (lon. × lat.)Institution/Country (Region)
BCC-CSM1.1 2.8° × 2.8° BCC/CN BCC-CSM2-MR 1.125° × 1.1213° BCC/CN 
BNU-ESM 2.8125° × 2.8125° BNU/CN FGOALS-g3 2° × 2.25° CAS/CN 
CanESM2 2.8° × 2.8° CCCma/CAN CanESM5 2.8125° × 2.7893° CCCma/CAN 
CNRM-CM5 1.4° × 1.4° CNRM-CERFACS/FR CNRM-CM6-1 1.4063° × 1.4004° CNRM-CERFACS/FR 
CSIRO-Mk3.6.0 1.8° × 1.8° CSIRO-QCCCE/AUS CNRM-ESM2-1 
EC-EARTH 1.125° × 1.125° EC-Earth-Consortium/EU ACCESS-ESM1-5 1.8750° × 1.25° CSIRO/AUS 
IPSL-CM5A-LR 3.75° × 1.8° IPSL/FR ACCESS-CM2 CSIRO-ARCCSS/AUS 
IPSL-CM5A-MR 2.5° × 1.25° EC-Earth3 0.7031° × 0.7017° EC-Earth-Consortium/EU 
FGOALS-g2 2.8125° × 3° CAS/CN EC-Earth3-Veg 
10 MIROC-ESM 2.8° × 2.8° MIROC/JPN INM-CM4-8 2° × 1.5° INM/R.F 
11 MIROC-ESM-CHEM  INM-CM5-0 
12 MPI-ESM-LR 1.8750° × 1.8750° MPI-M/DE IPSL-CM6A-LR 2.5° × 1.2676° IPSL/FR 
13 MPI-ESM-MR MIROC6 1.4063° × 1.4004° MIROC/JPN 
14 MRI-CGCM3 1.1° × 1.1° MRI/JPN MIROC-ES2L 2.8125° × 2.7893° 
15 CCSM4 1.25° × 0.9375° NCAR/U.S.A HadGEM3-GC31-LL 1.8750° × 1.25° MOHC/UK 
16 NorESM1-M 2.5° × 1.8750° NCC/NOR UKESM1-0-LL 
17 HadGEM2-AO 1.875° × 1.2414° NIMR-KMA/KOR MPI-ESM1-2-HR 0.9375° × 0.9349° MPI-M/DE 
18 GFDL-CM3 2.5° × 2.0° NOAA-GFDL/U.S.SA MPI-ESM1-2-LR 1.875° × 1.8647° 
19 GFDL-ESM2G MRI-ESM2-0 1.1250° × 1.1213° MRI/JPN 
20 GFDL-ESM2M NorESM2-MM 1.25° × 0.9424° NCC/NOR 
21 CESM1-CAM5 1.25° × 0.9375° NSF-DOE-NCAR/U.S.A GFDL-ESM4 1.25° × 1° NOAA-GFDL/U.S.A 

Bias correction methods

Since GCM simulations are too biased to be directly used as inputs in hydrological models for impact studies, two bias correction methods were applied to climate model simulations before driving hydrological models. These two methods include a univariate method (daily bias correction method, DBC) and a multivariate method (two-stage quantile mapping, TSQM).

DBC (Chen et al. 2013) and TSQM (Guo et al. 2019) are both quantile mapping-based methods. They correct the distribution of daily precipitation and temperature with two steps. In the first step, the frequency of simulated precipitation occurrence is corrected by determining a precipitation threshold ensuring that the frequency of simulated precipitation is equal to that of the observed one for the reference period. In the second step, the wet-day precipitation amounts and temperature are corrected by a quantile mapping method using 100 percentiles from the 1st to 100th to represent their distribution. Both methods first correct precipitation and temperature individually, while TSQM further introduces inter-variable correlations between daily precipitation TSQM shows better performance in simulating the mean temperature for wet and dry months than DBC (Guo et al. 2019).

Hydrological models

Two lumped hydrological models (HMETS and XAJ) with different structures and a number of parameters were employed for runoff simulations. HMETS (Hydrological Model of École de technologie supérieure) is a 21-parameter conceptual hydrological model, including four components that are snow accumulation and snowmelt module, evapotranspiration module, vertical water balance, and horizontal transportation (Martel et al. 2017). These four modules involve two connected reservoirs for representing unsaturated and saturated zones and captures basic hydrological processes. The snow accumulation and snowmelt module not only simulates snowmelt and refreezing processes, but also considers finer processes with 10 parameters in the snowmelt module. The actual evapotranspiration is estimated by a fraction (a parameter to be calibrated) of potential evapotranspiration. The vertical water balance module divides the effective rainfall into four sources of runoff (surface runoff, delayed runoff, hypodermic flow, and groundwater flow) based on two linear reservoirs with six free parameters. The first two runoff components which have four free parameters are routed through two gamma distributions in the horizontal transport.

XAJ (Xinanjiang) is a conceptual rainfall-runoff model with 15 parameters (Zhao Ren-Jun et al. 1980), which divides the watershed area into pervious and impervious areas and considers a 3-layer soil moisture store to simulate actual evapotranspiration. This model is physically based on the theory that runoff generation occurs since the repletion of storage, which means that runoff is not generated until the soil water deficit is replenished, that is impervious areas. All of the effective rainfall transfers to surface runoff in impervious areas. However, effective rainfall is separated into the surface runoff, interflow, and groundwater runoff in the previous area. Surface runoff routes to the outlet of catchments through the unit hydrograph. This model has been widely employed for streamflow simulation and flood forecasting, especially over the humid and semi-humid climate regions in China (Bárdossy et al. 2016; Li et al. 2017; Zhuo & Han 2017; Burek et al. 2020). In addition, the CemaNeige module (Valéry et al. 2014), which is used to simulate the snow accumulation and snowmelt processes in XAJ, has two parameters to be calibrated, divides precipitation into rainfall and snowfall, and calculates the snowmelt using a degree-day method.

The hydrological model parameters were calibrated using observed daily precipitation and potential evaporation, along with the daily streamflow time series. The potential evaporation was calculated by the Oudin equation (Oudin et al. 2005) based on temperatures. The observed data used for the calibration cover a length of 22–52 years depending on the length of the observed streamflow time series, and the odd years were used for calibration and the even years were used for validation. The optimal values of the free parameters were obtained by the algorithm of evolution strategy with covariance matrix adaptation (CMA-ES) (Merelo-Guervós & Castillo-Valdivieso 2004), which outperformed many other algorithms in calibrating hydrological models and it is especially efficient in handling higher-dimension parameter space (Arsenault et al. 2014). The Kling–Gupta efficiency (KGE) that considers the mean, variability, and dynamics of model errors was selected as the objective function for model calibration.

Data analysis

GCM simulations in CMIP5 and CMIP6 were compared in terms of multiple hydro-meteorological indices. Firstly, the mean value of daily mean temperature (tas, the mean value of daily maximum and minimum temperature) and mean annual precipitation (prcptot) were calculated using climate time series, and the mean daily streamflow (Qmean) and annual maximum daily streamflow (Qmax) were then calculated for hydrological simulations. The mean values were calculated for both temperature and precipitation. Since the precipitation at the daily time scale consists of nonzero and zero values, the annual mean is used. Secondly, the change of hydro-meteorological indices between future and historical periods was calculated. Specifically, the absolute change (future-historical) was calculated for the temperature index, and relative change [(future-historical)/historical] was calculated for precipitation and streamflow indices. In addition, the inter-model variability represented by the standard deviation (std) of hydro-meteorological indices across multiple GCM simulations was calculated for all indices to reflect the variability related to the choice of a climate model simulation. Since the results might slightly differ when using different methods, due to uncertainty related to the choice of bias correction methods and hydrological models, the ensembles mean was used to better represent the results. To obtain robust results, the ensemble means across two bias correction methods and two hydrological models were computed for all indices when calculating their change and inter-model variability.

Change and inter-model variability of climate simulations

Figure 2 presents the changes in daily mean temperature and mean annual precipitation for the future period (2071–2100) relative to the reference period (1976–2005) projected by climate models in CMIP5 and CMIP6 over all 343 catchments in terms of the mean. Generally, an overall increasing trend is observed for both temperature and precipitation over all catchments. The increase of daily mean temperature ranges between 4.2 and 6.4 °C (between 3.7 and 5.8 °C) for CMIP6 (CMIP5) with the average value of 5.3 °C (4.7 °C) across all 343 catchments. Moreover, GCMs in CMIP6 project more increase in daily mean temperature than those in CMIP5 with the mean difference ranging between 0.18 and 0.84 °C (Figure 2(e)). In other words, GCMs in CMIP6 project warmer temperatures than those in CMIP5, especially for northeastern China.
Figure 2

Change of daily mean temperature (tas) (a,c) and mean annual precipitation (prcptot) (b,d) for raw CMIP5 and CMIP6 multi-model means between the 2071–2100 and the 1976–2005 periods. Differences between CMIP6 (top row) and CMIP5 (center row) multi-model means are shown on the bottom row (e,f).

Figure 2

Change of daily mean temperature (tas) (a,c) and mean annual precipitation (prcptot) (b,d) for raw CMIP5 and CMIP6 multi-model means between the 2071–2100 and the 1976–2005 periods. Differences between CMIP6 (top row) and CMIP5 (center row) multi-model means are shown on the bottom row (e,f).

Close modal

The increase in mean annual precipitation ranges between 6.3 and 34.4% for CMIP6 and ranges between 4.1 and 20.9% for CMIP5 with the mean value of 16.5 and 11.4% across all watersheds, respectively. Similarly, GCMs in CMIP6 project larger increases in mean annual precipitation by up to 16% than those in CMIP5 for most watersheds, especially for northern China (Figure 2(f)). Only in southwestern China, GCMs in CMIP5 project a slightly larger increase in mean annual precipitation than those in CMIP6.

Figure 3 presents the inter-model variability of mean daily temperature simulated by GCMs in CMIP5 and CMIP6 over 343 catchments for both historical and future periods. Generally, the inter-model variability of mean daily temperature shows a large increase from the historical period to the future period for CMIP5 (Figure 3(c) and 3(d)). However, this is not the case for CMIP6, with exception of that in northeastern China (Figure 3(a) and 3(b)). Specifically, the average inter-model standard deviations are around 1.5 °C (historical period) and 1.6 °C (future period) for CMIP6, while the values are around 1.3 °C (historical period) and 1.9 °C (future period) for CMIP5. In the historical period, inter-model standard deviations of daily mean temperature simulated by GCMs in CMIP6 are larger than those simulated by GCMs in CMIP5, except for northeastern China, with a mean difference of 0.17 °C across all watersheds. However, an opposite pattern is found for the future period with a mean difference of −0.3 °C. These results indicate that GCMs in CMIP5 show larger variability than those in CMIP6 in simulating temperatures.
Figure 3

Inter-model standard deviation of daily mean temperature (tas) simulated by CMIP5 and CMIP6 climate models for 1976–2005 (his) (a,c) and 2071–2100 (fut) (b,d) periods.

Figure 3

Inter-model standard deviation of daily mean temperature (tas) simulated by CMIP5 and CMIP6 climate models for 1976–2005 (his) (a,c) and 2071–2100 (fut) (b,d) periods.

Close modal
Figure 4 presents the inter-model variability of mean annual precipitation simulated by GCMs in CMIP5 and CMIP6 over all 343 catchments for both historical and future periods. Generally, the inter-model variability of mean annual precipitation is projected to increase from the historical to the future periods for both CMIP5 and CMIP6. Specifically, in CMIP6, the inter-model standard deviations range between 76 and 473.7 mm for the historical period with a mean value of 266.6 mm and range between 90.8 and 575.1 mm for the future period with a mean value of 316.2 mm. In CMIP5, these values range between 60.7 and 354.8 mm with a mean value of 201.4 mm for the historical period and range between 64 and 428.1 mm with a mean value of 245.9 mm for the future period.
Figure 4

As Figure 3, but for mean annual precipitation (prcptot).

Figure 4

As Figure 3, but for mean annual precipitation (prcptot).

Close modal

For both historical and future periods, inter-model standard deviations of mean annual precipitation simulated by GCMs in CMIP6 is larger than those simulated by GCMs in CMIP5 for southern and northern China (Figure 4(e) and 4(f)). However, for central China, the inter-model standard deviations are larger for CMIP5 than CMIP6 with a mean difference of −13.8% for the historical period and −16.8% for the future period. The variation of mean annual precipitation values is shown in Appendix A in the Supplementary Material.

Performance of bias correction methods

Figure 5 illustrates the bias of daily mean temperature and mean annual precipitation simulated by GCMs in CMIP5 and CMIP6 before bias correction for the reference period over the 343 catchments. For most catchments, the daily mean temperature is underestimated by GCMs in both CMIP6 and CMIP5. However, there is a large spatial variability, especially for northwest China where the temperature is overestimated. Specifically, the daily mean temperature in CMIP6 is underestimated for 74% of watersheds and that in CMIP5 is underestimated for 63% of watersheds. For precipitation, GCMs in both CMIP6 and CMIP5 overestimated mean annual precipitation for most watersheds. This is especially true for CMIP6, as the mean annual precipitation is overestimated for 95% of watersheds, while the proportion is 70% for CMIP5. For most watersheds, GCMs in CMIP6 than CMIP5 show similar bias in simulating temperature, especially for central and northern China. However, for southern and northeastern China, GCMs in CMIP6 are more biased than those in CMIP5 in simulating daily mean temperature (Figure 5(e)). The mean annual precipitation simulated by GCMs in CMIP6 is more biased than that simulated by GCMs in CMIP5 for southern China, but an opposite pattern is observed for central China (Figure 5(f)).
Figure 5

Raw CMIP5 and CMIP6 multi-model mean of absolute bias (a,c) of daily mean temperature (tas) and relative bias (b,d) of mean annual precipitation (prcptot) and for the 1976–2005 period.

Figure 5

Raw CMIP5 and CMIP6 multi-model mean of absolute bias (a,c) of daily mean temperature (tas) and relative bias (b,d) of mean annual precipitation (prcptot) and for the 1976–2005 period.

Close modal
Figure 6 illustrates the bias of daily mean temperature and mean annual precipitation simulated by GCMs in CMIP5 and CMIP6 after bias correction over all the 343 catchments for the 1976–2005 period. Generally, the bias correction methods perform reasonably well for all watersheds. The bias is close to 0 °C for temperature (Figure 6(a) and 6(c)) and is less than 5% for precipitation (Figure 6(b) and 6(d)). The larger remaining bias in the northwest is because the value of annual precipitation is very small for this region, the calculation of relative bias seems large, even though the absolute bias is small. Therefore, the bias corrected climate simulations can be used for hydrological modeling.
Figure 6

As Figure 5, but for the GCMs after bias correction.

Figure 6

As Figure 5, but for the GCMs after bias correction.

Close modal

Hydrological models performance

Figure 7 illustrates the KGE values of two hydrological models for both calibration and validation periods over the 343 catchments. Generally, both hydrological models are well calibrated with the median KGE value of 0.85 for both HMETS and XAJ. In the validation period, the median KGE values are 0.82 and 0.81 for HMETS and XAJ, respectively. Approximate 80% catchments show KGE values being larger than 0.7 for both models in the calibration period, with a small drop in the validation period. Specifically, in the validation period, 77 and 78% catchments show KGE values larger than 0.7 for HMETS and XAJ, respectively. XAJ performs slightly better than HMETS model over the majority of catchments.
Figure 7

KGE values of two hydrological models for the calibration period (a,b) and the validation period (c,d).

Figure 7

KGE values of two hydrological models for the calibration period (a,b) and the validation period (c,d).

Close modal

Change and inter-model variability of predicted streamflow

Figure 8 presents changes in the streamflow indicators (i.e. mean and peak flows) simulated by hydrological models driven by bias corrected climate model simulations in CMIP5 and CMIP6 between the future (2071–2100) and reference (1976–2005) periods overall 343 catchments in terms of the MME mean. An overall increase is found for both mean and peak flow for 95% catchments, especially for CMIP5 with a proportion up to 99%. The mean change in mean flow is 12.2% for CMIP6 and 13.1% for CMIP5 across all catchments. Moreover, the mean flow in CMIP6 shows less increase than that in CMIP5 for southern China, while an opposite pattern is observed for northern China. A similar spatial pattern is also found for the difference in simulated peak flow between CMIP6 and CMIP5. However, the difference between CMIP5 and CMIP6 is minor for both mean and peak flow. In other words, GCM-simulated temperature and precipitation after bias correction project a similar increase in mean and peak flows between CMIP5 and CMIP6.
Figure 8

Changes of mean streamflow (Qmean) values (a,c) and peak streamflow (Qmax) values (b,d) as simulated by the CMIP6- and CMIP5-driven hydrological ensembles from the 1976–2005 period to the 2071–2100 period.

Figure 8

Changes of mean streamflow (Qmean) values (a,c) and peak streamflow (Qmax) values (b,d) as simulated by the CMIP6- and CMIP5-driven hydrological ensembles from the 1976–2005 period to the 2071–2100 period.

Close modal
Figure 9 shows the inter-model variability of the streamflow indicators (i.e. mean and peak flows) simulated by hydrological models driven by bias corrected GCM simulations in CMIP5 and CMIP6 for the future period. Generally, the inter-model variability of streamflow indicators shows a similar spatial pattern between CMIP5 and CMIP6. However, the magnitudes of inter-model variability are not the same between these two archives. Specifically, around 78% catchments show greater inter-model standard deviation in CMIP5 than that in CMIP6 for mean flow, while the proportion is 66% for peak flow. These catchments are mostly located in southern China and northeast China, implying that GCM simulations in CMIP6 are more reliable than those in CMIP5 for hydrological projections.
Figure 9

Standard deviation of mean streamflow (Qmean) values (a,c) and peak streamflow (Qmax) values (b,d) as simulated by the CMIP6- and CMIP5-driven hydrological ensembles over the 2071–2100 period.

Figure 9

Standard deviation of mean streamflow (Qmean) values (a,c) and peak streamflow (Qmax) values (b,d) as simulated by the CMIP6- and CMIP5-driven hydrological ensembles over the 2071–2100 period.

Close modal

This study compares the GCM-simulated precipitation and temperature in CMIP5 and CMIP6 in terms of climate change and inter-model variability. More importantly, the change and inter-model variability of simulated mean and peak flows are also compared between these two archives after driving hydrological models using bias corrected climate model simulations. The results show that the daily mean temperature and mean annual precipitation are presented to increase in the future period, while GCMs in CMIP6 project a larger increase than those in CMIP5 for almost all watersheds. For hydrological simulations, an increase in mean and peak flows are observed for both CMIP5 and CMIP6, while the difference between CMIP5 and CMIP6 is very limited, especially for southern China. This implies that the change of precipitation dominates the change of streamflow for almost all watersheds in China, because the bias correction method used in this study does not alter the change of precipitation and temperature. Even though the increase in temperature may result in the rise of evapotranspiration, its contribution to streamflow is less important than that of precipitation. However, for most of the southern watersheds, the difference between CMIP5 and CMIP6 does not emerge, which means that the limited difference in the increase of precipitation is offset by the increase of evapotranspiration caused by the rise in temperature.

In terms of the inter-model variability, GCMs in CMIP6 project less variability than those in CMIP5 for most watersheds. This implies that the former is more reliable than the latter in the projection of temperature for the future period. Especially taking into account the fact that the inter-model variability is larger for CMIP6 than CMIP5 for the historical period. However, this is not the case for precipitation, as GCMs in CMIP6 project larger inter-model variability of precipitation for watersheds in southern and northeastern China for both historical and future periods. For central China, the inter-model variability of precipitation is larger for CMIP5 than CMIP6. However, different results are observed for mean and peak flows. For most watersheds in southern and northeastern China, GCMs in CMIP6 project less inter-model variability in both mean and peak flows than those in CMIP5, while an opposite pattern is observed in central China. Apparently, the streamflow results are contradictory to precipitation and temperature. This is because the bias correction method changes the inter-model variability of precipitation for the future period. Moreover, the various hydrological conditions between southeast and northwest China also result in different results. Affected by the East Asian monsoon, precipitation gradually decreases from the southeast coast to the northwest inland, which leads to drought in the north and moisture in the south.

Figure 10 shows that after bias correction, for most watersheds in southern China and some watersheds in northeast China, the inter-model variability of precipitation simulated by GCMs in CMIP6 is smaller than that simulated by GCMs in CMIP5, which is consistent with that of streamflow. More reduction in inter-model variability of precipitation for CMIP6 than CMIP5 means that the bias of GCM simulations in CMIP6 is more stationary than that in CMIP5, because bias correction methods were developed based on the assumption of bias stationary of climate model simulations. However, some studies (Teutschbein & Seibert 2013; Velázquez et al. 2015) have found that the bias of GCM simulation is not stationary, especially for precipitation. When the biases of the future period are larger or smaller than those of the historical period, bias correction may improve the results, but the bias cannot be completely removed for the future period. This is the reason that large inter-model variability is found. When future biases reduce to less than half the calibration biases or when bias directions are different (different signs) between future and historical periods, the bias correction can deteriorate the original future simulation (Maraun 2012; Chen et al. 2021). As the bias nonstationarity is caused by natural climate variability and climate model sensitivity, this study implies that the climate model sensitivity is lower for CMIP6 than for CMIP5.
Figure 10

Inter-model standard deviation of mean annual precipitation (prcptot) (a,b) simulated by CMIP5 and CMIP6 climate models for 2071–2100 period after bias correction. Differences between CMIP6 (a) and CMIP5 (b) climate models are shown on the bottom row (c).

Figure 10

Inter-model standard deviation of mean annual precipitation (prcptot) (a,b) simulated by CMIP5 and CMIP6 climate models for 2071–2100 period after bias correction. Differences between CMIP6 (a) and CMIP5 (b) climate models are shown on the bottom row (c).

Close modal

This study uses the change and inter-model variability as the indices to compare the performance of GCM-simulated precipitation, temperature, and their hydrological impacts in CMIP5 and CMIP6. Following conclusions were drawn from this study.

  • 1.

    Daily mean temperature, and mean annual precipitation are projected to increase from historical to a future period, while GCMs in CMIP6 project a larger increase than those in CMIP5 for almost all watersheds.

  • 2.

    An increase in mean and peak flows are observed for both CMIP5 and CMIP6, while the difference between CMIP5 and CMIP6 is very limited, especially for southern China. The change in precipitation dominates the change of streamflow for almost all watersheds in China.

  • 3.

    GCMs in CMIP6 project less variability than those in CMIP5 in temperature for most watersheds, which shows that the former is more reliable than the latter in projection temperature for the future period.

  • 4.

    GCMs in CMIP6 project less inter-model variability in both mean and peak flows than those in CMIP5 for most watersheds in southern and northeastern China, while an opposite pattern is observed in central China.

Overall, this study indicates that GCMs in CMIP6 are more reliable than those in CMIP5 in simulating temperature. However, the reliability in projecting precipitation is dependent on regions. When applying a bias correction method to climate model simulations for hydrological impact studies, GCMs in CMIP6 are also more reliable in projecting future hydrological regimes.

This work was partially supported by the National Natural Science Foundation of China (Grant Nos. U2240201, 52079093), the Hubei Provincial Natural Science Foundation of China (Grant No. 2020CFA100), and the Overseas Expertise Introduction Project for Discipline Innovation (111 Project) funded by Ministry of Education and State Administration of Foreign Experts Affairs P.R. China (Grant B18037).

All relevant data are available from an online repository or repositories. The daily precipitation, maximum and minimum temperature of GCMs in CMIP6 is downloadable from http://esgf-node.llnl.gov/search/cmip6/, and that of GCMs in CMIP5 is downloadable from http://esgf-node.llnl.gov/search/cmip5/, and that of reference can be downloaded from http://data.cma.cn/).

The authors declare there is no conflict.

AghaKouchak
A.
,
Cheng
L.
,
Mazdiyasni
O.
&
Farahmand
A.
2014
Global warming and changes in risk of concurrent climate extremes: insights from the 2014 California drought
.
Geophysical Research Letters
41
(
24
),
8847
8852
.
https://doi.org/10.1002/2014GL062308
.
Arsenault
R.
,
Poulin
A.
,
Côté
P.
&
Brissette
F.
2014
Comparison of stochastic optimization algorithms in hydrological model calibration
.
Journal of Hydrologic Engineering
19
(
7
),
1374
1384
.
https://doi.org/10.1061/(asce)he.1943-5584.0000938
.
Asadieh
B.
&
Krakauer
N. Y.
2017
Global change in streamflow extremes under climate change over the 21st century
.
Hydrology and Earth System Sciences
21
(
11
),
5863
5874
.
https://doi.org/10.5194/hess-21-5863-2017
.
Bárdossy
A.
,
Huang
Y.
&
Wagener
T.
2016
Simultaneous calibration of hydrological models in geographical space
.
Hydrology and Earth System Sciences
20
(
7
),
2913
2928
.
https://doi.org/10.5194/hess-20-2913-2016
.
Bian
G.
,
Zhang
J.
,
Chen
J.
,
Song
M.
,
He
R.
,
Liu
C.
,
Liu
Y.
,
Bao
Z.
,
Lin
Q.
&
Wang
G.
2021
Projecting hydrological responses to climate change using CMIP6 climate scenarios for the Upper Huai River Basin, China
.
Frontiers in Environmental Science
9
(
December
),
1
14
.
https://doi.org/10.3389/fenvs.2021.759547
.
Burek
P.
,
Satoh
Y.
,
Kahil
T.
,
Tang
T.
,
Greve
P.
,
Smilovic
M.
,
Guillaumot
L.
,
Zhao
F.
&
Wada
Y.
2020
Development of the Community Water Model (CWatM v1.04) – a high-resolution hydrological model for global and regional assessment of integrated water resources management
.
Geoscientific Model Development
13
(
7
),
3267
3298
.
https://doi.org/10.5194/gmd-13-3267-2020
.
Chen
J.
,
Brissette
F. P.
,
Chaumont
D.
&
Braun
M.
2013
Performance and uncertainty evaluation of empirical downscaling methods in quantifying the climate change impacts on hydrology over two North American river basins
.
Journal of Hydrology
479
,
200
214
.
https://doi.org/10.1016/j.jhydrol.2012.11.062
.
Chen
J.
,
Brissette
F. P.
&
Caya
D.
2020
Remaining error sources in bias-corrected climate model outputs
.
Climatic Change
162
(
2
).
https://doi.org/10.1007/s10584-020-02744-z.
Chen
J.
,
Arsenault
R.
,
Brissette
F. P.
&
Zhang
S.
2021
Climate change impact studies: should we bias correct climate model outputs or post-process impact model outputs?
Water Resources Research
57
(
5
).
https://doi.org/10.1029/2020WR028638.
Choi
H. J.
,
Choi
S. J.
,
Koo
M. S.
,
Kim
J. E.
,
Kwon
Y. C.
&
Hong
S. Y.
2017
Effects of parameterized orographic drag on weather forecasting and simulated climatology over east Asia during boreal summer
.
Journal of Geophysical Research: Atmospheres
122
(
20
),
10,669
10,678
.
https://doi.org/10.1002/2017JD026696
.
Christiansen
B.
2020
Understanding the distribution of multimodel ensembles
.
Journal of Climate
33
(
21
),
9447
9465
.
https://doi.org/10.1175/JCLI-D-20-0186.1
.
Cook
B. I.
,
Mankin
J. S.
,
Marvel
K.
,
Williams
A. P.
,
Smerdon
J. E.
&
Anchukaitis
K. J.
2020
Twenty-first century drought projections in the CMIP6 forcing scenarios
.
Earth's Future
8
(
6
),
1
20
.
https://doi.org/10.1029/2019EF001461
.
Eyring
V.
,
Bony
S.
,
Meehl
G. A.
,
Senior
C. A.
,
Stevens
B.
,
Stouffer
R. J.
&
Taylor
K. E.
2016
Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization
.
Geoscientific Model Development
9
(
5
),
1937
1958
.
https://doi.org/10.5194/gmd-9-1937-2016
.
Fan
K.
,
Zhang
Q.
,
Singh
V. P.
,
Sun
P.
,
Song
C.
,
Zhu
X.
,
Yu
H.
&
Shen
Z.
2019
Spatiotemporal impact of soil moisture on air temperature across the Tibet Plateau
.
Science of the Total Environment
649
,
1338
1348
.
https://doi.org/10.1016/j.scitotenv.2018.08.399
.
Fan
X.
,
Miao
C.
,
Duan
Q.
,
Shen
C.
&
Wu
Y.
2020
The performance of CMIP6 versus CMIP5 in simulating temperature extremes over the global land surface
.
Journal of Geophysical Research: Atmospheres
125
(
18
),
1
16
.
https://doi.org/10.1029/2020JD033031
.
Fathalli
B.
,
Pohl
B.
,
Castel
T.
&
Safi
M. J.
2019
Errors and uncertainties in regional climate simulations of rainfall variability over Tunisia: a multi-model and multi-member approach
.
Climate Dynamics
52
(
1–2
),
335
361
.
https://doi.org/10.1007/s00382-018-4150-2
.
Greene
A. M.
,
Giannini
A.
&
Zebiak
S. E.
2009
Drought return times in the Sahel: a question of attribution
.
Geophysical Research Letters
36
(
12
).
https://doi.org/10.1029/2009GL038868.
Guo
Q.
,
Chen
J.
,
Zhang
X.
,
Shen
M.
,
Chen
H.
&
Guo
S.
2019
A new two-stage multivariate quantile mapping method for bias correcting climate model outputs
.
Climate Dynamics
53
(
5–6
),
3603
3623
.
https://doi.org/10.1007/s00382-019-04729-w
.
Her
Y.
,
Yoo
S. H.
,
Cho
J.
,
Hwang
S.
,
Jeong
J.
&
Seong
C.
2019
Uncertainty in hydrological analysis of climate change: multi-parameter vs. multi-GCM ensemble predictions
.
Scientific Reports
9
(
1
),
1
23
.
https://doi.org/10.1038/s41598-019-41334-7
.
Hirabayashi
Y.
,
Mahendran
R.
,
Koirala
S.
,
Konoshima
L.
,
Yamazaki
D.
,
Watanabe
S.
,
Kim
H.
&
Kanae
S.
2013
Global flood risk under climate change
.
Nature Climate Change
3
(
9
),
816
821
.
https://doi.org/10.1038/nclimate1911
.
Hou
Y.
,
Guo
H.
,
Yang
Y.
&
Liu
W.
2022
Global evaluation of runoff simulation from climate, hydrological and land surface models
.
Water Resources Research
1
25
.
https://doi.org/10.1029/2021WR031817
.
Huang
J.
,
Yu
H.
,
Dai
A.
,
Wei
Y.
&
Kang
L.
2017
Drylands face potential threat under 2 °C global warming target
.
Nature Climate Change
7
(
6
),
417
422
.
https://doi.org/10.1038/nclimate3275
.
Ishizaki
N. N.
,
Dairaku
K.
&
Ueno
G.
2017
Regional probabilistic climate projection for Japan with a regression model using multi-model ensemble experiments
.
Hydrological Research Letters
11
(
1
),
44
50
.
https://doi.org/10.3178/hrl.11.44
.
Jiang
T.
,
Chen
Y. D.
,
Xu
C. y.
,
Chen
X.
,
Chen
X.
&
Singh
V. P.
2007
Comparison of hydrological impacts of climate change simulated by six hydrological models in the Dongjiang Basin, South China
.
Journal of Hydrology
336
(
3–4
),
316
333
.
https://doi.org/10.1016/j.jhydrol.2007.01.010
.
Lehner
F.
,
Wood
A. W.
,
Vano
J. A.
,
Lawrence
D. M.
,
Clark
M. P.
&
Mankin
J. S.
2019
The potential to reduce uncertainty in regional runoff projections from climate models
.
Nature Climate Change
9
(
12
),
926
933
.
https://doi.org/10.1038/s41558-019-0639-x
.
Lehner
F.
,
Deser
C.
,
Maher
N.
,
Marotzke
J.
,
Fischer
E.
,
Brunner
L.
,
Knutti
R.
&
Hawkins
E.
2020
Partitioning climate projection uncertainty with multiple large ensembles and CMIP5/6
.
Earth System Dynamics Discussions
1
28
.
https://doi.org/10.5194/esd-2019-93
.
Li
P.
,
Mu
X.
,
Holden
J.
,
Wu
Y.
,
Irvine
B.
,
Wang
F.
,
Gao
P.
,
Zhao
G.
&
Sun
W.
2017
Comparison of soil erosion models used to study the Chinese Loess Plateau
.
Earth-Science Reviews
170
(
26
),
17
30
.
https://doi.org/10.1016/j.earscirev.2017.05.005
.
Li
Z.
,
Liu
T.
,
Huang
Y.
,
Peng
J.
&
Ling
Y.
2022
Evaluation of the CMIP6 precipitation simulations over global land
.
Earth's Future
10
(
8
),
1
21
.
https://doi.org/10.1029/2021EF002500
.
Lin
K.
,
Chen
H.
,
Xu
C. Y.
,
Yan
P.
,
Lan
T.
,
Liu
Z.
&
Dong
C.
2020
Assessment of flash flood risk based on improved analytic hierarchy process method and integrated maximum likelihood clustering algorithm
.
Journal of Hydrology
584
(
February
),
124696
.
https://doi.org/10.1016/j.jhydrol.2020.124696
.
Maraun
D.
2012
Nonstationarities of regional climate model biases in European seasonal mean temperature and precipitation sums
.
Geophysical Research Letters
39
(
6
).
https://doi.org/10.1029/2012GL051210.
Martel
J. L.
,
Demeester
K.
,
Brissette
F.
,
Poulin
A.
&
Arsenault
R.
2017
HMETS-A simple and efficient hydrology model for teaching hydrological modelling, flow forecasting and climate change impacts
.
International Journal of Engineering Education
33
(
4
),
1307
1316
.
Merelo-Guervós
J. J.
&
Castillo-Valdivieso
P.
2004
Conference paper assignment using a combined greedy/evolutionary algorithm
. In:
Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
, Vol.
3242
.
https://doi.org/10.1007/978-3-540-30217-9_61.
O'Neill
B. C.
,
Tebaldi
C.
,
van Vuuren
D. P.
,
Eyring
V.
,
Friedlingstein
P.
,
Hurtt
G.
,
Knutti
R.
,
Kriegler
E.
,
Lamarque
J.
,
Lowe
J.
,
Meehl
G. A.
,
Moss
R.
,
Riahi
K.
&
Sanderson
B. M.
2016
The Scenario Model Intercomparison Project (ScenarioMIP) for CMIP6
.
Geoscientific Model Development
9
(
9
),
3461
3482
.
https://doi.org/10.5194/gmd-9-3461-2016
.
Oudin
L.
,
Hervieu
F.
,
Michel
C.
,
Perrin
C.
,
Andréassian
V.
,
Anctil
F.
&
Loumagne
C.
2005
Which potential evapotranspiration input for a lumped rainfall-runoff model? part 2 – towards a simple and efficient potential evapotranspiration model for rainfall-runoff modelling
.
Journal of Hydrology
303
(
1–4
),
290
306
.
https://doi.org/10.1016/j.jhydrol.2004.08.026
.
Papadimitriou
L. V.
,
Koutroulis
A. G.
,
Grillakis
M. G.
&
Tsanis
I. K.
2016
High-end climate change impact on European runoff and low flows – exploring the effects of forcing biases
.
Hydrology and Earth System Sciences
20
(
5
),
1785
1808
.
https://doi.org/10.5194/hess-20-1785-2016
.
Paulo
J.
,
Fialho
L.
,
Cauduro
R.
,
Paiva
D. D.
,
Collischon
W.
,
Bravo
J. M.
&
Siqueira
V. A.
2020
Climate Change Impacts on South American Water Balance from a Continental-Scale Hydrological Model Driven by CMIP5 Projections Content Courtesy of Springer Nature, Terms of Use Apply . Rights Reserved. 503–522
.
Pokhrel
Y.
,
Felfelani
F.
,
Satoh
Y.
,
Boulange
J.
,
Burek
P.
,
Gädeke
A.
,
Gerten
D.
,
Gosling
S. N.
,
Grillakis
M.
,
Gudmundsson
L.
,
Hanasaki
N.
,
Kim
H.
,
Koutroulis
A.
,
Liu
J.
,
Papadimitriou
L.
,
Schewe
J.
,
Müller Schmied
H.
,
Stacke
T.
,
Telteu
C. E.
&
Wada
Y.
2021
Global terrestrial water storage and drought severity under climate change
.
Nature Climate Change
11
(
3
),
226
233
.
https://doi.org/10.1038/s41558-020-00972-w
.
Pörtner
H. O.
,
Roberts
D. C.
,
Adams
H.
,
Adler
C.
,
Aldunce
P.
,
Ali
E.
et al.
2022
Climate Change 2022: Impacts, Adaptation and Vulnerability. IPCC Sixth Assessment Report
.
Rasmussen
D. J.
,
Meinshausen
M.
&
Kopp
R. E.
2016
Probability-weighted ensembles of U.S. county-level climate projections for climate risk analysis
.
Journal of Applied Meteorology and Climatology
55
(
10
),
2301
2322
.
https://doi.org/10.1175/JAMC-D-15-0302.1
.
Stocker
T. F.
,
Qin
D.
,
Plattner
G. K.
,
Tignor
M. M. B.
,
Allen
S. K.
,
Boschung
J.
,
Nauels
A.
,
Xia
Y.
,
Bex
V.
&
Midgley
P. M.
2013
Climate change 2013 the physical science basis: Working Group I contribution to the fifth assessment report of the intergovernmental panel on climate change
. In:
Climate Change 2013 the Physical Science Basis: Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
, Vol.
9781107057
.
https://doi.org/10.1017/CBO9781107415324.
Syukuro
M.
2019
Role of greenhouse gas in climate change
.
Tellus A: Dynamic Meteorology and Oceanography
71
(
1
).
https://doi.org/10.1080/16000870.2019.1620078.
Teutschbein
C.
&
Seibert
J.
2013
Is bias correction of regional climate model (RCM) simulations possible for non-stationary conditions
.
Hydrology and Earth System Sciences
17
(
12
),
5061
5077
.
https://doi.org/10.5194/hess-17-5061-2013
.
Velázquez
J. A.
,
Schmid
J.
,
Ricard
S.
,
Muerth
M. J.
,
Gauvin St-Denis
B.
,
Minville
M.
,
Chaumont
D.
,
Caya
D.
,
Ludwig
R.
&
Turcotte
R.
2013
An ensemble approach to assess hydrological models’ contribution to uncertainties in the analysis of climate change impact on water resources
.
Hydrology and Earth System Sciences
17
(
2
),
565
578
.
https://doi.org/10.5194/hess-17-565-2013
.
Velázquez
J. A.
,
Troin
M.
,
Caya
D.
&
Brissette
F.
2015
Evaluating the time-invariance hypothesis of climate model bias correction: implications for hydrological impact studies
.
Journal of Hydrometeorology
16
(
5
),
2013
2026
.
https://doi.org/10.1175/JHM-D-14-0159.1
.
Xiao
M.
,
Zhang
Q.
,
Singh
V. P.
&
Chen
X.
2013
Regionalization-based spatiotemporal variations of precipitation regimes across China
.
Theoretical and Applied Climatology
114
(
1–2
),
203
212
.
https://doi.org/10.1007/s00704-013-0832-1
.
Yin
J.
,
Guo
S.
,
Gentine
P.
,
Sullivan
S. C.
,
Gu
L.
,
He
S.
,
Chen
J.
&
Liu
P.
2021
Does the hook structure constrain future flood intensification under anthropogenic climate warming?
Water Resources Research
57
(
2
).
https://doi.org/10.1029/2020WR028491.
Yin
J.
,
Guo
S.
,
Yang
Y.
,
Chen
J.
,
Gu
L.
,
Wang
J.
,
He
S.
,
Wu
B.
&
Xiong
J.
2022
Projection of droughts and their socioeconomic exposures based on terrestrial water storage anomaly over China
.
Science China Earth Sciences
65
(
9
),
1772
1787
.
https://doi.org/10.1007/s11430-021-9927-x
.
Zhang
S.
2021
Uncertainty in projection of climate extremes: a comparison of CMIP5 and CMIP6
.
Journal of Meteorological Research
35
(
August
),
646
.
https://doi.org/10.1007/s13351-021-1012-3.1.Introduction
.
Zhang
Q.
,
Xiao
M.
,
Singh
V. P.
,
Liu
L.
&
Xu
C. Y.
2015
Observational evidence of summer precipitation deficit-temperature coupling in China
.
Journal of Geophysical Research
120
(
19
).
https://doi.org/10.1002/2015JD023830.
Zhao Ren-Jun
Y.-L.
,
Fang Le-Run
Z.
,
Xin-Ren
L.
&
Sheng
Z. Q.
.
1980
The Xinanjiang model. Hydrological Forecasting. Proc. Oxford Symposium, April. International Association of Hydrological Sciences, Washington DC; IAHS-AISH Publication 129)
.
Zheng
Y.
,
Zhang
Q.
,
Luo
M.
,
Sun
P.
&
Singh
V. P.
2020
Wintertime precipitation in eastern China and relation to the Madden-Julian oscillation: spatiotemporal properties, impacts and causes
.
Journal of Hydrology
582
(
October 2019
),
124477
.
https://doi.org/10.1016/j.jhydrol.2019.124477
.
Zhou
X.
,
Ma
W.
,
Echizenya
W.
&
Yamazaki
D.
2020
Uncertainty in flood frequency analysis of hydrodynamic model simulations
.
Natural Hazards and Earth System Sciences Discussions
1
31
.
https://doi.org/10.5194/nhess-2020-272
.
Zhu
X.
,
Lee
S. Y.
,
Wen
X.
,
Ji
Z.
,
Lin
L.
,
Wei
Z.
,
Zheng
Z.
,
Xu
D.
&
Dong
W.
2021
Extreme climate changes over three major river basins in China as seen in CMIP5 and CMIP6
.
Climate Dynamics
57
(
3–4
),
1187
1205
.
https://doi.org/10.1007/s00382-021-05767-z
.
Zhuo
L.
&
Han
D.
2017
Multi-source hydrological soil moisture state estimation using data fusion optimisation
.
Hydrology and Earth System Sciences
21
(
7
),
3267
3285
.
https://doi.org/10.5194/hess-21-3267-2017
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).

Supplementary data