The GCM data of surface air temperature at 2 m and precipitation are derived from 42 CMIP5 GCMs for historical experiments and four sets of RCP experiments available online at http://pcmdi9.llnl.gov. The four RCPs include one mitigation scenario leading to a very low forcing level (RCP2.6), two medium stabilization scenarios (RCP4.5 and RCP6.0), and one very high baseline emissions scenario (RCP8.5) (van Vuuren et al., 2011). Basic information on the 42 GCMs and their experiments is provided in Table 1. The China Region 05.1 (CN05.1)

observational dataset, based on 2416 meteorological stations, is used to evaluate the ability of the GCMs in simulating the climatology over China for 1986-2005 (Wu and Gao, 2013).

Table 1

Table 1

Table 1 Information on the 42 CMIP5 models and their experiments, in which “ * ” denote the models with the length of the pre-industrial control run being at least 500 years, “ √ ” denotes the period 2006-2100, and “ × ” denote the period 2006-2009 for EC-EARTH, 2006-2035 for HadCM3 and MIROC4h, and 2006-2099 for HadGEM2-AO and HadGEM2-ES. No. Name Full name Country Atmospheric horizontal resolution Historical run RCP 2.6 RCP 4.5 RCP 6.0 RCP 8.5 * 01 ACCESS1.0 Australian Community Climate and Earth System Simulator version 1.0 Australia 192 × 145 1986-2005 √ √ * 02 ACCESS1.3 Australian Community Climate and Earth System Simulator version 1.3 Australia 192 × 145 1986-2005 √ √ * 03 BCC-CSM1.1 Beijing Climate Center Climate System Model version 1.1 China 128 × 64 1986-2005 √ √ √ √ 04 BCC-CSM1.1(m) Beijing Climate Center Climate System Model version 1.1 (m) China 320 × 160 1986-2005 √ √ √ √ * 05 BNU-ESM Beijing Normal University-Earth System Model China 128 × 64 1986-2005 √ √ √ * 06 CanESM2 Canadian Earth System Model version 2 Canada 128 × 64 1986-2005 √ √ √ * 07 CCSM4 Community Climate System Model version 4 USA 128 × 192 1986-2005 √ √ √ √ * 08 CESM1(BGC) Community Earth System Model version 1 (Biogeochemistry) USA 128 × 192 1986-2005 √ √ 09 CESM1(CAM5) Community Earth System Model version 1 (Community Atmosphere Model version 5) USA 128 × 192 1986-2005 √ √ √ √ 10 CMCC-CESM Centro Euro-Mediterraneo sui Cambiamenti Climatici-Carbon Earth System Model Italy 96 × 48 1986-2005 √ 11 CMCC-CM Centro Euro-Mediterraneo sui Cambiamenti Climatici-Climate Model Italy 480 × 240 1986-2005 √ √ * 12 CMCC-CMS Centro Euro-Mediterraneo sui Cambiamenti Climatici-Climate Model with the Stratospheric component Italy 192 × 96 1986-2005 √ √ * 13 CNRM-CM5 Centre National de Recherches Mé té orolgiques-Climate Model version 5 France 256 × 128 1986-2005 √ √ √ * 14 CSIRO-Mk3.6.0 Commonwealth Scientific and Industrial Research Organisation Mark version 3.6.0 Australia 192 × 96 1986-2005 √ √ √ √ 15 EC-EARTH European meteorological/Climate centres and universities-Earth system model European 320 × 160 1986-2005 × √ * 16 FGOALS-g2 Flexible Global Ocean-Atmosphere-Land System model, grid-point version 2 China 128 × 60 1986-2005 √ √ √ * 17 FIO-ESM First Institute of Oceanography-Earth System Model China 128 × 64 1986-2005 √ √ √ √ 18 GFDL-CM2.1 Geophysical Fluid Dynamics Laboratory-Climate Model version 2.1 USA 144 × 90 1986-2005 × * 19 GFDL-CM3 Geophysical Fluid Dynamics Laboratory-Climate Model version 3 USA 144 × 90 1986-2005 √ √ √ √ * 20 GFDL-ESM2G Geophysical Fluid Dynamics Laboratory-Earth Science Model version 2G USA 144 × 90 1986-2005 √ √ √ √ * 21 GFDL-ESM2M Geophysical Fluid Dynamics Laboratory-Earth Science Model version 2M USA 144 × 90 1986-2005 √ √ √ √ * 22 GISS-E2-H Goddard Institute for Space Studies-model E version 2-coupled with the HYCOM ocean model USA 144 × 90 1986-2005 √ √ √ √ 23 GISS-E2-H-CC Goddard Institute for Space Studies-model E version 2-coupled with the HYCOM ocean model, Carbon Cycle USA 144 × 90 1986-2005 √ √ * 24 GISS-E2-R Goddard Institute for Space Studies-model E version 2-coupled with Russell ocean model USA 144 × 90 1986-2005 √ √ √ √ 25 GISS-E2-R-CC Goddard Institute for Space Studies-model E version 2-coupled with Russell ocean model, Carbon Cycle USA 144 × 90 1986-2005 √ √ 26 HadCM3 Hadley Centre Climate Model version 3 UK 96 × 73 1986-2005 × * 27 HadGEM2-AO Hadley Centre Global Environmental Model version 2-Atmosphere-Ocean Korea 192 × 145 1986-2005 √ √ √ × 28 HadGEM2-CC Hadley Centre Global Environmental Model version 2-Carbon Cycle UK 192 × 145 1986-2005 √ √ * 29 HadGEM2-ES Hadley Centre Global Environmental Model version 2-Earth System UK 192 × 145 1986-2005 √ √ × √ * 30 INM-CM4 Institute for Numerical Mathematics-Climate Model version 4 Russia 180 × 120 1986-2005 √ √

Table 1 Information on the 42 CMIP5 models and their experiments, in which “ * ” denote the models with the length of the pre-industrial control run being at least 500 years, “ √ ” denotes the period 2006-2100, and “ × ” denote the period 2006-2009 for EC-EARTH, 2006-2035 for HadCM3 and MIROC4h, and 2006-2099 for HadGEM2-AO and HadGEM2-ES.

Table 1

Table 1

Table 1 (Continued) No. Name Full name Country Atmospheric horizontal resolution Historical run RCP 2.6 RCP 4.5 RCP 6.0 RCP 8.5 * 31 IPSL-CM5A-LR Institut Pierre-Simon Laplace-Coupled Model version 5A, coupled with Nucleus for European Modelling of the ocean-Low Resolution France 96 × 96 1986-2005 √ √ √ √ 32 IPSL-CM5A-MR Institut Pierre-Simon Laplace-Coupled Model version 5A, coupled with Nucleus for European Modelling of the ocean-Medium Resolution France 144 × 143 1986-2005 √ √ √ √ 33 IPSL-CM5B-LR Institut Pierre-Simon Laplace-Coupled Model version 5B, coupled with Nucleus for European Modelling of the ocean-Low Resolution France 96 × 96 1986-2005 √ √ 34 MIROC4h Model for Interdisciplinary Research on Climate version 4h Japan 640 × 320 1986-2005 × * 35 MIROC5 Model for Interdisciplinary Research on Climate version 5 Japan 256 × 128 1986-2005 √ √ √ √ * 36 MIROC-ESM Model for Interdisciplinary Research on Climate-Earth System Model Japan 128 × 64 1986-2005 √ √ √ √ 37 MIROC-ESM-CHEM Model for Interdisciplinary Research on Climate-Earth System Model-Chemistry coupled Japan 128 × 64 1986-2005 √ √ √ √ * 38 MPI-ESM-LR Max Planck Institute-Earth System Model-Low Resolution Germany 192 × 96 1986-2005 √ √ √ * 39 MPI-ESM-MR Max Planck Institute-Earth System Model-Medium Resolution Germany 192 × 96 1986-2005 √ √ √ * 40 MRI-CGCM3 Meteorological Research Institute-Coupled Atmosphere-Ocean General Circulation Model version 3 Japan 320 × 160 1986-2005 √ √ √ √ * 41 NorESM1-M Norwegian Earth System Model version 1-intermediate resolution Norway 144 × 96 1986-2005 √ √ √ √ 42 NorESM1-ME Norwegian Earth System Model version 1-intermediate resolution with carbon cycle Norway 144 × 96 1986-2005 √ √ √ √

Table 1 (Continued)

Considering the natural internal variability of a climate system occurs in the absence of external forcings (Deser et al., 2012; Hawkins and Sutton, 2012), the present internal variability of temperature and precipitation is obtained based on 27 GCMs with the length of the pre-industrial control simulation being at least 500 years. The first 100 years of the pre-industrial control run are not used to exclude the influence of initial conditions. The natural variability is calculated as the standard deviation of non-overlapping 20-year means after a quadratic fit is subtracted at every grid point to eliminate model drift (Collins et al., 2013). This is multiplied by the square root of 2, a factor that arises as the comparison is between two distributions of numbers. The median across all models of that quantity is used. In our work, regions where the multi-GCM median change exceeds two standard deviations of internal variability, and where at least 90% of the GCMs agree on the sign of change, are interpreted as ‘ large change with high model agreement’ ; and regions where the median change is less than one standard deviation of internal variability are interpreted as ‘ small signal or low agreement of models’ (Collins et al., 2013). In addition, based on the range of the horizontal resolutions of 42 GCMs, all the above data are re-gridded to a relatively mid-range horizontal grid resolution of 2° × 2° using a bilinear interpolation algorithm, which is roughly the average resolution of all the GCMs.

To quantitatively evaluate the capability of the 42 GCMs, spatial correlation coefficients (SCCs) and the centered root-mean-square error (RMSE) of annual and seasonal temperature and precipitation between each baseline simulation and the observation for the period 1986-2005 are calculated on the basis of 240 grid points over continental China. For temperature (precipitation), the SCCs range from 0.85 to 0.98 (0.03 to 0.93), and the centered RMSEs range from 0.19 to 0.67 (0.46 to 1.86). Since the abilities of the GCMs differ from one another, two thresholds are set to identify reliable GCMs. First, the SCCs must be positive and statistically significant at the 99% confidence level, and second, the RMSEs must be smaller than 1.5. For temperature, all GCMs pass the two criteria; where precipitation is concerned, BNU-ESM, FIO-ESM, and MIROC-ESM are excluded accordingly. Thus, a total of 39 GCMs are used in the subsequent analysis. Of note is that, compared with observation, the median of the 39 CMIP5 GCMs reproduces the spatial pattern and variability of climatological temperature and precipitation over China reasonably well. Nevertheless, there are still biases, such as cold biases over most of China, excessive precipitation on the Tibetan Plateau and northern East China, and underestimated precipitation in southeastern China and parts of Xinjiang.

The annual temperature over China is projected to continue to increase in the 21st century under all four RCPs, with warming in northern China being larger than in southern China (Fig. 1). Relative to 1986-2005, the increases of annual temperature for 2016-2035 under the four RCPs are close to each other, with regional averages of 1.0° C, 1.0° C, 0.9° C, and 1.1° C under RCP2.6, RCP4.5, RCP6.0, and RCP8.5, respectively. The magnitude of the warming becomes stronger as the radiative forcing increases, with regional averages of 1.5° C, 1.9° C, 1.7° C, and 2.6° C (1.4° C, 2.5° C, 3.0° C, and 4.7° C) under RCP2.6, RCP4.5, RCP6.0, and RCP8.5 for 2046-2065 (2081-2099), respectively. For 2016-2035, large annual temperature change with high model agreement occurs over most of China under all four RCPs. For 2081-2099, the medians of the annual changes are 9.9, 19.3, 22.8, and 35.9 times greater than one standard deviation of their internal variability under RCP2.6, RCP4.5, RCP6.0, and RCP8.5, respectively.

Figure 1

Figure Option ViewDownloadNew Window

Figure 1 Multi-model median of annual temperature changes (units: ° C) relative to 1986-2005 for 2016-2035, 2046-2065, and 2081-2099 under RCP2.6, RCP4.5, RCP6.0, and RCP8.5, respectively. Stippling indicates regions where the median change is greater than two standard deviations of internal variability and where at least 90% of models agree on the sign of change. Blank space indicates regions where the median change is between one and two standard deviations of internal variability. The number in the top-right corner indicates how many chosen models were available.

On the seasonal scale, large-scale warming patterns resemble the annual mean under all RCPs, and the warming also becomes stronger as the radiative forcing increases. However, for 2081-2099, a large temperature increase with high model agreement under RCP2.6 only occurs in the southeastern Tibetan Plateau in winter, while it appears over most of China in the other seasons (Fig. 2). Under RCP8.5, it occurs over the whole country during 2081-2099 for all seasons.

Figure 2

Figure Option ViewDownloadNew Window

Figure 2 Multi-model median of seasonal temperature changes (units: ° C) relative to 1986-2005 for 2081-2099 under RCP2.6 and RCP8.5. Stippling indicates regions where the median change is greater than two standard deviations of internal variability and where at least 90% of models agree on the sign of change. Blank space indicates regions where the median change is between one and two standard deviations of internal variability. The number in the top-right corner indicates how many chosen models were available.

The annual precipitation amount is projected to increase in the 21st century in most of China, and the magnitude of the changes becomes larger with increasing radiative forcing, especially in northern China (Fig. 3). Quantitatively, the increases of annual precipitation over China for 2016-2035 range from 2.5% to 3.5% under the four RCPs. For 2046-2065 (2081-2099), the regional averages of changes in annual precipitation are 6.2%, 6.5%, 5.3%, and 8.4% (6.1%, 9.3%, 9.6%, and 16.2%) under RCP2.6, RCP4.5, RCP6.0, and RCP8.5, respectively.

Figure 3

Figure Option ViewDownloadNew Window

Figure 3 Multi-model median of the percentage change in annual precipitation (units: %) relative to 1986-2005 for 2016-2035, 2046-2065, and 2081-2099 under RCP2.6, RCP4.5, RCP6.0, and RCP8.5, respectively. Stippling indicates regions where the median change is greater than two standard deviations of internal variability and where at least 90% of models agree on the sign of change. Blank space indicates regions where the median change is between one and two standard deviations of internal variability, and the crosses indicate regions where the median change is less than one standard deviation of internal variability. The number in the top-right corner means how many chosen models were available.

As indicated by the crosses in Fig. 3, a small signal or low agreement of models appears over most of the country under the four RCPs for 2016-2035. For 2046-2065, a small signal or low agreement of models is less apparent than for 2016-2035, and mainly appears in central and eastern China. A large annual precipitation change with high model agreement occurs in the northern Tibetan Plateau under RCP6.0 and RCP8.5. For 2081-2099, parts of central and eastern China still have a small signal or low agreement of models under the low and middle radiative forcing, while a large annual precipitation increase with high model agreement occurs in the northern Tibetan Plateau and the northern margin of Northeast China under the RCP6.0 and RCP8.5 scenarios (Fig. 3). For 2081- 2099, the medians of annual precipitation changes are 1.6, 2.3, 2.7, and 4.0 times greater than one standard deviation of their internal variability under RCP2.6, RCP4.5, RCP6.0, and RCP8.5, respectively. Seasonally, under the low and high radiative forcing, large-scale precipitation increases occur in China for 2081-2099, especially in northern China, but decreases in the central of East China in winter under RCP 8.5 (Fig. 4). A large change with high model agreement does not appear in any of the fourseasons under the RCP2.6 scenario. By contrast, it occurs in winter in parts of Xinjiang, Northeast China, and the northeastern Tibetan Plateau, and in spring in the northeastern Tibetan Plateau under the RCP8.5 scenario. For precipitation, a discrepancy exists between the observation and GCM simulations in East Asia, which need downscaling (Yu et al., 2010). Furthermore, the results from downscaling differ with GCMs (Gao et al., 2013).

Figure 4

Figure Option ViewDownloadNew Window

Figure 4 Multi-model median of the percentage change in seasonal precipitation (units: %) relative to 1986-2005 for 2081-2099 under RCP2.6 and RCP8.5. Stippling indicates regions where the median change is greater than two standard deviations of internal variability and where at least 90% of models agree on the sign of change. Blank space indicates regions where the median change is between one and two standard deviations of internal variability, and the crosses indicate regions where the median change is less than one standard deviation of internal variability. The number in the top-right corner means how many chosen models were available.