1. Introduction
With an average elevation over 4000 m, the Tibetan Plateau (TP) is the highest and largest plateau in the world. It contains the headwaters of nine major Asian rivers and supplies freshwater for 1.65 billion people and the surrounding ecosystems (Cuo and Zhang 2017; Gao et al. 2018b). Precipitation over the TP is vital to maintain the “water tower” of Asia (Immerzeel et al. 2010, 2013; Kääb et al. 2012). The rugged topography over the TP results in numerous finescale weather systems and diverse regional and local climate (Xue et al. 2014; Gao et al. 2015c). Over the past three decades, the TP has experienced faster warming than the global mean and other regions at the same latitude (Liu and Chen 2000; Rangwala et al. 2013; Gao et al. 2015b). A great concern is whether the faster local warming pace will continue in the future. If it does, the water resources and fragile ecosystems will be greatly disturbed due to global warming (You et al. 2011; Yang et al. 2014; Yao et al. 2007). The rapid warming and its environmental consequences have attracted much attention from both scientists and government departments (Yao et al. 2012; Guo et al. 2012; Dai 2006; Dai et al. 2019; Dai and Bloecker 2019; Gu and Adler 2013).
General circulation models (GCMs) are important tools for future climate change projections and provide useful climate information at global or continental scales (Flato et al. 2013; Giorgi et al. 2009; Seager et al. 2007, 2014). However, GCMs suffer from coarse horizontal resolution and representation of key physical processes (Han and Roads 2004; Liang et al. 2006), particularly for the TP, with its complex topography and land surface characteristics (Gao et al. 2015a,b,c, 2017, 2018b). For instance, most GCMs significantly overestimate the precipitation and underestimate the temperature over the TP in previous studies (Xu and Xu 2012; Su et al. 2013). Therefore, in many cases, their usefulness is limited for decision-making in local governments due to their inability to reproduce the climate at the regional scale. Downscaling has been developed to bridge the gap between GCMs and decision-maker needs. Statistical downscaling establishes empirical relationships between large-scale atmospheric circulation and local climate variation (Kim et al. 1984; Benestad et al. 2008). It is useful in regions with plenty of observations, such as eastern China, due to easy implementation (Chen and Chen 2003; Wetterhall et al. 2006; Dai et al. 2014). For the TP, however, observations are much scarcer than in eastern China. In particular, there is no observation site in the western TP (Gao et al. 2014, 2015a; Yin et al. 2015; X. Li et al. 2018). Dynamical downscaling modeling (DDM) utilizes a regional climate model (RCM) forced by a GCM to produce high-resolution climate information, and is used more extensively than statistical downscaling modeling because of the dynamical frameworks and physical processes included in RCMs (Gao et al. 2017). Therefore, DDM has been widely applied to projections of climate change, especially in complicated terrain regions (Gao et al. 2012, 2015a,b; Heikkilä et al. 2011; Ji and Kang. 2013; Zobel et al. 2018). RCMs usually have the ability to better represent the complex and rugged topography over the TP than GCMs (Figs. 1b,c).
(a) Simulation domain and topography in the DDM and different resolutions of topography over the TP for the (b) CCSM and (c) WRF-CCSM simulations (m).
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
Although computationally intensive, DDM has been used to study historical and future climate change in some regions—for instance, the North American Regional Climate Change Assessment Program (NARCCAP) in North America and the Coordinated Regional Climate Downscaling Experiment (CORDEX) in Asia (Mearns et al. 2009; Giorgi et al. 2009; Gao et al. 2011, 2012). Few simulations have focused on the TP; although, recently, a series of DDM studies by a group led by Gao (Gao et al. 2015a,b,c, 2017, 2018a) have been published. A 33-yr simulation was conducted with a 30-km horizontal resolution using the Weather Research and Forecasting Model (https://www.mmm.ucar.edu/weather-research-and-forecasting-model; Skamarock et al. 2008). The ERA-Interim dataset was chosen to force the DDM instead of three other well-known reanalysis datasets—NCEP–NCAR, NCEP–DOE, and ERA-40 (Gao et al. 2014, 2015a). We found that the DDM not only largely reduced temperature and precipitation biases, but also captured the observed elevation-dependent warming and the western-centered precipitation minus evaporation P − E changes pattern over the TP better than its forcing ERA-Interim reanalysis dataset (Gao et al. 2015a,c, 2018a).
To investigate the mechanisms of the pattern of drying and wetting in response to warming in the historical period for the TP, Gao et al. (2014, 2015a) decomposed the P − E changes into the dynamical contribution due to mean circulation changes, the thermodynamical contribution due to water vapor changes, and the transient eddy contribution. They found that the dynamical contribution dominates the P − E changes pattern over the TP from 1979 to 2011.
The projection performance of the 14 GCMs participating in phase 5 of the Coupled Model Intercomparison Project (CMIP5) was compared over the TP and its surroundings. The Centre National de Recherches Météorologiques climate model, version 5 (CNRM-CM5), and the Community Climate System Model, version 4 (CCSM4), were found to outperform the other GCMs (Xu et al. 2017). In addition, a historical simulation (1980–2005) using the WRF RCM driven by the CCSM4 model was conducted, and the impact of GCM biases on dynamical downscaling was analyzed by comparison with the same RCM driven by the ERA-Interim dataset. Future simulations using the WRF RCM driven by CCSM4 under representative concentration pathways (RCPs) 4.5 and 8.5 (2006–2100) were also conducted. Different projections were found between the DDM and CCSM4 over the TP. The CCSM4 projects a general wetting across the whole TP. However, DDM driven by the CCSM4 projects a wetting in the northern TP and a drying in the southern TP (Gao et al. 2018a). There were also differences in the nine extreme precipitation indices in the historical simulation period. However, the different P − E changes in response to warming were not analyzed.
This paper analyzes moisture flux changes over the TP in response to warming by comparing dynamic downscaling with its driving CCSM model for historical and future periods. The study aimed to determine 1) whether the P − E changes and associated mechanisms of the WRF RCM simulations driven by the CCSM model are the same as those in the WRF RCM driven by the ERA-Interim dataset over the historical period; and 2) the differences in the P − E changes and associated mechanisms between the CCSM and WRF RCM driven by the CCSM (WRF-CCSM). We follow the decomposition approach of Seager et al. (2010) and Gao et al. (2015a), and the mechanism of regional P − E change is investigated to explore the differences between the regional and global simulations. Section 2 briefly describes the data and methods. The results of historical and future projections of P − E changes are analyzed in section 3. Section 4 concludes the key findings.
2. Data and methods
The CCSM4 model was chosen as the GCM forcing for a long-term simulation by the WRF Model. CCSM is a coupled climate model for simulating Earth’s climate system, including atmosphere, ocean, land, and sea ice components. There are improvements related to higher resolution, the representation of regional topography, and the simulated sea surface temperature compared with the CCSM3 model (Cook et al. 2012). The gridded CCSM4 data have a spatial resolution of 1.25° × 0.9°. The simulated historical and future periods are 1979–2005 and 2006–2100, respectively.
The WRF Model has been widely used for regional climate predictions and dynamical downscaling simulations. The NARCCAP (Leung et al. 2006; Gao et al. 2011, 2012) used the WRF as one of multiple RCMs to produce high-resolution climate scenarios for the conterminous United States, northern Mexico, and most of Canada. The complex and rugged topography of the TP can be realistically represented in the WRF Model (Fig. 1c). In this study, WRF simulations were conducted for the historical and future periods under the RCP4.5 and RCP8.5 scenarios driven by CCSM4. The center of the simulation area is (35°N, 106°E), and the horizontal resolution is 30 km with 210 × 154 grid cells. The simulation area almost covers the whole of the East Asian region (Fig. 1a). The WRF configuration used the following physics schemes: the NCAR Community Atmospheric Model shortwave scheme and longwave scheme (Collins et al. 2004), the WRF single-moment 3-class microphysics scheme, the Grell–Devenyi ensemble convective scheme (Grell and Dévényi 2002), the Yonsei University planetary boundary layer scheme (Hong et al. 2006), and the Noah-MP land surface model (Niu et al. 2011). The historical simulation was initialized at 0000 UTC 1 January 1979 and ended at 2300 UTC 31 December 2005. The future simulation was initialized at 0000 UTC 1 January 2005 and ended at 2300 UTC 31 December 2099. More details of the configurations are described in Gao et al. (2017). The first years (1979 and 2005) of the two periods were taken as the spinup time and not used in the analysis.
The Global Land Data Assimilation System (GLDAS) dataset (Rodell et al. 2004), with a spatial resolution of 1° and temporal resolution of 3 h, was adopted as a reference for the P − E changes. GLDAS assimilates a variety of conventional data (radiosonde, buoy, ship, and airborne) and satellite-derived observations using a four-dimensional multivariate approach. Among the widely used global reanalysis products, GLDAS captures both daily and monthly precipitation and has satisfactory performance for other surface variables (Wang and Zeng 2012). It also reproduces the observed contrast in P − E changes between the northwestern and southeastern TP (Gao et al. 2014). Therefore, the GLDAS product is used in this study as the “ground truth,” although it contains uncertainties. Currently, the GLDAS product consists of four land surface models: Mosaic, Noah, the Community Land Model, and the Variable Infiltration Capacity. We used the ensemble mean of the four land surface model outputs.
To further examine moisture transport in and out of the TP in CCSM and WRF-CCSM, the ERA-Interim dataset was used as a reference in this study. ERA-Interim is an improved version of the ERA-40 dataset (Simmons et al. 2006). It ranks best among the examined reanalysis products in describing the temperature and water cycle over the TP (Gao et al. 2014). In addition, ERA-Interim has a horizontal resolution of 0.7° and is available at 6-h intervals.
For comparison, all of the datasets are regridded to the regular 1.25° × 0.9° grid in CCSM by bilinear interpolation, which is commonly applied to evaluate the dynamical downscaling method (Guo and Wang 2016; Lo et al. 2008). Mean value, bias, root-mean-square error (RMSE), and Pearson pattern correlation coefficient (CORR) are used to quantitatively evaluate the downscaling results in the historical period.
3. Results
a. Historical period evaluation
1) Climatology
For comparison, the climatologies of annual average precipitation P, evaporation E, and P − E from GLDAS, CCSM, and WRF-CCSM over the TP in the historical period from 1980 to 2005 are shown in Fig. 2. The annual average P, E, and P − E over the TP for GLDAS, CCSM, and WRF-CCSM in the historical period are listed in Table 1, which summarizes the mean values, biases, RMSEs, and CORRs between the simulations and the observations. CCSM and WRF-CCSM are able to reproduce the observed climatology according to GLDAS for annual P, E, and P − E over the TP. However, there is still a general overestimation of the P, E, and P − E values in WRF-CCSM and CCSM (Fig. 2 and Table 1). The overestimation is greatest in the southern TP, and especially in valleys. Furthermore, the overestimation in P − E mainly comes from the overestimation in P, and the overestimation of P in CCSM is greatly reduced in WRF-CCSM (Figs. 2a–c and Table 1). CCSM overestimates P, E, and P − E by 1.78, 0.57, and 1.23 mm day−1, respectively. The value of P is more substantially overestimated, with larger biases than E and P − E. However, the biases of the three variables are greatly reduced in WRF-CCSM compared with its coarse-resolution CCSM forcing (Table 1). Moreover, P, E, and P − E are redistributed in WRF-CCSM to have higher CORRs and lower RMSEs with GLDAS than CCSM. This indicates that the WRF-CCSM simulation is able to better represent the historical spatial distribution of P, E, and P − E than its coarse-resolution forcing.
Distribution of the historical climatology (1980–2005) for (a)–(c) P, (d)–(f) E, and (g)–(i) P − E over the TP for (left) GLDAS, (center) WRF-CCSM, and (right) CCSM (mm day−1).
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
Annual mean precipitation P, evaporation E, and P − E averaged in the TP for GLDAS, CCSM, and WRF-CCSM in 1980–2005 (mm day−1). Bias, root-mean-square error (RMSE), and Pearson pattern correlation coefficient (CORR) for CCSM and WRF-CCSM compared to GLDAS (mm day−1). CORRs with asterisks are statistically significant at the 95% confidence level based on the two-tailed t test.
From the seasonal-mean perspective, as shown in Fig. 3, GLDAS shows that the annual average P, E, and P − E are highest in summer (June–August) and lowest in winter (December–February). CCSM and WRF-CCSM both consistently capture the seasonal mean of the three variables over the TP in the historical period. However, there are still wet biases for all three variables in CCSM and WRF-CCSM compared with GLDAS, especially in summer for P and P − E. CCSM overestimates P, E, and P − E in summer by 3.76, 0.77, and 2.99 mm day−1, respectively, while WRF-CCSM reduces the biases to 1.43, 0.41, and 1.02 mm day−1, respectively. Compared with the coarse-resolution forcing, the WRF-CCSM simulation better captures the seasonal mean of climatological P, E, and P − E over the TP in the historical period.
Seasonal mean of (a) P, (b) E, and (c) P − E averaged over the TP for 1980–2005 for GLDAS, WRF-CCSM, and CCSM (mm day−1).
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
Figure 4 shows the spatial distribution of annual mean near-surface air temperature, latent heat flux and wind speed over the TP in GLDAS, WRF-CCSM, and CCSM. CCSM and WRF-CCSM reasonably reproduce the observed historical climatology for near-surface air temperature, latent heat flux, and wind speed, but larger biases of the three near-surface variables exist in CCSM than in WRF-CCSM; CCSM underestimates the near-surface air temperature by 3.14°C and overestimates near-surface latent heat flux and wind speed by 17.87 W m−2 and 1.72 m s−1, respectively (Table 2). WRF-CCSM better captures the observed near-surface air temperature, latent heat flux, and wind speed, with higher CORRs and lower biases and RMSEs, than CCSM. To examine moisture transport in and out of the TP, the vertically integrated water vapor flux over the TP and its surroundings in ERA-Interim are shown alongside those in WRF-CCSM and the CCSM forcing as shown in Fig. 5. Similar to ERA-Interim, CCSM and WRF-CCSM show that the majority of the moisture over the TP comes from the west, but a weaker moisture transport to the TP is simulated by CCSM and WRF-CCSM than ERA-Interim. However, the spatial correlation between WRF-CCSM and ERA-Interim is higher than between CCSM and ERA-Interim, being equal to 0.63 and 0.52, respectively. WRF-CCSM better simulates the water vapor flux over the TP and its surroundings than its CCSM forcing, indicating that WRF-CCSM may produce more realistic P. Consequently, the realistic land surface variables (i.e., near-surface air temperature, latent heat flux, and wind speed) and water vapor flux can play an important role in representing historical P − E over the TP in the WRF-CCSM simulation.
Distribution of the historical climatology for near-surface (a)–(c) air temperature (°C), (d)–(f) latent heat flux (W m−2), and (h)–(j) wind speed (m s−1) for (left) GLDAS, (center) WRF-CCSM, and (right) CCSM over the TP.
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
Annual mean near-surface air temperature (SAT), latent heat flux (SLH), and wind speed (SWS) averaged in the TP for GLDAS, CCSM, and WRF-CCSM during 1980–2005 (SAT in °C; SLH in W m−2; SWS in m s−1). Bias, RMSE, and CORR for CCSM and WRF-CCSM taken GLDAS as references. CORRs with asterisks are statistically significant at the 95% confidence level based on the two-tailed t test.
Distribution of the historical climatology for the vertically integrated water vapor flux from (a) ERA-Interim, (b) WRF-CCSM, and (c) CCSM over the TP and its surroundings.
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
2) P − E changes
Changes in P − E between the periods of 1998–2005 and 1980–97 over the TP from GLDAS are presented alongside those from CCSM and WRF-CCSM in Fig. 6. The P − E in GLDAS decreases in the southeastern TP and increases in the northwestern TP. This pattern of change was also found in ERA-Interim and most realistically reproduced in WRF-ERAI (see Figs. 2g–i; Gao et al. 2015a). Similar to ERA-Interim and WRF-ERAI, the WRF-CCSM simulation also captures the spatial pattern of P − E changes better than CCSM, with decreasing P − E in the southeastern TP and increasing P− E in the northwestern TP (Fig. 6). However, CCSM does not simulate the wetting over the northwestern TP. The CORR between GLDAS and CCSM for P − E changes is only 0.01 (lower than that between ERA-Interim and GLDAS), while it is 0.26 between GLDAS and WRF-CCSM, which passes the two-tailed t test at the 95% confidence level—even higher than WRF-ERAI (see Table 3; Gao et al. 2015a). Overall, compared with either CCSM or WRF-ERAI, WRF-CCSM better captures the spatial variations of P − E changes over the TP in the historical period.
Changes in P − E in 1998–2005 compared with 1980–97 for (a) GLDAS, (b) WRF-CCSM, and (c) CCSM over the TP (mm day−1). Positive values are marked by crosses.
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
Correlations between annual P − E changes (δP − E) and annual changes in the mean contributions of circulation (δMCD), thermodynamics (δTH), and transient eddies (δTE), averaged over the TP for 1998–2005 compared with 1980–97, for the WRF-CCSM simulation and CCSM. CORRs with asterisks are statistically significant at the 95% confidence level based on the two-tailed t test.
3) Contributions to P − E changes
Figure 7 shows the P − E changes and its three contributors (δMCD, δTH, and δTE) from the CCSM and WRF-CCSM simulations. The same analysis was also applied to analyze the mechanisms behind the changes in P − E from WRF-ERAI and ERA-Interim simulations in Gao et al. (2015a). Similar to WRF-ERAI and ERA-Interim, the changes in mean circulation dynamics (δMCD) match well with the P − E changes over the TP, indicating that δMCD is the dominant contributor to P − E changes for the historical period in WRF-CCSM and its CCSM forcing (Fig. 7). However, a more prominent contribution of δMCD to P − E changes occurs in CCSM, where the CORR is higher than in WRF-CCSM (Table 3). Similarly, unlike CCSM, it is notable that the contribution of the changes in thermodynamics (δTH) to P − E changes can also play an important role in the high-resolution simulation (Fig. 7e and Table 3 vs Fig. 7f), and the CORR of δTH with the P − E changes is 0.21, which is significant at the 95% confidence level (Table 3).
Distribution of (a),(b) P − E changes and contributions from (c),(d) mean circulation dynamics (δMCD), (e),(f) thermodynamics (δTH), and (g),(h) transient eddies (δTE), from WRF-CCSM and CCSM in 1998–2005 compared with 1980–97 (mm day−1). Positive values are marked by crosses.
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
The transient eddy changes (δTE) in CCSM and WRF-CCSM show a negative CORR with P − E changes over the TP. The CORRs between δTE and P − E changes are −0.28 and −0.38 in CCSM and WRF-CCSM, respectively, which are significant at the 95% confidence level (Table 3) and lower than the WRF simulation driven by ERA-Interim (see Table 3; Gao et al. 2015a). In addition, the WRF-CCSM-simulated δTE is negative in the central TP, where the terrain is relatively flat and positive along the Gandise, Hengduan, and Qilian Mountains in the TP where sharp gradients in terrain exist; δTE is usually associated with changes in local storms, which implies that storm activities may be stronger over rugged terrain in the TP (Figs. 1c and 7g). Therefore, relative to the coarse-resolution forcing, the complex and rugged terrain seems to play a more important role in the high-resolution dynamical downscaling simulation.
b. Future projections
Analysis of the performance of the WRF-CCSM simulation for the historical period provides confidence for projections of future changes. Precipitation and extreme precipitation indices have been analyzed in Gao et al. (2018a). In the following, we focus on the P − E changes and its three contributors in the near-term (2010–39) and long-term (2070–99) future for the CCSM and WRF-CCSM projections under the RCP4.5 and RCP8.5 scenarios compared with the historical period (1980–2005).
1) P − E changes
Figure 8 shows the spatial distribution of P − E changes over the TP for the near-term and long-term future under the RCP4.5 and RCP8.5 scenarios compared with the historical period. In the near-term future, CCSM projects slightly increasing P − E under the two scenarios, especially in the southeastern region (Figs. 8c,d). WRF-CCSM projects a larger spatial variation of P − E changes over the TP, with an increase in the north and a decrease in the south (Figs. 8a,b). That is, the CCSM and WRF-CCSM simulations project different signs of P − E changes across the TP, with the exception that both project increased P − E changes somewhere in the southeast.
Changes in P − E from WRF-CCSM and CCSM in (a)–(d) the near-term (2010–39) and (e)–(h) the long-term (2070–99) future under the RCP4.5 and RCP8.5 scenarios compared with the historical period of 1980–2005 (mm day−1). Black dots denote changes that passed the two-tailed t test with the statistically significant confidence level of 95%.
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
In the long-term future, CCSM projects increases in P − E in the southeast, which are much larger than for the near-term future (Figs. 8g,h). WRF-CCSM continues to project opposite changes in the northern and southern TP, and this is more apparent than for the near-term future (Figs. 8e,f), and for the RCP8.5 scenario than the RCP4.5 scenario. Therefore, the P − E changes projected by CCSM and WRF-CCSM are more significant for the long-term future and for the RCP8.5 scenario.
2) Contributions to P − E changes
Figure 9 shows the three contributors (δMCD, δTH, and δTE) to the P − E changes for the near-term future compared with 1980–2005 simulated by CCSM and WRF-CCSM under the RCP4.5 and RCP8.5 scenarios. Table 4 summarizes the three contributors averaged over the TP. In CCSM, δTH is the dominant contributor to P − E changes, while δMCD is the dominant mechanism in the historical period. The CORRs between TH changes and P − E changes are 0.65 and 0.6 under scenarios RCP4.5 and RCP8.5, respectively, which are significant at the 95% confidence level (Table 4). Unlike CCSM, WRF-CCSM has different mechanisms for the near-term future, with δMCD contributing the most to P − E changes, positively in the northwestern TP and negatively in the southeastern TP, which is consistent with the dominant mechanism in the historical period. In WRF-CCSM, the CORRs between MCD changes and P − E changes are 0.72 and 0.23 under the RCP4.5 and RCP8.5 scenarios, respectively, which are significant at the 95% confidence level (Table 4).
Changes in P − E contributions from (a)–(d) mean circulation dynamics (δMCD), (e)–(h) thermodynamics (δTH), and (i)–(l) transient eddies (δTE), for WRF-CCSM and CCSM in the near-term future (2010–39) under the RCP4.5 and RCP8.5 scenarios compared with the historical period of 1980–2005 (mm day−1). Positive values are marked by crosses.
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
Correlations between future annual P − E changes (δP − E) and annual changes in the mean contributions of circulation (δMCD), thermodynamics (δTH), and transient eddies (δTE), averaged over the TP in the near-term (2010–39) and long-term (2070–99) future compared with 1980–2005, for the WRF-CCSM simulation and CCSM under the RCP4.5 and RCP8.5 scenarios. Values with asterisks are statistically significant at the 95% confidence level based on the two-tailed t test.
In the long-term future, CCSM still shows that δTH is the dominant contributor to P − E changes, and the contribution is larger than in the near-term future under RCP4.5 and RCP8.5 (Figs. 10g,h). The CORRs between TH changes and P − E changes are 0.59 and 0.56 under the two scenarios, respectively, which are significant at the 95% confidence level (Table 4). However, WRF-CCSM still shows that δMCD dominates the P − E changes in the long-term future (Figs. 10a,b). The CORRs between MCD changes and P − E changes are 0.66 and 0.35 under the two scenarios in WRF-CCSM, respectively, which are significant at the 95% confidence level (Table 4).
As in Fig. 9, but for the long-term future (2070–99).
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
In summary, CCSM projects that thermodynamic changes (δTH) will be the dominant contributor to P − E changes in the near-term and the long-term future under RCP4.5 and RCP8.5, which is inconsistent with the dominant mechanism in the historical period. Unlike CCSM, WRF-CCSM presents a continuation from the historical period of changes in mean circulation dynamics (δMCD) being the dominant contributor to P − E changes for the two scenarios, as well as the near-term and long-term future. The different predominant contributors to P − E changes between CCSM and WRF-CCSM indicate their different mechanisms. The radiative forcing is directly input in CCSM, which approximately equals 4.5 and 8.5 W m−2 under RCP4.5 and RCP8.5, respectively (Moss et al. 2010), enhancing the contribution of δTH to P − E changes. Therefore, δTH contributes the most to P − E changes, with a small contribution from the dynamic component, especially in the long-term future under RCP8.5. Compared with the coarse-resolution CCSM forcing, the radiative forcing is passing through the forcing indirectly in the dynamical downscaling using WRF. The topography and land surface processes over the TP play a more critical role due to the strong heating effects on the atmospheric circulation from a vast area at exceptionally high elevations (Gao et al. 2017), enhancing the contribution of the dynamic component (δMCD) to P − E changes, hence δMCD contributes the most to P − E changes, with a small contribution from the thermodynamic component, especially in the near-term future under RCP4.5.
3) Contributions of changes in advection and convergence of moisture
Figure 11 shows the changes in P − E contributions from these four terms projected by CCSM and WRF-CCSM in the long-term future under RCP8.5 compared with the historical period. Table 5 summarizes the CORRs between the advection terms (δMCDA and δTHA) and the convergence or divergence terms (δMCDD and δTHD) with the changes in mean circulation and thermodynamics (δMCD and δTH), respectively. Given the high CORR between δP − E and δTH (Table 4) and the high CORR between δTH and δTHD (Table 5) in CCSM, δTHD provides moisture for almost the whole TP, except for Qaidam basin and the Himalayas (Fig. 11h). It is notable that changes in P − E are mainly attributable to the moisture changes of the thermodynamic component in CCSM (Figs. 8h, 10h, and 11h). However, the WRF projects δMCDD moistening in the northern TP and drying in the southern TP as well as a dominant contribution to δMCD (Fig. 11c and Table 5), and it is clear that circulation convergence or divergence changes contributes predominantly to P − E changes in the WRF-CCSM simulation.
Changes in P − E contributions from (a)–(d) mean circulation dynamics and (e)–(h) thermodynamics due to advection [in (a), (b), (e), and (f)] and convergence or divergence [in (c), (d), (g), and (h)], from WRF-CCSM and CCSM in the long-term future (2070–99) under the RCP8.5 scenario compared with the historical period 1980–2005 (mm day−1). Positive values are marked by crosses.
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
Correlations between annual changes in the mean contributions of circulation/thermodynamics (δMCD/δTH) and annual mean circulation dynamics/thermodynamics due to the advection of moisture (δMCDA/δTHA) and convergence or divergence of moisture (δMCDD/δTHD), in the long-term future (2070–99) compared with 1980–2005, for the WRF-CCSM simulation and CCSM under the RCP8.5 scenario. Values with asterisks are statistically significant at the 95% confidence level based on the two-tailed t test.
4) Responses to warming
Figures 12a–d shows the relative changes of P, E, P − E, and top-layer soil moisture projected by CCSM and WRF-CCSM for the near-term, medium-term, and long-term future under different forcing scenarios compared with the historical period, versus vertical averaged temperature changes averaged over the TP, respectively. Compared with the historical period, WRF-CCSM projects 1.1°, 1.73°, and 2.04°C increases under the RCP4.5 scenario and 1.14°, 2.6°, and 4.22°C under the RCP8.5 scenario for the near-term, medium-term, and long-term future, respectively. However, CCSM projects temperature increases 1.31°, 1.98°, and 2.44°C under the RCP4.5 scenario and 1.32°, 3°, and 4.58°C under the RCP8.5 scenario for the near-term, medium-term, and long-term future, respectively. Thus, the temperature changes projected in WRF-CCSM are slightly smaller than in its CCSM forcing (Fig. 12a). The P − E projected by CCSM and WRF-CCSM will increase in response to warming. According to the Clausius–Clapeyron (CC) relation, under the condition that the relative humidity in the lower troposphere stays constant, the water vapor content increases at a rate of about 7% K−1 of warming (Held and Soden 2006). However, the global average terrestrial precipitation and P − E show and project a slower increase rate of about 1%–3% K−1 (Wentz et al. 2007) and 2.7% K−1 of warming (Byrne and O’Gorman 2015). In this study, P − E increases at a rate of 5% K−1 in CCSM averaged over the TP. However, it is 2% K−1 in WRF-CCSM—half that in CCSM (Fig. 12a). This indicates that the P − E changes projected by WRF-CCSM are less sensitive to warming than those projected by the CCSM forcing over the TP. The lower sensitivity of P − E to warming in high-resolution dynamical downscaling than in GCMs is consistent with results for the Rocky Mountains (Gao et al. 2012). Moreover, the reason for the area-averaged P − E changes do not scale with the CC relation is due to circulation-induced changes to P − E changes playing a dominant role in WRF-CCSM—not just thermodynamic-induced P − E changes. Furthermore, we also explored the changes in P and E. CCSM and WRF-CCSM project 3%–4% K−1 changes in P due to warming in the future (Fig. 12b). However, E projections are 5% K−1 in WRF-CCSM, but only 2% K−1 in CCSM (Fig. 12c). The notably stronger E sensitivity to warming weakens the P − E changes and results in less sensitive P − E changes to warming in WRF-CCSM than CCSM.
Scatterplots of future relative changes in the annual mean (a) P − E, (b) P, (c) E, and (d) top-layer (0–10 cm) soil moisture averaged over the TP vs vertical averaged temperature changes compared with the historical period (1980–2005) for the near-term (2010–39), medium-term (2040–69), and long-term (2070–99) future under the RCP4.5 and RCP8.5 scenarios (%). Land cover types over the TP for the (e) WRF-CCSM and (f) CCSM simulations. The black lines are the regression lines for CCSM and WRF-CCSM, respectively.
Citation: Journal of Climate 32, 20; 10.1175/JCLI-D-19-0200.1
4. Summaries and discussions
The performance of dynamic downscaling using the WRF Model driven by CCSM was evaluated for P, E, and P − E in the historical period over the TP against the GLDAS dataset. Then, P − E changes and the associated mechanisms between CCSM and WRF-CCSM were compared with a WRF-ERAI and ERA-Interim pair for the historical period. Finally, P − E changes and the responsible mechanisms projected by CCSM and WRF-CCSM were analyzed for the near-term to the long-term future under the RCP4.5 and RCP8.5 scenarios. The following results were obtained:
Similar to WRF-ERAI and ERA-Interim, CCSM and WRF-CCSM overestimate the annual average and seasonal P, E, and P − E over the TP. Compared with its coarse-resolution forcing, WRF-CCSM better reproduces the historical spatial pattern and seasonal mean of annual average P, E, and P − E over the TP, with lower biases and RMSEs and higher CORRs. The overestimation of P − E in CCSM is greatest in the southern TP, which is greatly reduced in WRF-CCSM. Furthermore, WRF-CCSM also better reproduces the spatial distribution of P − E changes over the historical period than CCSM and WRF-ERAI.
CCSM projects a uniform increase in P − E over the TP for the near-term to long-term future under the RCP4.5 and RCP8.5 scenarios. However, WRF-CCSM projects spatial contrast of P − E changes, with an increase in the northern TP and decrease in the southern TP. Moreover, WRF-CCSM projects less sensitivity of P − E changes to warming than CCSM, which is mostly attributable to the weaker contribution of δTH to P − E changes in WRF-CCSM than CCSM.
Similar to WRF-ERAI and ERA-Interim, the dynamic component dominates P − E changes for the historical period in both CCSM and WRF-CCSM. However, CCSM projects that the thermodynamic component will dominate P − E changes in the near-term to the long-term future, and for both the RCP4.5 and RCP8.5 scenarios. WRF-CCSM reproduces the mechanism from the historical period, with the dynamic component being the dominant contribution to P − E changes.
The different predominant contributors to future P − E changes between CCSM and WRF-CCSM indicate their different mechanisms. The radiative forcing is directly input in CCSM, enhancing the contribution of δTH to P − E changes. However, in WRF-CCSM, the radiative forcing is passing through its forcing indirectly, the topography and land surface processes over the TP play a more critical role in WRF due to the strong heating effects on the atmospheric circulation from a vast area at exceptionally high elevations, enhancing the contribution of δMCD to P − E changes.
These differences in future E changes in response to warming are correspondingly related to the top-layer (0–10 cm) soil moisture responses due to land cover differences. Land cover types in the TP are dominated by grassland and shrubland in WRF-CCSM (Fig. 12e); however, sparsely vegetated land dominates in CCSM (Fig. 12f). Land cover types in WRF-CCSM are more realistic than CCSM compared to the vegetation map (Zhang 2007; R. Li et al. 2018). Grassland and shrubland have a stronger capacity for holding water than sparsely vegetated land. This feature of vegetated lands results in higher soil moisture responses in WRF_CCSM than the sparsely vegetated lands in CCSM (Fig. 12d).
This study suggests that the limitations of GCMs in terms of topography and its effects on moisture flux change have important implications for investigating the mechanisms of the dryness and wetness change pattern in response to warming in the future. This is crucial in the TP with its complex terrain and land surface processes. Larger differences are found in moisture flux projection over the TP between the coarse-resolution GCM and the high-resolution dynamical downscaling. Moreover, only one GCM as forcing and one RCM are used here because of computational constraints. More RCM–GCM pairings are necessary to analyze whether the conclusions of this study are valid. This work will be conducted in a subsequent investigation.
Acknowledgments
We appreciate the free access to the CMIP5 datasets, which is provided by the ESGF web portal (http://pcmdi9.llnl.gov/esgf-web-fe/). The GLDAS data are archived and distributed by NASA’s Goddard Earth Sciences (GES) Data and Information Services Center (DISC) and acquired through the Distributed Active Archive Center (DAAC). This research was jointly supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (XDA2006010202), the National Natural Science Foundation of China (91537105, 91537211) and Key Laboratory of Land Surface Process and Climate Change in Cold and Arid Regions, Chinese Academy of Sciences (Grant LPCC2018090). We thank the Super-Computing Center of the Chinese Academy of Sciences for computing the simulations.
REFERENCES
Benestad, R. E., I. Hanssen-Bauer, and D. Chen, 2008: Empirical–Statistical Downscaling. World Scientific, 300 pp.
Byrne, M. P., and P. A. O’Gorman, 2015: The response of precipitation minus evapotranspiration to climate warming: Why the “wet-get-wetter, dry-get-drier” scaling does not hold over land. J. Climate, 28, 8078–8092, https://doi.org/10.1175/JCLI-D-15-0369.1.
Chen, D., and Y. Chen, 2003: Association between winter temperature in China and upper air circulation over East Asia revealed by canonical correlation analysis. Global Planet. Change, 37, 315–325, https://doi.org/10.1016/S0921-8181(02)00206-0.
Collins, W. D., and Coauthors, 2004: Description of the NCAR Community Atmosphere Model (CAM 3.0). NCAR Tech. Note NCAR/TN-464+STR, 226 pp., https://doi.org/10.5065/D63N21CH.
Cook, K. H., G. A. Meehl, and J. M. Arblaster, 2012: Monsoon regimes and processes in CCSM4. Part II: African and American monsoon systems. J. Climate, 25, 2609–2621, https://doi.org/10.1175/JCLI-D-11-00185.1.
Cuo, L., and Y. Zhang, 2017: Spatial patterns of wet season precipitation vertical gradients on the Tibetan Plateau and the surroundings. Sci. Rep., 7, 5057, https://doi.org/10.1038/s41598-017-05345-6.
Dai, A., 2006: Precipitation characteristics in eighteen coupled climate models. J. Climate, 19, 4605–4630, https://doi.org/10.1175/JCLI3884.1.
Dai, A., and C. E. Bloecker, 2019: Impacts of internal variability on temperature and precipitation trends in large ensemble simulations by two climate models. Climate Dyn., 52, 289–306, https://doi.org/10.1007/s00382-018-4132-4.
Dai, A., R. M. Rasmussen, C. Liu, and K. Ikeda, 2019: A new mechanism for warm-season precipitation response to global warming based on convection-permitting simulations. Climate Dyn., https://doi.org/10.1007/S00382-017-3787-6, in press.
Dai, Y.-F., Y.-M. Liu, and J.-M. Jin, 2014: Statistical downscaling of FGOALS-s2 projected precipitation in Eastern China. Atmos. Ocean. Sci. Lett., 7, 388–394, https://doi.org/10.3878/j.issn.1674-2834.14.0025.
Flato, G., and Coauthors, 2013: Evaluation of climate models. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 741–866, http://www.climatechange2013.org/images/report/WG1AR5_Chapter09_FINAL.pdf.
Gao, Y., J. A. Vano, C. Zhu, and D. P. Lettenmaier, 2011: Evaluating climate change over the Colorado River basin using regional climate models. J. Geophys. Res., 116, D13104, https://doi.org/10.1029/2010JD015278.
Gao, Y., L. R. Leung, E. P. Salathé, F. Dominguez, B. Nijssen, and D. P. Lettenmaier, 2012: Moisture flux convergence in regional and global climate models: Implications for droughts in the southwestern United States under climate change. Geophys. Res. Lett., 39, L09711, https://doi.org/10.1029/2012GL051560.
Gao, Y., L. Cuo, and Y. Zhang, 2014: Changes in moisture flux over the Tibetan Plateau during 1979–2011 and possible mechanisms. J. Climate, 27, 1876–1893, https://doi.org/10.1175/JCLI-D-13-00321.1.
Gao, Y., L. R. Leung, Y. Zhang, and L. Cuo, 2015a: Changes in moisture flux over the Tibetan Plateau during 1979–2011: Insights from a high-resolution simulation. J. Climate, 28, 4185–4197, https://doi.org/10.1175/JCLI-D-14-00581.1.
Gao, Y., X. Li, L. R. Leung, D. Chen, and J. Xu, 2015b: Aridity changes in the Tibetan Plateau in a warming climate. Environ. Res. Lett., 10, 034013, https://doi.org/10.1088/1748-9326/10/3/034013.
Gao, Y., J. Xu, and D. Chen, 2015c: Evaluation of WRF mesoscale climate simulations over the Tibetan Plateau during 1979–2011. J. Climate, 28, 2823–2841, https://doi.org/10.1175/JCLI-D-14-00300.1.
Gao, Y., L. Xiao, D. Chen, F. Chen, J. Xu, and Y. Xu, 2017: Quantification of the relative role of land-surface processes and large-scale forcing in dynamic downscaling over the Tibetan Plateau. Climate Dyn., 48, 1705–1721, https://doi.org/10.1007/s00382-016-3168-6.
Gao, Y., L. Xiao, D. Chen, J. Xu, and H. Zhang, 2018a: Comparison between past and future extreme precipitations simulated by global and regional climate models over the Tibetan Plateau. Int. J. Climatol., 38, 1285–1297, https://doi.org/10.1002/joc.5243.
Gao, Y., F. Chen, D. P. Lettenmaier, J. Xu, L. Xiao, and X. Li, 2018b: Does elevation-dependent warming hold true above 5000 m elevation? Lessons from the Tibetan Plateau. npj Climate Atmos. Sci., 1, 19, https://doi.org/10.1038/S41612-018-0030-Z.
Giorgi, F., C. Jones, and G. R. Asrar, 2009: Addressing climate information needs at the regional level: The CORDEX framework. WMO Bull., 58, 175–183, http://wcrp.ipsl.jussieu.fr/RCD_Projects/CORDEX/CORDEX_giorgi_WMO.pdf.
Grell, G. A., and D. Dévényi, 2002: A generalized approach to parameterizing convection combining ensemble and data assimilation techniques. Geophys. Res. Lett., 29, 1693, https://doi.org/10.1029/2002GL015311.
Gu, G., and R. F. Adler, 2013: Interdecadal variability/long-term changes in global precipitation patterns during the past three decades: Global warming and/or Pacific decadal variability? Climate Dyn., 40, 3009–3022, https://doi.org/10.1007/s00382-012-1443-8.
Guo, D., and H. Wang, 2016: Comparison of a very-fine-resolution GCM with RCM dynamical downscaling in simulating climate in China. Adv. Atmos. Sci., 33, 559–570, https://doi.org/10.1007/s00376-015-5147-y.
Guo, D., H. Wang, and D. Li, 2012: A projection of permafrost degradation on the Tibetan Plateau during the 21st century. J. Geophys. Res., 117, D05106, https://doi.org/10.1029/2011JC007292.
Han, J., and J. O. Roads, 2004: U.S. climate sensitivity simulated with the NCEP regional spectral model. Climatic Change, 62, 115–154, https://doi.org/10.1023/B:CLIM.0000013675.66917.15.
Heikkilä, U., A. D. Sandvik, and A. Sorterberg, 2011: Dynamical downscaling or ERA-40 in complex terrain using WRF regional climate model. Climate Dyn., 37, 1551–1564, https://doi.org/10.1007/s00382-010-0928-6.
Held, I. M., and B. J. Soden, 2006: Robust responses of the hydrologic cycle to global warming. J. Climate, 19, 5686–5699, https://doi.org/10.1175/JCLI3990.1.
Hong, S.-Y., Y. Noh, and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 2318–2341, https://doi.org/10.1175/MWR3199.1.
Immerzeel, W. W., L. P. H. Van Beek, and M. F. P. Bierkens, 2010: Climate change will affect the Asian water towers. Science, 328, 1382–1385, https://doi.org/10.1126/science.1183188.
Immerzeel, W. W., F. Pellicciotti, and M. F. P. Bierkens, 2013: Rising river flows throughout the twenty-first century in two Himalayan glacierized watersheds. Nat. Geosci., 6, 742–745, https://doi.org/10.1038/ngeo1896.
Ji, Z., and S. Kang, 2013: Double-nested dynamical downscaling experiments over the Tibetan Plateau and their projection of climate change under two RCP scenarios. J. Atmos. Sci., 70, 1278–1290, https://doi.org/10.1175/JAS-D-12-0155.1.
Kääb, A., E. Berthier, C. Nuth, J. Gardelle, and Y. Arnaud, 2012: Contrasting patterns of early twenty-first-century glacier mass change in the Himalayas. Nature, 488, 495–498, https://doi.org/10.1038/nature11324.
Kim, J.-W., J.-T. Chang, N. L. Baker, D. S. Wilks, and W. L. Gates, 1984: The statistical problem of climate inversion: Determination of the relationship between local and large-scale climate. Mon. Wea. Rev., 112, 2069–2077, https://doi.org/10.1175/1520-0493(1984)112<2069:TSPOCI>2.0.CO;2.
Leung, L. R., Y. H. Kuo, and J. Tribbia, 2006: Research needs and directions of regional climate modeling using WRF and CCSM. Bull. Amer. Meteor. Soc., 87, 1747–1751, https://doi.org/10.1175/BAMS-87-12-1747.
Li, R., Y. Gao, D. Chen, Y. Zhang, and S. Li, 2018: Contrasting vegetation changes in dry and humid regions of the Tibetan Plateau over recent decades. Sci. Cold Arid. Reg., 10, 482–492.
Li, X., Y. Gao, Y. Pan, and Y. Xu, 2018a: Evaluation of near-surface wind speed simulations over the Tibetan Plateau from three dynamical downscalings based on WRF model. Theor. Appl. Climatol., 134, 1399–1411, https://doi.org/10.1007/s00704-017-2353-9.
Liang, X., J. Pan, J. Zhu, K. E. Kunkel, and J. X. L. Wang, 2006: Regional climate model downscaling of the U.S. summer climate and future change. J. Geophys. Res., 111, D10108, https://doi.org/10.1029/2005JD006685.
Liu, X., and B. Chen, 2000: Climatic warming in the Tibetan Plateau during recent decades. Int. J. Climatol., 20, 1729–1742, https://doi.org/10.1002/1097-0088(20001130)20:14<1729::AID-JOC556>3.0.CO;2-Y.
Lo, J. C. F., Z. Yang, and R. A. Pielke, 2008: Assessment of three dynamical climate downscaling methods using the Weather Research and Forecasting (WRF) model. J. Geophys. Res., 113, D09112, https://doi.org/10.1029/2007JD009216.
Mearns, L. O., W. Gutowski, R. Jones, R. Leung, S. McGinnis, A. Nunes, and Y. Qian, 2009: A regional climate change assessment program for North America. Eos, Trans. Amer. Geophys. Union, 90, 311–311, https://doi.org/10.1029/2009EO360002.
Moss, R. H., and Coauthors, 2010: The next generation of scenarios for climate change research and assessment. Nature, 463, 747–756, https://doi.org/10.1038/nature08823.
Niu, G., Z. Yang, K. E. Mitchell, F. Chen, M. B. Ek, M. Tewari, and Y. Xia, 2011: The community Noah land surface model with multiparameterization options (Noah-MP): 1. Model description and evaluation with local-scale measurements. J. Geophys. Res., 116, D12109, https://doi.org/10.1029/2010JD015139.
Rangwala, I., E. Sinsky, and J. R. Miller, 2013: Amplified warming projections for high altitude regions of the northern hemisphere mid-latitudes from CMIP5 models. Environ. Res. Lett., 8, 024040, https://doi.org/10.1088/1748-9326/8/2/024040.
Rodell, M., and Coauthors, 2004: The Global Land Data Assimilation System. Bull. Amer. Meteor. Soc., 85, 381–394, https://doi.org/10.1175/BAMS-85-3-381.
Seager, R., and Coauthors, 2007: Model projections of an imminent transition to a more arid climate in southwestern North America. Science, 316, 1181–1184, https://doi.org/10.1126/science.1139601.
Seager, R., N. Naik, and G. A. Vecchi, 2010: Thermodynamic and dynamic mechanisms for large-scale changes in the hydrological cycle in response to global warming. J. Climate, 23, 4651–4668, https://doi.org/10.1175/2010JCLI3655.1.
Seager, R., and Coauthors, 2014: Dynamical and thermodynamical causes of large-scale changes in the hydrological cycle over North America in response to global warming. J. Climate, 27, 7921–7948, https://doi.org/10.1175/JCLI-D-14-00153.1.
Simmons, A., S. Uppala, D. Dee, and S. Kobayashi, 2006: ERA-Interim: New ECMWF reanalysis products from 1989 onwards. ECMWF Newsletter, No. 110, ECMWF, Reading, United Kingdom, 25–35.
Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note, TN-475+STR, 113 pp., https://doi.org/10.5065/D68S4MVH.
Su, F., X. Duan, D. Chen, Z. Hao, and L. Cuo, 2013: Evaluation of the global climate models in the CMIP5 over the Tibetan Plateau. J. Climate, 26, 3187–3208, https://doi.org/10.1175/JCLI-D-12-00321.1.
Wang, A., and X. Zeng, 2012: Evaluation of multireanalysis products with in situ observations over the Tibetan Plateau. J. Geophys. Res., 117, 214–221, https://doi.org/10.1029/2011JD016553.
Wentz, F. J., L. Ricciardulli, K. Hilburn, and C. Mears, 2007: How much more rain will global warming bring? Science, 317, 233–235, https://doi.org/10.1126/science.1140746.
Wetterhall, F., A. Bádossy, D. Chen, S. Halldin, and C. Y. Xu, 2006: Daily precipitation-downscaling techniques in three Chinese regions. Water Resour. Res., 42, W11423, https://doi.org/10.1029/2005WR004573.
Xu, J., Y. Gao, D. Chen, L. Xiao, and T. Ou, 2017: Evaluation of global climate models for downscaling applications centered over the Tibetan Plateau. Int. J. Climatol., 37, 657–671, https://doi.org/10.1002/joc.4731.
Xu, Y., and C. Xu, 2012: Preliminary assessment of simulations of climate changes over China by CMIP5 multi-models. Atmos. Oceanic Sci. Lett., 5, 489–494, https://doi.org/10.1080/16742834.2012.11447041.
Xue, Y., Z. Janjic, J. Dudhia, R. Vasic, and F. De Sales, 2014: A review on regional dynamical downscaling in intraseasonal to seasonal simulation/prediction and major factors that affect downscaling ability. Atmos. Res., 147–148, 68–85, https://doi.org/10.1016/j.atmosres.2014.05.001.
Yang, K., H. Wu, J. Qin, C. Lin, W. Tang, and Y. Chen, 2014: Recent climate changes over the Tibetan Plateau and their impacts on energy and water cycle: A review. Global Planet. Change, 112, 79–91, https://doi.org/10.1016/j.gloplacha.2013.12.001.
Yao, T., J. Pu, A. Lu, Y. Wang, and W. Yu, 2007: Recent glacial retreat and its impact on hydrological processes on the Tibetan Plateau, China, and surrounding regions. Arct. Antarct. Alp. Res., 39, 642–650, https://doi.org/10.1657/1523-0430(07-510)[YAO]2.0.CO;2.
Yao, T., and Coauthors, 2012: Third Pole Environment (TPE). Environ. Dev., 3, 52–64, https://doi.org/10.1016/j.envdev.2012.04.002.
Yin, H., M. G. Donat, L. V. Alexander, and Y. Sun, 2015: Multi-dataset comparison of gridded observed temperature and precipitation extremes over China. Int. J. Climatol., 35, 2809–2827, https://doi.org/10.1002/joc.4174.
You, Q., and Coauthors, 2011: Changes in daily climate extremes in China and their connection to the large scale atmospheric circulation during 1961–2003. Climate Dyn., 36, 2399–2417, https://doi.org/10.1007/s00382-009-0735-0.
Zhang, W., T. Zhou, and L. Zhang, 2017: Wetting and greening Tibetan Plateau in early summer in recent decades. J. Geophys. Res., 122, 5808–5822, https://doi.org/10.1002/2017JD026468.
Zhang, X., 2007: Vegetation Map of the People's Republic of China (1:1000000). Geological Publishing House, 1270 pp.
Zobel, Z., J. Wang, D. J. Wuebbles, and V. R. Kotamarthi, 2018: Evaluations of high-resolution dynamically downscaled ensembles over the contiguous United States. Climate Dyn., 50, 863–884, https://doi.org/10.1007/s00382-017-3645-6.