Studies on wind have focused mainly on NSWS. NSWS has decreased by 5%–15% in the midlatitudes of the Northern Hemisphere from 1979 to 2008 (Vautard et al., 2010). The largest decrease was observed in central Asia (−0.16 m s−1 decade−1), followed by East Asia (−0.12 m s−1 decade−1) and Europe (−0.09 m s−1 decade−1), and the smallest decrease was observed in North America (−0.07 m s−1 decade−1; Vautard et al., 2010). Roderick et al. (2007) referred to this reduction in NSWS as “stilling.” At the regional scale, a slowdown of NSWS was also observed. NSWS in the Great Plains of the United States decreased by nearly 20% during the spring of 1971–2000 (Pryor et al., 2009; Green et al., 2012). The downward trend in NSWS over Australia reached −0.17 m s−1 decade−1 in 1975–2006 (McVicar et al., 2008; Troccoli et al., 2012). In Europe, NSWS in Switzerland decreased by −0.25 and −0.05 m s−1 decade−1 during the winter and summer of 1960–2006, respectively; and the rate of decrease in NSWS increased with the increasing altitude (McVicar et al., 2010). A decreasing trend in NSWS was also reported in southern and central France over 1974–2002 (Najac et al., 2011) and in Turkey during 1975–2006 (Dadaser-Celik and Cengiz, 2014). In addition to the overall reduction in NSWS, the probabilities of the occurrence of strong winds also showed decreasing trends. For instance, strong winds in the Netherlands decreased by 10% from 1910 to 2010 (Cusack, 2013). The frequency of extreme winds in Spain and Portugal decreased by 1.5 day yr−1 during 1961–2014 (Azorin-Molina et al., 2016). The 90th percentile of NSWS showed a decreasing trend in England from 1980 to 2010 (Earl et al., 2013). The trends of regional NSWS during different study periods are summarized in Fig. 1. Most studies present a reduced NSWS at the regional scale, although the study periods and regions are not consistent with one another (McVicar et al., 2012a). The trends in NSWS are not consistent across different study periods and regions, so the differences of the long-term changes in NSWS over different regions are considerable (Wu et al., 2018a).
Figure 1. Change trends of near-surface wind speed (NSWS) over different regions (sites) of the globe based on a review and synthesis of 148 regional studies (from McVicar et al., 2012a, with study number defined in their Table 2).
A stilling phenomenon has been observed in China (Li et al., 2004; Ren et al., 2005; Xu et al., 2006; Jiang et al., 2010a; Zha et al., 2017a). The relative change in magnitude of the annual mean NSWS decreased by more than 20% in most regions of China from 1966 to 2011, and strong reduction was found over Northwest China, Songhuajiang River, Yangtze River, and the river basins of Southeast China, which reached up to 80% (Fig. 2a; Liu et al., 2014). The trends of NSWS are sensi-tive to the selected time periods (Fu et al., 2011). The magnitudes of decreasing trends in NSWS vary among different studies because different periods, datasets, and methodologies are used. From 1951 to 2000, NSWS decreased by −0.11 m s−1 decade−1 based on the observed data (Wang et al., 2004). For 1956–2004, the annual mean NSWS decreased by −0.12 m s−1 decade−1 (Jiang et al., 2010a). From 1969 to 2005, the decreasing trend of the annual NSWS was −0.18 m s−1 decade−1 (Guo et al., 2011). Several studies discovered that the decreasing trend in the annual mean NSWS was −0.07 m s−1 decade−1 during 1960–2007 (Yin et al., 2010a, b; Liu et al., 2011a, b). The results concerning the trends in NSWS over China for different regions and time periods are summarized in Fig. 2b, with detailed information given in Table 1. Most studies have found a stilling in many regions of China in the past 40 years, although the magnitudes of the weakening NSWS trends vary in different studies.
Figure 2. (a) Distribution of relative change in magnitude (%) of the annual mean NSWS from 1966 to 2011 in China, overlapped with major river basins/regions of China (NW: Northwest China; YR: Yellow River; HAI: Haihe River; LR: Liao River; SHJ: Songhuajiang River; SW: Southwest China; CJ: Yangtze River; HUAI: Huai River; PR: Pearl River; SE: Southeast China). (b) Trends of observational NSWS (m s−1 decade−1) across different regions in China. The study number is given in Table 1. Panel (a) is taken from Liu et al. (2014).
No. Region (site) Period Trend Reference No. Region (site) Period Trend Reference 1 China (729) 1954–2000 −0.11 Wang et al. (2004) 21 Southwest China (110) 1969–2009 −0.24 Yang et al. (2012) 2 China (323) 1951–2002 −0.10 Ren et al. (2005) 22 Southwest China (110) 1969–2000 −0.37 Yang et al. (2012) 3 China (305) 1969–2000 −0.22 Xu et al. (2006) 23 China (540) 1971–2007 −0.17 Chen et al. (2013) 4 TP (75) 1966–2003 −0.17 Zhang et al. (2007) 24 China (472) 1960–2009 −0.10 Lin et al. (2013) 5 China (604) 1960–1999 −0.12 Li Y. et al. (2008) 25 TP (64) 1960–2009 −0.06 Lin et al. (2013) 6 Western deserts (23) 1973–2003 −0.29 Mahowald et al. (2009) 26 China (741) 1966–2011 −0.16 Liu et al. (2014) 7 HRB (45) 1957–2001 −0.14 Zheng et al. (2009) 27 East China (93) 1980–2011 −0.13 Wu et al. (2016) 8 China (317) 1956–2005 −0.11 Cong et al. (2009) 28 East China (93) 1980–2011 −0.13 Zha et al. (2016) 9 China (535) 1956–2004 −0.12 Jiang et al. (2010b) 29 Xinjiang (10) 1984–2013 −0.29 Liu et al. (2017) 10 Loess Plateau (82) 1960–2006 −0.14 McVicar et al. (2010) 30 East China (93) 1980–2011 −0.13 Wu et al. (2017) 11 China (595) 1961–2008 −0.09 Yin et al. (2010a) 31 China (492) 1979–2010 −0.11 Zha et al. (2017a) 12 China (603) 1971–2008 −0.12 Yin et al. (2010b) 32 China (580) 1970–2011 −0.15 Zha et al. (2017b) 13 TP (71) 1980–2005 −0.24 You et al. (2010) 33 BTH (154) 1978–2014 −0.10 Zhou et al. (2017) 14 China (597) 1961–2007 −0.13 Fu et al. (2011) 34 North China (155) 1971–2015 −0.17 Han et al. (2018) 15 China (726) 1969–2005 −0.18 Guo et al. (2011) 35 YRD (128) 1960–2015 −0.065 Li Y. J. et al. (2018) 16 China (518) 1960–1991 −0.12 Liu et al. (2011a) 36 East China (328) 1981–2011 −0.09 Zha et al. (2019a) 17 China (518) 1992–2007 −0.07 Liu et al. (2011b) 37 China (524) 1958–2015 −0.109 Zhang R. H. et al. (2019) 18 HRB (34) 1950–2007 −0.10 Tang et al. (2011) 38 BTH (223) 1980–2016 −0.25 Wang et al. (2020) 19 TP (78) 1984–2006 −0.30 Yang et al. (2011) 39 China (582) 1980–2017 −0.10 Zhang Y. et al. (2020) 20 Northeast China (87) 1961–2010 −0.25 Jin et al. (2012) 40 China (2333) 1960–2017 −0.15 Zhang and Wang (2020)
Table 1. A summary of observed NSWS trends (m s−1 decade−1) in China and its sub-regions over varied time periods, derived from different source studies/references. Note: Beijing–Tianjin–Hebei—BTH; Yangtze River Delta—YRD; Tibetan Plateau—TP; Haihe River basin—HRB
Some studies have pointed out that the reduction of NSWS is mainly reflected in the reduction of wind strength (Xu et al., 2006; Guo et al., 2011). For instance, gentle breezes (3.4–5.4 m s−1) and moderate breezes (5.5–7.9 m s−1) were reduced by 2.2% and 1.0% decade−1 during 1981–2011, respectively (Zha et al., 2016); however, the slowdown in NSWS did not occur in weak winds, and the probabilities of light air (0.3–1.5 m s−1) and light breeze (1.6–3.3 m s−1) showed increasing trends, at the rate of 1.1% and 3.1% decade−1, respectively (Fig. 3). Hence, the variations of different NSWS categories are not the same. A decrease in NSWS does not mean that all wind categories have decreased. The average NSWS mainly reflects the seasonal, interannual, and decadal variations, which cannot reveal the detailed characteristics of variations in NSWS ranges. Hence, studies on different wind categories are helpful in understanding the refined attributes of changes in NSWS.
Figure 3. Temporal changes in probabilities of six wind grades (%) in the observed NSWS across eastern China during 1980–2011. R2 is the correlation coefficient, which passed the significance t-test at the 0.01 level [Calm: 0–0.2 m s−1; LA: light air (0.3–1.5 m s−1); LB: light breeze (1.6–3.3 m s−1); GB: gentle breeze (3.4–5.4 m s−1); MB: moderate breeze (5.5–7.9 m s−1); WSGE8.0: wind speed ≥ 8.0 m s−1]. Adopted from Zha et al. (2016).
Regional differences of changes in NSWS are conside-rable (Shi et al., 2015). Large NSWS values were found in coastal regions and northern China, where the mean NSWS exceeded 2.0 m s−1. Small NSWS values were detected over the Tibetan Plateau, middle reaches of the Yangtze River, and southwestern and southeastern China, where the NSWS was lower than 1.8 m s−1 (Jiang et al., 2010a). The spatial pattern of NSWS trends was consistent with that of the mean NSWS, in which strong decreasing trend was accompanied by large NSWS and weak decreasing trend was accompanied by small NSWS. The reduction in NSWS was more significant in northern China than that in southern China (Yin et al., 2010a). The most significant reductions in NSWS were mainly located in northwestern China (McVicar et al., 2010; Lin et al., 2013), followed by northeastern China as well as the middle and lower reaches of the Yangtze River (Wang et al., 2004). To evaluate the impact of climate backgrounds on NSWS, variations of NSWS in se-veral climate zones of China have been studied (Bian et al., 2013; Zheng et al., 2013). The results showed that differences of the changes in NSWS were considerable over different zones: the strongest slowdown was in the extratropical monsoon region, and the weakest slowdown was in the subtropical monsoon region (Zha et al., 2017b). The NSWS over the Tibetan Plateau had a decreasing trend (Yang et al., 2011; Lin et al., 2013); however, the humidity condition and elevation of the Tibetan Plateau need to be considered in assessing the NSWS trend. The decrease of NSWS in the arid areas of the Tibetan Plateau was stronger than that in the humid areas (Yang et al., 2011), and was also greater over the high elevation areas around the Tibetan Plateau than those over the low elevation areas (Guo et al., 2017).
Seasonal differences of changes in NSWS are also considerable (Wang et al., 2004; Jiang et al., 2010a). From 1951 to 2000, across the whole of China, the most significant decrease in NSWS was observed in winter (−0.14 m s−1 decade−1), followed by spring (−0.12 m s−1 decade−1), autumn (−0.10 m s−1 decade−1), and summer (−0.08 m s−1 decade−1). In Hexi Corridor, the maximum NSWS occurred in summer (Li et al., 2004), whereas it occurred in spring over the Tibetan Plateau (Lin et al., 2013). Although the downward trends in NSWS for different seasons were pronounced, the trends began to slow down after 1990 (Fig. 4; Guo et al., 2011). It is obvious that NSWS presents seasonal, interannual, decadal, and multidecadal variations; but most studies mainly analyze the trend in NSWS, with the spatiotemporal characteristics of NSWS at multiple timescales rarely investigated. Investigating the multi-timescale variations of NSWS can help to assess many societal and economic issues related to NSWS at different timescales.
Figure 4. Temporal variations and trends (solid lines) in seasonal mean NSWS across China from 1969 to 2005 (from Guo et al., 2011).
A terrestrial stilling has been revealed in the last several decades, but some studies have also discovered a weak increase in NSWS over the past decade (Zeng et al., 2019; Zhang and Wang, 2020). Zeng et al. (2019) termed this increasing trend in NSWS “reversal.” The reversal of terrestrial stilling was also reported in China (Fig. 5; Zhang and Wang, 2020); however, the magnitudes and turning points of stilling are different due to differences in study regions, periods, datasets, and methods. Yang et al. (2012) analyzed the spatiotemporal characteristics of NSWS over southwestern China and found that NSWS increased after 2000 at rates of 0.60, 0.13, 0.82, and 0.65 m s−1 decade−1 in winter, spring, summer, and autumn, respectively. An increase in NSWS was also observed over northwestern China from 1993 to 2005, with a trend of 0.04 m s−1 decade−1 (Li Y. P. et al., 2018). Zha et al. (2019a) revealed a reversal of stilling in winter over eastern China since 2000 (0.0008 m s−1 decade−1). Potential causes for NSWS recovery include the increasing temperature difference between high and low latitudes (Yang et al., 2012; Li Y. P. et al., 2018), increasing sea-level pressure at high latitudes (Zha et al., 2019a), and increasing intensity of the Aleutian low over North Pacific (Zhang and Wang, 2020). The stilling and reversal could be a cyclical, decadal pattern of NSWS. However, recent studies mainly analyzed the trends in NSWS in China over the past several decades, while the multidecadal characteristics of NSWS have yet to be revealed. Furthermore, all studies have used the correlation analysis to discuss the causes of NSWS recovery in the recent decades, and the corresponding mechanisms were not revealed in detail. A reversal in terrestrial stilling and the corresponding mechanisms need to be revealed systema-tically in the future.
Figure 5. Time series of the annual mean NSWS and normalized geostrophic wind across (a–f) six sub-regions and (g) China from 1970 to 2017 (from Zhang and Wang, 2020).
Unlike research on historical NSWS, projection of changes in NSWS has been done rarely in China. It is especially important to select credible models to predict future changes in NSWS. To date, the arithmetic-mean ensemble method is mainly use to project such changes (Jiang et al., 2009, 2010b, 2010c, 2017, 2018; Jiang and Tian, 2013). The errors between observations and different models are different. Based on the arithmetic-mean ensemble method, a model with large errors has the same effect on the ensemble results as a model with small errors. To correct this, relative error between the observation and model is used as weighting coefficient of the model, and weighted ensemble mean is applied. It is thus found that NSWS will increase in the next two decades under the Representative Concentration Pathways 4.5 (RCP4.5) and RCP8.5 scenarios. Furthermore, based on statistical downscaling, the weakening trend in NSWS under RCP8.5 is 3.5 times greater than that under RCP4.5 (Zha et al., 2020). Similar results are also reported by Jiang et al. (2009, 2017, 2018) and Chen et al. (2012). The above studies mainly analyzed the performance of models in simulating long-term trends in NSWS. The ability of the models to simulate seasonal, interan-nual, decadal, and multidecadal variations in NSWS, as well as the causes of the differences in multi-timescale variations in NSWS between models and observations, has not yet been examined. In addition, the Coupled Model Intercomparison Project phase 3 (CMIP3) and phase 5 (CMIP5) do not perform well in reproducing the past long-term decreasing trends in NSWS (Chen et al., 2012; Jiang et al., 2018; Zha et al., 2020). Why cannot the CMIP models effectively capture the significantly decreasing trends in the observed NSWS? This needs to be investigated systematically, to reduce the uncertainty in NSWS projection.
It is well recognized that wind power is a virtually carbon-free and pollution-free electricity source, and the global wind resources greatly exceed the electricity demand. Hence, the installed capacity of wind turbines grew at an annual rate of > 20% from 2000 to 2019 and is projected to increase by a further 50% by the end of 2023 (Pryor et al., 2020). As a response to the development, wind power assessment and wind energy industry have received much attention in China (Li et al., 2007). The cumulative installed capacity of wind energy amounted to 180.4 GW by the end of 2015; the newly installed capacity reached 30.5 GW, which accounted for about 48.4% of new windmills globally (Zhang D. H. et al., 2017). Accompanied by the reduction in NSWS, wind energy significantly decreased across most regions of China. The mean wind energy decreased by −3.84 W m−2 decade−1 due to changes in anthropogenic land use, which was close to the observed climate change (−4.51 W m−2 decade−1; Li Y. et al., 2008). The significant decreases of wind-generated electricity were mainly found in western Inner Mongolia and northern Gansu, which are the two leading wind energy investment areas, with a decrease of −15% ± 7% and −17% ± 8%, respectively (Sherman et al., 2017). It is worth noting that the projected wind power density in northern China could increase by about 0.7% in the middle of the 21st century if greenhouse gas emissions stay at current levels, but could drop significantly by about −3.32% at the end of the century. Additionally, seasonal differences of changes in future wind energy are considerable. Winter wind energy is projected to increase significantly, and spring and summer wind energy resources are projected to generally decrease (Chen et al., 2020). Further details on wind energy and its demands and changes can be found in Yang et al. (2017) and Pryor et al. (2020).
In the last several years, significant advances have been made in studies of the changes in NSWS over China. Almost all the studies show that NSWS in China has been reduced significantly over the past 40 years, with marked regional and seasonal characteristics. How-ever, the results are inconsistent among different studies because of differences in study methods, datasets used, and time periods selected. Furthermore, a reversal of terrestrial stilling has been found in recent decades across China. NSWS is projected to decrease at the decadal timescale in the future and the downward trend could be strengthened with the increasing GHGs emissions, but the potential mechanisms remain unknown. The causes for the reduction in NSWS are given inconsistently. Some researchers proposed that the slowdown of NSWS across China is due to variations in LOACs under global warming (e.g., East Asian monsoons, PDO, and NAM), whereas others suggested that the slowdown is due to the increasing surface roughness induced by LUCC. Although progress has been made on studies of NSWS in a changing climate in the last few decades, further investi-gations on understanding the variation in NSWS and its underlying mechanisms are needed—especially investigations that address the following issues.
Wind observations are extremely sensitive to changes in anemometer height and in the location and exposure of the observing site, especially over complex terrains (Wan et al., 2010). Changes in these factors can cause large discontinuities in the wind data series; however, they are often inevitable, especially during a long period of record in China. Therefore, it is not easy to judge which one is correct, the observation or the model, as observational data may also contain large bias resulting from insufficient station representativeness. Consequently, correction and homogenization of wind data are imperative, in particular for assessment of the observed wind speed trends, data assimilation, and model performance. Some homogenization methods, such as the standard normal homogeneity test, the penalized maximal t test, the penalized maximal F test, the Buishand test, the Pettitt test, the multiple analysis of series for homogenization, and the two-phase regression, can be employed to correct the wind data series. Based on the homogenized NSWS series, the multi-timescale changes (seasonal, interannual, decadal, and multidecadal) in NSWS can be explored, and the key areas of LOACs that significantly affect multi-timescale changes in NSWS across China may be revealed; meanwhile, the contributions of LOACs to these changes can be quantitatively estimated.
The NSWS results from high-resolution RCMs are also important, which may be used in future for the application of NSWS (e.g., evaluation and development of wind energy). High-resolution RCMs are increasingly capable of simulating offshore winds and accurately depicting the influence of complex orography. NSWS is affected by both thermodynamic and dynamic factors. Increasing the horizontal resolution of models can better present the topography and land–sea contrast, which may improve the model performance in simulating the NSWS (Yu et al., 2014; Pryor et al., 2020). Actually, surface winds are much better simulated by a regional model at 6-km grid spacing compared to the reanalysis product at approximately 25-km grid spacing (Pryor et al., 2020). Furthermore, the NSWS simulation also needs to consider the assimilation of observed wind data and the improvement in physical processes, because variations in NSWS could be influenced by cumulus convection, planetary boundary layer, land surface, radiative transfer processes, and many others. The assimilation of observed wind data can correct the systematic errors of models and the improvement in physical processes can well capture some refined physical characteristics. Especially, for the simulation of diurnal cycle of NSWS (Yu et al., 2009, 2010), the improvement in physical parameterization is crucial (Li J. et al., 2008, 2011; Yu et al., 2014). Furthermore, the increasingly large ensembles of high-resolution climate simulations could also better sample uncertainties related to internal variability and model structu-ral uncertainty (Pryor et al., 2020). Land surface is the primary source of drag. LUCC can change the surface roughness, drag coefficient, albedo, and heat exchange, thereby changing the drag from the perspective of dynamics and thermal force. High-resolution RCM can better describe complex topography, so when it is coupled with a GCM and driven by high-precision land-cover data, it could effectively capture the relative contributions of the surface friction and turbulent mixing associated with LUCC to multi-timescale changes in NSWS.
Detection and attribution (D&A) of NSWS changes are required. Based on the results of D&A, an evaluation can be conducted of the contributions of natural forcing, anthropogenic forcing, and internal variability to changes in NSWS. To achieve this goal, several CMIP6 projects will prove useful—for instance, the D&A Model Intercomparison Project (DAMIP) and Land-Use Model Intercomparison Project (LUMIP). Comparative analysis of DAMIP and LUMIP can detect the effects of different external forcings and natural forcings on changes in NSWS (Table 2). Based on experiments using His-nat and historical results, natural forcing and anthropogenic forcing can be divided. Based on experiments in His-nat, His-GHG, His-aer, and historical results, contributions from natural forcing, GHGs, and aerosols to variation in NSWS can be divided.
Experiment Experiment description Simulation period Minimum ensemble member CMIP6 historical simulation and SSP2-4.5 Historical climate simulation in CMIP6 (1850–2014) and the SSP2-4.5 scenario simulation (2015–2020) 1850–2020 3 His-nat Historical simulations involving only natural forcing 1850–2020 3 His-GHG Historical simulations involving only sufficiently mixed greenhouse gas forcing 1850–2020 3 His-aer Historical climate simulations involving only anthropogenic aerosol forcing 1850–2020 3 His-sol Historical climate simulations involving only solar radiation forcing 1850–2020 3 His-volc Historical climate simulation involving only volcanic forcing 1850–2020 3 His-CO2 Historical climate simulation involving only CO2 1850–2020 3 SSP245-aer Future projections based on the His-aer test accompanied by the aerosol concentration or emission in tropospheric under the SSP2-4.5 2021–2100 1 SSP245-nat Future projections based on the His-nat test accompanied by solar forcing and volcanic forcing under the SSP2-4.5 2021–2100 1
Table 2. A summary of the experiments in Detection and Attribution Model Intercomparison Project (DAMIP) of CMIP6. Refer to Qian and Zhang (2019)
Projections and corresponding mechanism studies of future changes in NSWS over China are lacking. Projections of NSWS in China would aid the evaluation and exploitation of wind energy, and exploiting and utilizing wind energy will help China to fulfill the Paris Climate Agreement and its commitment to reduce emissions. In the mean time, the multi-timescale variations in NSWS need to be projected as well for different emission scenarios, with statistical and dynamical downscaling methods, across China and some representative regions (e.g., the Tibetan Plateau, Yangtze River Delta, Pearl River Delta, Beijing–Tianjin–Hebei region, and arid and semi-arid regions of Northwest China). GCMs with horizontal resolutions of 25–50 km are being included in CMIP6 under high resolution model intercomparison project (HighResMIP) and those with 5-km resolution are being actively developed and tested. These models can accommodate mesoscale and even convection-permitting resolutions in the refined domain while using coarser grid spacing in the outer domain. Hence, projections of NSWS across China and its sub-regions could be carried out based on the HighResMIP.
CO2 emissions increase air temperature. The related warming is temporally non-synchronized and spatially non-uniform, resulting in spatial differences in temperature changes in the horizontal direction. Accompanied by changes in CO2 emissions, the spatial difference in temperature could be in different degrees. Changes in NSWS are not consistent with the inhomogeneous warming under different social development scenarios. Investigating the mechanism of future changes in NSWS will need to estimate the performance of the models in simulating the variation in LOACs, as well as to analyze the effects of changes in CO2 emissions on the primary LOACs, affecting the changes in wind speed over China.
Studies of wind speed changes must consider the variation in the daily maximum wind speed (DMWS) and wind gust because the DMWS and wind gust influence soil erosion, pollutant diffusion, and damage to buildings, infrastructure, and crops, with large concomitant economic losses (Lin et al., 2015; Zhang J. Q. et al., 2017; Zhang G. F. et al., 2019). It is necessary to evaluate multi-timescale variation in DMWS and wind gust, then reveal the mechanisms of DMWS and wind gust changes. The occurrence and evolution of DMWS and wind gust are generated within the mean flows; therefore, to explore DMWS variation, one must compare DMWS with the mean wind field. Additionally, some human activities, such as LUCC and releases of AAs, GHGs, and AH, may influence DMWS and wind gust, especially in mega-city clusters. Future work needs to isolate and quantify the effects of such anthropogenic factors on changes in DMWS and wind gust.
Acknowledgments. This work is also supported by the Special Program for Research Assistants of Chinese Aca-demy of Sciences, Program for Key Laboratories in Universities of Yunnan Province, and Jiangsu Province Collaborative Innovation Center for Climate Change.
|No.||Region (site)||Period||Trend||Reference||No.||Region (site)||Period||Trend||Reference|
|1||China (729)||1954–2000||−0.11||Wang et al. (2004)||21||Southwest China (110)||1969–2009||−0.24||Yang et al. (2012)|
|2||China (323)||1951–2002||−0.10||Ren et al. (2005)||22||Southwest China (110)||1969–2000||−0.37||Yang et al. (2012)|
|3||China (305)||1969–2000||−0.22||Xu et al. (2006)||23||China (540)||1971–2007||−0.17||Chen et al. (2013)|
|4||TP (75)||1966–2003||−0.17||Zhang et al. (2007)||24||China (472)||1960–2009||−0.10||Lin et al. (2013)|
|5||China (604)||1960–1999||−0.12||Li Y. et al. (2008)||25||TP (64)||1960–2009||−0.06||Lin et al. (2013)|
|6||Western deserts (23)||1973–2003||−0.29||Mahowald et al. (2009)||26||China (741)||1966–2011||−0.16||Liu et al. (2014)|
|7||HRB (45)||1957–2001||−0.14||Zheng et al. (2009)||27||East China (93)||1980–2011||−0.13||Wu et al. (2016)|
|8||China (317)||1956–2005||−0.11||Cong et al. (2009)||28||East China (93)||1980–2011||−0.13||Zha et al. (2016)|
|9||China (535)||1956–2004||−0.12||Jiang et al. (2010b)||29||Xinjiang (10)||1984–2013||−0.29||Liu et al. (2017)|
|10||Loess Plateau (82)||1960–2006||−0.14||McVicar et al. (2010)||30||East China (93)||1980–2011||−0.13||Wu et al. (2017)|
|11||China (595)||1961–2008||−0.09||Yin et al. (2010a)||31||China (492)||1979–2010||−0.11||Zha et al. (2017a)|
|12||China (603)||1971–2008||−0.12||Yin et al. (2010b)||32||China (580)||1970–2011||−0.15||Zha et al. (2017b)|
|13||TP (71)||1980–2005||−0.24||You et al. (2010)||33||BTH (154)||1978–2014||−0.10||Zhou et al. (2017)|
|14||China (597)||1961–2007||−0.13||Fu et al. (2011)||34||North China (155)||1971–2015||−0.17||Han et al. (2018)|
|15||China (726)||1969–2005||−0.18||Guo et al. (2011)||35||YRD (128)||1960–2015||−0.065||Li Y. J. et al. (2018)|
|16||China (518)||1960–1991||−0.12||Liu et al. (2011a)||36||East China (328)||1981–2011||−0.09||Zha et al. (2019a)|
|17||China (518)||1992–2007||−0.07||Liu et al. (2011b)||37||China (524)||1958–2015||−0.109||Zhang R. H. et al. (2019)|
|18||HRB (34)||1950–2007||−0.10||Tang et al. (2011)||38||BTH (223)||1980–2016||−0.25||Wang et al. (2020)|
|19||TP (78)||1984–2006||−0.30||Yang et al. (2011)||39||China (582)||1980–2017||−0.10||Zhang Y. et al. (2020)|
|20||Northeast China (87)||1961–2010||−0.25||Jin et al. (2012)||40||China (2333)||1960–2017||−0.15||Zhang and Wang (2020)|