A Revisit of Direct and Diffuse Solar Radiation in ChinaBased on Homogeneous Surface Observations:Climatology, Trends, and TheirProbable CausesYawen Wang1,2, Su Yang3, Arturo Sanchez‐Lorenzo4, Wenping Yuan1,2 , and Martin Wild5

1Guangdong Province Key Laboratory for Climate Change and Natural Disaster Studies, School of Atmospheric Sciences,Sun Yat‐Sen University, Zhuhai, China, 2Southern Marine Science and Engineering Guangdong Laboratory, Zhuhai,China, 3National Meteorological Information Centre, China Meteorological Administration, Beijing, China, 4Departmentof Physics, University of Extremadura, Badajoz, Spain, 5Institute for Atmospheric and Climate Science, ETH Zurich,Zurich, Switzerland

Abstract Increasing energy and food demands require an understanding of not only the availabilityand variability of total solar radiation (R) but also its partitioning into direct (Rd) and diffuse (Rf)components, which are key elements for solar power generation and plant photosynthesis, respectively.Potential inhomogeneities in surface solar radiation observations, however, add uncertainties to thepublished patterns and trends of Rd and Rf over China. To exclude inhomogeneous time series andoutliers, a data quality control process has been applied to the surface observations of Rd and Rf. Thehomogeneous data show that, in response to the global dimming and brightening phenomena in R,decadal changes can also be observed in the Rd and Rf trends over China with the actual transitionoccurring in the mid‐2000s. Since the 1960s, Rd dimming has mainly distributed over eastern Chinawith intensive anthropogenic activities, with recoveries especially from 2008 onward over regions withhigh Rd energy potential. Rf, on the other hand, had further enhanced since the 1960s especially over thehigh Rf proportion areas but has decreased in the recent decade over the northern region. The summerseason shows the most significant decadal trends in Rd and Rf. Under clear‐sky conditions, that is,ruling out cloud effects, the transitions in Rd and Rf trends become more remarkable, pointing toatmospheric aerosols as the main driving force. In addition, dramatic Rd and Rf variations are observed inthe subsequent years after large volcanic eruptions. The varying Rd and Rf proportions are in line withrecent governmental control on SO2 emissions over China.

1. Introduction

Solar radiation reaching the Earth's surface, also known as surface solar radiation or total solar radiation (R),is the primary energy source that governs diverse physical, chemical, and biological processes on our planet(Wild, 2009). Worldwide measurements of R starting from the International Geophysical Year (1957/1958)revealed significant decadal changes in R with a decline on the order of 3–9 W m−2 up to 1990 and a subse-quent increase of 1–4 W m−2, coined as “global dimming and brightening” phenomena, respectively(Stanhill & Cohen, 2001; Wild, 2012; Wild et al., 2005). Consistent with the global trend, a similar transitionhas been widely reported in the R trend of China with the estimated changing rates of−2.5 to −12 and 0.4 to4 W m−2 decade−1 during the global dimming and brightening phases, respectively (Wang & Yang, 2014).Based on homogeneous data, recent studies further updated the decadal variations of R in China into threestages: a dimming until 1990, a leveling off afterward, and an actual brightening since 2008 (Wang,Trentmann et al., 2019; Yang et al., 2018, 2019).

The global dimming and brightening phenomena are functions of the changes in the sum of the direct (Rd)and diffuse (Rf) radiation components, which are respectively the fractions of R from the direction of the Sunand from the different directions in the rest of the sky dome scattered by clouds, aerosols, gases, and waterdroplets in the atmosphere (Calbó et al., 2017; Kvalevåg & Myhre, 2007; Ma et al., 2011). Rd, which is anenvironment‐friendly energy source and one of the most abundant renewable energy resources for Earth,is essential for the generation of solar power (Feng et al., 2017; Kambezidis, 2018). The development of

©2020. American Geophysical Union.All Rights Reserved.

RESEARCH ARTICLE10.1029/2020JD032634

Key Points:• Homogeneous observations show a

recent transition to brightening indirect radiation and to dimming indiffuse radiation over China since2008

• Clear‐sky trends suggest an impactof volcanic eruption on the dramaticchanges in solar radiation overChina during the early 1990s

• Governmental control on pollutantemissions may have contributed tothe variations of direct and diffuseradiation proportions over China

Supporting Information:• Supporting Information S1

Correspondence to:W. Yuan,yuanwp3@mail.sysu.edu.cn

Citation:Wang, Y., Yang, S., Sanchez‐Lorenzo,A., Yuan, W., & Wild, M. (2020). Arevisit of direct and diffuse solarradiation in China based on homoge-neous surface observations:Climatology, trends, and their probablecauses. Journal of Geophysical Research:Atmospheres, 125, e2020JD032634.https://doi.org/10.1029/2020JD032634

Received 18 FEB 2020Accepted 8 APR 2020Accepted article online 17 APR 2020

WANG ET AL. 1 of 19

solar energy is increasingly seen as important due to the rapid growth of the global energy demands and thedepletion of fossil energy (Jacovides et al., 2006; Wang, Lu et al., 2019). Rf, on the other hand, is of particularinterest for vegetation production, since Rf can increase the efficiency of plant photosynthesis by penetratingdeep into the vegetation canopy from all angles and reducing canopy photosynthetic saturation (Gu et al.,2002; Wild et al., 2012). Based onmodel reconstructions, the fertilization effect of Rf has enhanced the globalland carbon sink by approximately one quarter during the global dimming period (Mercado et al., 2009).From ground observations, however, as compared to the adequately discussed R, far less studies have inves-tigated the decadal changes of Rd and Rf (Wild, 2009).

Limited regional analyses on long‐term Rd and Rf variations show an overall tendency toward a decrease inRd and an increase in Rf during the global dimming phase in response to increasing cloudiness and/or aero-sol loads, and vice versa during the global brightening phase, over countries such as the former Soviet Union(Abakumova et al., 1996), Germany (Power, 2003), South Africa (Power & Mills, 2005), Estonia (Russak,2009), India (Soni et al., 2012), Spain (Calbó et al., 2017; Sanchez‐Lorenzo et al., 2013), or Greece(Kambezidis, 2018). Exceptions of opposite Rf trends were observed in sites from Ireland (Black et al.,2006; Stanhill, 1998) and the United States (Long et al., 2009). In China, the ground observations of Rdand Rf started in 1957 but have been examined by only very few studies. Three major issues remain in thepreviously analyzed Rd and Rf trends of China.

The first issue is a temporal inhomogeneity in the ground observations of solar radiation in China, especiallyduring 1990–1993 when a nationwide update in the measuring instruments and stations occurred (Wanget al., 2015). Efforts have been made by extensive research to diminish the inhomogeneity in the surfaceobservations of R (Shi et al., 2008; Tang et al., 2011; Wang & Wild, 2016; Yang et al., 2018). Unfortunately,there was not much done in terms of homogenization of Rd and Rf observations, which adds uncertaintiesin the published Rd and Rf trends. This is especially the case considering a potential sensitivity drift problemin the pyrheliometer measuring Rd during the pre‐1990 period (He &Wang, 2020). Besides, inconsistent sea-sonal trends of Rd were obtained by the studies of Zhao et al. (2009) and Ma et al. (2011). A revisit of the cli-matology and trends of Rd and Rf in China, and their spatial and temporal characteristics, is still necessary,based on homogeneous data.

The second issue is the neglect of the background of the global dimming and brightening phenomena, thatis, the decadal changes in R, when analyzing the trends of Rd and Rf components. Part of previous examina-tions of surface‐observed Rd and Rf trends in China were only for the dimming phase till the 1990s (e.g., Cheet al., 2005; Qian et al., 2007; Shi et al., 2008; Wang et al., 2015). The examined time series of Rd and Rf wereextended in the studies of Liang and Xia (2005), Ren et al. (2013), and Feng and Li (2018), which, however,lacked the investigation on possible decadal changes in the variations of Rd and Rf. The relative contributionsof the Rd and Rf components to the decadal changes of R, especially during the recent brightening since themid‐2000s in China, have not yet been clarified.

The third issue is that the reasons and mechanisms behind the variations of Rd and Rf have seldom beenexplored. Clouds and aerosols are the two factors playing the most important role in driving global dimmingand brightening (Wild, 2009, 2012). The changes in clouds and aerosols also greatly alter the relative fractionof diffuse and direct radiation (Feng et al., 2018). Despite the difference in the radiative effects of clouds andaerosols, an increase in the amount of clouds or aerosols generally tends to decrease Rd but increase Rf,though such effect may vary with the type and size of aerosols, other properties of clouds, the wavelengthof radiation, and air humidity (Kanniah et al., 2012; Power, 2003; Qian et al., 2007). To differentiate theeffects between clouds and aerosols, comparing the trends derived under clear‐sky and all‐sky conditionshas been increasingly proposed as a useful method (Qian et al., 2007; Wild et al., 2005; Yang et al., 2019).Besides, pollution emission controls, especially on SO2, a major atmospheric pollutant in China, becamemandatory since the implementation of the 11th Five‐Year Plan of China in 2006 (Jin et al., 2016; Maet al., 2019). This raises also the interest to investigate the implications of recent pollution control effortsin China for changes in the diffuse and direct radiation components.

This research therefore aims to clarify the evidence and causes of Rd and Rf trends under both all‐sky andclear‐sky conditions in China based on homogeneous surface data. First, a data quality control process willbe applied on the surface‐observed Rd and Rf data of China. Then, the responses of Rd and Rf to decadal

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 2 of 19

changes of R in China, especially to recent brightening since mid‐2000s, will be assessed at different tem-poral and spatial scales. Finally, attempts will be made to identify the main driving force behind the varia-tions of Rd and Rf in China, as well as discuss the implications of recent emission control efforts by theChinese government.

2. Materials and Methods2.1. Data Sources

Daily solar radiation data sets of R, Rd, and Rf over China were obtained from the China MeteorologicalData Service Center (http://data.cma.cn) of the China Meteorological Administration (CMA). The rawCMA solar radiation data set encompasses 130 stations across China, with only 83 of them providingRd and Rf observations. Specifically, the CMA Rd and Rf observations started in 1957 at 25 stations acrossChina. The number of observing stations had exceeded 60 since the 1960s but sharply decreased toaround 17 after 1993. To avoid the impact of changes in the number of stations on the composite timeseries, only 12 stations with continuous and complete records with more than 96% Rd and Rf observa-tions for the whole period of 1961–2016 were considered for further quality control. Among them, onestation, namely, Shanghai, was combined from two collocated stations, Xujiahui and Baoshan, to com-plete the recording period.

Basic data quality controls, including the tests on the climatic range, spatial consistency, and temporal con-tinuity, have been applied by the CMA (see the Data Description Document). Data quality of the releaseddaily observations of R, Rd, and Rf has been coded by the CMA as 0 (correct), 1 (suspicious), 2 (wrong), 8(missing), and 9 (without quality control); only the daily data flagged as correct were used in this study.However, uncertainties still exist in the quality of surface observations, especially considering the nationalupdate of the solar radiation monitoring network during the period 1990–1993 (Yang et al., 2018). The pyr-heliometers and pyranometers measuring Rd and Rf, respectively, were changed from the manualYanishevsky type with the errors not exceeding 3% and 5%, respectively, to automatic Chinese‐developedtype with the errors not exceeding 2% and 5%, respectively (Shi et al., 2008; Wang et al., 2015). To furtherreduce potential inhomogeneities in the Rd and Rf observations, a data quality control process was appliedas follows.

2.2. Data Quality Control

To further eliminate the outliers, the physical thresholds of daily R, Rd, and Rf were checked according to thefollowing criteria: (1) Rd and Rf cannot be bigger than R, and their sum should equal to R; (2) the measured Rshould neither exceed extraterrestrial radiation (Ra) nor fall below the lower boundary (0.03Ra) and shouldnot exceed the clear‐sky radiation (Rso) by more than 10% (Shi et al., 2008); (3) a large diffuse fraction Rf/Rmore than 0.9 should be expected for overcast skies when the clearness index R/Ra is less than 0.2, while Rf/Rshould be less than 0.8 for clear skies when R/Ra is over 0.6 (Reindl et al., 1990). The corresponding equationsare expressed as

R ¼ Rd þ Rf ; (1)

0:03Ra < R < Ra;

R < 1:1Rso;

�(2)

Rf=R ≥ 0:9; for R=Ra < 0:2;

Rf=R ≤ 0:8; for R=Ra > 0:6:

�(3)

Based on the FAO‐56 method (Allen et al., 1998), Ra (MJ m−2 day−1) can be calculated as

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 3 of 19

Ra ¼ 24 × 60π

Gscdr ωssinφsinδ þ cosφcosδsinωs½ �;

dr ¼ 1þ 0:033cos2π365

J

� �;

δ ¼ 0:409sin2π365

J − 1:39

� �;

ωs ¼ cos−1 −tanφtanδð Þ;

8>>>>>>>>><>>>>>>>>>:

(4)

where Gsc is the solar constant (0.082 MJ m−2 minute−1); φ is the latitude of the location; J ∈ [1,365/366] isthe number of the day in the year; and the calculated dr, δ, andωs are the inverse relative distance Earth‐Sun,solar declination, and sunset hour angle, respectively.

Setting z as the station elevation above sea level (m), Rso (MJ m−2 day−1) can then be calculated as

Rso ¼ 0:75þ 2 × 10−5 z� �

Ra: (5)

Only daily data of R, Rd, and Rf that passed the physical threshold test were averaged into monthly values forfurther homogenization. Of the calculated monthly means, 99.5% have more than 10 daily samplesper month.

2.3. Data Inhomogeneity Identification and Selection

Next, the homogeneity of Rd and Rf time series was checked by the Standard Normal Homogeneity Test(SNHT; Alexandersson, 1986), which has been largely used for studying inhomogeneities and discontinuitiesin the form of trends and jumps of climatological time series (Khaliq & Ouarda, 2007). The SNHT has beensuccessfully applied in building a homogeneous data set of Rf in Spain (Calbó et al., 2017; Sanchez‐Lorenzoet al., 2013) and Europe (Sanchez‐Lorenzo et al., 2015). The SNHT derives a statistic T(k) for eachmonth k ofa candidate time series {Xi}:

T kð Þ ¼ kz¯21 þ n − kð Þz¯22; (6)

where z¯1 and z¯2 are the averages of the Xi values before and after a possible change point, respectively, cal-culated based on the mean X¯ and standard deviation σ of the whole time series {Xi}:

z¯1 ¼ 1k∑k

i¼1

Xi − X¯σ

; (7)

z¯2 ¼ 1n − k

∑n

i¼kþ1

Xi − X¯σ

: (8)

Possible breaks/shifts in the time series can then be detected in the months when T(k) is above the criticalvalues, which can be acquired from Khaliq and Ouarda (2007) based on the length of the time series andthe confidence level (95% in this study).

Data of sunshine duration (SD), which quantifies the length of time in each day when Rd > 120Wm−2, werecollected from the CMA to serve as the reference for testing the homogeneity of Rd. The SNHT was thenapplied on the time series of Rd, Rf, and SD (known as “absolute” SNHT), as well as the time series of relativebias between Rd and SD (known as “relative” SNHT). Monthly anomalies from the 1961–2016 monthlymeans instead of absolute values were used here to reduce monthly differences and possible effects fromthe annual solar cycle. To assist the identification of inhomogeneous time series, the Pearson correlationsof the relative monthly anomalies between Rd and SD and between Rd and Rf were also assessed. In addition,the validity of each time series of Rd, Rf, and SDwas visually inspected. After all these inspections, the stationof Lhasa was further removed from this study, due to an abnormal decrease in Rd observations in the monthsfrom March 1988 to April 1992, which was not captured by the SD observations. Besides, 14 outlier monthsfrom two stations of Kunming and Wuhan were further eliminated from this study. The lack of good meta-data and the limited number of stations with long‐term observations added uncertainty to the inhomogene-ity identification and made it difficult to apply the data homogenization. Therefore, the inhomogeneous

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 4 of 19

series were deleted rather than corrected in this study. The reliability of data was instead ensured byselecting only the 11 stations over China with homogenous and continuous Rd and Rf observations for1961–2016 as shown in Figure 1.

Last, missing monthly data were filled to ensure 12 months was available for calculating annual means. Thisaims to avoid seasonal biases in averaging the annual values, as well as to minimize the effect of missing dataon the overall trends. Monthly data gaps (~0.6% of total) were filled based on the linear regression relation-ships established by the available data from the same month for the dimming (1961–1989)/leveling off(1993–2007)/brightening (2008–2016) periods, without changing the original decadal trend slopes foreach month.

2.4. Analysis Procedure

Based on homogeneous data, the climatology of Rd and Rf and its nondimensional form Rf% (the ratio of Rf/R) was evaluated over the selected 11 stations across China. To understand the spatial patterns of multiyearaveraged Rd, Rf, and Rf% over China, collocated total cloud cover (TCC) data were obtained from the CMAand estimated. TCC was visually observed according to the standards of the World MeteorologicalOrganization. Then, the variations of the Rd and Rf components in response to the decadal changes of Rwereinvestigated for the global dimming (1961–1989) and brightening (1993–2016) phases, as well as the recentlyproposed actual brightening phase in China (2008–2016), at different temporal (annual and seasonal) andspatial scales. The period between 1989 and 1993 with national reorganization of the solar radiation measur-ing network is avoided to further ensure the reliability of the examined trends. The seasons are defined asspring (March to May), summer (June to August), autumn (September to November), and winter(December to February).

Figure 1. Spatial distribution of the 11 solar radiation stations across China with homogeneous direct and diffuse radia-tion observations for 1961–2016. The gray scale of the star symbols indicates the elevation (m) of the stations. Thedivided four geographical regions of north China (NC), south China (SC), northwest China (NW), and southwest China(SW) are based on the three lines: Line A, also called Qinling‐Huaihe Line, approximates the 0 °C January isotherm andthe 800 mm isohyet in China; Line B represents the 400 mm isohyet and the boundary between monsoon and non-monsoon regions in China; Line C is the boundary between the first and second steps of China's terrain with averageelevation over 4,000 m and around 1000–2000 m, respectively.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 5 of 19

To identify the main driving force from the two most likely candidates of cloud and aerosol (Wild, 2009),three steps of analysis were taken as follows. First, the trends of TCC were examined and compared withthe Rd and Rf trends on the same temporal and spatial scales. Second, the individual and overall trends ofRd, Rf, and Rf% over the 11 homogenous stations were compared between all‐sky and clear‐sky conditions.Clear‐sky condition was defined as the days with TCC values lower than 50% of the monthly means of eachexamined station. The applied criterion for selecting clear‐sky days is less strict than that used in previousstudies commonly based on a daily mean TCC ≤ 10% or 0% (e.g., Manara et al., 2016; Qian et al., 2006,2007; Sprintsin et al., 2012; Wang et al., 2014). The higher threshold aims to ensure the data completeness,especially for the Guangzhou and Kunming stations (with humid subtropical monsoon climate), during thesummer season under clear‐sky conditions, as well as to reduce possible random effects on the sampling byenlarging the size of daily samples for calculating monthly means. High correlations in the monthly anoma-lies, annual fluctuations, and trends of Rd, Rf, and Rf% are shown for the clear‐sky conditions defined by dif-ferent TCC thresholds of 0%, 10%, and 50% monthly means (see Figures S1 and S2 and Tables S1–S3 in thesupporting information). This proves that the representativeness of clear sky has barely been reduced by thenew threshold, which in turn allows to increase the number of samples and enhance the reliability of thecalculated means and trends. Third, the data of SO2 emissions for 1993–2016 were collected from theChina Statistical Yearbook released by the National Bureau of Statistics of China (http://www.stats.gov.cn) for the provinces where the selected solar radiation stations are located. The individual and overall var-iations of SO2 emissions and Rf% were then compared over the 11 homogenous stations to investigate pos-sible implications of emission control efforts for the Rf% variations in China.

3. The Climatology of Direct and Diffuse Radiation in China

Figures 2a–2c reveal the climatology of Rd, Rf, and Rf%, respectively, over the 11 homogenous stations acrossChina for the period of 1961–2016. Seen from Figure 2a, Rd ranges from 56 Wm−2 (1,781 MJ m−2 year−1) to129 W m−2 (4,067 MJ m−2 year−1) over the 11 stations across China. Due to missing information over theSichuan basin (low Rd center) and Qinghai‐Tibet Plateau (high Rd zone), the derived Rd range is differentfrom that proposed by Zhao et al. (2009) (1,200–5,200 MJ m−2 year−1) based on the estimated Rd from SDover 743 stations across China. The average Rd for the whole China is 85Wm−2 (2,685MJm−2 year−1), simi-lar with the estimate of Shi et al. (2008) (88 W m−2) based on 64 stations with data records longer than25 years, suggesting a good representativeness of the selected 11 stations for China. The mean values areadditionally provided in the unit of MJ m−2 year−1 to enable comparison with previous analyses. In general,Rd enhances from the south and east of China featured by humid monsoon climate and the plain terrain tothe north and west characterized by continental or plateau climate and high elevation. Comparing Figures 1and 2a, geographical features play an important role in determining the spatial patterns of Rd, which shows apositive correlation of 0.42 with latitude and a negative correlation of −0.59 with longitude. Significant(p < 0.01) correlations with Rd are shown for elevation (0.86) and TCC (−0.61); Rd values lower than70Wm−2 and higher than 90Wm−2 are largely concentrated in the regions of south China with TCC valuesover 60% and west China with elevations over 2,000 m, respectively. The spatial distribution of Rd generatedover 11 homogeneous stations across China is consistent with that proposed by previous studies (Ma et al.,2011; Zhao et al., 2009; Zuo & Gong, 1962).

Almost opposite spatial patterns with low values in the north and west and high values in the south and eastare shown for Rf (Figure 2b), especially for its nondimensional form Rf% (Figure 2c). The average Rf value forthe whole China is 79 W m−2 (2,499 MJ m−2 year−1), close to the values of 78 W m−2 and2,477 MJ m−2 year−1 estimated by Shi et al. (2008) and Ren et al. (2013), respectively. Note that the averageRf almost equals Rd in China, resulting in a large proportion of Rf to R by 49%, which slightly increases fromthe estimation of 46% by Zhu (1984) for the pre‐1980 period. High Rf% values (>50%) are mainly distributedover south China (Figure 2c). Inferring from the high correlation (0.82; p< 0.01) between the spatial patternsof Rf% (Figure 2c) and TCC (Figure 2d), clouds as well as atmospheric transparency largely determine theclimatology of Rf%. Ranging from 64 W m−2 (2,033 MJ m−2 year−1) to 87 W m−2 (2,732 MJ m−2 year−1),Rf displays a more complex spatial pattern than does Rd (Figure 2b). The low Rf values generally distributein the areas with high latitude, dry climate, or low TCC, and vice versa. Similar spatial patterns of Rf werederived for China by Ma et al. (2011) and Ren et al. (2013) based on surface‐observed and SD‐estimated

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 6 of 19

spatial data of Rf, respectively. Despite the good representativeness of the 11 homogenous stations for China,to generate a full picture of the climatology of Rd and Rf in China, high‐quality Rd and Rf data with betterspatial representation are highly in need.

4. Trends of Direct and Diffuse Radiation in China4.1. Annual Scale

Figure 3 shows the annual time series of Rd, Rf, and Rf% from 1961 to 2016 averaged over the 11 homogenousstations across China. During 1961–2016, Rd significantly decreased by −1.4 W m−2 decade−1, while Rf andRf% generally leveled off with slight increases by 0.4 W m−2 decade−1 and 0.7% decade−1, respectively(Table 1).

During the global dimming phase of 1961–1989, Rd showed consistently a decreasing trend by−5.8 W m−2 decade−1, close to the magnitude of −6.6 W m−2 decade−1 calculated over 64 stations acrossChina by Che et al. (2005). The relative decreasing rate of Rd was estimated by previous studies to be inthe range of 7.5–8.6% decade−1 during the 1960s–1990s (Che et al., 2005; Liang & Xia, 2005; Shi et al.,2008). There was no significant trend in Rf, but a significant increasing trend in Rf% by 1.8% decade−1

(Table 1). A similar changing rate of Rf% by 1.7% decade−1 was presented by Che et al. (2005) for the globaldimming phase.

Figure 2. Spatial patterns of multiyear averaged (a) direct radiation (Rd, W m−2), (b) diffuse radiation (Rf, W m−2), (c)diffuse fraction (Rf fraction, %), and (d) total cloud cover (TCC, %) over 11 homogenous stations across China for 1961–2016. The scales are determined by the percentile method to enable equal number of samples in each category.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 7 of 19

The changes of Rd and Rf in China during the global brightening phase since the 1990s have been also inves-tigated. Seen from Figure 3 and Table 1, the reported leveling off/slight increasing trend of R over China dur-ing 1993–2016 (Tang et al., 2011; Wang &Wild, 2016; Yang et al., 2018) seemed an offset between increasingRf (2.7 W m−2 decade−1)/Rf% (1.3% decade−1) and decreasing Rd (−1.1 W m−2 decade−1).

Consistent with the proposed recent brightening phase of R in China based on homogeneous observations(Wang, Trentmann, et al., 2019; Yang et al., 2018, 2019), a transition in all three time series of Rd, Rf, andRf% occurred actually in 2008 (Figures 3a–3c). The actual brightening of Rd in China started from 2008 bya rate of 13.6 W m−2 decade−1 (Table 1). Meanwhile during 2008–2016, there was no significant trend inRf but a significant decreasing trend in Rf% by an average rate of–3.2% decade−1 over the 11 homogeneousstations across China.

Table 1Magnitudes of Linear Trends in Direct Radiation (Rd, W m−2 decade−1), Diffuse Radiation (Rf, W m−2 decade−1), Diffuse Fraction (Rf Fraction, % decade−1), andTotal Cloud Cover (TCC, % decade−1) for Different Periods Averaged Over the 11 Homogenous Stations Across China Under All‐Sky and Clear‐Sky Conditions asShown in Figure 3

Variables Stations 1961–2016 1961–1989 1993–2016 2008–2016

Rd (W m−2 decade−1) All‐sky −1.4 ± 0.4* −5.8 ± 0.7* −1.1 ± 1.1 13.6 ± 3.4*Clear‐sky −2.8 ± 0.7* −10.8 ± 0.9* 5.5 ± 1.2* 23.4 ± 4.3*

Rf (W m−2 decade−1) All‐sky 0.4 ± 0.2* 0.0 ± 0.4 2.7 ± 0.4* 0.3 ± 0.9Clear‐sky 1.3 ± 0.3* 3.4 ± 0.6* 1.4 ± 0.7* −3.7 ± 3.0

Rf fraction (% decade−1) All‐sky 0.7 ± 0.1* 1.8 ± 0.3* 1.3 ± 0.4* −3.2 ± 1.0*Clear‐sky 0.9 ± 0.2* 2.6 ± 0.3* −0.4 ± 0.3 −4.2 ± 1.4*

TCC (% decade−1) All‐sky 0.4 ± 0.2* −0.4 ± 0.3 3.0 ± 0.4* 4.2 ± 2.1*

*Significant trends at the 95% confidence level.

Figure 3. Annual variations of (a) direct radiation (Rd, W m−2), (b) diffuse radiation (Rf, W m−2), (c) diffuse fraction (Rf fraction, %) and (d) total cloud cover(TCC, %) averaged over 11 homogenous stations across China for 1961–2016. Red line = all‐sky condition; green line = clear‐sky condition.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 8 of 19

4.2. Seasonal Scale

Seasonal trends of Rd, Rf, and Rf% were derived in relative terms to exclude the effect from the annual solarcycle for enabling comparisons (Figure 4). During the dimming phase of 1961–1989, Rd significantlydecreased in all seasons with the largest magnitude in winter (−3.9% decade−1), followed by summer(−3.1% decade−1), spring (−2.6% decade−1), and autumn (−1.8% decade−1) (Figure 4a). Based on surfaceobservations from 14 stations across China, Ma et al. (2011) also noted the largest relative declining rateof Rd in winter during 1961–2009. Almost opposite patterns were observed in absolute seasonal Rd trendsby Zhao et al. (2009) based on SD‐estimated Rd over 743 stations across China for 1960–2005. Here we notedthat in absolute terms, Rd decreased themost in summer by –2.9Wm−2 decade−1 during 1961–1989, twice aslarge as the declining rates for autumn and winter. Consistently, Rf% significantly increased in all the sea-sons, especially in summer and winter with an average rate of around 0.8% decade−1 (Figure 4c). The seaso-nal trends of Rf are more complex with significant trends generally indicating a slight decrease in spring(−0.5% decade−1) and winter (−0.1% decade−1) but a slight increase in summer (0.2% decade−1) and autumn(0.1% decade−1) (Figure 4b). Similar seasonal Rf trend patterns were also noted by Ma et al. (2011) and Renet al. (2013) based on surface‐observed and SD‐estimated Rf for the periods of 1961–2009 and 1981–2010, respectively.

More consistent seasonal trends of Rf occurred in the global brightening phase of 1993–2016, which showedsignificant increases in all the seasons with the highest rate in winter (1.5% decade−1), followed by autumn(1.3% decade−1), summer (1.0% decade−1), and spring (0.9% decade−1). Similar seasonal patterns are

Figure 4. Comparisons in seasonal trends of (a) direct radiation (Rd, % decade−1), (b) diffuse radiation (Rf, % decade−1),(c) diffuse fraction (Rf fraction, % decade−1), and (d) total cloud cover (TCC, % decade−1) averaged over the 11homogenous stations across China for the periods of 1961–1989, 1993–2016, and 2008–2016. The symbol of asteriskdenotes significant trends at the 95% confidence level.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 9 of 19

presented in Rf% with the changing rate of 0.7% decade−1, 0.7% decade−1, 0.4% decade−1, and 0.2% decade−1

for winter, autumn, summer, and spring, respectively. Meanwhile, the seasonal trends of Rd generallydecreased with the trend in autumn (−1.2% decade−1) being significant.

The decline in seasonal Rd trends did not persist into the 2010s but had transitioned into brightening since2008, especially significant in summer (9.8% decade−1). Consistently, summer is the season showing adecreasing trend in Rf (−0.4% decade−1), as well as the significant decreasing trend in Rf% (−2.4% decade−1)during 2008–2016. Therefore, the recent transition to brightening in Rd and dimming in Rf mainly occurs insummer, themonsoon season of China with frequent rainfalls washing out air pollutants (Wang et al., 2012).Summer is a humid season with hygroscopic sulfate aerosols playing a main role in increasing aerosol opti-cal depths (AODs) and particle sizes (Wang, Xin, et al., 2011), which could have been further declined byrecent governmental control on SO2 emissions (Jin et al., 2016).

4.3. Spatial Scale

Spatial patterns of Rd, Rf, and Rf% trends are depicted in Figure 5 for the periods of 1961–1989, 1993–2016,and 2008–2016. The geographical distribution of the variations in Rd is very similar to that observed in R(Liang & Xia, 2005; Wang & Wild, 2016; Yang et al., 2018). During the global dimming phase of 1961–1989, the decreases of Rd cover all the examined stations, with 73% of the decreasing trends being significant(Figure 5a). Significant dimming trends of Rd are largely concentrated in the eastern part of China, that is,the regions of south and north China (Figure 1), with low Rd energy reserves (Figure 2a), consistent with pre-vious analyses on the spatial patterns of Rd trends during the global dimming phase (Liang & Xia, 2005; Maet al., 2011; Shi et al., 2008; Zhao et al., 2009). Opposite spatial patterns are shown for Rf% with overallincreases covering 91% of the stations (Figure 5g), which are especially significant over the eastern part ofChina with higher proportion of Rf (Figure 2c). Consistently, significant increasing trends of Rf are shownin eastern China, though Rf in general depicts a heterogeneous spatial pattern during the global dimmingphase, as observed also by Liang and Xia (2005) and Shi et al. (2008) for the periods of 1961–2000 and1961–1992, respectively.

The spatial Rf trends had becomemore consistent since the 1990s with increases covering 91% of the stations,which are especially significant in the eastern region with a high proportion of Rf resources (Figures 2c and5e). Since 2008, Rf had transitioned to decrease in 55% of the stations mainly distributed in the north ofChina (Figure 5f). Similar spatial patterns are shown for Rf% trends, which had continued to increase in73% of the stations especially over the eastern region since 1993 (Figure 5h) but transitioned to decreasein 45% of the stations mainly distributed in the north of China from 2008 onward (Figure 5i). Rd had contin-ued to dimming since the 1990s in 64% of the stations (Figure 5b) but largely transitioned to brightening after2008 (Figure 5c), especially in the north and west with high Rd energy reserves (Figure 2a). In the region ofsouth China, the dimming of Rd continued even into the 2010s.

All in all, Rd tends to decrease especially in south China with low Rd energy reserves but further increase inother regions relatively rich in Rd energy, which would then benefit solar production especially over thenorthwest region. Meanwhile, Rf mainly increases in eastern China with a high proportion of Rf resourcesand vice versa. These result in a further increase of Rf% in the south and east of China but a further decreasein the north and west, which would in turn generate implications for plant growth patterns in China.

5. The Probable Causes5.1. Cloud Effects

Clouds, which can reduce R but increase the proportion of Rf reaching the Earth's surface, play a vital role inshaping the geographical distributions of Rd and Rf resources in China (Figure 2). Contrary to the spatial pat-terns that would result from extraterrestrial radiation, Rd in the east side of China increases from low to highlatitudes, while Rf and Rf% decrease with increasing latitude. This can be largely explained by the high TCCvalue (>60%) zone in south China characterized by subtropical and tropical monsoon climates with annualprecipitation over 800 mm (Figure 2), which would enhance the scattering of incoming solar radiation andthus lead to high Rf/Rf% but low Rd in this region.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 10 of 19

Cloud variations, however, can hardly cause the observed decadal changes of Rd, Rf, and Rf% since the 1960s.During the dimming phase of 1961–1989, there were no significant annual or seasonal TCC trends, whichcould not result in significant dimming trends of Rd and significant brightening trends of Rf% for the

Figure 5. Spatial trends of (a–c) direct radiation (Rd, W m−2 decade−1), (d–f) diffuse radiation (Rf, W m−2 decade−1), (g–i) diffuse fraction (Rf fraction,% decade−1), and (j–l) total cloud cover (TCC, % decade−1) over 11 homogenous stations across China for the periods of 1961–1989, 1993–2016, and 2008–2016. Red and blue circles denote positive and negative trends, respectively, marked with rhombs denoting significant trends at the 95% confidence level.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 11 of 19

collocated stations at different temporal scales (Figures 3 and 4). At the spatial scale, a low correlation (abso-lute coefficient < 0.2) is shown between the spatial trends of TCC and those of Rd, Rf, and Rf% (Figure 5),further indicating insignificant total cloud effects in the dimming phase as proposed by previous studies(Liang & Xia, 2005; Qian et al., 2006; Wang et al., 2013).

Since 1993, TCC had started to significantly increase, which might have partially contributed to the paralleldecreasing trend in Rd and increasing trend in Rf/Rf% (Table 1). Previous studies have also indicated that halfof the decadal variability of R since the 1990s should be attributed to the changes of TCC (Wang, Zhang,et al., 2011; Xia, 2010a). TCC increased the most in autumn and winter, the seasons with the most obviousdecrease in Rd and increase in Rf/Rf% (Figure 4). Besides, spatial consistency between the trends of TCC andradiation components is shown in Figure 5, with increases of TCC and Rf and decreases of Rd mainly distrib-uted over the east side of China.

TCC kept increasing during 1993–2016, which cannot lead to the transitions to brightening in Rd and to dim-ming in Rf/Rf% over China from 2008 onward (Figure 3). The transition in Rd and Rf trends is most signifi-cant in summer, the only season with stabilized TCC trend during 2008–2016 (Figure 4). In the otherseasons, increasing TCC might have helped in counteracting the brightening of Rd and dimming of Rf/Rf%. Seen from Figure 5, only one station, Shanghai, showed a transition in the TCC trend since 2008, incon-sistent with the widespread transitions in the Rd and Rf trends over the north of China. These pieces of evi-dence indicate that some other factor rather than TCC was playing the main role in inducing the recent Rdbrightening and Rf dimming in China.

5.2. Aerosol Effects

To investigate the effects from aerosols, clear‐sky trends of Rd, Rf, and Rf% were examined over the 11 homo-genous stations across China for 1961–2016. Seen from Figure 3a, two drops in the clear‐sky Rd are shown in1983 and 1992, corresponding to the subsequent years of the El Chichón eruption in April 1982 and thePinatubo eruption in June 1991 (Ren et al., 2013; Wu & Fu, 2011), indicating a notable effect from volcaniceruptions on solar radiation variability in China. Correspondingly, clear‐sky Rf and Rf% showed the sharpestincrease in 1983 and 1992 (Figures 4b and 4c), as observed also in previous studies (Fu et al., 2015; Qian et al.,2007; Ren et al., 2013). The Rf increase induced by volcanic aerosols was reported to have enhanced the ter-restrial carbon sink (Gu et al., 2003). The strong responses of clear‐sky Rd and Rf to volcanic eruptions sug-gest that part of the evidenced dramatic variations of solar radiation in China during 1990–1993 (Wang &Wild, 2016) might be driven by the sharply increased volcanic aerosols after the Pinatubo eruption in 1991.

Ruling out the effects from clouds, the dimming rate of Rd almost doubled during 1961–1989 (Figure 3a andTable 1). The higher dimming rate of Rd under clear‐sky than all‐sky conditions is prevalent over all homo-genous stations, except Kashgar, a county‐side station located in remote northwestern China and thus lessinfluenced by anthropogenic aerosols (Figures 1 and 6). Though there was no significant trend in all‐skyRf during 1961–1989, clear‐sky Rf significantly increased by 3.4 W m−2 decade−1 (Table 1), consistent withthe finding of Qian et al. (2007). Under all‐sky conditions, significant increasing Rf trends only cover 18%of the stations; the ratio increases to 64% under clear‐sky conditions (Figure 6b). Rf% also shows higherincreasing trends under clear‐sky than all‐sky conditions over all examined homogenous stations, exceptKashgar, with the average increase by 2.6% decade−1 under clear‐sky condition (Figure 6c and Table 1).As hypothesized, this suggests that aerosols have contributed in decreasing Rd but increasing Rf duringthe global dimming phase.

During the global brightening phase of 1993–2016, the mean clear‐sky series shows a significant increasingtrend in Rd by 5.5 Wm−2 decade−1, which quadrupled to 23.4 Wm−2 decade−1 after 2008 over the 11 homo-genous stations across China (Figure 3a and Table 1). Two exceptions were in Kunming and Shanghai(Figure 6a). For Kunming, Rd showed a less significant brightening trend under clear‐sky conditions, indi-cating the key role of cloud in driving recent all‐sky Rd brightening in this site. For Shanghai, the dimmingof Rd continued into the 1990s especially under clear‐sky conditions, suggesting a pollution‐driven Rd dim-ming in this site. The transition in Rf/Rf% from leveling off/increase to decrease since 2008 is more obviousunder clear‐sky than all‐sky conditions (Figures 3b and 3c). The most obvious transition occurs in the sta-tions of Beijing and Shenyang in the region of north China (Figures 6b and 6c). Meanwhile, multiple efforts

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 12 of 19

have been made by the Chinese government to bring the blue skies back to this region for the 2008 BeijingOlympic Games (Jin et al., 2016). The most influential regulation was a nationwide control on SO2 emissionssince 2006 implemented by the 11th Five‐Year Plan of China (http://www.gov.cn). Its implication for theshift proportion of the Rd and Rf components is discussed in the next section. Overall, it is very likely thataerosols have played the main role in driving the recent reversals of the Rd and Rf trends since 2008.

5.3. SO2 Emissions

With the economic growth powered by coal‐dominated fossil fuel consumption, China was the largest SO2

emitter in the world for many years (Gao et al., 2009). SO2, when formed into sulfate aerosols, may scatterincoming solar radiation and produce less Rd and more Rf reaching the Earth's surface (Kanniah et al.,2012). The reflection of solar radiation back to space by sulfate aerosols and their indirect effects on cloudalbedo, exerting one of the largest anthropogenic radiative forcing of the atmosphere, could cancel virtuallyall the greenhouse forcing (Kiehl & Briegleb, 1993; Mitchell et al., 1995). In order to control SO2 emissions,which count for amajority of the total pollutant emissions in China, the Chinese government has implemen-ted numerous regulations, with the 11th Five‐Year Plan for the first time compulsorily giving quantitative

Figure 6. Comparison of clear‐sky and all‐sky trends of (a) direct radiation (Rd, W m−2 decade−1), (b) diffuse radiation (Rf, W m−2 decade−1), (c) diffuse fraction(Rf fraction, % decade−1) over the 11 first‐class stations across China for the periods of 1961–1989, 1993–2016, and 2008–2016.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 13 of 19

emission control targets (Jin et al., 2016). Consequently, a transition to decrease can be noted in the trend ofSO2 emissions with the peak in 2006 averaged over the provinces that contain the homogenous stationsacross China (Figure 7a). A corresponding turning point is shown in the all‐sky Rf% trend sincethe mid‐2000s.

A consistency in the trends of SO2 emission and Rf% can also be generally observed for the individual sta-tions (Figures 7b–7l). In Beijing (Figure 7b), the capital of China, a declining trend in SO2 emissions can evenbe traced back to the middle of 1990s, when the Air Pollution Prevention and Control Law was amendedwith the key to control SO2 emitted by coal burning (Gao et al., 2009). SO2 emissions in Beijing reboundedin 2001 but had turned to decrease again since around 2006 when the 11th Five‐Year Plan was released. As aresponse, the increasing all‐sky Rf% trend in Beijing had reversed to decrease since around 2008, which couldnot be explained by the continued increasing TCC trend (Figure 5). A better agreement in the trendsbetween Rf% and SO2 is shown for clear‐sky conditions when the cloud effects are largely removed; theclear‐sky Rf% trend of Beijing had decreased even since the 1990s (Figure 6c), more consistent with thedeclining SO2 emissions since the 1990s (Figure 7b). For the surrounding stations of Beijing, that is,

Figure 7. Comparisons in the (a) overall and (b–l) individual trends between all‐sky diffuse fraction (Rf%, or Rf fraction, %) and SO2 emission (million ton) overthe 11 homogenous stations across China for 1993–2016. The data for diffuse fraction and corresponding SO2 emission are collected at local and provincial scales,respectively. Names of the stations and their belonged provinces are shown in the subtitles and legends of the panels, respectively.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 14 of 19

Harbin (Figure 7d), Zhengzhou (Figure 7e), and Shenyang (Figure 7g), SO2 emissions tend to slightlydecrease only after 2006 with the drop mainly in the end of the examined period of 2016, when the 13thFive‐Year Plan was released with an augmented control target on main pollutant emissions. Consistently,the Rf% trends in the three stations change from an increase to leveling off since the mid‐2000s with anobvious decrease mainly shown in the last year of 2016. The obvious drop of SO2 emissions in 2016 mightalso relate to special clean air actions for the 2016 Hangzhou G20meetings (Tilt, 2019). This can be indicatedin the SO2 emission trend in its surrounding city, Shanghai (Figure 7i), which shows a drop in 2016, as wellas in 2010, when the Shanghai 2010 World EXPO was held. The transition to decrease in the SO2 emissiontrend of Shanghai starts from 2006, when the increasing Rf% trend stabilized (Figure 7i). A similar phenom-enon can be found in another station located in the Yangtze River basin, namely, Wuhan (Figure 7f).

Among all examined stations, the most obvious transition since 2006 in the SO2 emission trend from a steepincrease to a steep decrease occurs in Guangzhou (Figure 7c). There is, however, no change in the Rf% trendof Guangzhou, which generally leveled off with a slight increase during 1993–2016. The continued increas-ing trend in TCC (Figures 5k and 5l) might have partially compensated for the effect from decreasing SO2

emissions on the Rf% trend, which can be indicated by the more obvious decreasing Rf% trend inGuangzhou under clear‐sky conditions (Figure 6c). By contrast, little has been done by the Chinese govern-ment to control SO2 emissions in remote regions. SO2 emissions generally keep increasing in the Qinghai(Figure 7h) and Xinjiang (Figures 7j and 7k) provinces, with a sharp decrease only shown in 2016. Theall‐sky Rf% trends of Geermu (Figure 7h), Urumqi (Figure 7j), and Kashgar (Figure 7k) show no obviouschanges; a transition to decrease in the Rf% trends can be detected under clear‐sky conditions (Figure 6c).In the Yunnan (Figure 7l) province, SO2 emissions control became effective only since 2011, the implemen-tation of the 12th Five‐Year Plan. An earlier transition is, however, noted in the Rf% trend over its capital ofKunming (Figure 7l), possibly due to the decrease in the TCC trend (Figure 5). To summarize, the Chinesegovernment has executed effective control on SO2 emission since the 11th Five‐Year Plan especially overregions with dense sulfate aerosols, which seems to largely help in altering the Rd and Rf proportions inChina as well as the transition in recent Rd and Rf trends.

6. Discussions

The existent inconsistency between the Rf% and SO2 emission trends points to impacts from other types ofaerosols and possible changes in cloud properties due to the indirect and semidirect aerosol effects.Besides the effects from scattering aerosols, for example, SO2, the reduction of black carbon, a type of absorb-ing aerosol, has been suggested to play a main role in the variations of Rf% over the North China Plain (Liet al., 2018). Direct observations have also indicated a significant role of nitrogen dioxide in the absorptionof solar radiation in both clear and cloudy skies (Solomon et al., 1999). Dust aerosols may alter the atmo-spheric radiation budget by directly scattering and partly absorbing shortwave and longwave radiation(Choobari et al., 2014). Effective regulations on the national emissions of NOX, smoke, and dust have alsobeen reported during the 11th and 12th Five‐Year Plan periods (Jin et al., 2016). Accordingly, various datasources, such as reanalysis data (e.g., MERRA‐2, the Modern Era Retrospective‐Analysis for Research andApplications) and satellite data (e.g., MODIS, the MODerate resolution Imaging Spectroradiometer;MISR, theMulti‐angle Imaging Spectroradiometer; and ATSR, the Along‐Track Scanning Radiometer), haveindicated a reduction in AOD over China since the mid‐2000s (Filonchyk et al., 2019; He et al., 2016; Qinet al., 2018; Sogacheva et al., 2018; Zhang et al., 2017; Zhao et al., 2017).

In principle, through interactions with clouds, aerosols may induce an indirect radiative forcing by acting ascloud condensation nuclei, thereby influencing cloud radiative properties and lifetimes; furthermore,absorbing aerosols cause a semidirect radiative forcing by heating up the atmosphere to inhibiting cloud for-mation or dissolving existing clouds (Wild, 2009). The complex aerosol‐cloud‐radiation interactions make itdifficult to completely discriminate the radiative effects from clouds and aerosols, though an attempt hasbeen made by comparing the clear‐sky and all‐sky Rd and Rf trends. Furthermore, TCC may not fully repre-sent the cloud radiative effects. Besides cloud cover, a full understanding of the cloud type with physical andradiative properties is required to completely characterize the cloud effects on solar radiation (Liang & Xia,2005; Qian et al., 2006; Xia, 2010b). For example, Tang et al. (2017) highlighted the roles of cloud propertiesand the interactions between clouds and aerosols in determining the long‐term variations of R estimated

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 15 of 19

from various meteorological variables (e.g., SD) over 519 stations across China. Since 9 out of the selected 11homogenous stations are located in urban areas, which serve as a main source for anthropogenic aerosols(Imamovic et al., 2016; Wang et al., 2017), the applied attribution analysis might be influenced by a potentialurbanization effect on the derived trends.

Though efforts have been made by this research to ensure the analyzed time series of Rd and Rf are homo-geneous, the proposed trends and patterns are unfortunately still limited by the spatial representation of sur-face observations. Clarifying the climatology and trends of Rd and Rf is essential for a more effectiveproduction of solar energy and agricultural crops. Based on surface observations of R, it has been estimatedthat if the radiation levels in China could revert back to the 1960s, electricity generation may increase by 12–13%, with the corresponding economic benefits amounting to USD 1.9 billion (Sweerts et al., 2019). In thisresearch, we evidenced that the recovery of Rd has almost reached the level of the 1960s, and such recoverymainly occurred in the season and region with abundant Rd resources optimal for solar power generation.These pieces of evidence provide a bright prospect for future solar energy development in China.Increasing Rf with a fertilization effect on agricultural productivity has been estimated by the Joint UKLand Environment Simulator (JULES) model to enhance the global land carbon sink by approximatelyone quarter, despite the global dimming in R between 1960 and 1999 (Mercado et al., 2009). An enhancedplant productivity can also be assumed in China, in view of the general increasing trend in the surface obser-vations of Rf since the 1960s, especially over the east region optimum for crop planting. Such assumption onthe agricultural implications, however, still needs to be proven by further investigations, with the considera-tion also of the recent transition to decrease in the Rf trend of China.

7. Conclusions

The quality of daily Rd and Rf measurements across China from 1961 to 2016 was controlled using a numberof tests. After excluding inhomogeneous time series and outliers, this study revisited the Rd and Rf climatol-ogy in China, clarified the relative contributions of the Rd and Rf components to the decadal changes of R(i.e., the global dimming and brightening phenomena) in China, and made a first step forward in identifyingthe main driving force of the variations in the Rd and Rf components in China from the two most likely can-didates (i.e., clouds and aerosols). The main findings of this study are summarized as follows:

1. The decadal variations in R, that is, the global dimming and brightening phenomena, are mainly a con-sequence of the changes in the Rd component. Opposite decadal changes from a brightening to dimmingare generally shown for Rf, especially Rf%. Different from the transition year of 1990 proposed for the glo-bal trend of solar radiation, the actual transition to brightening in Rd and to dimming in Rf occurred in2008 over China, in line with recent evidence for a transition of R. Summer is the season showing themost significant decadal changes in Rd and Rf.

2. Of relevance for regional environmental implications are the findings that during the past half century,Rd tended to further decrease over regions with low Rd energy reserves but further increase especiallysince the mid‐2000s over western China with high Rd energy potential beneficial for solar power genera-tion. Meanwhile, the proportion of Rf resources has been further enhanced over eastern China, the areawell suited for crop planting, with a high proportion of Rf resources, though the trend of Rf has reversed todecrease in the very recent decade especially over the northern region.

3. Aerosols have been suggested as the main driving force of the variations in Rd and Rf over China based onthree lines of evidence: first, clear‐sky trends indicate a notable impact of volcanic aerosols on the dra-matic variations of Rd and Rf for 1990–1993, the period showing an unusual jump in most published timeseries of R over China; second, the recent Rd brightening and Rf dimming since the mid‐2000s shows agood consistency with recent reduction in pollutant emissions over China; third, TCC kept increasingwithout showing any transition in 2008, which might have even partially counteracted the tendency tobrightening in Rd and to dimming in Rf.

ReferencesAbakumova, G., Feigelson, E., Russak, V., & Stadnik, V. (1996). Evaluation of long‐term changes in radiation, cloudiness, and surface

temperature on the territory of the former Soviet Union. Journal of Climate, 9(6), 1319–1327.Alexandersson, H. (1986). A homogeneity test applied to precipitation data. International Journal of Climatology, 6(6), 661–675.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 16 of 19

AcknowledgmentsThis research was funded by theNational Natural Science Foundation ofChina (41971018, 41805128, and41501036), the National Key BasicResearch Program of China (Grant2016YFA0602701), the National YouthTop‐Notch Talent Support Program(2015‐48), the Changjiang YoungScholars Program of China (Q2016161),and the Fundamental Research Fundsfor the Central Universities (19lgjc02).A. Sanchez‐Lorenzo was supported by apostdoctoral fellowship RYC‐2016–20784 funded by the Spanish Ministryof Science and Innovation. GlobalDimming/Brightening research at ETHZurich was funded by the SwissNational Science Foundation Grants200021_135395, 200020_159938, and200020_188601. Data supporting thisresearch are available from the ChinaMeteorological Data Service Center(CMDC, http://data.cma.cn) of theChina Meteorological Administration(CMA).

Allen, R. G., Pereira, L. S., Raes, D., & Smith, M. (1998). FAO irrigation and drainage paper no. 56, (Vol. 56, e156). Rome: Food andAgriculture Organization of the United Nations.

Black, K., Davis, P., Lynch, P., Jones, M., McGettigan, M., & Osborne, B. (2006). Long‐term trends in solar irradiance in Ireland and theirpotential effects on gross primary productivity. Agricultural and Forest Meteorology, 141(2–4), 118–132.

Calbó, J., González, J.‐A., & Sanchez‐Lorenzo, A. (2017). Building global and diffuse solar radiation series and assessing decadal trends inGirona (NE Iberian Peninsula). Theoretical and Applied Climatology, 129(3–4), 1003–1015.

Che, H., Shi, G., Zhang, X., Arimoto, R., Zhao, J., Xu, L., et al. (2005). Analysis of 40 years of solar radiation data from China, 1961–2000.Geophysical Research Letters, 32, L06803. https://doi.org/10.1029/2004GL022322

Choobari, O. A., Zawar‐Reza, P., & Sturman, A. (2014). The global distribution of mineral dust and its impacts on the climate system: Areview. Atmospheric Research, 138, 152–165. http://www.sciencedirect.com/science/article/pii/S0169809513003281

Feng, L., Lin, A., Wang, L., Qin, W., & Gong, W. (2018). Evaluation of sunshine‐based models for predicting diffuse solar radiation inChina. Renewable and Sustainable Energy Reviews, 94, 168–182. http://www.sciencedirect.com/science/article/pii/S1364032118304477

Feng, Y., Cui, N., Zhang, Q., Zhao, L., & Gong, D. (2017). Comparison of artificial intelligence and empirical models for estimation of dailydiffuse solar radiation in North China Plain. International Journal of Hydrogen Energy, 42(21), 14,418–14,428. http://www.sciencedirect.com/science/article/pii/S0360319917314738

Feng, Y., & Li, Y. (2018). Estimated spatiotemporal variability of total, direct and diffuse solar radiation across China during 1958–2016.International Journal of Climatology, 38(12), 4395–4404. https://rmets.onlinelibrary.wiley.com/doi/abs/10.1002/joc.5676

Filonchyk, M., Yan, H., Zhang, Z., Yang, S., Li, W., & Li, Y. (2019). Combined use of satellite and surface observations to study aerosoloptical depth in different regions of China. Scientific Reports, 9(1), 6174. https://doi.org/10.1038/s41598‐019‐42466‐6

Fu, C., Dan, L., Chen, Y., & Tang, J. (2015). Trends of the sunshine duration and diffuse radiation percentage on sunny days in urbanagglomerations of China during 1960–2005. Journal of Environmental Sciences, 34, 206–211.

Gao, C., Yin, H., Ai, N., & Huang, Z. (2009). Historical analysis of SO2 pollution control policies in China. Environmental Management,43(3), 447–457. https://doi.org/10.1007/s00267‐008‐9252‐x

Gu, L., Baldocchi, D., Verma, S. B., Black, T., Vesala, T., Falge, E. M., & Dowty, P. R. (2002). Advantages of diffuse radiation for terrestrialecosystem productivity. Journal of Geophysical Research‐Atmospheres, 107(D6). https://doi.org/10.1029/2001JD001242

Gu, L., Baldocchi, D. D., Wofsy, S. C., Munger, J. W., Michalsky, J. J., Urbanski, S. P., & Boden, T. A. (2003). Response of a deciduous forestto the Mount Pinatubo eruption: Enhanced photosynthesis. Science, 299(5615), 2035–2038. https://doi.org/10.1126/science.1078366

He, Q., Zhang, M., & Huang, B. (2016). Spatio‐temporal variation and impact factors analysis of satellite‐based aerosol optical depth overChina from 2002 to 2015. Atmospheric Environment, 129, 79–90.

He, Y., &Wang, K. (2020). Variability in direct and diffuse solar radiation across China from 1958 to 2017. Geophysical Research Letters, 47,e2019GL084570. https://doi.org/10.1029/2019GL084570

Imamovic, A., Tanaka, K., Folini, D., &Wild, M. (2016). Global dimming and urbanization: Did stronger negative SSR trends collocate withregions of population growth? Atmospheric Chemistry and Physics, 16(5), 2719–2725. http://www.atmos‐chem‐phys.net/16/2719/2016/

Jacovides, C., Tymvios, F., Assimakopoulos, V., & Kaltsounides, N. (2006). Comparative study of various correlations in estimating hourlydiffuse fraction of global solar radiation. Renewable Energy, 31(15), 2492–2504.

Jin, Y., Andersson, H., & Zhang, S. (2016). Air pollution control policies in China: A retrospective and prospects. International Journal ofEnvironmental Research and Public Health, 13(12), 1219.

Kambezidis, H. D. (2018). The solar radiation climate of Athens: Variations and tendencies in the period 1992–2017, the brightening era.Solar Energy, 173, 328–347. http://www.sciencedirect.com/science/article/pii/S0038092X18307400

Kanniah, K. D., Beringer, J., North, P., & Hutley, L. (2012). Control of atmospheric particles on diffuse radiation and terrestrial plantproductivity: A review. Progress in Physical Geography, 36(2), 209–237.

Khaliq, M. N., & Ouarda, T. B. M. J. (2007). On the critical values of the standard normal homogeneity test (SNHT). International Journal ofClimatology, 27(5), 681–687. https://doi.org/10.1002/joc.1438

Kiehl, J. T., & Briegleb, B. P. (1993). The relative roles of sulfate aerosols and greenhouse gases in climate forcing. Science, 260(5106),311–314. https://doi.org/10.1126/science.260.5106.311

Kvalevåg, M. M., & Myhre, G. (2007). Human impact on direct and diffuse solar radiation during the industrial era. Journal of Climate,20(19), 4874–4883.

Li, J., Jiang, Y. W., Xia, X. G., & Hu, Y. Y. (2018). Increase of surface solar irradiance across East China related to changes in aerosolproperties during the past decade. Environmental Research Letters, 13(3), 9.

Liang, F., & Xia, X. A. (2005). Long‐term trends in solar radiation and the associated climatic factors over China for 1961–2000. AnnalesGeophysicae, 23(7), 2425–2432.

Long, C. N., Dutton, E. G., Augustine, J., Wiscombe, W., Wild, M., McFarlane, S. A., & Flynn, C. J. (2009). Significant decadal brightening ofdownwelling shortwave in the continental United States. Journal of Geophysical Research, 114, D00D06. https://doi.org/10.1029/2008JD011263

Ma, J., Liang, H., Luo, Y., & Li, S.‐K. (2011). Variation trend of direct and diffuse radiation in China over recent 50 years. Acta PhysicaSinica, 60(6), 069601.

Ma, Z., Liu, R., Liu, Y., & Bi, J. (2019). Effects of air pollution control policies on PM2.5 pollution improvement in China from 2005 to2017: A satellite‐based perspective. Atmospheric Chemistry and Physics, 19(10), 6861–6877. https://www.atmos‐chem‐phys.net/19/6861/2019/

Manara, V., Brunetti, M., Celozzi, A., Maugeri, M., Sanchez‐Lorenzo, A., & Wild, M. (2016). Detection of dimming/brightening in Italyfrom homogenized all‐sky and clear‐sky surface solar radiation records and underlying causes (1959–2013). Atmospheric Chemistry andPhysics, 16(17), 11,145–11,161. https://www.atmos‐chem‐phys.net/16/11145/2016/

Mercado, L. M., Bellouin, N., Sitch, S., Boucher, O., Huntingford, C., Wild, M., & Cox, P. M. (2009). Impact of changes in diffuse radiationon the global land carbon sink. Nature, 458(7241), 1014–1017. https://doi.org/10.1038/nature07949

Mitchell, J. F. B., Johns, T. C., Gregory, J. M., & Tett, S. F. B. (1995). Climate response to increasing levels of greenhouse gases and sulphateaerosols. Nature, 376(6540), 501–504. https://doi.org/10.1038/376501a0

Power, H. (2003). Trends in solar radiation over Germany and an assessment of the role of aerosols and sunshine duration. Theoretical andApplied Climatology, 76(1–2), 47–63.

Power, H., & Mills, D. (2005). Solar radiation climate change over southern Africa and an assessment of the radiative impact of volcaniceruptions. International Journal of Climatology, 25(3), 295–318.

Qian, Y., Kaiser, D. P., Leung, L. R., & Xu, M. (2006). More frequent cloud‐free sky and less surface solar radiation in China from 1955 to2000. Geophysical Research Letters, 33, L01812. https://doi.org/10.1029/2005GL024586

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 17 of 19

Qian, Y., Wang, W., Leung, L. R., & Kaiser, D. P. (2007). Variability of solar radiation under cloud‐free skies in China: The role of aerosols.Geophysical Research Letters, 34, L12804. https://doi.org/10.1029/2006GL028800

Qin, W., Liu, Y., Wang, L., Lin, A., Xia, X., Che, H., et al. (2018). Characteristic and driving factors of aerosol optical depth over mainlandChina during 1980–2017. Remote Sensing, 10(7), 1064. https://doi.org/10.3390/rs10071064

Reindl, D. T., Beckman, W. A., & Duffie, J. A. (1990). Diffuse fraction correlations. Solar Energy, 45(1), 1–7. http://www.sciencedirect.com/science/article/pii/0038092X9090060P

Ren, X., He, H., Zhang, L., Zhou, L., Yu, G., & Fan, J. (2013). Spatiotemporal variability analysis of diffuse radiation in China during 1981–2010. Annales Geophysicae, 31(2), 277–289.

Russak, V. (2009). Changes in solar radiation and their influence on temperature trend in Estonia (1955–2007). Journal of GeophysicalResearch, 114, D00D01. https://doi.org/10.1029/2008JD010613

Sanchez‐Lorenzo, A., Calbó, J., & Wild, M. (2013). Global and diffuse solar radiation in Spain: Building a homogeneous dataset andassessing their trends. Global and Planetary Change, 100, 343–352. http://www.sciencedirect.com/science/article/pii/S0921818112002238

Sanchez‐Lorenzo, A., Wild, M., Brunetti, M., Guijarro, J. A., Hakuba, M. Z., Calbó, J., et al. (2015). Reassessment and update of long-termtrends in downward surface shortwave radiation over Europe (1939–2012). Journal of Geophysical Research:Atmospheres, 120(18),9555–9569. https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1002/2015JD023321

Shi, G.‐Y., Hayasaka, T., Ohmura, A., Chen, Z.‐H., Wang, B., Zhao, J.‐Q., et al. (2008). Data quality assessment and the long‐term trend ofground solar radiation in China. Journal of Applied Meteorology and Climatology, 47(4), 1006–1016. https://doi.org/10.1175/2007JAMC1493.1

Sogacheva, L., Rodriguez, E., Kolmonen, P., Virtanen, T. H., & Saponaro, G. (2018). Spatial and seasonal variations of aerosols over Chinafrom two decades of multi‐satellite observations. Part II: AOD time series for 1995–2017 combined from ATSR ADV and MODIS C6. 1for AOD tendencies estimation. Atmospheric Chemistry and Physics, 18, 16631–16652.

Solomon, S., Portmann, R. W., Sanders, R. W., Daniel, J. S., Madsen,W., Bartram, B., & Dutton, E. G. (1999). On the role of nitrogen dioxidein the absorption of solar radiation. Journal of Geophysical Research, 104(D10), 12,047–12,058. https://doi.org/10.1029/1999JD900035

Soni, V., Pandithurai, G., & Pai, D. (2012). Evaluation of long‐term changes of solar radiation in India. International Journal of Climatology,32(4), 540–551.

Sprintsin, M., Chen, J. M., Desai, A., & Gough, C. M. (2012). Evaluation of leaf‐to‐canopy upscaling methodologies against carbon flux datain North America. Journal of Geophysical Research, 117, G01023. https://doi.org/10.1029/2010JG001407

Stanhill, G. (1998). Long‐term trends in, and spatial variation of, solar irradiances in Ireland. International Journal of Climatology, 18(9),1015–1030.

Stanhill, G., & Cohen, S. (2001). Global dimming: A review of the evidence for a widespread and significant reduction in global radiationwith discussion of its probable causes and possible agricultural consequences. Agricultural and Forest Meteorology, 107(4), 255–278.

Sweerts, B., Pfenninger, S., Yang, S., Folini, D., van der Zwaan, B., & Wild, M. (2019). Estimation of losses in solar energy production fromair pollution in China since 1960 using surface radiation data. Nature Energy, 4(8), 657–663. https://doi.org/10.1038/s41560‐019‐0412‐4

Tang, W., Yang, K., Qin, J., Niu, X., Lin, C., & Jing, X. (2017). A revisit to decadal change of aerosol optical depth and its impact on globalradiation over China. Atmospheric Environment, 150, 106–115.

Tang, W. J., Yang, K., Qin, J., Cheng, C. C. K., & He, J. (2011). Solar radiation trend across China in recent decades: A revisit withquality‐controlled data. Atmospheric Chemistry and Physics, 11(1), 393–406.

Tilt, B. (2019). China's air pollution crisis: Science and policy perspectives. Environmental Science & Policy, 92, 275–280.Wang, C. H., Zhang, Z. F., & Tian, W. S. (2011). Factors affecting the surface radiation trends over China between 1960 and 2000.

Atmospheric Environment, 45(14), 2379–2385.Wang, K., Ma, Q., Li, Z., & Wang, J. (2015). Decadal variability of surface incident solar radiation over China: Observations, satellite

retrievals, and reanalyses. Journal of Geophysical Research‐Atmospheres, 120, 6500–6514. https://doi.org/10.1002/2015JD023420Wang, L., Lu, Y., Zou, L., Feng, L., Wei, J., Qin, W., & Niu, Z. (2019). Prediction of diffuse solar radiation based on multiple variables in

China. Renewable and Sustainable Energy Reviews, 103, 151–216. http://www.sciencedirect.com/science/article/pii/S136403211830827XWang, Y., Trentmann, J., Pfeifroth, U., Yuan, W., &Wild, M. (2019). Improvement of air pollution in China inferred from changes between

satellite‐based and measured surface solar radiation. Remote Sensing, 11(24), 2910. https://doi.org/10.3390/rs11242910Wang, Y., & Wild, M. (2016). A new look at solar dimming and brightening in China. Geophysical Research Letters, 43, 11,777–11,785.

https://doi.org/10.1002/2016GL071009Wang, Y., Wild, M., Sanchez‐Lorenzo, A., & Manara, V. (2017). Urbanization effect on trends in sunshine duration in China. Annales

Geophysicae, 35(4), 839–851. https://www.ann‐geophys.net/35/839/2017/Wang, Y., Xin, J., Li, Z., Wang, S., Wang, P., Hao, W.M., et al. (2011). Seasonal variations in aerosol optical properties over China. Journal of

Geophysical Research, 116, D18209. https://doi.org/10.1029/2010JD015376Wang, Y., & Yang, Y. (2014). China's dimming and brightening: Evidence, causes and hydrological implications. Annales Geophysicae,

32(1), 41–55.Wang, Y., Yang, Y., Han, S., Wang, Q., & Zhang, J. (2013). Sunshine dimming and brightening in Chinese cities (1955–2011) was driven by

air pollution rather than clouds. Climate Research, 56(1), 11–20. http://www.int‐res.com/abstracts/cr/v56/n1/p11‐20/Wang, Y., Yang, Y., Zhao, N., Liu, C., &Wang, Q. (2012). Themagnitude of the effect of air pollution on sunshine hours in China. Journal of

Geophysical Research, 117, D00V14. https://doi.org/10.1029/2011JD016753Wang, Y., Yang, Y., Zhou, X., Zhao, N., & Zhang, J. (2014). Air pollution is pushing wind speed into a regulator of surface solar irradiance in

China. Environmental Research Letters, 9(5), 054004.Wild, M. (2009). Global dimming and brightening: A review. Journal of Geophysical Research, 114, D00D16. https://doi.org/10.1029/

2008JD011470Wild, M. (2012). Enlightening global dimming and brightening. Bulletin of the American Meteorological Society, 93(1), 27–37.Wild, M., Gilgen, H., Roesch, A., Ohmura, A., Long, C. N., Dutton, E. G., et al. (2005). From dimming to brightening: Decadal changes in

solar radiation at Earth's surface. Science, 308(5723), 847–850. https://science.sciencemag.org/content/308/5723/847, https://doi.org/10.1126/science.1103215

Wild, M., Roesch, A., & Ammann, C. (2012). Global dimming and brightening–Evidence and agricultural implications. CAB Reviews,7(003), 1–7.

Wu, F., & Fu, C. (2011). Assessment of GEWEX/SRB version 3.0 monthly global radiation dataset over China.Meteorology and AtmosphericPhysics, 112(3‐4), 155–166. https://doi.org/10.1007/s00703‐011‐0136‐x

Xia, X. (2010a). A closer looking at dimming and brightening in China during 1961–2005. Annales Geophysicae, 28(5), 1121–1132.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 18 of 19

Xia, X. A. (2010b). Spatiotemporal changes in sunshine duration and cloud amount as well as their relationship in China during 1954–2005.Journal of Geophysical Research, 115, D00K06. https://doi.org/10.1029/2009JD012879

Yang, S., Wang, X. L., & Wild, M. (2018). Homogenization and trend analysis of the 1958–2016 in situ surface solar radiation records inChina. Journal of Climate, 31(11), 4529–4541.

Yang, S., Wang, X. L., & Wild, M. (2019). Causes of dimming and brightening in China inferred from homogenized daily clear‐sky andall‐sky in situ surface solar radiation records (1958–2016). Journal of Climate, 32(18), 5901–5913. https://doi.org/10.1175/JCLI‐D‐18‐0666.1

Zhang, J., Reid, J. S., Alfaro‐Contreras, R., & Xian, P. (2017). Has China been exporting less particulate air pollution over the past decade?Geophysical Research Letters, 44, 2941–2948. https://doi.org/10.1002/2017GL072617

Zhao, B., Jiang, J. H., Gu, Y., Diner, D., Worden, J., Liou, K.‐N., et al. (2017). Decadal‐scale trends in regional aerosol particle properties andtheir linkage to emission changes. Environmental Research Letters, 12(5), 054021. https://doi.org/10.1088/1748‐9326/aa6cb2

Zhao, D., Luo, Y., Gao, G., & Zhu, C. (2009). Essential characteristics of solar direct radiation over recent 50 years in China. Acta EnergiaeSolaris Sinica, 30(7), 946–952.

Zhu, C.‐H. (1984). The calculation and distribution of diffuse radiation in China. Acta Energiae Solaris Sinica, 3, 242–249.Zuo, D., & Gong, R. (1962). The correlations among direct, diffuse, and total solar radiation in China. Acta Geographica Sinica, 3, 175–186.

10.1029/2020JD032634Journal of Geophysical Research: Atmospheres

WANG ET AL. 19 of 19