A revisit to decadal change of aerosol optical depth and its impact on global radiation over China

Accepted Manuscript A revisit to decadal change of aerosol optical depth and its impact on global radiation over China Wenjun Tang, Kun Yang, Jun Qin, Xiaolei Niu, Changgui Lin, Xianwen Jing PII:

S1352-2310(16)30927-X

DOI:

10.1016/j.atmosenv.2016.11.043

Reference:

AEA 15036

To appear in:

Atmospheric Environment

Received Date: 18 September 2016 Revised Date:

14 November 2016

Accepted Date: 16 November 2016

Please cite this article as: Tang, W., Yang, K., Qin, J., Niu, X., Lin, C., Jing, X., A revisit to decadal change of aerosol optical depth and its impact on global radiation over China, Atmospheric Environment (2016), doi: 10.1016/j.atmosenv.2016.11.043. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT A revisit on decadal change of aerosol optical depth and its impact on global radiation over China Wenjun Tang1,2, Kun Yang1,2, Jun Qin1，Xiaolei Niu1, Changgui Lin3, Xianwen Jing4

RI PT

1. Key Laboratory of Tibetan Environment Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing 100101, China.

SC

2. CAS Center for Excellence in Tibetan Plateau Earth Sciences, Chinese Academy

M AN U

of Sciences, Beijing 100101, China.

3. Regional Climate Group, Department of Earth Sciences, University of Gothenburg, Gothenburg, Sweden.

4. Laboratory for Climate Studies, National Climate Center, China Meteorological

TE D

Administration, Beijing 100081, China. Corresponding author and address: Wenjun Tang, Dr.

EP

Institute of Tibetan Plateau Research, Chinese Academy of Sciences

AC C

Building 3, Courtyard 16, Lin Cui Road, Chaoyang District, Beijing 100101, China Email: [email protected] Tel: +86-10-84097046 Fax: +86-10-8409707

1

ACCEPTED MANUSCRIPT Abstract

2

Global radiation over China decreased between the 1960s and 1990, since when it has

3

remained stable. As the total cloud cover has continued to decrease since the 1960s,

4

variations in aerosols were suggested in previous studies to be the primary cause for

5

variations in global radiation over China. However, the effect of aerosols on global

6

radiation on a decadal scale has not been physically quantified over China. In this

7

study, aerosol optical depth (AOD) data since 1980 are estimated by combining

8

horizontal visibility data at stations in China and AOD observed by the moderate

9

resolution imaging spectroradiometer (MODIS). It is found that the AOD exhibits

10

decadal changes, with two decreasing periods (before the end of 1980s and after 2006)

11

and one increasing period (from 1990 to 2006). With the derived AOD, a clear-sky

12

model is then applied to quantify the role of aerosols in the variations in global

13

radiation over China. The results show that aerosol direct effect cannot fully explain

14

the decadal variations in the global radiation over China between 1980 and 2010,

15

though it has a considerable effect on global radiation climatology. There are

16

significant differences between the trends of clear-sky global radiation impacted by

17

aerosols and those of all-sky global radiation impacted by aerosols and clouds, and the

18

correlation coefficient for the comparison is very low. Therefore, the variations in

19

all-sky global radiation over China are likely to be due to changes in cloud properties

20

and to interactions between clouds and aerosols.

21

Keywords: Global radiation; aerosol optical depth; clear-sky; visibility

AC C

EP

TE D

M AN U

SC

RI PT

1

2

ACCEPTED MANUSCRIPT 22

1. Introduction Global radiation E g ↓ over most regions of the Earth experienced a transition

24

from dimming to brightening around the late 1980s or early 1990s, based on ground

25

observations and satellite retrievals [Stanhill and Moreshet, 1994; Stanhill and Cohen,

26

1997; Stanhill and Cohen, 2001; Liepert, 2002; wild et al. 2005; Che et al., 2005;

27

Pinker et al., 2005; Shi et al., 2008; Gilgen et al., 2009; Stanhill and Cohen, 2009;

28

Wild et al., 2009; Wild, 2012a]. Variations in global radiation E g ↓ have profound

29

influences on the environmental, societal, and economic aspects of our habitats.

30

Decadal variations in global radiation E g ↓ originate from changes in the

31

transparency of the atmosphere, which are mainly attributable to changes in clouds,

32

aerosols and water vapor [Wild, 2012a]. For example, a change in aerosol loading has

33

been suggested as the dominant factor for long-term variations in global radiation

34

over Europe [Norris and Wild, 2007; Ohmura, 2009; Folini and Wild, 2011], while

35

changes in cloud cover are the determining factor for long-term variations in global

36

radiation over the United States [Liepert, 2002; Long et al., 2009; Augustine and

37

Dutton, 2013].

SC

M AN U

TE D

EP

AC C

38

RI PT

23

In China, several studies [e.g., Zhang et al., 2004; Che et al. 2005; Liang and Xia,

39

2005; Xia et al., 2006a; Streets et al., 2006; Wang et al., 2012; Wang and Yang, 2014]

40

speculated that changes in atmospheric aerosols are the dominant cause for global

41

radiation variations over China, given that the total cloud cover (TCC) measured at

42

China Meteorological Administration (CMA) stations has decreased since 1954

43

[Kaiser, 1998; Kaiser, 2000; Qian et al., 2006; Xia, 2010]. However, aerosol 3

ACCEPTED MANUSCRIPT emissions over China continued increasing around 1990 [Wang et al., 2009] and so its

45

variation does not explain the change in the radiation transition from dimming to

46

slightly brightening around 1990. Indeed, Lin et al. [2015] established a statistical

47

wind speed –global radiation relationship and with which derived that aerosol direct

48

effect can only explain 20% of the decadal change in global radiation over China.

49

Furthermore, it should be noted that a decrease in TCC does not necessarily

50

correspond to a decrease in cloud optical thickness because of the diversity of possible

51

cloud shapes and types. For example, Yang et al. [2012] found that the TCC over the

52

Tibetan Plateau has decreased since 1984, while deep cloud cover has increased.

M AN U

SC

RI PT

44

Aerosols over China have drawn much attention over the past decade, due to

54

their potential effects on climate and the environment [Ramanathan et al. 2001; Wu et

55

al., 2013; Rosenfeld et al. 2014]. The annual mean aerosol optical depth (AOD)

56

averaged over China is about three times that averaged over all the Aerosol Robotic

57

Network (AERONET) sites [Li et al., 2011]. The aerosol radiative effect over China

58

can even reach to several tens of W m−2 on global radiation climatology [Li et al.,

59

2011], but physically-based quantitative evaluation of aerosol effects on long-term

60

variations in global radiation over China is challenging for the following reasons.

EP

AC C

61

TE D

53

First, a long-term observational dataset of aerosols over China is not available.

62

However, visibility is a parameter that is routinely measured at weather stations. This

63

parameter is often used to estimate near surface AOD [Vautard et al., 2009; Wang et

64

al., 2009; Wang et al., 2012]. Some radiation transfer software packages, such as

65

MODerate resolution atmospheric TRANsmission (MODTRAN) and Second 4

ACCEPTED MANUSCRIPT Simulation of a Satellite Signal in the Solar Spectrum (6S) [Berk et al., 1998;

67

Vermote et al., 1997] also use visibility to characterize near surface AOD information.

68

Second, it is difficult to distinguish the effects of clouds, aerosols, water vapor

69

and their interactions on long-term changes in global radiation [Wild et al., 2012b].

70

An analysis of long-term clear-sky global radiation may provide information that is

71

useful for the evaluation of the effects of aerosols on variations in global radiation.

72

Alternatively, using clear-sky radiation model with reliable source of aerosols can

73

quantitatively evaluate the contribution of aerosols to long-term changes in clear-sky

74

global radiation.

M AN U

SC

RI PT

66

In this study, we first use the observed visibility data to characterize long-term

76

variations in near surface AOD over China. We then use the MODIS monthly mean

77

AOD to correct the visibility-based AOD before application. Although the

78

visibility-based AOD may have great uncertainty, the long-term variation in the

79

visibility-based AOD is relatively reliable [Wang et al., 2009]. With the derived AOD

80

data, we try to quantify the effect of aerosols on changes of clear-sky global radiation

81

so as to evaluate the role of aerosols in global radiation variation over China. This

82

may help us to further evaluate the causes of the variations in global radiation over

83

China.

85

EP

AC C

84

TE D

75

2. Data

86

Three types of data are used in this study. The first is the CMA routine weather

87

data recorded at 519 stations where the data contain records for at least 20 days per 5

ACCEPTED MANUSCRIPT month from 1980 to 2010. The CMA observation data comprise air temperature,

89

relative humidity, surface pressure, sunshine duration, TCC, low cloud cover (LCC),

90

and visibility. These data are used to estimate clear-sky and all-sky global radiation

91

with Yang’s hybrid model [Yang et al., 2006]. Visual observations of cloud covers

92

(includes TCC and LCC) and visibility are taken by trained observer following the

93

World Meteorological Organizations (WMO) standards every six hours (0:00, 6:00,

94

12:00, and 18:00 GMT) at each station. Daily mean values of these CMA data are

95

used except visibility, for which the 6:00 GMT (14:00 Beijing Standard Time) values

96

is used. The geographical distribution of the 519 stations is shown in Figure 1. The

97

second is the MODIS Level-3 monthly-mean AOD data (Aqua) from 2002 to 2010,

98

which is used to correct the visibility-based AOD. The half-year climatology of

99

MODIS AOD is also presented in Figure 1. To investigate the decadal change of

100

aerosols and its effect on long-term variations in global radiation at a regional scale,

101

we divide mainland China into three sub-regions (see Figure 1), following Lin et al.

102

[2015], who defined the sub-regions according to the distribution of MODIS

103

multi-year mean AOD. The region of central-eastern China (CE) exhibits the highest

104

AOD values, and moderately lower AOD values are found in the region of southern

105

China (SC). The other regions (OT), including northeastern China, northwestern

106

China and the Tibetan Plateau have the lowest AOD among the three sub-regions. The

107

third is the quality-controlled all-sky global radiation dataset ( E qc ↓ , available at

108

http://dam.itpcas.ac.cn), which is developed with two datasets by Tang et al. [2013].

109

One is estimated with routine meteorological variables by Yang’s hybrid model at 716

AC C

EP

TE D

M AN U

SC

RI PT

88

6

ACCEPTED MANUSCRIPT individual CMA stations, and the other is estimated with the Artificial Neural

111

Network (ANN) based model at 96 individual CMA radiation stations. The former is

112

dynamically corrected by the latter at a monthly scale because the accuracy of the

113

latter is generally higher than the former. The quality-controlled all-sky global

114

radiation dataset ( Eqc ↓ ) was validated over China by Tang et al. [2013], and the

115

relative mean bias error (MBE) and root mean square error (RMSE) to the

116

measurements are about 0.8% and 12.7%, respectively. Tang et al. [2011] selected ten

117

CMA radiation stations with continuous and quality-consistent measurements to

118

validate the long-term variations in the all-sky global radiation estimated by Yang’s

119

hybrid model and the ANN-based model, and found that the long-term variations in

120

the estimated global radiation are generally reliable. Thus, we may believe that the

121

long-term variations of the quality-controlled all-sky global radiation dataset ( Eqc ↓ )

122

are reliable. These all-sky global radiation data are compared with the clear-sky global

123

radiation to quantify the role of aerosols in global radiation over China.

TE D

M AN U

SC

RI PT

110

Two issues should be noted when deriving aerosol information from visibility

125

data. One is that the haze information in visibility data is affected by the water vapor

126

and naturally occurring hydrometeors (such as fog, rain and snow), which should be

127

corrected and eliminated, respectively. Che et al. [2007] proposed a method to filter

128

visibility data, in which three rules were used: (i) Only use the 6:00 GMT values to

129

represent daily visibility because the noon observation of visibility is more

130

representative than the other three ones (two in night and one in morning) [Wu et al.,

131

2012]; (ii) filter the visibility measurements for natural events such as fog, rain, snow;

AC C

EP

124

7

ACCEPTED MANUSCRIPT and (iii) exclude visibility measurements that correspond to a relative humidity of

133

higher than 90%. In addition, the impact of relative humidity on visibility was also

134

corrected following to the method reported in Che et al. [2007]. Another issue is that

135

different observation methods were used before and after 1980. Prior to 1980,

136

visibility was recorded using 10 distance ranks, while after 1980 real distances were

137

used. To avoid the uncertainty introduced by this difference in observation methods,

138

we only use data acquired since 1980.

SC

RI PT

132

M AN U

139 140

3. Methods

141

3.1 Determination of Aerosol Optical Depth

In this study, the AOD at 550 nm is determined by two steps. Firstly, the

143

traditional Elterman AOD retrieval method [Elterman, 1970; Zhao et al., 1986] is used

144

to calculate AOD from horizontal visibility data using Equation (1):

145

′ τ 550

TE D

142

− − − − 3.912 0.75 2−ν * H H H H = 0.733( − 0.0116)( ) [H 1(e 1 − e 1 ) + 12.5e 1 + H 2e 1 ], (1) 5.5

5.5

5.5

λ

EP

V

z

′ H 1 = 0.886 + 0.0222V [km], H 2 = 3.77 [km]. τ 550 is AOD at the

where

147

wavelength of 550 nm, Z [km] the site altitude, ν * (=3) the Junge spectral parameter,

148

and V [km] the visibility at sea level, which is derived from the measured visibility

149

Vz [km] by Elterman [1970], as shown in Equation (2):

150

AC C

146

V z = 3.912[0.0116 − 0.00099z + (

3.912

− 0.0116)e

−

z (0.886 + 0.222V ) −1

]

(2)

V 151

Qiu and Lin [2001] pointed out that the aerosol vertical distribution assumed in

152

Elterman [1970] is not applicable to China and needs to be corrected. Therefore, they 8

ACCEPTED MANUSCRIPT introduced a correction coefficient f, which depends on the distribution of aerosol

154

particle number density with height. f =1 means that the aerosol vertical distribution is

155

the same as the assumption in Elterman [1970]; f >1 means that the attenuation of

156

aerosol particle number density with height is slower than that assumed in Elterman

157

[1970]; and f <1 means that the attenuation of aerosol particle number density with

158

height is faster than that assumed in Elterman [1970].

RI PT

153

Secondly, following the method of Qiu and Lin [2001], the final AOD ( τ 550 ) is

160

derived from Equation (3) by implementing a correction coefficient f to account for

161

the impact of the aerosol vertical profile.

M AN U

162

SC

159

′ ⋅f τ 550 = τ 550

(3)

Qiu and Lin [2001] determined the correction coefficient f by comparing the AODs

164

retrieved from Equations (1) and (2) with those retrieved from direct solar radiation

165

measurements at 16 stations over China. However the small number of stations

166

considered means that the correction method presented in Qiu and Lin [2001] requires

167

local calibration in order to prevent large errors when applied to other stations. To

168

address this, we use MODIS Level-3 AOD data (Aqua) from 2002 to 2010 to calibrate

169

Equation (4) at a monthly mean scale,

EP

AC C

170

TE D

163

2 (d ⋅Vz / Pw )

f = e(a +b ⋅Pw +c ⋅Vz )⋅e

,

(4)

171

where Pw [hPa] is surface vapor pressure, and the formula is similar to the one of

172

Qiu and Lin [2001]. The parameters ( a , b ,

173

shuffled complex evolution method [Duan et al., 1993]. Once these parameters have

174

been determined for each station, the final AOD from 1980 to 2010 can be derived. It 9

c and d ) are optimized using the

ACCEPTED MANUSCRIPT should be noted that the spatial resolution difference between level-3 MODIS AOD

176

and the station AOD would introduce non-negligible uncertainty into the accuracy of

177

the final AOD. However, the effect of the spatial resolution difference would be

178

reduced when the calibration procedure is done at the monthly-mean scale.

RI PT

175

As shown in Figure 2, we compare the visibility-based monthly-mean AODs at

180

550 nm before and after correction with the observed AOD recorded at the Beijing

181

AERONET station [Xia et al., 2006b]. The AERONET AODs at 550 nm are

182

converted from the observed values at 440 nm using the corresponding Ångström

183

exponent. It can be seen that the visibility-based AODs before correction are

184

substantially underestimated and no obvious seasonal variation is apparent. In contrast,

185

the visibility-based AODs after correction exhibit an obvious seasonal pattern, and the

186

variations are relatively consistent with the observed data. In Figure 2, we may find

187

some large discrepancy between the corrected visibility-based AOD and the

188

AERONET AOD, which may be due to the accuracy of the monthly-mean MODIS

189

AOD. As well known, the algorithm for MODIS AOD retrieval works only under

190

clear-sky condition. If there are not enough effective retrievals in a month due to

191

cloudy conditions, the monthly-mean MODIS AOD may not be representative and

192

thus would lead to large discrepancy. To further demonstrate the applicability of the

193

above calibrated method, we separate the MODIS AOD data into two groups: data

194

between 2005 and 2010 are used to calibrate the parameters, and data between 2002

195

and 2004 are used to evaluate the performance of the calibration method. Figure 3

196

shows the validation of the visibility-based monthly-mean AODs before and after

AC C

EP

TE D

M AN U

SC

179

10

ACCEPTED MANUSCRIPT correction against the MODIS AODs between 2002 and 2004 at all CMA stations.

198

After correction, the accuracy of the visibility-based AOD is obviously improved,

199

with the mean bias error (MBE) reducing from 0.1 to 0.04, root mean square error

200

(RMSE) reducing from 0.31 to 0.19 and correlation coefficient increasing from 0.44

201

to 0.74. This demonstrates that the correction method is feasible and can improve the

202

accuracy of the visibility-based AOD.

RI PT

197

An issue should be noted that the correction method presented in this study

204

improves the absolute accuracy of the visibility-based AOD with MODIS AOD, but

205

does not change the relative variations in the visibility-based AOD.

206

3.2 Modeling aerosols and water vapor direct effects on global radiation

M AN U

SC

203

Yang’s hybrid model [2006] is used in this study. The model contains two parts.

208

One is the calculation of transmittances for clear sky, which is converted from a

209

radiative transfer model and has accuracy comparable to that of spectral radiative

210

transfer models [Gueymard, 2003a; 2003b], based on local geographical and

211

meteorological conditions. The other is the calculation of transmittance for cloud,

212

which is parameterized with sunshine duration data. The inputs of the Yang’s model

213

contain surface pressure, water vapor, ozone thickness and Ångström turbidity and

214

sunshine duration. Therefore, using the hybrid model, we can estimate clear-sky

215

global radiation and all-sky global radiation simultaneously. In order to quantify the

216

effects of aerosols and water vapor on all-sky global radiation climatology over China,

217

three control experiments were designed to simulate the all-sky global radiation using

218

the hybrid model of Yang et al. [2006]. First is to simulate the all-sky global radiation

AC C

EP

TE D

207

11

ACCEPTED MANUSCRIPT 219

without considering aerosol attenuation; second is to simulate the all-sky global

220

radiation without considering water vapor absorption; and third is to simulate the

221

all-sky global radiation ( E g ↓ ). It is difficult to directly quantify the role of aerosols and water vapor in

223

long-term variations of global radiation because we cannot distinguish the effects of

224

clouds, aerosols and water vapor on long-term variations in global radiation. However,

225

it may be a feasible method to evaluate the role of aerosols and water vapor in

226

long-term variations of global radiation, by comparing long-term variations (or trends)

227

between all-sky global radiation and clear-sky global radiation impacted by aerosols

228

or water vapor. The clear-sky part of Yang’s hybrid model is used to estimate the

229

clear-sky daily global radiation at the 519 stations. In order to separate the impact of

230

aerosols and water vapor on clear-sky global radiation, three simulation experiments

231

were also designed. The first experiment simulates the clear-sky global radiation

232

(termed E a ↓ ) with water vapor fixed at 1980 levels, and so investigates the impact of

233

aerosols on variations in clear-sky global radiation. The second experiment simulates

234

clear-sky global radiation (termed E w ↓ ) with aerosol loading fixed at 1980 levels, and

235

so investigates the impact of water vapor on variations in clear-sky global radiation.

236

The third experiment simulates clear-sky global radiation (termed E(a +w )↓ ) impacted

237

by both aerosols and water vapor.

AC C

EP

TE D

M AN U

SC

RI PT

222

238

The Ångström turbidity is converted from the AOD at 550 nm, which was first

239

estimated from visibility observed at CMA stations [Elterman, 1970] and then

240

corrected with MODIS AOD data on the monthly mean scale. The ozone thickness is 12

ACCEPTED MANUSCRIPT derived from the climatological data based on Total Ozone Mapping Spectrometer

242

zonal means, which are provided by the National Aeronautics and Space

243

Administration Goddard Space Flight Center. Water vapor can be derived from the

244

re-analysis data, but their spatial resolutions are too coarse (larger than 0.5o) and

245

applying re-analysis data at certain stations through interpolation may bring some

246

uncertainty. On the other hand, water vapor can be easily derived at routine weather

247

stations with the semi-empirical formula. The accuracy of the water vapor estimated

248

by this way is not high, but the long-term variations in the water vapor estimated from

249

air temperature and relative humidity is reliable, which is accordant with the main

250

objective of this study, i.e., to discuss the role of aerosols and water vapor on the

251

long-term changes of global radiation over China. Therefore, the water vapor is

252

determined approximately from the air temperature and relative humidity according to

253

Yang et al. [2006] as Equation (5):

255

SC

M AN U

TE D

Pw = 0.00493RH ⋅ T −1 exp(26.23 − 5416T −1 ), where RH [%] is relative humidity and T [K] is air temperature.

AC C

256 257

4. Results

258

4.1 AOD variations

259

(5)

EP

254

RI PT

241

Figure 4 shows the annual mean time series of AOD ( τ 550 ) between 1980 and

260

2010 averaged over the whole China (CN), the CE region, the SC region and the OT

261

region, respectively. Averaged over the whole China (CN), the AOD decreases from

262

1980 to 1989 at a rate of around −0.002 per year, increases since 1990 until 2006 at a 13

ACCEPTED MANUSCRIPT rate of around 0.0014 per year, and then decreases after 2006 at a rate of around

264

-0.0113 per year. The variations of aerosols over CE regions are similar to the one

265

over the whole China with the same two increasing periods (before the end of 1980s

266

and after 2006) and one decreasing period (from 1990 to 2006). Averaged over the SC

267

region, the AOD slightly increases from 1980 to 1989, rapidly increases since 1990

268

until 2006, and then decreases after 2006. The increasing rate between 1990 and 2006

269

is obviously larger than the ones averaged over the whole China and the CE region.

270

Averaged over the OT region, the AOD decreases from 1980 to 1989, and then

271

slightly increases since 1990. But the slightly increasing trend does not pass the

272

significance test (p < 0.05).

M AN U

SC

RI PT

263

As seen from Figure 4, one may question that the AOD derived here is not

274

significantly increased during the last decade in China, and especially, the AOD

275

decreases after 2006. These seem to conflict with the impression that the haze days

276

increased significantly over China in recent years. But this impression may be not true.

277

Actually, there are severally studies analyzing the long-term variations of haze days in

278

China (Fu and Dan [2014]; Fu et al. [2014]). Fu and Dan [2014] analyzed the haze

279

days over China (except Xinjiang and Tibet) during 1960–2010. Their results show

280

that haze days decrease weakly since the end of 1970s, and increase rapidly since the

281

middle of 1990s, but decrease after the middle of 2000s (See Figure 6 of Fu and Dan

282

[2014]). This is generally consistent with the variations of AOD in Figure 4. All the

283

four urban agglomerations (The Jing-Jin-Ji region, Yangtze River Delta region, Pearl

284

River Delta Region and Sichuan Basin region) show a slow decrease in haze days

AC C

EP

TE D

273

14

ACCEPTED MANUSCRIPT after the middle of 2000s (See Figure 6(b) of Fu and Dan [2014]). Fu et al. [2014]

286

analyzed the trends of fog and haze in the North China Plain over the past 30 years

287

and also found a slow decrease in haze days after the middle of 2000s, which is

288

consistent with the variation of SO2 emissions in Hebei Province (See Figure 3 of Fu

289

et al. [2014]). Liu et al. [2015] investigated the emissions from China’s coal-fired

290

power plants during 1990–2010, and found that the emissions of SO2, PM2.5 and PM10

291

are decreased since 2006 (See Figure 4 of Liu et al. [2015]). These results are

292

consistent with the conclusion that the AOD derived in this study is decreased after

293

2006.

M AN U

SC

RI PT

285

Li et al. [2014] derived recent trends in aerosol optical properties from

295

AERONET measurements and found that the 440 nm AOD at Beijing station has a

296

decreasing trend since 2002, though the trend is not statistically significant (See

297

Figure 2 and Table 1 in Li et al. [2014]). Xu et al. [2015] retrieved the 750 nm AOD

298

over 14 first-class solar radiation stations (contains continuous measurements of

299

global, direct and diffuse radiation) in China based on direct radiation measurements

300

during 1993–2012, and found decreasing trends of AOD at two stations of Beijing and

301

Guangzhou (See Figure 9 of Xu et al. [2015]). These two studies also suggest that the

302

AOD is not continuously increased in the last decade in CE and SC regions, despite

303

the AOD values at these regions are general large.

304

4.2 Validity of the estimated AOD for radiation variation study

AC C

EP

TE D

294

305

It is hard to directly validate the estimated AOD due to lack of observed aerosols

306

data. Instead, we present an indirect validation through comparing observed clear-sky 15

ACCEPTED MANUSCRIPT global radiation with the simulated one by Yang’s hybrid model, in which AOD (or

308

Ångström turbidity) is an important input. In order to evaluate the accuracy of the

309

estimated clear-sky global radiation ( E(a +w )↓ ), we chose the clear days at all CMA

310

radiation stations. A clear-sky day is defined as one with daily TCC being less than

311

10%, according to Qian et al. [2007]. The instrument to measure the radiation in CMA

312

stations and its calibration procedures are introduced in Tang et al. [2010]. A quality

313

control scheme was applied to the CMA observed radiation data to exclude the

314

erroneous and suspected data [Tang et al., 2010]. Figure 5 presents the validation

315

results of the estimated daily clear-sky global radiation against the observed one at all

316

CMA radiation stations during 1993-2000. The relative MBE and RMSE of the

317

estimated clear-sky daily global radiation to the observations are about 1.9% and

318

8.9%, respectively. The slightly overestimation of the daily clear-sky radiation may

319

partly be attributed to the cloud contamination in the observed clear sky global

320

radiation, which was identified based on daily TCC observations.

TE D

M AN U

SC

RI PT

307

Furthermore, we evaluate the applicability of the estimated AOD in representing

322

the inter-annual variability of clear-sky radiation. We count the number of clear-sky

323

days in each year at each CMA radiation station, and find that there are 32 radiation

324

stations, each of which contains at least 20 clear-sky days in each year during

325

1980-2010. Considering the inhomogeneous distribution of clear-sky days and

326

seasonal cycle of solar radiation, Figure 6 shows the comparisons of the variations in

327

solar transmittance under clear-sky days between the observed one and the estimated

328

one averaged over the 32 CMA radiation stations during 1980-2010. It can be found

AC C

EP

321

16

ACCEPTED MANUSCRIPT that the inter-annual variability of the estimated clear-sky transmittance is similar to

330

that of the observed one, though the estimated values are slightly underestimated. This

331

indicates that the long-term variations in the estimated AOD are relatively reliable,

332

which is consistent with the conclusion of Wang et al. [2009].

333

4.3 Aerosol effect on global radiation variations

RI PT

329

Considering the potential transition for global radiation over China, Tang et al.

335

[2011] divided the data into two time periods: pre- and post- 1989. For consistency

336

with the analysis of global radiation over China, we also separate the time period into

337

two sub-periods (1980 – 1989 and 1990 – 2010) and investigate the decadal change in

338

each period. Figure 7 presents the trends comparisons between the clear-sky global

339

radiation ( E a ↓ ) impacted by aerosols and the corresponding all-sky global radiation

340

( E qc ↓ ) at the 519 individual CMA stations for time periods 1980 – 1989 and 1990 –

341

2010, respectively. It is shown that there are significant differences between the trends

342

in E a ↓ and E qc ↓ , and the correlation coefficient between them is 0.01 and -0.03 for

343

the two time periods respectively. These low correlation values demonstrate that

344

long-term variations in E qc ↓ are not related to those of E a ↓ , indicating that aerosols

345

are unlikely to be the main cause for the long-term variations in global radiation over

346

China.

M AN U

TE D

EP

AC C

347

SC

334

To further investigate the impacts of aerosols on long-term variations in global

348

radiation at a regional scale, Figure 8 (a) shows the anomalies of annual mean

349

clear-sky global radiation ( E a ↓ ) and all-sky global radiation ( E qc ↓ ) averaged over the

350

whole China (CN) for 1980-2010. The clear-sky global radiation ( E a ↓ ) increases 17

ACCEPTED MANUSCRIPT from 1980 to 1989 at a rate of about 0.13 W m−2 yr−1, and then decreases from 1990 to

352

2010 at a rate of about -0.07 W m−2 yr−1. Obviously, the variations of the all-sky

353

global radiation ( E qc ↓ ) are quite different from the variations of the clear-sky global

354

radiation ( E a ↓ ). Their trends are opposite to each other for both time periods (1980 –

355

1989 and 1990 – 2010). Moreover, the trends seen in E a ↓ for the two time periods

356

are statistically significant, while the trends seen in E qc ↓ do not pass the significance

357

test (p < 0.05). This may indicate that variations in aerosols are not the primary causes

358

for variations in global radiation over China. Figure 8 (b)–(d) presents the results

359

averaged over the CE region, the SC region and the OT region, respectively. In the

360

CE region, the trends in E a ↓ and E qc ↓ are opposite to each other during 1980 –

361

1989, but they are both decreasing during 1990 – 2010, and the former can seem to

362

explain about 54% (-0.07/-0.13) of the latter. Similarly to the results averaged over

363

the whole China, the aerosols cannot account for the variations of the global radiation

364

over the SC and the OT regions.

365

4.4 Water vapor on global radiation variations

EP

TE D

M AN U

SC

RI PT

351

In addition to aerosols, water vapor may have a considerable effect on global

367

radiation climatology. To present the effects of aerosols and water vapor on global

368

radiation climatology, Figure 9 shows the difference between the all-sky global

369

radiation without considering aerosol attenuation and all-sky global radiation, and the

370

one between all-sky global radiation without considering water vapor absorption and

371

all-sky global radiation averaged over all CMA stations from 1980 to 2010,

372

respectively. The effects of aerosols and water vapor on global radiation climatology

AC C

366

18

ACCEPTED MANUSCRIPT are about 30 W m-2 and 25 W m-2, respectively. This indicates that water vapor may

374

have equally important role in determining the variations of global radiation as

375

aerosols. Figure 10 compares the trends of the clear-sky global radiation ( Ew ↓ )

376

impacted by water vapor with those of all-sky global radiation ( E qc ↓ ) at the 519

377

individual CMA stations for the time periods 1980 – 1989 and 1990 – 2010. It can be

378

seen that fluctuations in Ew ↓ are significantly lower in magnitude than the

379

corresponding fluctuations in E qc ↓ , in fact Ew ↓ remains close to zero and within the

380

range -0.1 to 0.1. This indicates that the variations in E qc ↓ over China cannot be

381

accounted for by the variations in Ew ↓ , suggesting that water vapor is not the primary

382

factor for long-term variations in global radiation over China.

M AN U

SC

RI PT

373

383

5. Discussions

TE D

384

Based on above results, aerosols and water vapor are not the primary causes for

386

the long-term variations in global radiation over China. Alternatively, the changes of

387

cloud properties and the interactions between clouds and aerosols may be the main

388

causes for the variations in global radiation over China. The effect of cloud on global

389

radiation climatology over China can be derived by using the clear-sky global

390

radiation ( E(a +w )↓ ) minus the all-sky global radiation ( E g ↓ ) averaged over China, and

391

its value is about 65 W m-2, which is much greater than the effects of aerosols and

392

water vapor as shown in Figure 9. This indicates that the cloud effect may be the

393

primary factor for long-term variations in global radiation over China. To further

394

discuss the role of cloud in the long-term changes of global radiation over China,

AC C

EP

385

19

ACCEPTED MANUSCRIPT Figure 11 shows anomalies in TCC, LCC and all-sky global radiation ( E qc ↓ ),

396

normalized and averaged over the whole China (CN), CE, SC and OT regions during

397

1980 – 2010. In general, there is negative correlation between global radiation and

398

cloud (TCC or LCC). Especially, the negative correlation coefficient between global

399

radiation and cloud is -0.76 and -0.82 over the SC region for TCC and LCC,

400

respectively. This may indicate that the changes of cloud (including aerosol indirect

401

effect) could explain the variations of global radiation over the SC region.

SC

RI PT

395

The aerosol indirect effect on long-term variations in global radiation cannot be

403

ignored since aerosols can indirectly attenuate global radiation through altering cloud

404

reflectivity and lifetime. In pristine regions, aerosol–cloud interactions may amplify

405

the variations in global radiation; while in polluted regions, aerosol–cloud interactions

406

may suppress the variations in global radiation [Wild et al. 2012a]. Stanhill et al.

407

[2014] presented a novel idea for separating aerosol direct and indirect effects on

408

global radiation. They analyzed the Angstrom-Prescott relationship between

409

normalized global radiation and sunshine duration at five stations with different

410

climate, and found that the parameters of this relationship were rather stable across

411

sub-periods related to the dimming and brightening, thus they concluded that changes

412

in the cloud cover rather than anthropogenic aerosol emissions was the major cause

413

for the variations in global radiation. However, this hypothesis has been tested and

414

rejected at Potsdam station (one of the five stations used in Stanhill et al. [2014]) by

415

Vetter and Wechsung [2015], who found that the residuals of the Angstrom-Prescott

416

relationship at Potsdam station are not only nonrandom but also contain variations

AC C

EP

TE D

M AN U

402

20

ACCEPTED MANUSCRIPT that are consistent with the variations in the aerosol emissions around Potsdam.

418

Another point is that cloud types, which have different impacts on global radiation,

419

may change with climate conditions. For example, the cloud types over the Tibetan

420

Plateau have changed since 1984 with less TCC but more LCC [Yang et al., 2012].

421

Therefore, it is very difficult to quantify the effect of aerosols on cloud and the effect

422

of cloud on long-term variations in global radiation because variations in cloud cover

423

do not represent variations in cloud optical properties since the same cloud cover

424

amount may correspond to clouds with different optical depths and different

425

microphysical properties [Stubenrauch et al., 2013]. Therefore, the uncertainties in

426

explaining the variations of global radiation with cloud cover should be investigated

427

and quantified in future studies.

429

6. Conclusions

TE D

428

M AN U

SC

RI PT

417

The global radiation over China has undergone substantial decadal changes over

431

the past five decades, and this may be attributed to variations in water vapor, clouds,

432

aerosols and interactions among them. In this study, we first derive the long-term

433

aerosol information by estimating the AOD from the visibility data and correcting

434

with the MODIS AOD data. AOD averaged over China exhibits decadal changes,

435

with two decreasing periods (before the end of 1980s and after 2006) and one

436

increasing period (from 1990 to 2006). Then the effect of aerosols on long-term

437

variations in global radiation over China is quantified, using a physical radiation

438

model for clear skies. It is found that aerosols have a considerable effect on global

AC C

EP

430

21

ACCEPTED MANUSCRIPT radiation climatology over China, but aerosols cannot wholly explain the variations in

440

global radiation over China. Generally, the aerosol direct effect can explain about 54%

441

of the variations in global radiation over the CE region during 1990–2010. In CE

442

region during 1980–1989 and in CE region during 1980–1989 and 1990–2010, the

443

aerosol direct effect suppresses the decadal variations in global radiation. This result

444

is consistent with the finding of Lin et al. (2015), who indicated that aerosol direct

445

effect can only explain 20% of the decadal change in global radiation over China

446

according to a statistical method. Therefore, it is reasonable to speculate that

447

variations in clouds and interactions between clouds and aerosols are the main causes

448

of the long-term variations in global radiation over China. As this study only

449

considers the direct effect of aerosols on global radiation variation, the aerosol

450

indirect effect also warrants further investigations.

451 452

Acknowledgments

TE D

M AN U

SC

RI PT

439

This work was supported by the National Natural Science Foundation of China

454

(Grants No. 41301359), “Strategic Priority Research Program (B)” of the Chinese

455

Academy of Sciences (Grant No. XDB03030306), Open Fund from the State Key

456

Laboratory of Remote Sensing Science (Grant No. OFSLRSS201303) that is

457

cosponsored by the Institute of Remote Sensing and Digital Earth, Chinese Academy

458

of Sciences and Beijing Normal University, and China Postdoctoral Science

459

Foundation (Grant No. 2014T70127). The CMA station data were obtained from the

460

National Meteorological Information Center.

AC C

EP

453

22

ACCEPTED MANUSCRIPT 461

References

463

Augustine, J. A., and E. G. Dutton (2013), Variability of the surface radiation budget

464

over the United States from 1996 through 2011 from high-quality

465

measurements,

466

doi:10.1029/2012JD018551.

J.

Geophys.

Res.

RI PT

462

Atmos.,

118,

43–53,

Berk, A., L. S. Bernstein, G. P. Anderson, P. K. Acharya, D. C. Robertson, J. H.

468

Chetwynd, and S. M. Adler‐Golden (1998), MODTRAN cloud and multiple

469

scattering upgrades with application to AVIRIS, Remote Sens. Environ., 65(3),

470

367–375, doi:10.1016/S0034-4257(98)00045-5.

M AN U

SC

467

Che, H. Z., Shi, G. Y., Zhang, X. Y., Arimoto, R., Zhao, J. Q., Xu, L., Wang, B., and

472

Chen, Z. H. (2005), Analysis of 40 years of solar radiation data from China,

473

1961–2000, Geophys. Res. Lett., 32, L06803, doi:10.1029/2004GL022322.

474

Che, H., X. Zhang, Y. Li, Z. Zhou, and J. J. Qu (2007), Horizontal visibility trends in

TE D

471

China

1981–2005,

476

doi:10.1029/2007GL031450.

Geophys.

Res.

Lett.,

34,

L24706,

AC C

EP

475

477

Duan, Q.Y., Gupta, V.K., Sorooshian, S. (1993), Shuffled complex evolution

478

approach for effective and efficient global minimization, J. Optim. Theory Appl.

479

76, 501–521.

480

E. F. Vermote, N. Z. Tanré, J. L. Deuzé, M. Herman, J. J. Morcette (1997), Second

481

Simulation of the Satellite Signal in the Solar Spectrum, 6S: An Overview,

482

IEEE Trans. Geosci. Remote Sens., 35, 675-686. 23

ACCEPTED MANUSCRIPT 483 484

Elterman, L. (1970), Relationships between vertical attenuation and surface meteorological range, Appl. Optics., 9, 1804-1810. Folini, D., and M. Wild (2011), Aerosol emissions and dimming/ brightening in

486

Europe: Sensitivity studies with ECHAM5-HAM, J. Geophys. Res., 116,

487

D21104, doi:10.1029/2011JD016227.

RI PT

485

Fu, C. & Dan, L. (2014), Spatiotemporal characteristics of haze days under heavy

489

pollution over central and eastern China during 1960–2010, Climatic and

490

Environmental Research 19, 219–226, 10.3878/j.issn.1006-9585.2014.13213

491

[in Chinese].

M AN U

SC

488

Fu, G. Q., Xu, W. Y., Yang, R. F., Li, J. B., and Zhao, C. S. (2014), The distribution

493

and trends of fog and haze in the North China Plain over the past 30 years,

494

Atmos. Chem. Phys., 14, 11949-11958, doi:10.5194/acp-14-11949-2014.

TE D

492

Gueymard, C.A. (2003a), Direct solar transmittance and irradiance predictions with

496

broadband models—Part I: Detailed theoretical performance assessment, Solar

497

Energy, 74, 355–379.

499 500

Gueymard, C.A. (2003b), Direct solar transmittance and irradiance predictions with

AC C

498

EP

495

broadband models—Part II: Validation with high-quality measurements, Solar Energy, 74, 381–395.

501

Gilgen, H., Roesch, A. Wild, M., and Ohmura, A. (2009), Decadal changes of

502

shortwave irradiance at the surface in the period 1960 to 2000 estimated from

503

Global

504

doi:10.1029/2008JD011383.

Energy Balance Archive,

24

J.

Geophys.

Res.,

114,

D00D08,

ACCEPTED MANUSCRIPT

506 507 508

Kaiser, D. P. (1998), Analysis of total cloud amount over China, 1951–1994, Geophys. Res. Lett., 25(19), 3599–3602, doi:10.1029/98GL52784. Kaiser, Dale P.: Decreasing cloudiness over China (2000), An updated analysis examining additional variables, Geophys Res Lett, 27(15), 2193.

RI PT

505

Li, J., Carlson, B. E., Dubovik, O., and Lacis, A. A. (2014), Recent trends in aerosol

510

optical properties derived from AERONET measurements, Atmos. Chem. Phys.,

511

14, 12271-12289, doi:10.5194/acp-14-12271-2014.

SC

509

Li, Z., et al. (2011), East Asian Studies of Tropospheric Aerosols and their Impact on

513

Regional Climate (EASTAIRC): An overview, J. Geophys. Res., 116, D00K34,

514

doi:10.1029/2010JD015257.

M AN U

512

Liang, F., and Xia, X. A. (2005), Long-term trends in solar radiation and the

516

associated climatic factors over China for 1961–2000, Ann. Geophys., 23,

517

2425–2432.

TE D

515

Liepert, B. G. (2002), Observed reductions of surface solar radiation at sites in the

519

United States and worldwide from 1961 to 1990, Geophys. Res. Lett., 29(10),

520

1421, doi:10.1029/2002GL014910.

AC C

EP

518

521

Lin, C., K. Yang, J. Huang, W. Tang, et al. (2015), Impacts of wind stilling on solar

522

radiation variability in China, Sci. Rep., 5, 15135, doi: 10.1038/srep15135.

523

Liu, F., Zhang, Q., Tong, D., Zheng, B., Li, M., Huo, H., and He, K. B. (2015),

524

High-resolution inventory of technologies, activities, and emissions of

525

coal-fired power plants in China from 1990 to 2010, Atmos. Chem. Phys., 15,

526

13299-13317, doi:10.5194/acp-15-13299-2015. 25

ACCEPTED MANUSCRIPT Long, C. N., Dutton, E. G., Augustine, J. A., Wiscombe, W., Wild, M., McFarlane, S.

528

A. and Flynn, C. J. (2009), Significant decadal brightening of downwelling

529

shortwave in the continental United States, J. Geophys. Res., 114, D00D06,

530

doi:10.1029/2008JD011263.

RI PT

527

Norris, J. R., and M. Wild (2007), Trends in aerosol radiative effects over Europe

532

inferred from observed cloud cover, solar dimming, and solar brightening, J.

533

Geophys. Res., 112, D08214, doi:10.1029/2006JD007794.

535 536 537

Ohmura, A. (2009), Observed decadal variations in surface solar radiation and their

M AN U

534

SC

531

causes, J. Geophys. Res., 114, D00D05, doi:10.1029/2008JD011290. Pinker, R. T., Zhang, B., and Dutton, E. G. (2005), Do satellites detect trends in surface solar radiation? Science, 308, 850–854.

Qian, Y., D. P. Kaiser, L. R. Leung, and M. Xu (2006), More frequent cloud-free sky

539

and less surface solar radiation in China from 1955 to 2000, Geophys Res Lett,

540

33, L01812.

TE D

538

Qian, Y., W. Wang, L. R. Leung, and D. P. Kaiser (2007), Variability of solar

542

radiation under cloud-free skies in China: The role of aerosols, Geophys Res

544 545 546 547 548

AC C

543

EP

541

Lett, 34, L12804.

Qiu, J. H., and Y. R. Lin (2001), A parameterization model of aerosol optical depths [in Chinese], Acta Meteorol. Sin., 59(3), 368–372.

Ramanathan, V., Crutzen, P. J., Kiehl, J. T., and Rosenfeld, D. (2001), Aerosols, climate, and the hydrological cycle, Science, 294, 2119– 2124. Rosenfeld, D., S. Sherwood, R. Wood, and L. Donner (2014), Climate effects of 26

ACCEPTED MANUSCRIPT 549

aerosol-cloud interactions, Science, 343(6169), 379–380. Shi, G. Y., Hayasaka, T., Ohmura, A., Chen, Z. H., Wang, B., Zhao, J. Q., Che, H. Z.,

551

and Xu, L. (2008), Data quality assessment and the long-term trend of ground

552

solar radiation in China, J. Appl. Meteorol. Climatol., 47, 1006–1016.

553 554

RI PT

550

Stanhill, G., and S. Cohen (1997), Recent changes in solar irradiance in Antarctica, J. Clim., 10, 2078–2086.

Stanhill, G., and S. Cohen (2009), Is solar dimming global or urban? Evidence from

556

measurements in Israel between 1954 and 2007, J. Geophys. Res., 114, D00D17,

557

doi:10.1029/2009JD011976.

M AN U

SC

555

Stanhill, G., and S. Cohen (2001), Global dimming: A review of the evidence for a

559

widespread and significant reduction in global radiation with discussion of its

560

probable causes and possible agricultural consequences. Agric. For. Meteor.,

561

107, 255–278.

564 565 566 567

from surface sources of pollution, Clim. Change, 26, 89–103.

EP

563

Stanhill, G., and S. Moreshet (1994), Global radiation changes at seven sites remote

Stanhill, G., O. Achiman, R. Rosa, and S. Cohen (2014), The cause of solar dimming

AC C

562

TE D

558

and brightening at the Earth’s surface during the last half century: Evidence from measurements of sunshine duration, J. Geophys. Res. Atmos., 119, 10,902–10,911, doi:10.1002/2013JD021308.

568

Streets, D. G., Y. Wu, and M. Chin (2006), Two-decadal aerosol trends as a likely

569

explanation of the global dimming/brightening transition, Geophys. Res. Lett.,

570

33, L15806, doi:10.1029/2006GL026471. 27

ACCEPTED MANUSCRIPT 571 572

Tang, W., Yang, K., He, J., and Qin, J. (2010). Quality control and estimation of global solar radiation in China. Solar Energy, 84, 466–475. Tang, W., K. Yang, J. Qin, C. Cheng, and J. He (2011), Solar radiation trend across

574

China in recent decades: a revisit with quality-controlled data, Atmos. Chem.

575

Phys, 11, 393-406.

RI PT

573

Tang, W. J., Yang, K., Qin, J. & Min, M. (2013), Development of a 50-year daily

577

surface solar radiation dataset over China, Sci. China Earth Sci., 56, 1555–

578

1565, doi: 10.1007/S11430-012-4542-9.

M AN U

579

SC

576

Vautard, R., Yiou, P., and van Oldenborgh, G. J.(2009), Decline of fog, mist and haze

580

in

Europe

over

the

581

doi:10.1038/ngeo414.

past

30

years,

Nature

Geosci.,

2,

115–119,

Vetter, T., and F. Wechsung (2015), Direct aerosol effects during periods of solar

583

dimming and brightening hidden in the regression residuals: Evidence from

584

Potsdam

585

doi:10.1002/2015JD023669.

587 588 589 590 591 592

J.

Geophys.

Res.

Atmos.,

120,

EP

measurements,

Wild, M., Gilgen, H., and Roesch, A. (2005), From dimming to brightening: Decadal

AC C

586

TE D

582

changes in solar radiation at earth’s surface, Science, 308, 847–850.

Wild, M. (2009), Global dimming and brightening: A review, J. Geophys. Res., 114, D00D16, doi:10.1029/2008JD011470.

Wild, M. (2012a), Enlightening global dimming and brightening, Bull. Am. Meteorol. Soc., 93(1), 27–37. Wild

M.,

A.

Roesch,

C.

Amman 28

(2012b),

Global

dimming

and

ACCEPTED MANUSCRIPT 593

brightening―evidence and agricultural implications. CAB Rev, 7, 1-7. Wang, K. C., Dickinson, R. E., and Liang, S. L. (2009), Clear sky visibility has

595

decreased over land globally from 1973 to 2007, Science, 323, 1468–1470.

596

Wang, K., R. Dickinson, M. Wild, and S. Liang (2012), Atmospheric impacts on

597

climatic variability of surface incident solar radiation, Atmos. Chem. Phys., 12,

598

9581-9592.

601 602

SC

600

Wang, Y. W. and Y. H. Yang (2014), China’s dimming and brightening: evidence, causes and hydrological implications, Ann. Geophys., 32, 41–55.

M AN U

599

RI PT

594

Wu, P. L., N. Christidis and P. Stott (2013), Anthropogenic impact on Earth’s hydrological cycle, Nature Climate Change, 3, 807-810.

Wu, J., Fu, C.B., Zhang, L.Y., Tang, J.P., 2012. Trends of visibility on sunny days in

604

China in the recent 50 years. Atmospheric Environment 55, 339e342. http://

605

dx.doi.org/10.1016/j.atmosenv.2012.03.037.

TE D

603

Xia, X. A., Wang, P. C., Chen, H. B., and Liang, F. (2006a), Analysis of downwelling

607

surface solar radiation in China from National Center for Environmental

608

Prediction reanalysis, satellite estimates, and surface observations, J. Geophys.

AC C

609

EP

606

Res., 111, D09103, doi:10.1029/2005JD006405.

610

Xia, X. A., H. B. Chen, P. C. Wang, W. X. Zhang, P. Goloub, B. Chatenet, T. F. Eck,

611

and B. N. Holben (2006b), Variation of column-integrated aerosol properties in a

612

Chinese

613

doi:10.1029/2005JD006203.

614

urban

region,

J.

Geophys.

Res.,

111,

D05204,

Xia, X. (2010), Spatiotemporal changes in sunshine duration and cloud amount as 29

ACCEPTED MANUSCRIPT 615

well as their relationship in China during 1954–2005, J. Geophys. Res., 115,

616

D00K06, doi:10.1029/2009JD012879. Xu, X. F., Qiu, J. H., Xia, X. A. & Min, M. (2015), Characteristics of atmospheric

618

aerosol optical depth variation in China during 1993-2012, Atmospheric

619

Environment, 119, 82-94.

RI PT

617

Yang, K., T. Koike, and B.-S. Ye (2006), Improving estimation of hourly, daily, and

621

monthly solar radiation by importing global data sets, Agric. For. Meteorol.,

622

137, 43–55, doi:10.1016/j.agrformet.2006.02.001.

M AN U

SC

620

623

Yang, K., B.H. Ding, J. Qin, W.J. Tang, N. Lu (2012), Can aerosol loading explain

624

the solar dimming over the Tibetan Plateau? Geophys. Res. Lett. 39, L20710,

625

doi:10.1029/2012GL053733.

Zhao, B. L., P. F. Zhang, and G. M. Gao (1986), Study of the characteristic of

627

atmospheric aerosol thickness over China. Acta Meteorologica Sinica, 44(2),

628

235-241 (in Chinese).

TE D

626

Zhang, Y. L., Qin, B. Q., and Chen, W. M. (2004), Analysis of 40 year records of

630

solar radiation data in Shanghai, Nanjing and Hangzhou in eastern China, Theor.

AC C

631

EP

629

Appl. Climatol., 78, 217–227.

30

ACCEPTED MANUSCRIPT Figure captions

633

Figure 1 The distribution of the CMA routine meteorological stations where the data

634

contain records for at least 20 days per month from 1980 to 2010. The

635

color filling represents the MODIS winter half-year AOD climatology

636

according to Lin et al. [2015], and the grey indicates missing AOD values.

RI PT

632

Figure 2 Comparisons between monthly mean AOD estimated from visibility data

638

before and after correction, and AOD recorded at Beijing station. The

639

observation data are from Beijing AERONET station.

641

M AN U

640

SC

637

Figure 3 Validation of monthly mean visibility-based AOD before and after correction against MODIS AOD for 2002 – 2004 at all CMA stations. Figure 4 Annual mean time series of AOD ( τ 550 ) averaged over the whole China

643

(CN), the CE region, the SC region and OT region for 1980 – 2010. The

644

star symbol (*) denotes a trend that passes the significance test (p <0.05).

647 648 649

and the estimated one at all CMA radiation stations during 1993–2000.

EP

646

Figure 5 Comparison between the observed daily clear-sky global radiation ( E(a +w )↓ )

Figure 6 Comparison of the variations in solar transmittance under clear-sky days

AC C

645

TE D

642

between the estimated one and the observed one averaged over the 32 CMA radiation stations during 1980–2010.

650

Figure 7 Comparison between the trends in clear-sky global radiation impacted by

651

aerosols ( E a ↓ ) and the trends in all-sky global radiation ( E qc ↓ ) at 519

652

individual CMA stations for (a) 1980 – 1989 and (b) 1990 – 2010.

653

Figure 8 Anomalies of annual mean clear-sky global radiation ( E a ↓ ，blue Line) and 31

ACCEPTED MANUSCRIPT 654

all-sky global radiation ( E qc ↓ ，red Line) averaged over (a) CN, (b) CE, (c)

655

SC, and (d) OT for 1980 – 2010. The star symbol (*) denotes a trend that

656

passes the significance test (p <0.05). Figure 9 The effects of aerosols (with solid marker) and water vapor (with circle

658

marker) on global radiation climatology averaged over all CMA stations

659

during 1980 – 2010. The effect of aerosols is the difference between the

660

all-sky global radiations estimated with/without considering aerosol

661

attenuation, and the effect of water vapor is the difference between the

662

all-sky global radiations estimated with/without considering water vapor

663

absorption.

M AN U

SC

RI PT

657

Figure 10 Comparison between the trends in clear-sky global radiation impacted by

665

water vapor ( E w ↓ ) and the trends in all-sky global radiation ( E qc ↓ ) at 519

666

individual CMA stations for (a) 1980 – 1989 and (b) 1990 – 2010.

667

Figure 11 Normalized anomaly time series of TCC, LCC and all-sky global radiation

EP

( E qc ↓ ) averaged over (a) CN, (b) CE, (c) SC, and (d) OT for 1980 – 2010.

AC C

668

TE D

664

32

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

669

Figure 1 The distribution of the CMA routine meteorological stations where the data

671

contain records for at least 20 days per month from 1980 to 2010. The color

672

filling represents the MODIS winter half-year AOD climatology according

673

to Lin et al. [2015], and the grey indicates missing AOD values.

AC C

EP

TE D

670

33

ACCEPTED MANUSCRIPT

RI PT

674

Figure 2 Comparisons between monthly mean AOD estimated from visibility data

676

before and after correction, and AOD recorded at Beijing station. The

677

observation data are from Beijing AERONET station.

AC C

EP

TE D

M AN U

SC

675

34

RI PT

ACCEPTED MANUSCRIPT

TE D

M AN U

correction against MODIS AOD for 2002 – 2004 at all CMA stations.

EP

680

Figure 3 Validation of monthly mean visibility-based AOD before and after

AC C

679

SC

678

35

681

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

682

Figure 4 Annual mean time series of AOD ( τ 550 ) averaged over the whole China

683

(CN), the CE region, the SC region and OT region for 1980 – 2010. The star

684

symbol (*) denotes a trend that passes the significance test (p <0.05).

36

SC

RI PT

ACCEPTED MANUSCRIPT

686 687

M AN U

685

Figure 5 Comparison between the estimated daily clear-sky global radiation ( E(a +w )↓ ) and the observed one at all CMA radiation stations during 1993–2000.

AC C

EP

TE D

688

37

RI PT

ACCEPTED MANUSCRIPT

SC

689

Figure 6 Comparison of the variations in solar transmittance under clear-sky days

691

between the estimated one and the observed one averaged over the 32 CMA

692

radiation stations during 1980–2010.

AC C

EP

TE D

M AN U

690

38

RI PT

ACCEPTED MANUSCRIPT

SC

693

Figure 7 Comparison between the trends in clear-sky global radiation impacted by

695

aerosols ( E a ↓ ) and the trends in all-sky global radiation ( E qc ↓ ) at 519

696

individual CMA stations for (a) 1980 – 1989 and (b) 1990 – 2010.

AC C

EP

TE D

M AN U

694

39

697

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

698

Figure 8 Anomalies of annual mean clear-sky global radiation ( E a ↓ ，blue Line) and

699

all-sky global radiation ( E qc ↓ ，red Line) averaged over (a) CN, (b) CE, (c)

700

SC, and (d) OT for 1980 – 2010. The star symbol (*) denotes a trend that

701

passes the significance test (p <0.05).

40

RI PT

ACCEPTED MANUSCRIPT

SC

702

Figure 9 The effects of aerosols (with solid marker) and water vapor (with circle

704

marker) on global radiation climatology averaged over all CMA stations

705

during 1980 – 2010. The effect of aerosols is the difference between the

706

all-sky global radiations estimated with/without considering aerosol

707

attenuation, and the effect of water vapor is the difference between the

708

all-sky global radiations estimated with/without considering water vapor

709

absorption.

AC C

EP

TE D

M AN U

703

41

RI PT

ACCEPTED MANUSCRIPT

SC

710

Figure 10 Comparison between the trends in clear-sky global radiation impacted by

712

water vapor ( E w ↓ ) and the trends in all-sky global radiation ( E qc ↓ ) at 519

713

individual CMA stations for (a) 1980 – 1989 and (b) 1990 – 2010.

AC C

EP

TE D

M AN U

711

42

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

714

TE D

( E qc ↓ ) averaged over (a) CN, (b) CE, (c) SC, and (d) OT for 1980 – 2010.

EP

716

Figure 11 Normalized anomaly time series of TCC, LCC and all-sky global radiation

AC C

715

43

ACCEPTED MANUSCRIPT AOD is retrieved by combining visibility data and MODIS aerosol data. Aerosols have considerable effect on global radiation climatology over China. Aerosol direct effect is not adequate to explain the decadal variations in global

AC C

EP

TE D

M AN U

SC

RI PT

radiation over China.