- Email: [email protected]
A critical assessment of direct radiative effects of different aerosol types on surface global radiation and its components
Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
Contents lists available at ScienceDirect
Journal of Quantitative Spectroscopy & Radiative Transfer journal homepage: www.elsevier.com/locate/jqsrt
A critical assessment of direct radiative effects of different aerosol types on surface global radiation and its components Xiangao Xia n LAGEO, Institute of Atmospheric Physics, China
a r t i c l e in f o
abstract
Article history: Received 7 June 2014 Received in revised form 19 July 2014 Accepted 21 July 2014 Available online 1 August 2014
A critical assessment of direct radiative effects of different aerosol types on surface global, direct and diffuse radiation is presented. The analysis is based on measurements of aerosol optical properties and surface solar radiation (SSR) of cloud-free days at the Baseline Surface Radiation Network (BSRN) and Aerosol Robotic Network station (AERONET) of Xianghe over the North China Plain between October 2004 and May 2012. Six aerosol types are classified based on aerosol size and absorption from the AERONET retrieval products, including two coarse-mode dominated aerosol types: dust (DU: fine mode fraction (FMF) o0.4) and polluted dust (PD: FMF within 0.4–0.7) and four fine-mode dominated aerosol types (FMF 40.7) but with different single scattering albedo (SSA): highly absorbing (HA: SSA o0.85), moderately absorbing (MA: SSA within 0.85–0.90), slightly absorbing (SA: SSA within 0.90–0.95) and very weakly absorbing (WA: SSA4 0.95). Dramatic differences in aerosol direct radiative effect (ADRE) on global SSR and its components between the six aerosol types have been revealed. ADRE efficiency on global SSR for solar zenight angle (SZA) between 551 and 651 ranges from 106 W m 2 for WA to 181 W m 2 for HA. The minimum ADRE efficiency on diffuse SSR is derived for HA aerosols, being 113 W m 2 that is about half of that by DU, the maximum value of six aerosol types. ADRE efficiency on global SSR by DU and PD ( 141 to 150 W m 2 for SZA between 551 and 651) is comparable to that by MA, although 100 W m 2 more direct SSR is extincted by DU and PD than by MA. DU and PD induce more diffuse SSR than MA that offsets larger reduction of direct SSR by DU and PD. Implications of the results to related researches are detailed discussed. The results are derived from aerosol and radiation data in the North China Plain, however the method can be used to any other stations with similar measurements. & 2014 Elsevier Ltd. All rights reserved.
Keywords: Aerosol Direct radiative effect Aerosol type
1. Introduction Tropospheric aerosols are minute particles suspended in the troposphere that scatter and absorb sunlight. Their extinction of sunlight can result in a reduction of surface solar radiation (SSR), which has the potential to be interesting in global warming, water cycling and solar energy
n
Tel.: þ86 10 82995071; fax: þ 86 90 82995073. E-mail address: [email protected]
http://dx.doi.org/10.1016/j.jqsrt.2014.07.020 0022-4073/& 2014 Elsevier Ltd. All rights reserved.
application, just to name a few [1]. Both natural and anthropogenic processes produce aerosols that vary in size and composition and thereby in their optical properties. Emissions of aerosols into the troposphere from the major sources include sulfates from the oxidation of sulphurcontaining gases, nitrates from gaseous nitrogen species, organic materials from biomass combustion and oxidation of volatile organic compounds, soot from combustion, and mineral dust from Aeolian processes. Natural aerosols are generally composed of these aerosol species but vary with their percentages. Owing to the short lifetime of aerosol
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
particles in the tropospheric atmosphere, and the nonuniform distribution of sources, their geographical distribution is highly non-uniform. As a consequence, the relative importance of the numerous sources shows considerably spatial and temporal variations, which results in substantial variation of aerosol optical properties across the world [2]. The North China Plain (NCP) is a densely populated region of the world and has experienced unprecedented economic and population growth during the past three decades, which has resulted in a remarkable increase of anthropogenic aerosols [3,4]. Additionally, the dust originates in the deserts and Gobi deserts of Mongolia, northern China and Kazakhstan where dense clouds of fine, dry soil particles are sporadically emitted into the atmosphere under favorable weather conditions. These dust clouds are then carried eastward by prevailing winds and pass over the NCP [5]. Physical and optical properties between dust and anthropogenic aerosols are quite different. Dust aerosols are major coarse particles that absorb solar radiation in the ultraviolet range more than in the visible spectrum [6,7]. Anthropogenic aerosols are major fine-mode particles and their optical properties are highly dependent on their physical and chemical properties. The mixture of coarse dust particles and anthropogenic pollution has resulted in a rather complex nature of aerosol physical and optical properties [8]. The external linear mixing of both fine- and coarsemode components dominates variations in the complex refractive index and single scattering albedo (SSA) in spring and winter when the fine-mode fraction of extinction is less than about 0.6 [6,9]. The complexity of aerosols in the NCP provides us a natural laboratory to study how aerosols affect SSR. More importantly, we can learn how aerosol effects on SSR vary with aerosol types. Sunphotometer has been widely used to investigate various column-integrated aerosol properties. The Aerosol Robotic Network (AERONET) program provides a long-term, continuous and readily accessible public domain database of aerosol optical, microphysical and radiative properties across the world for aerosol characterization research, validation of satellite retrievals, and synergism with satellite and model databases [6,10,11]. The AERONET station has been established since the beginning of this century in the Beijing and Xianghe, an urban and a suburban station in the NCP. More importantly, continuous SSRs have been measured since the fall of 2004, which provides use a good opportunity to study aerosol effects on SSR via its scattering and absorption, not only on global SSR, but also on its components, i.e., direct and diffuse SSR. Aerosol direct effects on global and diffuse SSR were detailed studied in Refs. [12,13]. However, there was still no attempt reported in the literature, as far as I know, to reveal considerable difference in aerosol effects on SSR by different aerosol types, which, however, has the potential to be interesting in researches such as aerosol effects on plant production, carbon cycling and solar energy application. 2. Site, data and methodology 2.1. Site The sunphotometer and a set of pyranometers were established in fall of 2004 at Xianghe (XH) station, a
73
suburban station in the NCP. XH, a county of Hebei Province close to Beijing, is characterized by agricultural land and light industries. The site experiences both natural aerosols and anthropogenic urban and rural pollutants depending on meteorological parameters such as wind speed, rainfall, relative humidity, cloud amount and type [14]. 2.2. Data 2.2.1. AERONET data The CIMEL sunphotometer, which is the standard instrument of the AERONET, is able to measure direct radiation from the sun at wavelengths ranging from 340 nm to 1020 nm and the angular distribution of sky radiance at four wavelengths (440, 675, 870, and 1020 nm). In addition to the continuous cloud-screened measurements of aerosol optical depth (AOD) with an accuracy of 0.01–0.02 [10,15], the inversion algorithm retrieves aerosol physical and optical properties from the spectral AODs and almucantar scans of radiances as a function of scattering angle, such as aerosol size distribution, aerosol complex refractive index, SSA (the ratio of scattering to extinction), and asymmetry factor (ASY) at four wavelengths. The aerosol inversion algorithm was developed by Dubovik and King [16] and was then further improved to take into account non-spherical shapes of aerosol particles [17,18]. The AERONET inversion algorithm also calculates SSR and aerosol direct radiative effect (ADRE) using the DISORT module with the retrieved aerosol size distribution and complex refractive index as inputs [19]. ADRE is defined here as follows: ADRE ¼ SSRWA SSRNA ;
ð1Þ
where SSRWA and SSRNA represent SSR with and without the presence of aerosols in the atmosphere, respectively. The ADRE efficiency is defined for the case of no aerosols relative to the value at AOD at 550 nm of 1.0. 2.2.2. Radiation data An aerosol–radiation–cloud platform was established at XH in September 2004 as part of the “East Asian Study of Tropospheric Aerosols: an International Regional Experiment” project (EAST-AIRE) [14], and has since taken continuous measurements of various radiative and aerosol quantities. SSR was measured by both independent and redundant pyranometers for quality control purposes. Kipp and Zonne's CM21 and CM11 radiometers were used to measure global SSR, respectively. Diffuse and direct SSRs were measured by a black and white pyranometer and a normal incidence pyrheliometer, both mounted on the solar track. All SSR measurements were taken at a 1-min temporal resolution. The data were quality-checked using the Baseline Surface Radiation Network (BSRN) quality control procedures and then submitted to the BSRN data archive. Measurements of SSR from October 2004 to May 2012 were used in this study. Clear-sky SSR measurements were firstly collected based on an empirical clear-sky detection algorithm [20] that was modified to cope with the specific conditions under study [12]. The rationale behind this algorithm is
74
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
contain AOD measurements at wavelengths from the ultraviolet to infrared spectrum, the closest AERONET inversion products, i.e., SSA and ASY at four AERONET diffuse wavelengths, are interpolated and extrapolated into values at the spectral divisions of the SBDART model to simulate SSR. Surface albedo data at seven wavelengths from the MODIS are used to derive the SBDART simulations [24]. The solar zenith angle dependence of surface albedo is not considered. The column-integrated ozone amounts are from the ozone overpass data based on the Ozone Monitoring Instrument (http://www.temis.nl/protocols/o3col/overpass_omi.html). Instantaneous AERONET level 2.0 water vapor data are used to scale water vapor content of the standard atmospheric model. Uncertainty of input parameters for SBDART calculations and the resulting errors in computed fluxes were estimated to be 8.767 3.44 W m 2 [22]. Fig. 1. The scatter-plot of fine mode fraction (FMF) and single scattering albedo (SSA) at 550 nm. Five aerosol types are defined according to this combination. The occurrence frequency, the mean and one standard deviation of FMF and SSA for each aerosol types are also included.
that the temporal variability of clear-sky SSR should be much less than that of cloudy-sky. A few tests were used to check whether the 1-min measurement was made under clear-sky conditions. A running standard deviation within half hour is calculated to check the variability of the measurement. The global and diffuse measurements should be greater than a limit for the maximum global and diffuse radiation. The standard deviations within 1-min period should be less than 0.02 [12]. About 20% of measurements are identified as clear-sky data. 2.3. Radiative transfer model calculation The AERONET inversion and thereby ADRE calculation are only available for the solar zenith angle (SZA) between 501 and 751. More importantly, we cannot derive the ADRE of different aerosol types and estimate aerosol effects on diffuse SSR from the AERONET retrieval products. Therefore, the Santa Barbara DISORT Atmospheric Radiative Transfer model (SBDART) is used to calculate SSR using AERONET level 2.0 direct sun aerosol products to extend the ADRE analysis to a wider range of SZA based on much more data points. This is further beneficial for the statistical analysis of the relationships between AOD and SSR for different aerosol types and for global, direct and diffuse SSR, which are then used to estimate the ADREs on global SSR and its components by different aerosol types. The SBDART model relies on LOWTRAN-7 atmospheric transmission data and the radiative transfer equation is numerically integrated with the DISORT radiative transfer module [21]. It has been shown that SBDART simulations of SSR within 0.2–5.0 mm agree at better than 3% with surface measurements of SSR [22,23]. It should be noted that the spectral range of pyranometer is within 0.305–2.8 mm. However, the pyranometer is calibrated through side-by-side comparison with cavity radiometer measurements of solar radiation within 0.2–5.0 mm. Therefore, pyranometer measurements are partially corrected to 0.2–5.0 mm by the calibration. Because the AERONET direct sun products only
2.4. Classification of aerosols Aerosol size and absorption are widely used to separate among aerosol type clusters [7,25]. Fine mode fraction (FMF) and SSA at 550 nm are used to classify aerosols into six aerosol types. Note that FMF and SSA at 550 nm are derived from interpolation of values at 440 and 670 nm. Dust aerosol (DU) is characterized by FMFo0.4. Aerosols with FMF within 0.4–0.7 are classified to polluted dust (PD). Aerosols with FMF larger than 0.7 are classified into four sub-groups based on their SSA values, i.e., highly absorbing (HA: SSA o0.85), moderately absorbing (MA: SSA within 0.85–0.90), slightly absorbing (SA: SSA within 0.90–0.95) and very weak absorbing (WA: SSA 40.95). Fig. 1 presents the scatterplot of FMF and SSA at 550 nm, in which the occurrence frequencies of the six aerosol types as well as their mean FMF and SSA values are also presented. The most frequent observed aerosol type is SA ( 32%) that is followed by MA ( 27%) and PD ( 20%). The occurrence percentages of the remaining 3 types are from 6% (DU) to 9% (WA). Occurrence of MA is not dependent season, but 80% of MA cases occur in fall and winter. DU and PD is mainly observed in spring and winter. HA aerosols mainly occur in fall and winter, on the contrary, WA mainly in summer, the rainy season. The dominant aerosol types are DU in spring, SA and HA in summer, MA and SA in fall and winter, respectively. 2.5. Methodology The AERONET AOD data are firstly collocated with SSR data for the analysis of ADRE. Given the fact that water vapor absorbs solar radiation in the infrared spectrum which shows a distinct seasonal variation, water vapor absorption should be considered when we study the relationship of AOD to SSR and thereby ADRE. The observations of SSR are scaled to values with the same water vapor content using the following equation: SSRscaled ¼ ½SSRfSBDART =SSRvSBDART SSRobs ;
ð2Þ
where SSRfSBDART and SSRvSBDART represent SBDART simulations of SSR using fixed water vapor and AERONET water
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
vapor contents as input, respectively. SSRscaled and SSRobs represent scaled and observed SSR, respectively. Since the SZA is one of the dominant factors deteriming SSR, the effect of SZA shold be isolated. This is achieved by grouping the aerosol and radiation measurements according to the SZA. The SSR at time with AOD measurement are firstly interpolated from SSR measurements within 30 min of AOD measurement and thereby AOD and SSR data are collocated. The data points are then subdivided into 20 groups with the same data points based on their SZA. It should also be noted that SSR has been normalized to the same SZA and solar distance. Finally, empirical equations describing the relationships between AOD and global, direct and diffuse SSRs are developed for the each narrow range of the SZA, which is then used to study the instantenous ADRE. Diurnal mean ADRE is then calculated from integration of instantenous ADRE. SSR for zero AOD, which cannot be directly measured, is estimated from measurements, which is one of most important advantages of this method. The second feature is that water vapor effect is isolated by Eq. (2) in the analysis of ADRE. 3. Comparison of measured and modelled SSR Fig. 2 presents a comparison of measured global SSR and its components measured by pyranometers to SBDART calculations. The observations and calculations are in good agreement. It should be noted that global, direct and diffuse SSR are measured by three independent solar radiometers, i.e., CM21, NIP and B&W. The mean biases are 0.1, 4.4 and 5.0 W m 2 for global, direct and diffuse SSR, respectively. The standard deviations are within 6.9– 12.2 W m 2. Differences are well within the measurement and model simulation uncertainties [22].
4. Analysis of ARF
Simulation radiation (W m−2)
Fig. 3 presents a scatter plot of AOD and measurements of global SSR at 20 specified SZAs for MA aerosols. The analysis is performed seperately for the six aerosol types. Similar analysis is also performed based on SBDART model simulations. It has been shown that an exponential equation is more sutiable than a linear equation for describing the Global 1000 Bias=0.1[0.0%] STD=12.2[3.3%] 800
75
relationship between AOD and global SSR if AOD exceeds 0.5 [12], which is supported by the fact that the standard error of the regression using the exponential equation (values in brackets) is always less than that using the linear equation. Therefore, the exponential equation is used here to discuss how global SSR changes with AOD: SSRglobal ¼ a expð b AODÞ
ð3Þ
where parameter a resprensts expected global SSR at a specified SZA with the absence of aerosols in the atmosphere (i.e., AOD¼zero). Parameter b deterimines how SSR varies with AOD. Both parameters have a close relationship to the SZA and the relationship is well described by a power law equation (not shown). Absolute values of b decrease with the SZA, indicating the larger the SZA, the greater the level of aerosol extinction. This is not surprising since the solar light transfer path increases with the SZA. Based on Eq. (3), we can calculate the ADRE efficiency for any given SZA as follows: ADRE ¼ a expð bÞ a
ð4Þ
where the former item of the left side represents global SSR with the presence of one unit of AOD aerosols and the second item represents the expected globsl SSR with the absence of aerosols in the atmosphere. Similarly, ADRE on direct SSR is studied based on AOD and direct SSR measurements. Fig. 4 presents a scatter plot of AOD and direct SSR for MA aerosols as an example. The same equations as that of global SSR is used to describe the relationship of direct SSR to AOD. With regard to ADRE on diffuse SSR, diffuse SSR increases as AOD increases as a result of increased aerosol scattering, which is clearly shown in Fig. 5, a scatter plot of AOD and diffuse SSR for MA aerosols. Diffuse SSR linearly increases with AOD for AODo0.25. The increasing rate of diffuse SSR with AOD decreases as AOD continously increases. Diffuse SSR generally levels off as AOD exceeds 0.5. This behavior of diffuse SSR to AOD can be well described by a power law equation as follows: SSRdif f use ¼ a AODb
ð5Þ
where parameter a represents diffuse SSR for one unit of AOD. Given the fact that we cannot derive the expected diffuse SSR for zero AOD from this equation, a linear equation 600 Diffuse Bias=5.0[4.2%] 500 STD=6.9[5.8%]
Direct 1000 Bias=−4.4[−1.7%] STD=10.9[4.3%] 800
400 600
600
400
400
200
200
300
200 400 600 800 1000
200 100 200 400 600 800 1000
200
400
600
−2
Observation radiation (W m ) Fig. 2. Density plot of measured and modelled SSR for global (left), direct (middle) and diffuse SSR (right). The mean bias and one standard deviation between them are also included.
Global SSR
800
Global SSR
800
Global SSR
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
800
Global SSR
76
800
600 400 200
600 400 200
600 400 200
600 400 200 0
0.5 AOD
1
550nm
0
0.5 AOD
1
0
550nm
0.5 AOD
1
550nm
0
0.5 AOD
1
550nm
0
0.5 AOD
1
550nm
Fig. 3. The scatter-plot of AOD at 550 nm and global SSR for MA aerosol types at twenty SZA ranges. The red and blue points represent measured and modelled SSR, respectively. An exponential equation is used to describe the variation of global SSR as a function of AOD. Values in the bracket represent the standard deviation of the regression analysis. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
is then developed based on diffuse SSR and AOD for measurments with AODo0.25, i.e., SSRdif f use ¼ c þ d AODðAOD o 0:25Þ. Thereafter, diffuse SSR for zero AOD is represented by parameter c (the intercept of the equation). Based on the combination of the linear equation and the power law equation for diffuse SSR, the ADRE on diffuse SSR is then calculated by subtracting a by c. Fig. 6 presents ADRE efficiency for global (ADREglobal, left), direct (ADREdirect, middle) and diffuse SSR (ADREdiffuse, right) for the six aerosol types, respectively. One of the interesting feature is that ADRE varies dramatically with aerosol types and this difference increases as the SZA increases. ADRE (the absolute magnitude) of the four fine-mode dominated aerosol types increases as SSA decreases. The largest ADREglobal is produced by HA aerosols that varies from about 100 W m 2 for SZA of 781 to 220 W m 2 for SZA of 501. In order to make a consistent compasion between the ADRE of the six aerosol types, the mean ADRE efficiency values have been provided, if not otherwise specified, for SZA between 551 and651. The mean ADREglobal for SZA between 551 and 651 is about 181 W m 2, which is larger than that of WA (with the minimum ADRE efficiency) by 75 W m 2. ADRE2 , respectively, which is global by DU and PD is 141 and 150 W m comparable to that by MA. The results are consistent with
the results by Garcia et al. [19] based on the AERONET inversion products. It is interesting to note that ADREdirect of HA is in the middle of the six aerosol types. ADREdirect of HA is generally less than that of DU and PD but larger than that of the three other fine-mode dominated aerosol types. The largest ADREdirect is derived for DU aerosols. The mean ADREdirect of DU aerosols is 373 W m 2, which is larger than that of WA (with the minimum ADREdirect) by 100 W m 2. The extinction of DU shows much less wavelength dependence. Dust extinction in the infrared spectrum should be much larger than that of fine particles for the same AOD, which results in much more solar direct radiation is scattered by dust aerosols (larger ADREdirect). However, scattering solar radiation reaching the surface by dust aerosols should be larger than that of fine-mode dominated aerosol types because of their larger size, which is supported by the fact that the maximum increase rate of diffuse SSR is derived for the DU aerosols that is followed by PD aerosols. The mean ADREdiffuse by DU aerosols is 230 W m 2, which is nearly double larger than that of HA (with the minimum ADREdiffuse). ADREdiffuse for SZAr651 was estimabed to be about 220 W m 2 at the Southern Great Plains [13], which is slightly smaller than
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
Direct SSR Direct SSR
800
Direct SSR
800
Direct SSR
SZA=44.4
800
800
Y Y
SZA=53.2
=698.5e =715.5e
[15.7] [11.6]
Y Y
SZA=57.6
=556.4e =564.6e
[13.5] [10.6]
Y Y
77
SZA=60.6
=479.8e =492.6e
[13.8] [11.9]
Y Y
SZA=62.9
=432.0e =443.7e
[11.0] [9.1]
Y Y
=391.4e =404.9e
[10.4] [8.4]
600 400 200 SZA=64.6 Y Y
SZA=65.7
=355.9e =368.1e
[11.6] [9.1]
Y Y
SZA=67.0
=343.3e =352.2e
[9.7] [8.0]
Y Y
SZA=67.9
=317.7e =330.2e
[10.2] [7.8]
Y Y
SZA=68.9
=304.5e =314.2e
[9.4] [7.5]
Y Y
=285.7e =296.3e
[9.2] [7.4]
600 400 200 SZA=69.9 Y Y
SZA=70.8
=268.6e =277.4e
[9.4] [8.4]
Y Y
SZA=71.8
=252.6e =259.2e
[8.8] [7.6]
Y Y
SZA=72.7
=235.3e =243.9e
[7.9] [7.3]
Y Y
SZA=73.5
=220.6e =228.1e
[7.7] [6.2]
Y Y
=208.2e =215.3e
[7.5] [6.3]
600 400 200 SZA=74.0 Y Y
SZA=74.9
=196.7e =204.1e
[7.3] [5.6]
Y Y
SZA=75.6
=183.0e =188.4e
[7.4] [5.6]
Y Y
SZA=76.9
=173.4e =179.9e
[5.6] [4.7]
Y Y
SZA=78.2
=152.9e =158.6e
[5.3] [3.8]
Y Y
=135.7e =140.4e
[4.3] [3.3]
600 400 200 0
0.5 AOD
1
550nm
0
0.5 AOD
1
0
550nm
0.5 AOD
1
0
550nm
0.5 AOD
1
550nm
0
0.5 AOD
1
550nm
Fig. 4. Similar to Fig. 3 but for direct SSR.
that by dust aerosols but larger than that by the four finemode dominated aerosol types. 100 W m 2 more direct SSR is extincted by DU and PD than by MA, which is offset by a larger increase of diffuse SSR induced by DU and PD. Therefore, ADREglobal of DU and PD is comparable to that by MA. The scattering and absorbing of direct solar radiation by HA aerosols is less than that of dust aerosols, however, its strong absorption results in much less diffuse solar radiation reaching the surface. Therefore, it is not surprising that the maximum ADREglobal is derived for HA aerosols. ADREs derived from SBDART model simulations are in good agreement with those from measurements. The difference in ADRE efficiency derived from measuremens and SBDART simulations is generally not larger than 10 W m 2 and the mean difference is 7.0 W m 2 (73.8 W m 2). Fig. 7 presents seasonal mean of the daily integrations of the ADRE for each aerosol type. A distinct seasonal variation of ADRE is evident as a result of seasonal variation of solar radiation available for the scattering and absorbing by aerosols. DU and WA showed the maximum and minimum ADREdirect (in absolute value). ADREdirect of DU varied from 95 W m 2 in winter to 220 W m 2 in summer, the corresponding value of WA ranged from 80 W m 2 to 168 W m 2. As a result of very strong absorption, HA showed the minimum ADREdiffuse, being 30 W m 2 in winter and 82 W m 2 in
summer which was less than that of DU (with the maximum ADREdiffuse) by 21 W m 2 and 63 W m 2. The largest ADREglobal was observed for HA, being from 65 W m 2 in winter to 120 W m 2 in summer, which was about two times larger than that of WA (with the minimum ADREglobal) in seasons except winter. 5. Discussion It is widely suggested that a high proportion of diffuse SSR increases the radiation use efficiency (RUE) of photosynthesis since diffuse SSR is able to penerate vegetative layers. Therefore, much attention has been paid to the effect of diffuse SSR on crop growth and carbon cycling [26]. Significant increase of diffuse SSR induced by aerosols were reported [12,13], but theses studies did not consider potential different performance of aerosol types. Dramatical difference in aerosol effect on diffuse SSR between the six aerosol types was clearly revealed in this paper. More specific, long-range transportion of dust from the Gobi Deserts to the NCP is likely benefical for plant production since the increasing rate of diffuse SSR by dust scattering is nearly double larger than that of anthropogenic fine-mode dominated particles with strong absorption. Therefore, we should take aerosol types into consideration when potential aerosol effects on plant production and carbon cycling are studied.
Diffuse SSR
300
Diffuse SSR
300
Diffuse SSR
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
300
Diffuse SSR
78
300
200 100
200 100
200 100
200 100 0
0.5 AOD
1
550nm
0
0.5 AOD
1
0
550nm
0.5 AOD
1
0
550nm
0.5 AOD
1
550nm
0
0.5 AOD
1
550nm
Fig. 5. Similar to Fig. 3 but for diffuse SSR. A power law equation is used to describe the variation of diffuse SSR as a function of AOD.
Global SSR and direct SSR are of particular interest to concentrating solar thermal installations and installations that track the position of the sun [27]. Since aerosol effects on SSR differ dramatically between aerosol types, evaluation of global SSR as the main data input in the calculation of photovoltaic (PV) power production and direct SSR as the most important parameter for concentrated PV should also take aerosol types into consideration. Additionally, multiple junction and thin film PV technologies respond differently to varying spectral distributions of solar energy [28]. Much attention should also be paid to varying effects of different aerosol types on spectral solar energy that needs further study. Knowledge of long-term changes in aerosols is fundamental for a better understanding of climate change. However, little information on changes in aerosol concentration, especially prior to the 1980s, is availabe. It has been suggested to use long-term measurements of sunshine duration, especially at sunrise and sunset, to detect changes in aerosol loading [29]. The World Meteorological Organization uses the term “sunshine duration” to mean the cumulative time during which an area receives direct irradiance from the sun of at least 120 W m 2. Analysis of ADRE on direct SSR has revealed signficant difference in ADRE on direct SSR between the six aerosol types. Therefore, aerosol type should be considered when we attempt to derive aerosol loading from sunshine duration data.
6. Conclusions Using the AERONET aerosol products and the BSRN SSR data at Xianghe over the NCP, aerosol direct radiative effects of different aerosol types on global SSR and its components are critically assessed. Aerosols are firstly classified into six types based on aerosol size and absorption. Aerosol optical data and SSR measurements are then collocated for different aerosol types to derive empirical relationships between AOD and SSR. ADRE is finally derived from these equations for the six aerosol types. Dramatic variation of ADRE between aerosol types is clearly revealed. The major conclusions can be summarized as follows: A statistical method is developed to establish empirical relationships between SSR and AOD based on collocated measurements of aerosol optical properties and SSR data, which are then used to study ADRE effects on global SSR and its components. This method can be potentially applied to regions with the presence of both the AERONET and BSRN data. ADRE varies dramatically with aerosol types and this difference increases as the SZA increases. HA aerosols had the largest global ADRE efficiency, being 181 W m 2 for SZA between 551 and 651, which is larger than the minimum ADRE effciency of WA aerosol by 75 W m 2. DU aerosols showed the largest direct ADRE efficiency,
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
−50
−100
Global
Direct
ADRE (W m−2)
−100 −300 −400 −500 −600 x: observation +: simulation 0.4
0.6
0.8
−700 0.2
Diffuse
150
−200
−250 0.2
400 350 DU PD 300 HA MA 250 SA 200 WA
−200
−150
79
0.4
100 0.6
0.8 The cosine of the solar zenith angle
50 0.2
0.4
0.6
0.8
Fig. 6. The scatter-plot between ADRE efficiency and the cosine of the solar zenith angle for global (left), direct (middle) and diffuse (right) SSR and for six aerosol types. Values derived from measurements and model simulations are represented by “plus” and “cross”, respectively.
Fig. 7. Seasonal mean of the daily integrations of the ADRE on global, direct and diffuse irradiance for six aerosol types derived from measurements.
being 373 W m 2 for SZA between 551 and 651, which is larger than that of WA aerosols (with the minimum ADRE efficiency) by 100 W m 2. The mean ADRE on diffuse SSR by DU aerosols for SZA between 551 and 651 is 230 W m 2, which is nearly double larger than that of HA (with the minimum ADRE on diffuse SSR). The scattering and absorbing of direct solar radiation by HA aerosols is less than that of dust aerosols, however, its strong absorption results in much less diffuse solar radiation reaching the
surface. Therefore, the maximum global ADRE efficiency is derived for HA aerosols. ADRE efficiency on global SSR by DU and PD is 141 and 150 W m 2, respectively, which is comparable to that by MA. However, a reduction of dirct SSR induced by DU and PD is larger than that by MA 100 W m 2, which is offset by a larger increase of diffuse SSR induced by DU and PD. The results are derived from aerosol and surface irradiance measurements at Xianghe, a suburban station in
80
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
the North China Plain. The method, however, is potentially useful for researches at any stations with similar measurements. Dramatic difference in ADRE on global SSR and its components between aerosol types revealed in this paper is of particular interest to aerosol potential effects on plant production and carbon cycling, to the application of solar energy and to detecting long-term changes in aerosols from sunshine duration measurements.
[12]
[13]
[14]
Acknowledgments The AERONET and BSRN data used in the paper are publicly available from http://aeronet.gsfc.nasa.gov/ and http://www.bsrn.awi.de/, respectively. This research was supported by the National Basic Research Program of China (2010CB950804), the National Natural Science Foundation of China (4085011), and the Strategic Priority Research Program of the Chinese Academy of Sciences, China (XDA050100301). References [1] Wild M. Global dimming and brightening: a review. J Geophys Res 2009;114:D00D16. http://dx.doi.org/10.1029/2008JD011470. [2] Dubovik O, Holben B, Eck T, Smirnov A, Kaufman Y, King M, et al. Variability of absorption and optical properties of key aerosol types observed in worldwide locations. J Atmos Sci 2002;59:590–608. [3] Luo Y, Lu D, Zhou X, Li W, He Q. Characteristics of the spatial distribution and yearly variation of aerosol optical depth over China in last 30 years. J Geophys Res 2001;106:14501–13. [4] Qiu J, Yang L. Variation characteristics of atmospheric aerosol optical depths and visibility in North China during 1980–1994. Atmos Environ 2000;34:603–9. [5] Huang J, Minnis P, Chen B, Huang Z, Liu Z, Zhao Q, et al. Long-range transport and vertical structure of Asian dust from CALIPSO and surface measurements during PACDEX. J Geophys Res 2008;113: D23212. http://dx.doi.org/10.1029/2008JD010620. [6] Eck T, Holben B, Sinyuk A, Pinker R, Goloub P, Chen H, et al. Climatological aspects of the optical properties of fine/coarse mode aerosol mixtures. J Geophys Res 2010;115:D19205. http://dx.doi.org/10.1029/ 2010JD014002. [7] Russell P, Bergstrom R, Shinozuka Y, Clarke A, DeCarlo P, Jimenez J, et al. Absorption Angstrom Exponent in AERONET and related data as an indicator of aerosol composition. Atmos Chem Phys 2010;10: 1155–69. http://dx.doi.org/10.5194/acp-10-1155-2010. [8] Li Z, Li C, Chen H, Tsay S, Holben B, Huang J, et al. East Asian Studies of Tropospheric Aerosols and their Impact on Regional Climate (EAST‐AIRC): an overview. J Geophys Res 2011;116:1–15, http://dx.doi.org/10.1029/2010JD015257. [9] Xia X, Chen H, Goloub P, Zong X, Zhang W, Wang P. Climatological aspects of aerosol optical properties in North China Plain based on ground and satellite remote-sensing data. J Quant Spectrosc Radiat Transf 2013;127:12–23. [10] Holben B, Eck T, Slutsker I, Tanre D, Buis J, Setzer A, et al. AERONET – a federated instrument network and data archive for aerosol characterization. Remote Sens Environ 1998;66:1–16. [11] Che H, Xia X, Zhu J, Li Z, Dubovik O, Holben B, et al. Column aerosol optical properties and aerosol radiative forcing during a serious
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
haze-fog month over North China Plain in 2013 based on groundbased sunphotometer measurements. Atmos Chem Phys 2014;14: 2125–38. http://dx.doi.org/10.5194/acp-14-2125-2014. Xia X, Li Z, Wang P, Chen H, Cribb M. Estimation of aerosol effects on surface irradiance based on measurements and radiative transfer model simulations in northern China. J Geophys Res 2007;112: 1–11. http://dx.doi.org/10.1029/2006JD008337. Creekmore T, Joseph E, Long C, Li S. Quantifying aerosol direct effects from broadband irradiance and spectral aerosol optical depth observations. J Geophys Res 2014;119:1–15, http://dx.doi.org/10.1002/2013JD021217. Li Z, Xia X, Cribb M, Mi W, Holben B, Wang P, et al. Aerosol optical properties and their radiative effects in northern China. J Geophys Res 2006;112:1–11. http://dx.doi.org/10.1029/2006JD007382. Eck T, Holben B, Reid J, Dubovik O, Smirnov A, O'Neill N, et al. Wavelength dependence of the optical depth of biomass burning, urban, and desert dust aerosol. J Geophys Res 1999;104:31333–49. Dubovik O, King M. A flexible inversion algorithm for retrieval of aerosol optical properties from Sun and sky radiance measurements. J Geophys Res 2000;105:20673–96. Dubovik O, Sinyuk A, Lapyonok T, Holben B, Mishchenko M, Yang P, et al. The application of spheroid models to account for aerosol particle nonsphericity in remote sensing of desert dust. J Geophys Res 2006;111:D11208. http://dx.doi.org/10.1029/2005JD006619. Mishchenko M, Travis D, Mackowski D. T-matrix computations of light scattering by nonspherical particles: a review. J Quant Spectrosc Radiat Transfer 1996;55:535–75. Garcia O, Diaz J, Exposito F, Diaz A, Dubovik O, Derimian Y, et al. Shortwave radiative forcing and efficiency of key aerosol types using AERONET data. Atmos Chem Phys 2012;12:5129–45, http://dx.doi.org/10.5194/acp-12-5129-2012. Long C, Ackerman T. Identification of clear skies from broadband pyranometer measurements and calculation of downwelling shortwave cloud effects. J Geophys Res 2000;105:15609–26. Ricchiazzi P, Yang S, Gautier C, Sowle D. SBDART: a research and teaching software tool for plane-parallel radiative transfer in the Earth's atmosphere. Bull Am Meteorol Soc 1998;79:2101–14. Li Z, Lee K, Wang Y, Xin X, Hao W. First observation‐based estimates of cloud‐free aerosol radiative forcing across China. J Geophys Res 2010;115:D00K18. http://dx.doi.org/10.1029/2009JD013306. Michalsky J, Dolce R, Dutton E, Haeffelin M, Major G, Schlemmer J, et al. Results from the first ARM diffuse horizontal shortwave irradiance comparison. J Geophys Res 2003;108:4108, http://dx.doi.org/10.1029/2002JD002825. Schafer C, Gao F, Strahler A, Lucht W, Li W, Tsang T, et al. First operational BRDF, albedo nadir reflectance products from MODIS. Remote Sens Environ 2002;83:135–48. Gilles D, Holben B, Eck T, Sinyuk A, Smirnov A, Slutsker I, et al. An analysis of AERONET aerosol absorption properties and classifications representative of aerosol source regions. J Geophys Res 2012;117:D17203. http://dx.doi.org/10.1029/2012JD018127. Mercado L, Bellouin N, Sitch S, Boucher O, Huntingford C, Wild M, et al. Impact of changes in diffuse radiation on the global land carbon sink. Nature 2009;458:1014–7. http://dx.doi.org/10.1039/nature07949. Suri M, Thomas H, Ewan D, Heinz O. Potential of solar electricity generation in the European Union member states and candidate countries. Sol Energy 2007;81:1295–305. Krishnan P, Schüttauf J, van der Werf C, Hassanzadeh B, van Sark W, Schropp R. Response to simulated typical daily outdoor irradiation conditions of thin film silicon based triple band gap, triple junction solar cells. Sol Energy Mater Sol Cells 2009;93:691–7. Sanchez-Romero A, Sanchez-Lorenzo A, Calbo J, Gonzalez J, ArorinMolina C. The signal of aerosol induced changes in sunshine duration records: a review of the evidence. J Geophys Res 2014;119: 1–17. http://dx.doi.org/10.1002/2013JD021393.
Contents lists available at ScienceDirect
Journal of Quantitative Spectroscopy & Radiative Transfer journal homepage: www.elsevier.com/locate/jqsrt
A critical assessment of direct radiative effects of different aerosol types on surface global radiation and its components Xiangao Xia n LAGEO, Institute of Atmospheric Physics, China
a r t i c l e in f o
abstract
Article history: Received 7 June 2014 Received in revised form 19 July 2014 Accepted 21 July 2014 Available online 1 August 2014
A critical assessment of direct radiative effects of different aerosol types on surface global, direct and diffuse radiation is presented. The analysis is based on measurements of aerosol optical properties and surface solar radiation (SSR) of cloud-free days at the Baseline Surface Radiation Network (BSRN) and Aerosol Robotic Network station (AERONET) of Xianghe over the North China Plain between October 2004 and May 2012. Six aerosol types are classified based on aerosol size and absorption from the AERONET retrieval products, including two coarse-mode dominated aerosol types: dust (DU: fine mode fraction (FMF) o0.4) and polluted dust (PD: FMF within 0.4–0.7) and four fine-mode dominated aerosol types (FMF 40.7) but with different single scattering albedo (SSA): highly absorbing (HA: SSA o0.85), moderately absorbing (MA: SSA within 0.85–0.90), slightly absorbing (SA: SSA within 0.90–0.95) and very weakly absorbing (WA: SSA4 0.95). Dramatic differences in aerosol direct radiative effect (ADRE) on global SSR and its components between the six aerosol types have been revealed. ADRE efficiency on global SSR for solar zenight angle (SZA) between 551 and 651 ranges from 106 W m 2 for WA to 181 W m 2 for HA. The minimum ADRE efficiency on diffuse SSR is derived for HA aerosols, being 113 W m 2 that is about half of that by DU, the maximum value of six aerosol types. ADRE efficiency on global SSR by DU and PD ( 141 to 150 W m 2 for SZA between 551 and 651) is comparable to that by MA, although 100 W m 2 more direct SSR is extincted by DU and PD than by MA. DU and PD induce more diffuse SSR than MA that offsets larger reduction of direct SSR by DU and PD. Implications of the results to related researches are detailed discussed. The results are derived from aerosol and radiation data in the North China Plain, however the method can be used to any other stations with similar measurements. & 2014 Elsevier Ltd. All rights reserved.
Keywords: Aerosol Direct radiative effect Aerosol type
1. Introduction Tropospheric aerosols are minute particles suspended in the troposphere that scatter and absorb sunlight. Their extinction of sunlight can result in a reduction of surface solar radiation (SSR), which has the potential to be interesting in global warming, water cycling and solar energy
n
Tel.: þ86 10 82995071; fax: þ 86 90 82995073. E-mail address: [email protected]
http://dx.doi.org/10.1016/j.jqsrt.2014.07.020 0022-4073/& 2014 Elsevier Ltd. All rights reserved.
application, just to name a few [1]. Both natural and anthropogenic processes produce aerosols that vary in size and composition and thereby in their optical properties. Emissions of aerosols into the troposphere from the major sources include sulfates from the oxidation of sulphurcontaining gases, nitrates from gaseous nitrogen species, organic materials from biomass combustion and oxidation of volatile organic compounds, soot from combustion, and mineral dust from Aeolian processes. Natural aerosols are generally composed of these aerosol species but vary with their percentages. Owing to the short lifetime of aerosol
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
particles in the tropospheric atmosphere, and the nonuniform distribution of sources, their geographical distribution is highly non-uniform. As a consequence, the relative importance of the numerous sources shows considerably spatial and temporal variations, which results in substantial variation of aerosol optical properties across the world [2]. The North China Plain (NCP) is a densely populated region of the world and has experienced unprecedented economic and population growth during the past three decades, which has resulted in a remarkable increase of anthropogenic aerosols [3,4]. Additionally, the dust originates in the deserts and Gobi deserts of Mongolia, northern China and Kazakhstan where dense clouds of fine, dry soil particles are sporadically emitted into the atmosphere under favorable weather conditions. These dust clouds are then carried eastward by prevailing winds and pass over the NCP [5]. Physical and optical properties between dust and anthropogenic aerosols are quite different. Dust aerosols are major coarse particles that absorb solar radiation in the ultraviolet range more than in the visible spectrum [6,7]. Anthropogenic aerosols are major fine-mode particles and their optical properties are highly dependent on their physical and chemical properties. The mixture of coarse dust particles and anthropogenic pollution has resulted in a rather complex nature of aerosol physical and optical properties [8]. The external linear mixing of both fine- and coarsemode components dominates variations in the complex refractive index and single scattering albedo (SSA) in spring and winter when the fine-mode fraction of extinction is less than about 0.6 [6,9]. The complexity of aerosols in the NCP provides us a natural laboratory to study how aerosols affect SSR. More importantly, we can learn how aerosol effects on SSR vary with aerosol types. Sunphotometer has been widely used to investigate various column-integrated aerosol properties. The Aerosol Robotic Network (AERONET) program provides a long-term, continuous and readily accessible public domain database of aerosol optical, microphysical and radiative properties across the world for aerosol characterization research, validation of satellite retrievals, and synergism with satellite and model databases [6,10,11]. The AERONET station has been established since the beginning of this century in the Beijing and Xianghe, an urban and a suburban station in the NCP. More importantly, continuous SSRs have been measured since the fall of 2004, which provides use a good opportunity to study aerosol effects on SSR via its scattering and absorption, not only on global SSR, but also on its components, i.e., direct and diffuse SSR. Aerosol direct effects on global and diffuse SSR were detailed studied in Refs. [12,13]. However, there was still no attempt reported in the literature, as far as I know, to reveal considerable difference in aerosol effects on SSR by different aerosol types, which, however, has the potential to be interesting in researches such as aerosol effects on plant production, carbon cycling and solar energy application. 2. Site, data and methodology 2.1. Site The sunphotometer and a set of pyranometers were established in fall of 2004 at Xianghe (XH) station, a
73
suburban station in the NCP. XH, a county of Hebei Province close to Beijing, is characterized by agricultural land and light industries. The site experiences both natural aerosols and anthropogenic urban and rural pollutants depending on meteorological parameters such as wind speed, rainfall, relative humidity, cloud amount and type [14]. 2.2. Data 2.2.1. AERONET data The CIMEL sunphotometer, which is the standard instrument of the AERONET, is able to measure direct radiation from the sun at wavelengths ranging from 340 nm to 1020 nm and the angular distribution of sky radiance at four wavelengths (440, 675, 870, and 1020 nm). In addition to the continuous cloud-screened measurements of aerosol optical depth (AOD) with an accuracy of 0.01–0.02 [10,15], the inversion algorithm retrieves aerosol physical and optical properties from the spectral AODs and almucantar scans of radiances as a function of scattering angle, such as aerosol size distribution, aerosol complex refractive index, SSA (the ratio of scattering to extinction), and asymmetry factor (ASY) at four wavelengths. The aerosol inversion algorithm was developed by Dubovik and King [16] and was then further improved to take into account non-spherical shapes of aerosol particles [17,18]. The AERONET inversion algorithm also calculates SSR and aerosol direct radiative effect (ADRE) using the DISORT module with the retrieved aerosol size distribution and complex refractive index as inputs [19]. ADRE is defined here as follows: ADRE ¼ SSRWA SSRNA ;
ð1Þ
where SSRWA and SSRNA represent SSR with and without the presence of aerosols in the atmosphere, respectively. The ADRE efficiency is defined for the case of no aerosols relative to the value at AOD at 550 nm of 1.0. 2.2.2. Radiation data An aerosol–radiation–cloud platform was established at XH in September 2004 as part of the “East Asian Study of Tropospheric Aerosols: an International Regional Experiment” project (EAST-AIRE) [14], and has since taken continuous measurements of various radiative and aerosol quantities. SSR was measured by both independent and redundant pyranometers for quality control purposes. Kipp and Zonne's CM21 and CM11 radiometers were used to measure global SSR, respectively. Diffuse and direct SSRs were measured by a black and white pyranometer and a normal incidence pyrheliometer, both mounted on the solar track. All SSR measurements were taken at a 1-min temporal resolution. The data were quality-checked using the Baseline Surface Radiation Network (BSRN) quality control procedures and then submitted to the BSRN data archive. Measurements of SSR from October 2004 to May 2012 were used in this study. Clear-sky SSR measurements were firstly collected based on an empirical clear-sky detection algorithm [20] that was modified to cope with the specific conditions under study [12]. The rationale behind this algorithm is
74
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
contain AOD measurements at wavelengths from the ultraviolet to infrared spectrum, the closest AERONET inversion products, i.e., SSA and ASY at four AERONET diffuse wavelengths, are interpolated and extrapolated into values at the spectral divisions of the SBDART model to simulate SSR. Surface albedo data at seven wavelengths from the MODIS are used to derive the SBDART simulations [24]. The solar zenith angle dependence of surface albedo is not considered. The column-integrated ozone amounts are from the ozone overpass data based on the Ozone Monitoring Instrument (http://www.temis.nl/protocols/o3col/overpass_omi.html). Instantaneous AERONET level 2.0 water vapor data are used to scale water vapor content of the standard atmospheric model. Uncertainty of input parameters for SBDART calculations and the resulting errors in computed fluxes were estimated to be 8.767 3.44 W m 2 [22]. Fig. 1. The scatter-plot of fine mode fraction (FMF) and single scattering albedo (SSA) at 550 nm. Five aerosol types are defined according to this combination. The occurrence frequency, the mean and one standard deviation of FMF and SSA for each aerosol types are also included.
that the temporal variability of clear-sky SSR should be much less than that of cloudy-sky. A few tests were used to check whether the 1-min measurement was made under clear-sky conditions. A running standard deviation within half hour is calculated to check the variability of the measurement. The global and diffuse measurements should be greater than a limit for the maximum global and diffuse radiation. The standard deviations within 1-min period should be less than 0.02 [12]. About 20% of measurements are identified as clear-sky data. 2.3. Radiative transfer model calculation The AERONET inversion and thereby ADRE calculation are only available for the solar zenith angle (SZA) between 501 and 751. More importantly, we cannot derive the ADRE of different aerosol types and estimate aerosol effects on diffuse SSR from the AERONET retrieval products. Therefore, the Santa Barbara DISORT Atmospheric Radiative Transfer model (SBDART) is used to calculate SSR using AERONET level 2.0 direct sun aerosol products to extend the ADRE analysis to a wider range of SZA based on much more data points. This is further beneficial for the statistical analysis of the relationships between AOD and SSR for different aerosol types and for global, direct and diffuse SSR, which are then used to estimate the ADREs on global SSR and its components by different aerosol types. The SBDART model relies on LOWTRAN-7 atmospheric transmission data and the radiative transfer equation is numerically integrated with the DISORT radiative transfer module [21]. It has been shown that SBDART simulations of SSR within 0.2–5.0 mm agree at better than 3% with surface measurements of SSR [22,23]. It should be noted that the spectral range of pyranometer is within 0.305–2.8 mm. However, the pyranometer is calibrated through side-by-side comparison with cavity radiometer measurements of solar radiation within 0.2–5.0 mm. Therefore, pyranometer measurements are partially corrected to 0.2–5.0 mm by the calibration. Because the AERONET direct sun products only
2.4. Classification of aerosols Aerosol size and absorption are widely used to separate among aerosol type clusters [7,25]. Fine mode fraction (FMF) and SSA at 550 nm are used to classify aerosols into six aerosol types. Note that FMF and SSA at 550 nm are derived from interpolation of values at 440 and 670 nm. Dust aerosol (DU) is characterized by FMFo0.4. Aerosols with FMF within 0.4–0.7 are classified to polluted dust (PD). Aerosols with FMF larger than 0.7 are classified into four sub-groups based on their SSA values, i.e., highly absorbing (HA: SSA o0.85), moderately absorbing (MA: SSA within 0.85–0.90), slightly absorbing (SA: SSA within 0.90–0.95) and very weak absorbing (WA: SSA 40.95). Fig. 1 presents the scatterplot of FMF and SSA at 550 nm, in which the occurrence frequencies of the six aerosol types as well as their mean FMF and SSA values are also presented. The most frequent observed aerosol type is SA ( 32%) that is followed by MA ( 27%) and PD ( 20%). The occurrence percentages of the remaining 3 types are from 6% (DU) to 9% (WA). Occurrence of MA is not dependent season, but 80% of MA cases occur in fall and winter. DU and PD is mainly observed in spring and winter. HA aerosols mainly occur in fall and winter, on the contrary, WA mainly in summer, the rainy season. The dominant aerosol types are DU in spring, SA and HA in summer, MA and SA in fall and winter, respectively. 2.5. Methodology The AERONET AOD data are firstly collocated with SSR data for the analysis of ADRE. Given the fact that water vapor absorbs solar radiation in the infrared spectrum which shows a distinct seasonal variation, water vapor absorption should be considered when we study the relationship of AOD to SSR and thereby ADRE. The observations of SSR are scaled to values with the same water vapor content using the following equation: SSRscaled ¼ ½SSRfSBDART =SSRvSBDART SSRobs ;
ð2Þ
where SSRfSBDART and SSRvSBDART represent SBDART simulations of SSR using fixed water vapor and AERONET water
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
vapor contents as input, respectively. SSRscaled and SSRobs represent scaled and observed SSR, respectively. Since the SZA is one of the dominant factors deteriming SSR, the effect of SZA shold be isolated. This is achieved by grouping the aerosol and radiation measurements according to the SZA. The SSR at time with AOD measurement are firstly interpolated from SSR measurements within 30 min of AOD measurement and thereby AOD and SSR data are collocated. The data points are then subdivided into 20 groups with the same data points based on their SZA. It should also be noted that SSR has been normalized to the same SZA and solar distance. Finally, empirical equations describing the relationships between AOD and global, direct and diffuse SSRs are developed for the each narrow range of the SZA, which is then used to study the instantenous ADRE. Diurnal mean ADRE is then calculated from integration of instantenous ADRE. SSR for zero AOD, which cannot be directly measured, is estimated from measurements, which is one of most important advantages of this method. The second feature is that water vapor effect is isolated by Eq. (2) in the analysis of ADRE. 3. Comparison of measured and modelled SSR Fig. 2 presents a comparison of measured global SSR and its components measured by pyranometers to SBDART calculations. The observations and calculations are in good agreement. It should be noted that global, direct and diffuse SSR are measured by three independent solar radiometers, i.e., CM21, NIP and B&W. The mean biases are 0.1, 4.4 and 5.0 W m 2 for global, direct and diffuse SSR, respectively. The standard deviations are within 6.9– 12.2 W m 2. Differences are well within the measurement and model simulation uncertainties [22].
4. Analysis of ARF
Simulation radiation (W m−2)
Fig. 3 presents a scatter plot of AOD and measurements of global SSR at 20 specified SZAs for MA aerosols. The analysis is performed seperately for the six aerosol types. Similar analysis is also performed based on SBDART model simulations. It has been shown that an exponential equation is more sutiable than a linear equation for describing the Global 1000 Bias=0.1[0.0%] STD=12.2[3.3%] 800
75
relationship between AOD and global SSR if AOD exceeds 0.5 [12], which is supported by the fact that the standard error of the regression using the exponential equation (values in brackets) is always less than that using the linear equation. Therefore, the exponential equation is used here to discuss how global SSR changes with AOD: SSRglobal ¼ a expð b AODÞ
ð3Þ
where parameter a resprensts expected global SSR at a specified SZA with the absence of aerosols in the atmosphere (i.e., AOD¼zero). Parameter b deterimines how SSR varies with AOD. Both parameters have a close relationship to the SZA and the relationship is well described by a power law equation (not shown). Absolute values of b decrease with the SZA, indicating the larger the SZA, the greater the level of aerosol extinction. This is not surprising since the solar light transfer path increases with the SZA. Based on Eq. (3), we can calculate the ADRE efficiency for any given SZA as follows: ADRE ¼ a expð bÞ a
ð4Þ
where the former item of the left side represents global SSR with the presence of one unit of AOD aerosols and the second item represents the expected globsl SSR with the absence of aerosols in the atmosphere. Similarly, ADRE on direct SSR is studied based on AOD and direct SSR measurements. Fig. 4 presents a scatter plot of AOD and direct SSR for MA aerosols as an example. The same equations as that of global SSR is used to describe the relationship of direct SSR to AOD. With regard to ADRE on diffuse SSR, diffuse SSR increases as AOD increases as a result of increased aerosol scattering, which is clearly shown in Fig. 5, a scatter plot of AOD and diffuse SSR for MA aerosols. Diffuse SSR linearly increases with AOD for AODo0.25. The increasing rate of diffuse SSR with AOD decreases as AOD continously increases. Diffuse SSR generally levels off as AOD exceeds 0.5. This behavior of diffuse SSR to AOD can be well described by a power law equation as follows: SSRdif f use ¼ a AODb
ð5Þ
where parameter a represents diffuse SSR for one unit of AOD. Given the fact that we cannot derive the expected diffuse SSR for zero AOD from this equation, a linear equation 600 Diffuse Bias=5.0[4.2%] 500 STD=6.9[5.8%]
Direct 1000 Bias=−4.4[−1.7%] STD=10.9[4.3%] 800
400 600
600
400
400
200
200
300
200 400 600 800 1000
200 100 200 400 600 800 1000
200
400
600
−2
Observation radiation (W m ) Fig. 2. Density plot of measured and modelled SSR for global (left), direct (middle) and diffuse SSR (right). The mean bias and one standard deviation between them are also included.
Global SSR
800
Global SSR
800
Global SSR
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
800
Global SSR
76
800
600 400 200
600 400 200
600 400 200
600 400 200 0
0.5 AOD
1
550nm
0
0.5 AOD
1
0
550nm
0.5 AOD
1
550nm
0
0.5 AOD
1
550nm
0
0.5 AOD
1
550nm
Fig. 3. The scatter-plot of AOD at 550 nm and global SSR for MA aerosol types at twenty SZA ranges. The red and blue points represent measured and modelled SSR, respectively. An exponential equation is used to describe the variation of global SSR as a function of AOD. Values in the bracket represent the standard deviation of the regression analysis. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
is then developed based on diffuse SSR and AOD for measurments with AODo0.25, i.e., SSRdif f use ¼ c þ d AODðAOD o 0:25Þ. Thereafter, diffuse SSR for zero AOD is represented by parameter c (the intercept of the equation). Based on the combination of the linear equation and the power law equation for diffuse SSR, the ADRE on diffuse SSR is then calculated by subtracting a by c. Fig. 6 presents ADRE efficiency for global (ADREglobal, left), direct (ADREdirect, middle) and diffuse SSR (ADREdiffuse, right) for the six aerosol types, respectively. One of the interesting feature is that ADRE varies dramatically with aerosol types and this difference increases as the SZA increases. ADRE (the absolute magnitude) of the four fine-mode dominated aerosol types increases as SSA decreases. The largest ADREglobal is produced by HA aerosols that varies from about 100 W m 2 for SZA of 781 to 220 W m 2 for SZA of 501. In order to make a consistent compasion between the ADRE of the six aerosol types, the mean ADRE efficiency values have been provided, if not otherwise specified, for SZA between 551 and651. The mean ADREglobal for SZA between 551 and 651 is about 181 W m 2, which is larger than that of WA (with the minimum ADRE efficiency) by 75 W m 2. ADRE2 , respectively, which is global by DU and PD is 141 and 150 W m comparable to that by MA. The results are consistent with
the results by Garcia et al. [19] based on the AERONET inversion products. It is interesting to note that ADREdirect of HA is in the middle of the six aerosol types. ADREdirect of HA is generally less than that of DU and PD but larger than that of the three other fine-mode dominated aerosol types. The largest ADREdirect is derived for DU aerosols. The mean ADREdirect of DU aerosols is 373 W m 2, which is larger than that of WA (with the minimum ADREdirect) by 100 W m 2. The extinction of DU shows much less wavelength dependence. Dust extinction in the infrared spectrum should be much larger than that of fine particles for the same AOD, which results in much more solar direct radiation is scattered by dust aerosols (larger ADREdirect). However, scattering solar radiation reaching the surface by dust aerosols should be larger than that of fine-mode dominated aerosol types because of their larger size, which is supported by the fact that the maximum increase rate of diffuse SSR is derived for the DU aerosols that is followed by PD aerosols. The mean ADREdiffuse by DU aerosols is 230 W m 2, which is nearly double larger than that of HA (with the minimum ADREdiffuse). ADREdiffuse for SZAr651 was estimabed to be about 220 W m 2 at the Southern Great Plains [13], which is slightly smaller than
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
Direct SSR Direct SSR
800
Direct SSR
800
Direct SSR
SZA=44.4
800
800
Y Y
SZA=53.2
=698.5e =715.5e
[15.7] [11.6]
Y Y
SZA=57.6
=556.4e =564.6e
[13.5] [10.6]
Y Y
77
SZA=60.6
=479.8e =492.6e
[13.8] [11.9]
Y Y
SZA=62.9
=432.0e =443.7e
[11.0] [9.1]
Y Y
=391.4e =404.9e
[10.4] [8.4]
600 400 200 SZA=64.6 Y Y
SZA=65.7
=355.9e =368.1e
[11.6] [9.1]
Y Y
SZA=67.0
=343.3e =352.2e
[9.7] [8.0]
Y Y
SZA=67.9
=317.7e =330.2e
[10.2] [7.8]
Y Y
SZA=68.9
=304.5e =314.2e
[9.4] [7.5]
Y Y
=285.7e =296.3e
[9.2] [7.4]
600 400 200 SZA=69.9 Y Y
SZA=70.8
=268.6e =277.4e
[9.4] [8.4]
Y Y
SZA=71.8
=252.6e =259.2e
[8.8] [7.6]
Y Y
SZA=72.7
=235.3e =243.9e
[7.9] [7.3]
Y Y
SZA=73.5
=220.6e =228.1e
[7.7] [6.2]
Y Y
=208.2e =215.3e
[7.5] [6.3]
600 400 200 SZA=74.0 Y Y
SZA=74.9
=196.7e =204.1e
[7.3] [5.6]
Y Y
SZA=75.6
=183.0e =188.4e
[7.4] [5.6]
Y Y
SZA=76.9
=173.4e =179.9e
[5.6] [4.7]
Y Y
SZA=78.2
=152.9e =158.6e
[5.3] [3.8]
Y Y
=135.7e =140.4e
[4.3] [3.3]
600 400 200 0
0.5 AOD
1
550nm
0
0.5 AOD
1
0
550nm
0.5 AOD
1
0
550nm
0.5 AOD
1
550nm
0
0.5 AOD
1
550nm
Fig. 4. Similar to Fig. 3 but for direct SSR.
that by dust aerosols but larger than that by the four finemode dominated aerosol types. 100 W m 2 more direct SSR is extincted by DU and PD than by MA, which is offset by a larger increase of diffuse SSR induced by DU and PD. Therefore, ADREglobal of DU and PD is comparable to that by MA. The scattering and absorbing of direct solar radiation by HA aerosols is less than that of dust aerosols, however, its strong absorption results in much less diffuse solar radiation reaching the surface. Therefore, it is not surprising that the maximum ADREglobal is derived for HA aerosols. ADREs derived from SBDART model simulations are in good agreement with those from measurements. The difference in ADRE efficiency derived from measuremens and SBDART simulations is generally not larger than 10 W m 2 and the mean difference is 7.0 W m 2 (73.8 W m 2). Fig. 7 presents seasonal mean of the daily integrations of the ADRE for each aerosol type. A distinct seasonal variation of ADRE is evident as a result of seasonal variation of solar radiation available for the scattering and absorbing by aerosols. DU and WA showed the maximum and minimum ADREdirect (in absolute value). ADREdirect of DU varied from 95 W m 2 in winter to 220 W m 2 in summer, the corresponding value of WA ranged from 80 W m 2 to 168 W m 2. As a result of very strong absorption, HA showed the minimum ADREdiffuse, being 30 W m 2 in winter and 82 W m 2 in
summer which was less than that of DU (with the maximum ADREdiffuse) by 21 W m 2 and 63 W m 2. The largest ADREglobal was observed for HA, being from 65 W m 2 in winter to 120 W m 2 in summer, which was about two times larger than that of WA (with the minimum ADREglobal) in seasons except winter. 5. Discussion It is widely suggested that a high proportion of diffuse SSR increases the radiation use efficiency (RUE) of photosynthesis since diffuse SSR is able to penerate vegetative layers. Therefore, much attention has been paid to the effect of diffuse SSR on crop growth and carbon cycling [26]. Significant increase of diffuse SSR induced by aerosols were reported [12,13], but theses studies did not consider potential different performance of aerosol types. Dramatical difference in aerosol effect on diffuse SSR between the six aerosol types was clearly revealed in this paper. More specific, long-range transportion of dust from the Gobi Deserts to the NCP is likely benefical for plant production since the increasing rate of diffuse SSR by dust scattering is nearly double larger than that of anthropogenic fine-mode dominated particles with strong absorption. Therefore, we should take aerosol types into consideration when potential aerosol effects on plant production and carbon cycling are studied.
Diffuse SSR
300
Diffuse SSR
300
Diffuse SSR
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
300
Diffuse SSR
78
300
200 100
200 100
200 100
200 100 0
0.5 AOD
1
550nm
0
0.5 AOD
1
0
550nm
0.5 AOD
1
0
550nm
0.5 AOD
1
550nm
0
0.5 AOD
1
550nm
Fig. 5. Similar to Fig. 3 but for diffuse SSR. A power law equation is used to describe the variation of diffuse SSR as a function of AOD.
Global SSR and direct SSR are of particular interest to concentrating solar thermal installations and installations that track the position of the sun [27]. Since aerosol effects on SSR differ dramatically between aerosol types, evaluation of global SSR as the main data input in the calculation of photovoltaic (PV) power production and direct SSR as the most important parameter for concentrated PV should also take aerosol types into consideration. Additionally, multiple junction and thin film PV technologies respond differently to varying spectral distributions of solar energy [28]. Much attention should also be paid to varying effects of different aerosol types on spectral solar energy that needs further study. Knowledge of long-term changes in aerosols is fundamental for a better understanding of climate change. However, little information on changes in aerosol concentration, especially prior to the 1980s, is availabe. It has been suggested to use long-term measurements of sunshine duration, especially at sunrise and sunset, to detect changes in aerosol loading [29]. The World Meteorological Organization uses the term “sunshine duration” to mean the cumulative time during which an area receives direct irradiance from the sun of at least 120 W m 2. Analysis of ADRE on direct SSR has revealed signficant difference in ADRE on direct SSR between the six aerosol types. Therefore, aerosol type should be considered when we attempt to derive aerosol loading from sunshine duration data.
6. Conclusions Using the AERONET aerosol products and the BSRN SSR data at Xianghe over the NCP, aerosol direct radiative effects of different aerosol types on global SSR and its components are critically assessed. Aerosols are firstly classified into six types based on aerosol size and absorption. Aerosol optical data and SSR measurements are then collocated for different aerosol types to derive empirical relationships between AOD and SSR. ADRE is finally derived from these equations for the six aerosol types. Dramatic variation of ADRE between aerosol types is clearly revealed. The major conclusions can be summarized as follows: A statistical method is developed to establish empirical relationships between SSR and AOD based on collocated measurements of aerosol optical properties and SSR data, which are then used to study ADRE effects on global SSR and its components. This method can be potentially applied to regions with the presence of both the AERONET and BSRN data. ADRE varies dramatically with aerosol types and this difference increases as the SZA increases. HA aerosols had the largest global ADRE efficiency, being 181 W m 2 for SZA between 551 and 651, which is larger than the minimum ADRE effciency of WA aerosol by 75 W m 2. DU aerosols showed the largest direct ADRE efficiency,
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
−50
−100
Global
Direct
ADRE (W m−2)
−100 −300 −400 −500 −600 x: observation +: simulation 0.4
0.6
0.8
−700 0.2
Diffuse
150
−200
−250 0.2
400 350 DU PD 300 HA MA 250 SA 200 WA
−200
−150
79
0.4
100 0.6
0.8 The cosine of the solar zenith angle
50 0.2
0.4
0.6
0.8
Fig. 6. The scatter-plot between ADRE efficiency and the cosine of the solar zenith angle for global (left), direct (middle) and diffuse (right) SSR and for six aerosol types. Values derived from measurements and model simulations are represented by “plus” and “cross”, respectively.
Fig. 7. Seasonal mean of the daily integrations of the ADRE on global, direct and diffuse irradiance for six aerosol types derived from measurements.
being 373 W m 2 for SZA between 551 and 651, which is larger than that of WA aerosols (with the minimum ADRE efficiency) by 100 W m 2. The mean ADRE on diffuse SSR by DU aerosols for SZA between 551 and 651 is 230 W m 2, which is nearly double larger than that of HA (with the minimum ADRE on diffuse SSR). The scattering and absorbing of direct solar radiation by HA aerosols is less than that of dust aerosols, however, its strong absorption results in much less diffuse solar radiation reaching the
surface. Therefore, the maximum global ADRE efficiency is derived for HA aerosols. ADRE efficiency on global SSR by DU and PD is 141 and 150 W m 2, respectively, which is comparable to that by MA. However, a reduction of dirct SSR induced by DU and PD is larger than that by MA 100 W m 2, which is offset by a larger increase of diffuse SSR induced by DU and PD. The results are derived from aerosol and surface irradiance measurements at Xianghe, a suburban station in
80
X. Xia / Journal of Quantitative Spectroscopy & Radiative Transfer 149 (2014) 72–80
the North China Plain. The method, however, is potentially useful for researches at any stations with similar measurements. Dramatic difference in ADRE on global SSR and its components between aerosol types revealed in this paper is of particular interest to aerosol potential effects on plant production and carbon cycling, to the application of solar energy and to detecting long-term changes in aerosols from sunshine duration measurements.
[12]
[13]
[14]
Acknowledgments The AERONET and BSRN data used in the paper are publicly available from http://aeronet.gsfc.nasa.gov/ and http://www.bsrn.awi.de/, respectively. This research was supported by the National Basic Research Program of China (2010CB950804), the National Natural Science Foundation of China (4085011), and the Strategic Priority Research Program of the Chinese Academy of Sciences, China (XDA050100301). References [1] Wild M. Global dimming and brightening: a review. J Geophys Res 2009;114:D00D16. http://dx.doi.org/10.1029/2008JD011470. [2] Dubovik O, Holben B, Eck T, Smirnov A, Kaufman Y, King M, et al. Variability of absorption and optical properties of key aerosol types observed in worldwide locations. J Atmos Sci 2002;59:590–608. [3] Luo Y, Lu D, Zhou X, Li W, He Q. Characteristics of the spatial distribution and yearly variation of aerosol optical depth over China in last 30 years. J Geophys Res 2001;106:14501–13. [4] Qiu J, Yang L. Variation characteristics of atmospheric aerosol optical depths and visibility in North China during 1980–1994. Atmos Environ 2000;34:603–9. [5] Huang J, Minnis P, Chen B, Huang Z, Liu Z, Zhao Q, et al. Long-range transport and vertical structure of Asian dust from CALIPSO and surface measurements during PACDEX. J Geophys Res 2008;113: D23212. http://dx.doi.org/10.1029/2008JD010620. [6] Eck T, Holben B, Sinyuk A, Pinker R, Goloub P, Chen H, et al. Climatological aspects of the optical properties of fine/coarse mode aerosol mixtures. J Geophys Res 2010;115:D19205. http://dx.doi.org/10.1029/ 2010JD014002. [7] Russell P, Bergstrom R, Shinozuka Y, Clarke A, DeCarlo P, Jimenez J, et al. Absorption Angstrom Exponent in AERONET and related data as an indicator of aerosol composition. Atmos Chem Phys 2010;10: 1155–69. http://dx.doi.org/10.5194/acp-10-1155-2010. [8] Li Z, Li C, Chen H, Tsay S, Holben B, Huang J, et al. East Asian Studies of Tropospheric Aerosols and their Impact on Regional Climate (EAST‐AIRC): an overview. J Geophys Res 2011;116:1–15, http://dx.doi.org/10.1029/2010JD015257. [9] Xia X, Chen H, Goloub P, Zong X, Zhang W, Wang P. Climatological aspects of aerosol optical properties in North China Plain based on ground and satellite remote-sensing data. J Quant Spectrosc Radiat Transf 2013;127:12–23. [10] Holben B, Eck T, Slutsker I, Tanre D, Buis J, Setzer A, et al. AERONET – a federated instrument network and data archive for aerosol characterization. Remote Sens Environ 1998;66:1–16. [11] Che H, Xia X, Zhu J, Li Z, Dubovik O, Holben B, et al. Column aerosol optical properties and aerosol radiative forcing during a serious
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
haze-fog month over North China Plain in 2013 based on groundbased sunphotometer measurements. Atmos Chem Phys 2014;14: 2125–38. http://dx.doi.org/10.5194/acp-14-2125-2014. Xia X, Li Z, Wang P, Chen H, Cribb M. Estimation of aerosol effects on surface irradiance based on measurements and radiative transfer model simulations in northern China. J Geophys Res 2007;112: 1–11. http://dx.doi.org/10.1029/2006JD008337. Creekmore T, Joseph E, Long C, Li S. Quantifying aerosol direct effects from broadband irradiance and spectral aerosol optical depth observations. J Geophys Res 2014;119:1–15, http://dx.doi.org/10.1002/2013JD021217. Li Z, Xia X, Cribb M, Mi W, Holben B, Wang P, et al. Aerosol optical properties and their radiative effects in northern China. J Geophys Res 2006;112:1–11. http://dx.doi.org/10.1029/2006JD007382. Eck T, Holben B, Reid J, Dubovik O, Smirnov A, O'Neill N, et al. Wavelength dependence of the optical depth of biomass burning, urban, and desert dust aerosol. J Geophys Res 1999;104:31333–49. Dubovik O, King M. A flexible inversion algorithm for retrieval of aerosol optical properties from Sun and sky radiance measurements. J Geophys Res 2000;105:20673–96. Dubovik O, Sinyuk A, Lapyonok T, Holben B, Mishchenko M, Yang P, et al. The application of spheroid models to account for aerosol particle nonsphericity in remote sensing of desert dust. J Geophys Res 2006;111:D11208. http://dx.doi.org/10.1029/2005JD006619. Mishchenko M, Travis D, Mackowski D. T-matrix computations of light scattering by nonspherical particles: a review. J Quant Spectrosc Radiat Transfer 1996;55:535–75. Garcia O, Diaz J, Exposito F, Diaz A, Dubovik O, Derimian Y, et al. Shortwave radiative forcing and efficiency of key aerosol types using AERONET data. Atmos Chem Phys 2012;12:5129–45, http://dx.doi.org/10.5194/acp-12-5129-2012. Long C, Ackerman T. Identification of clear skies from broadband pyranometer measurements and calculation of downwelling shortwave cloud effects. J Geophys Res 2000;105:15609–26. Ricchiazzi P, Yang S, Gautier C, Sowle D. SBDART: a research and teaching software tool for plane-parallel radiative transfer in the Earth's atmosphere. Bull Am Meteorol Soc 1998;79:2101–14. Li Z, Lee K, Wang Y, Xin X, Hao W. First observation‐based estimates of cloud‐free aerosol radiative forcing across China. J Geophys Res 2010;115:D00K18. http://dx.doi.org/10.1029/2009JD013306. Michalsky J, Dolce R, Dutton E, Haeffelin M, Major G, Schlemmer J, et al. Results from the first ARM diffuse horizontal shortwave irradiance comparison. J Geophys Res 2003;108:4108, http://dx.doi.org/10.1029/2002JD002825. Schafer C, Gao F, Strahler A, Lucht W, Li W, Tsang T, et al. First operational BRDF, albedo nadir reflectance products from MODIS. Remote Sens Environ 2002;83:135–48. Gilles D, Holben B, Eck T, Sinyuk A, Smirnov A, Slutsker I, et al. An analysis of AERONET aerosol absorption properties and classifications representative of aerosol source regions. J Geophys Res 2012;117:D17203. http://dx.doi.org/10.1029/2012JD018127. Mercado L, Bellouin N, Sitch S, Boucher O, Huntingford C, Wild M, et al. Impact of changes in diffuse radiation on the global land carbon sink. Nature 2009;458:1014–7. http://dx.doi.org/10.1039/nature07949. Suri M, Thomas H, Ewan D, Heinz O. Potential of solar electricity generation in the European Union member states and candidate countries. Sol Energy 2007;81:1295–305. Krishnan P, Schüttauf J, van der Werf C, Hassanzadeh B, van Sark W, Schropp R. Response to simulated typical daily outdoor irradiation conditions of thin film silicon based triple band gap, triple junction solar cells. Sol Energy Mater Sol Cells 2009;93:691–7. Sanchez-Romero A, Sanchez-Lorenzo A, Calbo J, Gonzalez J, ArorinMolina C. The signal of aerosol induced changes in sunshine duration records: a review of the evidence. J Geophys Res 2014;119: 1–17. http://dx.doi.org/10.1002/2013JD021393.