Chinese Science Bulletin © 2007

SCIENCE IN CHINA PRESS

Springer

Validation of MODIS aerosol products by CSHNET over China WANG LiLi1, XIN JinYuan1, WANG YueSi1†, LI ZhanQing2, WANG PuCai1, LIU GuangRen1 & WEN TianXue1 1 Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China; 2 Department of Meteorology, University of Maryland, College Park, MD 20782, USA

The Chinese Sun Hazemeter Network (CSHNET) provides the necessary ground-based observation to validate and assess the applicability of MODIS aerosol optical depth (AOD) products over different ecological and geographic regions in China for the first time. The validation results show that the comprehensive utilization ratio and applicability of MODIS products varied very much over different regions and seasons from August 2004 to July 2005. On the Tibetan Plateau, the comprehensive utilization ratio of MODIS data was low: MODIS products only accounted for 16% of the ground-based observation; on average, 31% to 45% of MODIS products fell within the retrieval errors issued by NASA. A similar result was found in northern desert areas with the ratio of MODIS to observation ranging from 15% to 55%, with 7% to 39% of MODIS products within errors. In the remote northeast corner of China, low ratios of MODIS to observation were also found ranging from 14% to 46%, with 49% to 69% of MODIS within errors. The forested sites exhibited moderate ratios of MODIS to observation ranging from 46% to 65%, with 30% to 59% of MODIS within errors. This was similar to numbers observed at sites along eastern seashore of China and inland urban sites with the ratio of MODIS to observation between 63% to 75%, with 25% to 67% of MODIS within errors for sites along eastern seashore of China and 43% to 78%, with 35% to 75% of MODIS within errors for inland urban sites. The ratio of MODIS to observation over agricultural areas ranged from 61% to 89%; 59%－88% of MODIS fell within the retrieval errors. At homogeneous and well vegetated areas, the comprehensive utilization ratio of MODIS products was over 80% and above 70% of MODIS products fell within the retrieval errors in growing season. Chinese Sun Hazemeter Network (CSHNET), aerosol optical depth (AOD), MODIS

The ambient aerosols effect on climatic change has become an important area of atmospheric science research[1,2]. Tropospheric aerosol has a substantial climatic influence through direct and indirect radiactive ― forcing and taking part in cloud processes[3 6], and aerosol also plays a most important role in environmental change[7,8]. However, climatic and environment influence of aerosol is still uncertain due to the incomplete knowledge of the microphysical and optical properties of aerosols and their extreme heterogeneous spatial distribution[9,10]. The ground-based network observation and satellite remote sensing have become the most important basic research methods in the aerosol www.scichina.com

www.springerlink.com

field[11 13], so that they can provide basic data for further research on climatic change and environmental study[14,15]. Due to the limitation of observation condition, equipment, and capital, ground-based network observation can only monitor in a restricted spatial and temporal region. Although ground-based observations can provide aerosol features with high temporal resolution in a particular region where the station is located, it ―

Received August 7, 2006; accepted December 11, 2006 doi: 10.1007/s11434-007-0222-0 † Corresponding author (email: [email protected]) Supported by the National Natural Science Foundation of China (Grants Nos. 40675073, 40525016 and 40520120071) and the National Basic Research Program (973) of China (Grant No. 2007CB407303)

Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718

ARTICLES

modify and improve the satellite retrieval algorithm[21,35,36]. This article validates and assesses the applicability of MODIS AOD products by analyzing them in comparison with the Chinese Sun Hazemeter Network (CSHNET) ground-based observation (including 21 stations) over different ecological and geographic regions in China for the first time, which provides scientific gist for applicability and further research of MODIS over China.

1 Methodology and data 1.1 The Chinese Sun Hazemeter Network (CSHNET) The Chinese Sun Hazemeter Network (CSHNET) was initiated in August, 2004. This network includes 19 CERN (the Chinese Ecosystem Research Network) stations representing some typical ecosystems, four urban sites, two instrument calibration centers, and one data collection/processing center. Figure 1 shows the locations of the sites in the CSHNET. The network uniformly uses the LED (Light-emitting Diode) hazemeter also called sun hazemeter, which have been widely used in the GLOBE program[37] and generally recognized by international scientists[38,39]. The measurement period is from 10 am to 2 pm (local time), encompassing MODIS satellite overpassing time. Measurements are taken three times every half an hour and at least 15―20 times a day, and weather condition and cloud amount (not observing if cloud amount is more than four fifths) are synchronously recorded during the measurement period. The LED hazemeters are uniformly calibrated by the Langley plot calibration and compared with CEMIL annually. The details of CSHNET about running mode, calibration approaches, and running conditions can be found in refs. [40―42]. 1.2 Introduction of MODIS aerosol products MODIS onboard the EOS-Terra satellite launched on December 18, 1999 has 36 spectral bands (0.41―14 μm) at three different spatial resolutions (250 m, 500 m and 1 km) with a scanning width of 2330 km[43], which provides an available method for land aerosol remote sensing; high spatial resolution and near daily global coverage data make observation on atmospheric components more convenient[44]. MODIS Collection 4 Level 2 (MOD04_L2) (10 km×10 km) granule-based (granule: a 5-minute segment of one MODIS orbital datum) aero-

WANG LiLi et al. Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718

1709

ATMOSPHERIC SCIENCES

cannot acquire good spatial features, which goes against research and analysis about aerosol on a global scale[15,16]. Satellite remote sensing can make up for the shortage of the ground-based network observation, and its predominance is more obvious, especially in the widest ocean and remote regions with atrocious environment[17]. However, satellite remote sensing and retrieval of the aerosol optical properties are influenced by many external factors, such as surface reflectance, aerosol models, and sub-pixel clouds[18,19], so that satellite remote sensing must be evaluated with ground-based measurements to increase the accuracy of the retrieval[20,21]. MODIS (the Moderate Resolution Imaging Spectrometer) aerosol products provided by NASA (National Aeronautics and Space Administration) firstly provide global aerosol optical properties with high spatial resolution. Recently, these products have achieved a certain precision, which can be applied to climatic and environmental research[22]. Aerosol Robotic Network (AERONET)[11] stations founded by NASA are operational worldwide, which provide multi-spectral channels validation data for MODIS-derived AOD to complete synthetical measurements on a global scale[21]. In the last few years, Chinese scientists have gradually used geostationary and polar-orbiting satellites to derive aerosol ― optical properties over land[23 25], especially, application of MODIS measurements has gotten primary achievements and retrieved AOD over certain regions in ― China[26 29]. In addition, Chinese scientists have used MODIS aerosol products by validating with groundbased observation over certain regions, such as Beijing, Hong Kong and the Yangtze River delta, to analyze cli― matic change and air quality[30 34]. However, most of the researches lack validation of large-scale and long-term ground-based observation data. Although the Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia) field experiment added some ground-based equipments to Asia[18], there has been so few equipments that the application of MODIS products is more uncertain over China[18,21], thus the satellite retrieval aerosol optical properties need further research. Due to the significant diversity of vegetation distribution and aerosol types over land in China, setting up ground-based network sun-photometers is very important to evaluate MODIS aerosol products according to different ecological and geographic regions[30,33] to

Figure 1 Geographical locations of observation sites in the CERN sun hazemeter network.

sol products provide aerosol optical depths (τa) over land at 0.47 and 0.66 µm wavelengths. The MODIS retrieval of τa over land employs primarily three spectral channels centered at 0.47, 0.66 and 2.1 µm wavelength[45]. First, cloud-free pixels are selected using the multi-spectral MODIS cloud mask, which uses more than twenty tests including two cirrus detection tests to make up a clear pixel at 1 km×1 km resolution according to the clear pixels at 250 m and 500 m resolutions[46]. Then, to estimate the surface reflectance (ρ ) at 0.47 µm and at 0.66 μm, the empirical relationships (ρ0.47 μm /ρ2.1 μm= 0.25, ρ0.66 μm/ρ2.1 μm=0.5,)[47] developed over vegetated surfaces are used. In addition, to minimize the error, ρ2.1 μm are limited to less than 0.25 and only the 30 percentile of MODIS measured radiance is averaged as the representative value in a 10 km×10 km grid box[20]. To choose aerosol models, Kaufman et al.[45] first presume the continental aerosol model to derive, then the ratio of aerosol path radiance at 0.47 and 0.66 μm is calculated to modify the aerosol model and τa; Non-dust aerosols are separated a priori according to the geographic locations and seasons of their emission sources[21]. Finally, τa is interpolated at 550 nm wavelength using τa at 0.47 and 0.66 μm[48]. 1710

2 Data processing This article uses the MODIS Collection 4 Level 2 aerosol products (short for MODIS aerosol products) and CSHNET-derived AOD data from August, 2004 to July, 2005. CSHNET-derived AOD data, which are cloud screened by manual record table and are chosen under such weather condition with free cloud or cloud amount less than one-half (in order to assure no cloud within a field of sun/view angle of 30℃)[42], are interpolated at 550 nm wavelength by τa at 0.65 and 0.5 µm, which are ― correspondent well with CIMEL-derived data[40 42]. MODIS has almost complete global coverage every 1 or 2 days, thus we can regard MODIS as a set of spatially variational and temporally fixed distribution data in the region centered at the station; however, CSHNET data are acquired at 30-min intervals at the fixed point, which can be regarded as a set of spatially fixed and temporally variational distribution data. To make the two sets of data more comparative and representative in time and space, MODIS data are averaged over 50 km area centered at the ground stations (including at least 5 pixels), while CSHNET data are averaged over one-hour interval

WANG LiLi et al. Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718

3 Results and discussion Statistics parameters of the CSHNET assessing MODIS AOD products are delineated in Table 1. Table 2 shows seasonal statistics parameters at different sites in the CSHNET. We will make discussion separately according to the sites located in different ecosystems and geographic regions as follows.

ARTICLES

centered at the overpassing time of the TERRA satellite[21,48]. We did some correlation statistics analyses between the two sets of data, and validated MODIS products over different stations with the retrieval errors of Δτa=±0.05±0.2 τa[21] (τa is aerosol optical depth) over land issued by NASA in order to assess MODIS AOD products by CSHNET.

Table 1 Statistics parameters of the CSHNET assessing MODIS aerosol optical depth products

Table 2

Percentages of the retrieval errors calculated by NASA at different sites in the CSHNET for four seasons Spring Summer Autumn Winter Site n r1 (%) r4 (%) n r1 (%) r4 (%) n r1 (%) r4 (%) n r1 (%) Lhasa 9 20 11 11 48 73 9 14 44 NAN ― Haibei 1 6 0 8 47 38 3 18 33 1 4 Shapotou 1 2 0 1 4 0 23 42 13 NAN ― Ordos 12 19 25 9 39 56 14 39 43 1 4 Fukang 18 42 6 32 78 3 30 71 13 NAN ― Hailun 6 17 67 3 25 67 7 33 71 NAN ― Sanjiang 9 53 22 15 79 33 16 64 75 1 3 Changbai Mountain 8 35 38 8 40 0 30 68 47 12 32 Beijing Forest 34 67 74 22 67 23 31 56 74 10 19 Xishuangbanna 31 89 48 1 10 0 NAN 21 95 ― ― Taihu Lake 14 78 43 4 57 50 18 67 78 9 64 Jiaozhou Bay 29 64 52 8 62 88 31 72 48 24 52 Shanghai 33 85 18 11 79 18 12 86 50 7 44 Lanzhou 12 32 67 19 90 42 21 68 47 NAN ― Beijing 17 63 24 15 83 40 24 80 42 8 32 Xianghe 24 96 75 13 100 77 DIV 3 23 ― ― Shenyang 33 73 45 12 67 58 44 81 68 4 7 Ansai 4 80 75 13 57 54 DIV DIV ― ― ― Fengqiu 23 85 65 9 100 56 22 96 68 9 75 Yanting 17 85 76 13 76 77 8 80 75 8 80 Taoyuan 20 95 85 8 57 75 15 65 100 15 100 NAN = no MODIS data, DIV = no ground-based observation data, r1 = ratio of MODIS to observation, r4 = percentage within errors.

WANG LiLi et al. Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718

r4 (%) ― 0 ― 0 ― ― 100 17 40 5 89 42 29 ― 25 67 25 ― 56 75 87

1711

ATMOSPHERIC SCIENCES

r2 (%) r3 (%) r4 (%) Site Ecosystem n r1(%) Lhasa plateau agriculture 29 16 0 55 45 Haibei plateau grassland 13 16 31 38 31 Shapotou desert 25 15 0 88 12 Ordos desert 36 24 6 55 39 Fukang desert 80 55 0 93 7 Hailun agriculture 16 14 25 6 69 Sanjiang marsh 41 42 15 36 49 Changbai Mountain temperate Forest 58 46 38 29 33 Beijing Forest warm Temperate Forest 97 50 31 10 59 Xishuangbanna tropic Forest 53 65 68 2 30 Taihu Lake lake 45 68 0 33 67 Jiaozhou Bay marginal sea 92 63 3 46 51 Shanghai urban along the sea 63 75 2 73 25 Lanzhou urban in the north 52 43 29 21 50 Beijing urban in the east 64 64 6 59 35 Xianghe suburban 40 78 15 10 75 Shenyang suburban agriculture 93 53 25 18 57 Ansai agriculture 17 61 41 0 59 Fengqiu agriculture 63 89 2 35 63 Yanting agriculture 46 81 20 4 76 Taoyuan agriculture 58 79 2 10 88 n = number of MODIS products during ground-based observing period, r1 = ratio of MODIS to observation, r2 = percentage of τaMODIS< τaCSHNET outside the NASA evaluation errors, r3=percentage of τaMODIS>τaCSHNET outside errors, r4 = percentage within errors.

Figure 2 shows the comparison of MODIS- and CSHNET-derived AOD at 550 nm wavelength at sites on the Qinghai-Tibet Plateau. Due to high altitude, less rain, less vegetation, and long-term snow coverage, especially in mountainous regions, pixel points satisfying the “dark background” condition and matching the ground-based measurements are spare (no more than 5 pixels) in the 50 km×50 km area. Thus, we compare the origin MODIS data at the two sites. However, ratios of MODIS to observation are both only 16%, and correlation coefficients (R) both never exceed the 95% confidence level at the two sites. τa derived by CSHNET and MODIS is both very low, and τaMODIS is distinctly larger than τaCSHNET in general. MODIS within the error range are primarily in summer and autumn (33%≤r4≤73%); in spring and winter, MODIS are few because of sparse vegetation and large-scale snow coverage. Due to low AOD which is less than 0.15 yearly on average[42] on the Qinghai-Tibet Plateau, surface reflectance error contributes to large errors in MODIS retrieval comparing with ground-based observation, so that satellite retrieval results do not have representation and reliability and ground-based observation is very necessary in this region. Figure 3 is the same comparison as Figure 2 at northern desert stations. MODIS products account for 15%, 24% and 55% of ground-based observation, and only 12%, 39% and 7% of MODIS are within the error range at Shapotou, Ordos, and Fukang stations, respectively. MODIS-derived results have large biases comparing with CSHNET, with the root mean square (RMS) errors >0.2 in this region. MODIS retrieval has larger system-

atic biases with the slope of linear regression (Sl) > 1.17[21], and τaMODIS is larger than τaCSHNET in general. As shown in the figure, the spatial standard deviation of MODIS is larger and the temporal standard deviation of CSHNET is smaller, which illuminates the fact that MODIS retrievals have larger errors for different pixel points in the same region due to sparse vegetation, although aerosol optical characteristics are stable in the desert region[41,42]. Shapotou, lying in an arid part of the Tengger Desert, has few comprehensive utilization data all the year round, except a few data in autumn (r1=13%); Ordos in a sandy grassland ecosystem within a semi-arid region and Fukang in the desert-oasis ecosystem both have higher comprehensive utilization ratios (39%≤r1 ≤78%) in summer and autumn, whereas fewer ratios in winter; and the percentages within errors are smaller with r4≤56% at these three sites in four seasons. In conclusion, due to few pixels satisfying the “dark background” condition in the desert, MODIS products have poor application and no regional representation, which need to modify the retrieval algorithm to acquire more accurate results. Figure 4 shows the same results as those at sites in Northeast China. MODIS products account for 14%, 42% and 46% of ground-based observation, and 69%, 49% and 33% of MODIS products are within the error range at Hailun, Sanjiang and Changbai Mountains stations, respectively. MODIS retrievals have larger systematic biases with Sl<0.64 in all of three sites and R does not exceed the 95% confidence level in Hailun and Changbai Mountains. The three sites are situated in the far northeast corner of China and are exposed to clean

Figure 2 Comparisons of the CSHNET- and MODIS-derived AOD at 550 nm wavelength at sites on the Qinghai-Tibet Plateau. The dash lines represent y=0.05+1.2 x and the dotted lines y=−0.05+0.8 x, which is the retrieval errors lines estimated by NASA. The solid lines represent the linear regression fits. Temporal and spatial standard deviations are shown as the error bars in x (CSHNET)- and y(MODIS)-direction respectively. 1712

WANG LiLi et al. Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718

ARTICLES Comparisons of the CSHNET- and MODIS-derived AOD at 550 nm wavelength at northern desert stations.

ATMOSPHERIC SCIENCES

Figure 3

Figure 4 Comparisons of the CSHNET- and MODIS-derived AOD at 550 nm wavelength at sites in Northeast China. WANG LiLi et al. Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718

1713

air with an annual mean AOD < 0.2[42], in addition, the temporal standard deviation of CSHNET is small, whereas spatial standard deviation of MODIS is large; especially at Hailun Station, due to very low AOD, the deviation of estimating surface reflectance may bring errors to satellite retrieval, which is concluded by Zhu et al.[31] at a test of sensitivity to the satellite apparent reflectance and AOD. As far as seasons are concerned, in winter, there are few MODIS’s in Hailun and Sanjiang due to long-term ice/snow coverage in the northeast of China; though Changbai Mountain is located in a forested area, the ratio of MODIS to observation is only 32%, along with 17% of MODIS within errors, which suggests that MODIS products do not have obvious applicability in winter in this region. In other seasons, the results of comparisons are satisfactory on Northeast Plain (Hailun and Sanjiang), and particularly in autumn, more than 70% of MODIS products are within errors. Therefore, satellite retrieval AOD algorithm can be corrected better because surface reflectance and aerosol models in general vary regularly[41,42]. Figure 5 is the results at forest stations. In Beijing Forest located in mid-latitude area, the comprehensive utilization ratio of MODIS products is over 50% and linear correlation is significant with R >0.8 exceeding the 95% confidence level in terms of dense vegetation and more pixels satisfying the “dark background” conditions. With the latitude decreasing, τaMODIS is obviously less than τaCSHNET and the percentage within errors is also dropping, Beijing Forest has 59% MODIS products, whereas Xishuangbanna 30%. Although MODIS products achieve a certain precision at Beijing Forest region, satellite retrieval algorithm can be better optimized due

Figure 5 1714

to regular variation of aerosol models[41,42]. Whereas MODIS products have no region representative (r1<10%) in summer and autumn at Xishuangbanna station which is tropical rainforest, due to high humidity, more cloud and rain, and uncertainty of organic aerosol elimination[41,42], thus the algorithm needs to be improved. The comparisons near large water bodies (Taihu Lake, Shanghai City and Jiaozhou Bay) are depicted in Figure 6. At the three sites, quantities of satellite retrieval data are almost equal to those of CSHNET with the ratio of MODIS to observation > 60% and R all exceeding the 95% confidence level. Due to complexity of aerosol models and inhomogeneity of surface reflectance in Shanghai near the eastern seashore of China, this regional satellite retrieval error with only 25% of MODIS satisfying the errors is distinctly larger than that in non-urban places near the seashore. Jiaozhou Bay is also near the seashore, however, seawater is full of sand, which primarily results in less evaluation of surface reflectance and increases the satellite retrieval error[21] with 50% of MODIS within the error range. The water pollution in Taihu Lake is the primary factor added to the satellite retrieval error, which is less than that of regions near the sea compared with 67% of MODIS within errors in Taihu Lake. As the figures show, τaMODIS is larger than τaCSHNET, and spatial errors are also larger near the sea regions like Shanghai and Jiaozhou Bay stations. Complexity of aerosol models[42] increases the satellite retrieval systematic biases in this region. In addition, due to sandy seawater, water contamination, and less evaluation of surface reflectance, the retrieval algorithm of MODIS in this region can be modified in the future to improve the retrieval precision.

Comparisons of the CSHNET- and MODIS-derived AOD at 550 nm wavelength at forest stations.

WANG LiLi et al. Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718

ARTICLES

The results at a few major urban and suburban stations are shown in Figure 7. The ratio of MODIS to observation at Lanzhou is 43%. MODIS products do not match well with ground-based observation with R not exceeding the 95% confidence level, in addition, spatial and temporal deviations are both large; only 50% of MODIS products within the error range and few MODIS products can be utilized in winter. The reason for the large satellite retrieval errors is that Lanzhou is located in a valley basin with mountains around it and the city is full of high concentration dust particles[42] and a contamination layer above the city (especially in winter)[49,50] through out the year. Except the Lanzhou City site located in western China, the ratios of MODIS to observation in other three urban sites are over 50%, especially over 70% in suburban sites, and linear correlation is significant with R both over 0.7 exceeding the 99% confidence level; however, number of MODIS having larger retrieval errors is distinctly less than that of CSHNET in winter. The comparison between the Beijing urban site and the Xianghe suburban site about 70 km from the Beijing indicates an overestimation for Bei-

jing site of AOD by MODIS with respect to CSHNET with only 35% of MODIS within the error range, whereas applicability of MODIS products at Xianghe site is much better with 75% within errors, which indicates better region representative. MODIS products have larger errors at Beijing site in terms of complicated surface condition resulting in heterogeneous surface reflectance[31,51] and aerosol models mixed dust and smoke aerosols[41,42]. Therefore, it is necessary to use ground-based observation data to modify the satellite retrievals in urban region due to complexity of surface characteristic and aerosol models. The results obtained at four agricultural stations are shown in Figure 8. The stations except for Ansai Station have mainly farmland around them, thus homogeneous and vegetated surfaces give the best satellite retrieval conditions with the ratio of MODIS to observation over 80% obviously higher than those of other regions; linear correlation is both significant with R-0.8 exceeding the 99% confidence level, and RMS-0.15―0.25. Ansai Station in Chinese Loess Plateau has relatively less vegetation, which results in a systematic underestimation of

WANG LiLi et al. Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718

1715

ATMOSPHERIC SCIENCES

Figure 6 Comparisons of the CSHNET- and MODIS-derived AOD at 550 nm wavelength at sites along the eastern seashore of China.

Figure 7 Comparisons of the CSHNET- and MODIS-derived AOD at 550 nm wavelength at urban and suburban stations.

Figure 8 Comparisons of the CSHNET- and MODIS-derived AOD at 550 nm wavelength at agricultural stations.

1716

WANG LiLi et al. Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718

4 Conclusions MODIS aerosol products have a lack of applicability in general over China, due to extensive terrain, complexity of surface conditions, diverse ecological types, great difference of surface reflectance and aerosol models in different regions in China. MODIS products have better representation at agricultural regions, then a certain representative is at warm temperate forest, bay and lake. 1

Houghton J T, Filho M L G, Bruce J P, et al, eds. IPCC, Climate

The authors are grateful to NASA/GSFC and the CERN stations of the Chinese Academy of Sciences for contribution to this research.

13

Change 1994: Radiative Forcing of Climate Change. Cambridge:

King M D, Kaufman Y J, Tannre D, et al. Remote sensing of tropospheric Aerosols from Space: Past, Present and Future. Bull Amer

Cambridge University Press, 1995. 127―162

Meteorol Soc, 1999, 80(11): 2229―2259

2

Anderson T L, Charlson R J, Schwartz S E, et al. Climate forcing by aerosols—A hazy picture. Science, 2003, 300: 1103―1104

strategies for quantifying the relationships between Asian aerosols

3

Satheesh S K, Moorthy K K. Radiative effects of natural aerosols: A

and their climatic impacts. J Geophys Res, 2003, 108(D23):

14

review. Atm Environ, 2005, 39(11), 2089―2110 4

Houghton J T, Ding Y, Griggs D J, et al, eds. IPCC, Climate Change

Breon F-M, Tanre D, Generoso S. Aerosol effect on cloud droplet size

6

Boucher O, Anderson T L. General circulation model assessment of

Geophys Res, 2001, 106: 12067―12097 16

ties, 1992―1994. J Geophys Res, 1996, 101(D14): 19465―19482 17

the sensitivity of direct climate forcing by anthro-pogenic sulfate

Science Foundation of China (in Chinese). 2001 18

Hemisphere during ACE-Asia: results from global model, satellite

tributions of ambient aerosols must be studied in order to determine

observations, and Sun photometer measurements. J Geophy Res, 2004,

Chu D A, Kaufman Y J, Zibordi G, et al. Global monitoring of air

109(D23): D23S90-1-15 19

pollution over land from the Earth Observing System-Terra Moderate 2003, 108(D21): ACH4-1-18 9

Charlson R J, Schwartz S E, Hales J M, et al. Climate forcing by an-

10

Haywood J M, Boucher O. Estimates of the direct and indirect radia-

and properties? Geophys Res Lett, 2000, 27(23): 3861―3864 20

Remer L A, Kaufman Y J, Tanre D, et al. The MODIS aerosol algorithm, products, and validation. J ATMOS SCI, 2005, 62(4):

21

Chu D A, Kaufman Y J, Ichoku C, et al. Validation of MODIS aerosol

947―973

thropogenic aerosol. Science, 1992, 255: 423―430

optical depth retrieval over land. Geophys Res Lett, 2002, 29(12):

tive forcing due to tropospheric aerosols: a review. Rev Geophys,

1617―1621

2000, 38: 513―543 Holben B N, Eck T F, Slutsker I, et al. AERONET―A federated in-

22

Sens Environ, 1998, 66: 1―16

Ichoku C, Kaufman Y J, Remer L A, et al. Global aerosol remote sensing from MODIS. Adv Space Res, 2004, 34: 820―827

strument network and data archive for aerosol characterization. Rem 12

Kaufman Y J, Holben B N. Will aerosol measurements from Terra and Aqua polar orbiting satellites represent the daily aerosol abundance

Resolution Imaging Spectroradiometer (MODIS). J Geophy Res,

11

Chin M, Chu A, Levy R, et al. Aerosol distribution in the Northern

Lee R E Jr. The size of suspended particulate matter in air: size distheir effects on the environment. Science, 1972, 178: 567―575

8

Mao J T. Research of radiative characteristics of aerosols in China. The Conclusive Report of the Key Subject of the National Natural

aerosols to aerosol size and chemistry. J Geophy Res, 1995, 100(D12): 7

Holben B N, Setzer A, Eck T F, et al. Effect of dry-season biomass burning on Amazon basin aerosol concentrations and optical proper-

monitored from satellite. Science, 2002, 295: 834―838

26117―26134

Holben B N, Tanre D, Smirnov A, et al. An emerging ground-based aerosol climatology: aerosol optical depth from AERONET. J

Third Assessment Report of the Intergovernmental Panel on Climate 5

Huebert B J, Bates T, Russell P B, et al. An overview of ACE-Asia:

ACE1-1-20 15

2001: The Scientific Basis-contribution of Working Group I to the Change. New York: Cambridge University Press, 2001

ARTICLES

However, the larger errors and weak representative and applicability of MODIS AOD exist on the Qinghai-Tibet Plateau, in northern desert, urban and suburban areas, and the remote northeast corner of China and tropic rainforest. As far as MODIS products errors are concerned, those in southern areas are less than those of northern areas, east is less than west, and growing season is less than non-growing season. Currently, the retrieval algorithm of MODIS aerosol products needs to be modified in terms of large errors. Thus, it is necessary to systematically correct satellite retrieval process using the ground-based observation data and develop new algorithms according to different surface conditions before MODIS products can be applied widely.

23

Zhao B L, Yu X D. The remote sensing research of oceanic aerosol

Kaufman Y J, Tanre D, Boucher O. A satellite view of aerosols in the

using satellite data. Chin Sci Bull (in Chinese), 1986, 31(21),

climate system. Nature, 2002, 419: 215―223

1645―1649

WANG LiLi et al. Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718

1717

ATMOSPHERIC SCIENCES

MODIS AOD, with 59% within errors and 63%―88% are within errors at other three stations. The results indicate that MODIS products have the highest applicability in agricultural areas in China. However, due to disciplinarily variations of surface reflectance and aerosol models[42], satellite retrieval AOD algorithm can be better corrected.

24

Zhang J H, Si Z J, Mao J T, et al. Remote Sensing Aerosol Optical

and Forest Service handheld sun photometers in southern Africa

Depth over China with GMS – 5 Satellite. Chin J Atm Sci (in Chinese),

during the SAFARI 2000 campaign. Int J Remote Sens, 2005, 26(19):

2003, 27 (1): 23―35 25

Li C C, Mao J T, Lau A K-H. Remote sensing aerosol with MODIS

39

and the application of MODIS aerosol products. Acta Sci Nat Uni Pek (in Chinese), 39(suppl): 108―117 26

Li X J, Liu Y J, Qiu H, et al. Retrieval method for optical thickness of

40

aerosols over Beijing and its vicinity by using the MODIS data. Acta Meteor Sin (in Chinese), 2003, 61(5): 581―592 27

Xu X D, Zhou X J, Weng Y H, et al. Study on variational aerosol

41

fields over Beijing and its adjoining areas derived from Terra-MODIS and ground sunphotometer observation. Chin Sci Bull, 2003, 48(18): 2010―2017 28

Wang X Q, Yang S Z, Zhu Y H, et al. Aerosol optical thickness

42

retrieval over land from MODIS data based on the inversion of the 6S 29

30

31

model. Chin J Quantum Electron (in Chinese), 2003, 20(5): 629―634 Li C C, Mao J T, Liu Q H, Remote sensing of high spatial resolution aerosol optical depth with MODIS data over Hong Kong. Chin J Atm Sci (in Chinese), 2005, 29(3): 335―342 Mao J T, Li C C, Zhang J H, et al. The comparison of remote sensing aerosol optical depth from MODIS data and ground sun-photometer observations. J Appl Meteor Science (in Chinese), 2002, 13 (suppl): 127―135

43

44

Zhu A H, Li C C, Liu G Q, et al. Validation and application of MODIS remote aerosol information in Beijing. Acta Sci Circ (in Chinese),

45

2004, 24(1): 86―90 32

Li C C, Lau A K-H, Mao J T, et al. Retrieval, validation, and application of the 1-km aerosol optical depth from MODIS measurements over Hong Kong. IEEE Trans on Geosci Remote Sensing, 2005, 43(11): 2650―2658

33

Li C C, Mao J T, Lau K A, et al. Characteristics of distribution and

46 47

seasonal variation of aerosol optical depth in eastern China with MODIS products. Chin Sci Bull, 2003, 48(22): 2488―2495 34

Liu G Q, Li C C, Zhu A H, et al. Optical depth research of atmospheric aerosol in the Yangtze River Delta region. Environment Protection

48

(in Chinese), 2003(8): 50―54 35

Li Z Q. Aerosol and climate: A perspective from East Asia. In: Zhu X, eds. Observation, theory, and modeling of the atmospheric variability.

49

Singapore: World Scientific Publishing Company, 2004. 501―525 36

Kaufman Y J, Tanr’e D, Gordon H R, et al. Passive remote sensing of Tropospheric aerosols and atmospheric correction for the aerosol effect. J Geophy Res, 1997, 102 (D14): 16815―16830

37

50

Brooks D R, Mims F M III. Development of an inexpensive handheld LED-based Sun photometer for the GLOBE program. J Geophys Res, 2001, 106(D5): 4733―4740

38

Hao W M, Ward D E, Susott R A, et al. Comparison of aerosol optical thickness measurements by MODIS, AERONET sun photometers,

1718

51

4169―4183 Acharya Y B. Spectral and emission characteristics of LED and its application to LED-based sun-photometry. Opt Laser Technol, 2005, 37(7): 547―550 Xin J Y, Wang Y S, Li Z Q, et al. Introduction and calibration of the Chinese sun hazemeter network. Environ Sci (in Chinese), 2006, 27 (9): 1697―1702 Xin J Y, Wang Y S, Li Z Q, et al. AOD and Angstrom exponent of aerosols observed by the Chinese Sun Hazemeter Network from August 2004 to September 2005. J Geophy Res, 2006, 112, D05203-1-13 Wang Y S, Xin J Y, Li Z Q, et al. AOD and angstrom parameters of aerosols observed by the Chinese sun hazemeter network from August to December 2004. Environ Sci (in Chinese), 2006, 27 (9): 1703―1711 Salomonson V V, Barnes W L, Maymon P W, et al. MODIS advanced facility instrument for studies of the earth as a system. IEEE Trans Geosci Remote Sensing, 1989, 27(2): 145―153 Barnes W L, Pagano T S, Salomonson V V. Prelaunch characteristics of the Moderate Resolution Imaging Spectror Diameter(MODIS) on EOS-AM1. IEEE Trans Geosci Remote Sensing, 1998, 36: 1088―1100 Kaufman Y J, Tanre D, Remer L A, et al. Operational remote sensing of tropospheric aerosol over land from EOS moderate resolution imaging spectroradiometer. J Geophys Res, 1997, 102(D14): 17051―17061 Ackerman S A, et al. Discriminating clear-sky from clouds with MODIS. J Geophys Res, 1998, 103: 32141―32158 Kaufman Y J, Wald A E, Remer L A, et al. The MODIS 2.1 mm channel―correlation with visible reflectance for use in remote sensing of aerosol. IEEE Trans Geosci Remote Sensing, 1997, 35(5): 1286―1298 Ichoku C, Chu D A, Mattoo S, et al. A spatio-temporal approach for global validation and analysis of MODIS aerosol products. Geophys Res Lett, 2002, 29(12): MOD1-1-4 Wang S G, Yang D B, Chen C H. Comparison analysis of spatial and temporal variation disciplinarian of atmosphere pollution over different seasons in Lanzhou. J Lanzhou Univ (Natural Sciences) (in Chinese), 1994, 30(3): 150―155 Wang B J, Xu D P, Pu Y L, et al. A study on synoptic cause of air pollution and forecast methods of air pollutant Concentrations in Lanzhou Proper in Winter. Gansu Meteor (in Chinese), 2001, 19 (4): 18―21 Li C C, Mao J T, Lau A K-H, et al. Research on the air pollution in Beijing and its surroundings with MODIS AOD products. Chin J Atm Sci (in Chinese), 2003, 27(5): 869―880

WANG LiLi et al. Chinese Science Bulletin | June 2007 | vol. 52 | no. 12 | 1708-1718