REPORTS Chinese Science Bulletin 2003 Vol. 48 No. 182010-2017
Study on variational aerosol fields over Beijing and its adjoining areas derived from Terra-MODIS and ground sunphotometer observation 1
1
1
XU Xiangde , ZHOU Xiuji , WENG Yonghui , 2 3 1 TIAN Guoliang , L1U YUjie , YAN Peng , 3 DING Guoan 1, ZHANG Yuxiang , MAO Jietai 4 3 & QIU Hong 1. Chinese Academy of Meteorological Sciences, Beijing 100081, China; 2. Institute of Remote Sensing, Chinese Academy of Sciences, Beijing
100101, China; 3. National Satellite Meteorological Center, Beijing 100081, China; 4. Department of Atmospheric Sciences, Peking University, Beijing 100871, China
Abstract This paper presents a comprehensive observation technique on derived aerosols data from mobile sunphotometer graph and Terra-Moderate Resolution Imaging Spectroradiometer (MODIS) observation. Research results suggest that after being treated by the variational technique with respect to sunphotometer observations, the TerraMODIS remote sensing aerosol data are remarkably improved, thus for the first time revealing features of the influence of aerosols and pollution emissions of Beijing and its adjoining areas (Hebei, Shandong, etc. provinces). The regional impact features of aerosols are related with the peripheral V-shape topography of Beijing. Analyses with Hybrid Single Particle Lagrangian Integrated Trajectory model (HYSPLIT-4) and meteorological data in the case studied confirm the pollutants diffusion process along the trajectory from the sources in the south-west region, and the regional aerosol impact features. Keywords: MODIS, sunphotometer, variationl correction, optical depth. DOl: lO.1360/03wd0477
East Asia is one of areas with larger sulfide emissions, and in recent years, more sulfur-rich gases are emitted into the atmosphere along with the rapid growth of economy. A great amount of sulfate aerosol produced in industrial activities not only degrades the environment but also affects the regional climate. In China, besides the sulfate aerosol produced, the dust particle is also one of main aerosols in the atmosphere. According to observations, the soil-origin aerosol still poses the largest portion over Chongqing--the severest industrial pollution city in China. Moreover, land desertification tends to become severe, small sand and loess aerosols originated from northwest China seriously affect East China and several peripheral countries especially in spring each year[l]. 2010
However, there are few available aerosol observations in China, especially lacking a quantitative understanding of the regional distribution and secular variations of atmospheric aerosols. Therefore, how to reveal the spatial-temporal evolution features of regional aerosols in China has become a basic research topic urgent to be solved[2]. The World Meteorological Organization (WMO) set up the Global Atmosphere Watch Urban Research Meteorology and Environment (GURME) to study the adaptive measures[3]. At present, many cities environmental observation projects in many countries mainly concentrate in urban ground meteorological observation experiment, ground air pollution observation and the relation between aerosols and visibility[4---6]. The method for urban aerosol observations from satellite remote sensing (SRS) has long been a key technique unsolved. The satellite observation of aerosol distribution over city is subject to the influence of the complicated urban ground features. Especially for a metropolitan such as Beijing, due to the extended range of building limited numbers of surface observation sites, and the restrictions of environmental representative for urban observation sites, finding an effective approach to monitor the spatial-temporal distribution of urban pollution gases and aerosols has long puzzled scientists. At movement, the processing technique of SRS aerosol data has drawn much attention in international scientific community. The Earth Science Enterprise (ESE) has formed an international research contingent of the four disciplinary groups of atmosphere, land, ocean and calibration for the processing and application of Moderate Resolution Imaging Spectroradiometer (MODIS) data[7]. During the period of the intensified experiment of Beijing City Air Pollution Observation Field Experiment (BECAPEX), using a "point-surface" comprehensive sounding technique (PSCST), the evolutional features and their regularity of the 3-D dynamic-chemical processes of urban air pollution in Beijing winter have been analyzed from the atmospheric chemical pollutant observations at limited urban sites, the single-site lidar aerosol sounding, as well the large range aerosol remote sensing data of Terra-MODIS and Total Ozone Mapping Spectrometer (TOMS), thus furnishing important scientific evidence for the design of the city environmental observation system[8]. This paper focuses on discussing the above PSCST of SRS aerosol observation. The SRS aerosol optical depth (AOD) of MODIS is retrieved by jointly solving the ground surface and atmospheric spectrum models and numerically calculating pixel aerosol turbidity, yielding the AOD distribution of a higher resolution than the ground-surface limited sites, and thus revealing the large-scale features associated with regional atmospheric pollution diffusion processes. The ground multi-site sun-
Chinese Science Bulletin
Vol. 48
No. 18
September
2003
REPORTS photometer observation in this paper are used in the variational correction of the high resolution field of SRS aerosols, thus solving the difficulties in the monitoring and analyses of aerosols over Beijing and its peripheral areas, and forming a comprehensive approach of SRS-ground sunphotometer sounding. 1
The voltage measured by a sunphotometer is proportional to the incident solar radiation. Let V(A.) be the measurement value of sunphotometer, Vo(A.) the calibration constant, m( e) the atmospheric mass number, and T (A.) the atmospheric optical depth at wavelength A., then the atmospheric optical depth of the whole layer may be given by the following expression:
SRS aerosol optical depth fields
Acquisition of the SRS information used in AOD retrieval is restricted by regional sky condition and satellite orbit, that is to say, the effective information of satellite observation in a certain region can be acquired only under the condition that the sky is clear or cloudless in the period of over-the-ground observation of the satellite. Therefore, in clear day, the remote sensing information of the AOD of MODIS is objective only for a limited strip region near the sub satellite point of the orbit. Considering the reliability of the Terra-MODIS AOD data of the Beijing region, according to the transit orbit of the satellite and the condition of clear air, 28 January and 3 February 2002 were selected as the optimal analysis periods in the observational period of the experiment from the end of January to the beginning of February. The distributions of AOD, retrieved from the SRS data, indicate that the concentration of aerosols on 28 January (Fig. lea)) is low, and the main body of relatively high value lies over southern Hebei Province and northern Shandong Province, with a high value of smaller area over the Beijing urban area; the atmospheric AOD field on 3 February (Fig. l(b)) exhibits a SW-NE zonal distribution of aerosols from the Shijiazhuang region to the south suburb region of Beijing, and the pollution emissions in southern Hebei Province may enter Beijing along the south entrance of the V-shaped "valley" topography of its periphery, i.e. along a S-N zonal diffusion trajectory. 2 A variational correction approach of combining the aerosol SRS with the automatic and portable sunphotometer observation The reflectivity in the SRS aerosol model is not only a function of atmospheric AOD, but also a response function of complex underlying surface, therefore there still exist many uncertain factors in using the high resolution retrieval information from SRS data to describe the aerosol distribution in a large area, and the retrieval field must be corrected using true values at multi-stations on the ground. Automatic and portable sunphotometers were used in the experiment to obtain the aerosol observational data at multiple-point or mobile-point, and then the data were corrected by using a variational method. The reanalysis of ground-based and airborne remote sensing data was performed to get more objective aerosol distribution features of high resolution. Chinese Science Bulletin
Vol. 48
No. 18
September
T(A.)
= _l_ln(VO(A.)] m(e)
(1)
V(A.)'
where T(A.) consists of three parts: T(A.)
= T aero (A.) + Tr (A.) + Tab (A.),
(2)
where Taero (A.) is the atmospheric AOD, -z;. (A.) the Rayleigh scattering optical depth, and Tab (A.) the atmospheric absOlption gas optical depth. If the atmospheric molecule Rayleigh scattering optical depth and the absorption gas optical depth at wave length A. are known, then the atmospheric AOD at the wave length can be calculated from (2pJ. A comprehensive technical approach of combining the fixed-point measurement in the Beijing urban area with the mobile measurement in the distant suburbs was adopted in the experiment to acquire the distribution of the extinction optical features of ground aerosols in Beijing city and its peripheral suburbs. Sunphotometer automatic CE-3l8 and sunphotometer portable CE-3l7 were used during the period of observational experiment. The dynamic measurement of sunphotometer portable CE-317 along a path of the west-south-east suburb was conducted in 7 days of 28-30 January and 3-6 February, yielding the dynamic measurement data of 10 times successive measurements in a time interval of 2 min at 12 different sites. Another two times mobile measurements along the ~ourth ring road were also performed in the period, yieldmg a measurement data at 19 sites. The fixed-point successive measurement was conducted at the sites in Fangshan, Daxing, Fengtai, and Huairou using CE-3l7, and at Qianmen and Baishiqiao in the urban area using sunphotometer automatic CE-3l8. The above multi-point ground sunphotometer aerosol measurements and the AOD retrievals of SRS data during BECAPEX were comprehensively analyzed to solve the objective correction problem of satellite aerosol retrieval fields. According to the variation principle and its applied techniques[IO,llJ, if the deviation field of SRS AOD from the ground automatic and portable sunphotometer measurements is ~ T(l\x, y), in terms of variation correction factor field function ~i'(l)(x,y) functional f(l) can be constructed to satisfy the following condition that f(l) vanishes.
2003
~T(l)(X,y)
= TM(X,Y)-Tv(X,y),
(3) 2011
REPORTS /(1)
=
~min,
(4)
where A is a constraint coefficient. Expression (4) can be rewritten as 5J*(1)
=
and the computation of the reanalysis field of SRS atmospheric composition fields in dust storm processes, and has achieved a distinct effect[l2J. For the validity check of the variation field of Terra-MODIS data, the ground sunphotometer measurements on 3 February 2002 were used in the analysis test of the MODIS retrieval AOD values with and without the variational processing. For each point tested, the variational correction of MODIS retrievals was performed on 0.90,---------------------, 0.85 0.80 0.75 0.70 0.65 0.60
where i is a constraint function. The above equations can be numerically solved using the iteration method to get the first variation-corrected AOD field, T~1)(X,y). The variational correction processing scheme adopted in this paper has been applied to the processing of the SRS retrieval data of urban heat island Table 1
0.55 0.50 '--_ _.l...-_ _- ' - -_ _...l..-_ _- ' -_ _- - ' -_ _--' 0.55 0.60 0.65 0.70 0.75 0.80 0.85
Fig. 2. Correlation of the sunphotometer measurements and variation-corrected MODIS retrievals of AODs on 3 February 2002 (horizontal coordinate is sunphotometer measurements, vertical coordinate is variation-corrected MODIS retrievals of AODs).
Sunphotometer measurements, MODIS retrievals, and first variation-corrected MODIS retrievals of AODs on 3 February 2002
Sunphotometer measurements
MODIS retrievals
Variation-corrected values
Lat. (ON)
Long. (OE)
AOD
Retrievals
Deviation
Corrected values
Deviations
39.9000
116.3964
0.67200
0.89600
0.224
0.77357
0.10157
39.9483
116.3206
0.71590
0.87040
0.1545
0.71348
0.00242
39.9861
116.3589
0.82870
0.89600
0.0673
0.72715
0.10155
39.9753
116.4561
0.75030
0.89600
0.1457
0.73512
0.01518
39.9214
116.4842
0.71240
0.87040
0.158
0.71729
0.00489
39.8572
116.4811
0.74960
0.87040
0.1208
0.72412
0.02548
39.8317
116.4544
0.77850
0.87040
0.0919
0.73152
0.04698
39.8306
116.4317
0.76590
0.89600
0.1301
0.78429
0.01839
39.8300
116.3761
0.87420
0.87040
0.0038
0.71349
0.16071
39.8289
116.3139
0.77210
0.89600
0.1239
0.75553
0.01657
39.8650
116.2528
0.76370
0.89600
0.1323
0.72989
0.03381
39.8483
116.2081
0.78510
0.89600
0.1109
0.72479
0.06031 0.04643
39.8747
116.1631
0.72520
0.87040
0.1452
0.67877
39.8994
116.1042
0.68370
0.84480
0.1611
0.63986
0.04384
39.9325
116.1061
0.58650
0.87040
0.2839
0.67643
0.08993
39.9256
116.1369
0.60260
0.87040
0.2678
0.64916
0.04656
39.9233
116.1778
0.63110
0.92160
0.2905
0.70467
0.07357
39.9236
116.1994
0.64440
0.89600
0.2516
0.67004
0.02564
39.9250
116.2350
0.67390
0.89600
0.2221
0.69702
0.02312
39.9283
116.2686
0.68230
0.89600
0.2137
0.71166
0.02936
39.9440
116.2675
0.67570
0.87040
0.1947
0.67374
0.00196
0.7178
0.88381
0.1663714
0.711028
0.046108
Means AOD : Aerosol Optical Depth.
2012
Chinese Science Bulletin
Vol. 48
No. 18 September 2003
REPORTS the basis of the ground sunphotometer measurements except one at this point to get its corrected value of MODIS retrieval, then it was compared with the direct retrieval value of MODIS of the point. Table 1 shows that the mean deviation of corrected values from the sunphotometer measurements is 0.046, and the maximum error at sample points is 0.161; the mean deviation of direct retrieval values is 0.166, and the maximum error is 0.291. Those indicate that the effect of the variational correction of MODIS retrievals is distinct, since the mean error of corrected values is one fourth that of direct retrieval values, and the
maximum error is almost one half. Fig. 2 exhibits the linear correlation between variationcorrected Terra-MODIS remote sensing AOD values and objective analysis values of ground sunphotometer measurements (y = 0.5662x + 0.2856). Analysis results show that the aerosol concentration over Beijing region on 3 February is higher than that on 28 January, and the area of the aerosol concentration of high value on 3 February is far larger than that on 28 January. Those indicate that overall, the aerosol concentration field associated with the atmospheric pollution on
40' 39.5'
38.5' 38'
37' =-----'----"-->-.o..LL-~~L-~-=__'__.._~____,t::'___._L'------"-----~~"____._~~____'__L.L_____' 113.5' 114 114.5 115' 115.5' 116 116.5 117' 117.5 118'E 0
0
41.5 N
0
0
0
0
w--.-c<='----;OC==-:c:-:c;c--c---c-----"."...---c----c---c---,
41 ° 40.5° o
40' 39.5° 39° 38.5° 38° 37.5° 37' 113.5'
116'
114'
116.5'
117'
117.5'
118°E
Fig. 3. MODIS retrieval AODs and its ground sunphotometer vatiational corrected field. (a) MODIS retrieval AOD field; (b) the ground sunphotometer variation-corrected field of MODIS retrieval AOD. Inset in the figures denotes the urban observational experiment area where automatic and portable sunphotometer measurements were performed.
Chinese Science Bulletin
Vol. 48
No. 18
September
2003
2013
REPORTS 3 February is more severe than that on 28 January (figure omitted). Especially, a south-north direction zonal diffusion trajectory of high aerosol concentration on 3 February is remarkable from Shijiazhuang of Hebei Province to the south suburb of Beijing, but no similar distributive feature of aerosol anomaly exists on 28 January. The variational analysis of ground sunphotometer measurements in this paper will further trace the above aerosol zonal diffusion trajectory possibly associated with the atmospheric pollution diffusion of Beijing's peripheral regions such as Hebei Province, etc. 3
-42
Regional analysis of AODs
It can be found from comparing and analyzing Fig. 3(a) and (b) that sunphotometer variational correction reduces the systematic error of original MODIS retrieval aerosol values. Particularly after variational processing the extensive enhanced reflectivity (false high value) area resulting from the mountainous complex topography is deleted over the west and north parts of the Beijing region, therefore, the high value center of aerosols over the south part and the non-uniform distribution over the urban area of Beijing become prominent. Besides, the south-north zonal aerosol diffusion/transfer features from Shijiazhuang, etc. regions of Hebei Province to the urban area of Beijing is distinct, and the high concentration center of aerosols associated with the Shijiazhuang strong pollution emission source is even more remarkable; an aerosol pollution diffusion/transfer charmel along the valley from Datong of Shanxi Province to Beijing is also obvious. The above computational results reveal the westerly and southerly transfer charmels through which peripheral pollution emission sources affect the Beijing region.
4 Peripheral aerosol impact trajectories The computational method of backward trajectory was used to trace aerosol long-range diffusion propagation paths. In the simulation experiments in this paper, Draxler's HYSPLIT-4 trajectory model[13] was employed to calculate the backward trajectories on 3 February. Fig. 4 shows that the Terra satellite transited over Beijing at about 11: 00 am on 3 February, the 12 h backward trajectory of air mass is consistent with the south-north zonal distribution feature of the high value area of variation-corrected AOD, and also in general agreement with the stream field on the lower atmospheric layer in the corresponding period. 5 Variational analysis of the PMIO concentration field on lower atmospheric layer It was assumed that the ground observed PMlO concentration is correlated with the whole layer atmospheric aerosol field observed by the Terra-MODIS. The PMlO observations on the lower atmospheric layer of ground observational stations in the peripheral area such as Hebei 2014
06
18
00
12
Fig. 4. The 12 h backward trajectory on 3 February, 2002 computed by the HYSOLIT-4 trajectory model. Source location at 39.90 o N, 116.300E (in meter AGL).
Province, Tianjin City, etc. were used in the variational correction of Terra-MODIS remote sensing optical depth field to get the extensive high resolution aerosol distribution on the lower atmospheric layer over Beijing and its peripheral areas, and to explore the effect of the surface atmospheric pollutants or aerosols on the lower atmospheric layer of peripheral areas on the Beijing urban area. The variational correction scheme of SRS-ground PMlO field: the difference/deviation field ~T(2)(X,y) was calculated from the first variation-corrected AOD field of Terra-MODIS, ~T2)(x,y), minus the ground air quality PMlO particle concentration field measured at ground air quality monitoring stations in Beijing's peripheral area such as Shijiazhuang, Tangshan, Chengde, Tianjin, etc. According to the above-mentioned equations, the numerical solution of the variation-corrected field of PMlO concentration, i.e. ground PMlO-associated extensive high resolution particle concentration analysis (x, y), can be obtained by using the iteration field, method. It can be found from the objective analysis field (Fig. 5(a)) of PMlO concentration at the Terra satellite transit time (11:00 LT) on 3 February over Beijing and its peripheral areas such as Hebei Province, Tianjin, etc. that the PMIO analysis field (without variational processing) shows a similar transfer channel feature from Shijiazhuang to Beijing, but the PMlO distribution only delineates an approximate trend. The trajectory trend is close to the stream field on 850 hPa (figure omitted) at 08:00-
T;M
Chinese Science Bulletin
Vol. 48
No. 18
September
2003
REPORTS 41.5° N 41 °
40.5° 40°
39.5°
39° 38.5° 38° 37.5°
3T 113.5°
114°
114.5°
115°
115.5°
116°
116.5°
lIT
117.5°
118° E
114°
114.5'
115'
115.5'
116°
116.5°
lIT
117.5'
118° E
41.5° N 41 °
40.5° 40° 39.5°
39° 38.5°
38° 37.5°
3T 113.5°
Fig. 5. The ground-observed PMIO analysis field at the sub satellite point on 3 January and the variational corrected field of Terra-MODIS remote sensing AOD. (a) The PMlO analysis field measured for Beijing's peripheral area in air quality monitoring stations; (b) the variation-corrected field of Terra-MODIS remote sensing AOD.
11:00 (LT) on 3 February, and to the extensive distributive trend of Terra-MODIS remote sensing aerosols (Fig. 3). In comparison with this, it can also be seen from the variation-corrected field of Terra-MODIS remote sensing aerosols (Fig. 5(b)) that the Shijiazhuang-Beijing zonal channel and its high resolution distribution are very distinct, thus clearly revealing the diffusion path of ground atmospheric pollution of the peripheral area and its effect on Beijing. Chinese Science Bulletin
Vol. 48
No. 18
September
2003
6
Discussion
Beijing City is located in a V-shaped valley with the Yanshan Mountains in its north and plain in its south, therefore the diurnal variation of topographic mountain-valley wind, and the dynamic transport and diffusion of pollutants from far/near sources may appear under the regional climate background. Their comprehensive effect leads to the dynamic transport and diffusion of pollutants 2015
REPORTS in the Beijing suburb V-shaped valley and the "pile-up" of pollutants over the Beijing urban area[14J. The climatic statistics of dominant wind in the Beijing region show that northwest and southerly winds dominated in 1959-1980, while northerly and southwesterly winds prevailed in the recent decade of 1981-1998. Therefore, the southerly dominant wind is the major climatic feature of Beijingregion's wind fields. Besides, urban building complexes also exhibit a V -shaped distribution similar with this megarelief pattern. Vnder the background of southerly dominant wind over Beijing, such a V-shaped "multiple-scale-nesting" structure of peripheral megarelief and urban building complexes will restrain and retard the further "migration" of the south coming air pollutants from the peripheral areas, resulting in the pile-up of the pollutants over the Beijing region. Obviously, the comprehensive impact of the above topographic, climatic, and urban development situations may result in the feature of the effect of peripheral emission sources on the atmospheric pollution over Beijing. It is found from the above analysis that under the condition of southerly stream field, a SRS variation-corrected PMlO high concentration trajectory pattern similar with the case of 3 February 2002 may be associated with the impact of the peripheral topography, and the urban development layout and the particular synoptic system background on the atmospheric pollution diffusion. The variational processing technique of SRS ground sunphotometer observation may offer a new clue and a technical processing method to design the monitoring system on the aerosol distribution over Beijing and its peripheral areas and the impact of peripheral emission sources. Acknowledgements We would like to thank the Institute of Remote Sensing of the CAS, National Satellite Meteorological Center, and the meteorological and environmental sectors of Hebei Province for full support and valuable data of scientific experimental observation. Thanks also extend to Lin Yan of the Meteorological Bureau of Hebei Province and Liu Qinhuo of the Institute of Remote Sensing of CAS for their hard work in data collecting and processing. This work was jointly supported by the National Basic Theoretical Research Project (Grant No. GI999045700), and the Hi-Tech Research and Development Program of China (Grant No. 2002AA135360).
References I.
2. Luo Yunfeng, Lu Daren, Li Weiliang et aI., Variational characteristics of atmospheric aerosols optical depths of the recent 30 years in China, Chin. Sci. Bul. (in Chinese), 2000, 45: 549-554. 3. WMO, The Global Atmosphere Watch (GAW) Strategic Plan 2001-2007,2001. 4. Apple, B. R. et aI., Berkeley visibility as related to atmospheric aerosol constituents, Atmospheric Environment, 1985, 19: 15251534. 5.
Sisler, 1. E, Maim, W. C., The relative importance of soluble aerosols to spatial and seasonal trends of impaired visibility in the United States, Atmospheric Environment, 1994,28: 851-862.
6. Maim, W. C., Gebhart, K. A., Examining the relationship between atmospheric aerosols and light extinction at Mount Rainier and North Cascades National Parks, Atmospheric Environment, 1994, 28: 347-360. 7. Liu Yujie, Yang Zhongdong et aI., MODIS Remote Sensing Information Processing Principles and Algorithms (in Chinese), Beijing: Science Press, 200 I. 8. Xu Xiangde, Localized 3D-structural features of dynamic-chemical processes of urban air pollution in Beijing winter, Chin. Sci. Bul., 2003,48(8): 819-825. 9. Mao Jitai, Li Chengcai, Zhang Junhua et aI., The comparison of remote sensing aerosol optical depth from MODIS data and ground sunphotometer observations, 1. Appl. Meteor. Sci. (in Chinese), 2002, 13(Suppl.): 27-135. 10. Weng Yonghui, Xu Xiangde, Numerical simulation over the Tibetan plateau by using variation technique revised TOVS data, Chin. 1. Atmos. Sci., 1999,23(6): 703-712. II. Weng Yonghui, Xu Xiangde, Bai Jingyu et aI., Applying the variational method of TOVS data over the Tibetan Plateau to improving model's initial fields, Acta Meteorologica Sinica, 2000, 58(6): 679 -691. 12. Zhou Xiuji, Xu Xiangde, Yan Peng et aI., Dynamic characteristics of spring sand/dust storms in 2000, Science in China, Ser. D, 2002, 32(4): 327-334. 13. Draxler, R. R, Hess, G D., Description of the Hysplit_4 modeling system. 1997, NOAA Tech Memo ERLARL-224, 24p.
Luo Yunfeng, Zhou Xiuji, Li Weiliang, Advances in the study of atmospheric aerosol radiative forcing and climate change, Adv.
14. Xu Xiangde, Dynamic issues of urban atmospheric pollution models, 1. Appl. Meteor. Sci. (in Chinese), 2002, 13(Suppl.): 1-12. (Received January 14, 2003; accepted June 1,2003)
Earth Sci. (in Chinese), 1998,572-581.
2016