WANG Han,SUN Xiaobing,SUN Bin,3,LIANG Tianquan,3,LI Cuili,3,and HONG Jin1Anhui Institute of Optics and Fine Mechanics,Chinese Academy of Sciences,Hefei 230031

2Key Laboratory of Optical Calibration and Characterization,Chinese Academy of Sciences,Hefei 230031

3University of Chinese Academy of Sciences,Beijing 100049

1.Introduction

Climate(Duan et al.,2012;Hansen et al.,1998)and environment(Brunekreef and Holgate,2002)are affected by aerosols because they change the energy cycle(Haywood and Boucher,2000).The quantity of aerosol is an important parameter.However,the large spatial and temporal variability of aerosol properties(Dai et al.,2014)make assessing the effect of aerosol particles on the local air quality and global climate diff i cult.Though scientists have made many attempts to quantify aerosol properties,remote sensing is one of the most useful methods to date.

Satellite remote sensing is a signif i cant way to detect global aerosol optical properties(Diner et al.,2005;Hauser et al.,2005;Remer et al.,2005).The upwelling total radiance which is used to derive the aerosol optical properties includes atmospheric ref l ectance,land surface ref l ectance,and the surface–atmosphere interactions(Waquet et al.,2009a).All three parts need to be separated from the total ref l ectance,since each part is the function of various parameters.Accurate estimation of the land surface and aerosol ref l ectance is important in aerosol retrieval as it is a challenging problem for intensity measurement.However,it can be better solved by using a polarization measurement.Lower relative contribution,less spectral dependence,and less spatial contrast are three obvious advantages of using a polarization measurement to estimate the surface contribution(Deuz′e et al.,1993;Waquet et al.,2007,2009b).Moreover,many experimental and theoretical studies have shown that polarization measurements exhibit a higher sensitivity to aerosol properties than to land surface(Mishchenko and Travis,1997;Cairns et al.,1997;Chowdhary et al.,2005).

Taking advantage of these features of polarization,Deuz′e et al.(2001)developed a method based on polarized radiance measurements to retrieve the aerosol optical thickness over lands at 670 and 865 nm with an a priori knowledge of the land surface.The measurements were provided by the Polarization and Directionality of the Earth Ref l ectance instruments(POLDER).Combined polarization and intensity measurements were used to retrieve ocean aerosol properties from Polarization and Anisotropy of Ref l ectances for Atmospheric Sciences Coupled with Observations from Lidar(PARASOL)multi-angle photopolarimetric measurements(Hasekamp et al.,2011)which improved the agreement with the Aerosol Robotic Network(AERONET)retrievals com-pared to intensity-only retrievals.A general applied method has been developed by Dubovik et al.(2011)in which both the aerosol and surface models are used to simulate total and polarized ref l ectance.It can be used to retrieve aerosol and surface properties simultaneously from multi-spectral and multi-angular data and is derived from POLDER measurements.Before obtaining satellite observations,airborne measurements are essential to develop and improve the retrieval algorithms.Airborne simulators based on the Research Scanning Polarimeter(RSP)(Cairns et al.,1997;Chowdhary et al.,2002;Cairns et al.,2003;Waquet et al.,2009b)and the microPOL(Waquet et al.,2005,2007)have been developed.The RSP provides multi-angular,multi-spectral,and polarized data.Measurements at 2250 nm are used to estimate the land surface polarized ref l ectance,and the aerosol optical properties are retrieved using an optimal estimation method(Waquet et al.,2009a).Knobelspiesse et al.(2008)introduced an iterative method to solve the diffuse and multiple interaction terms and the kernel values of the Ross–Li surface ref l ectance model using RSP data.MICROPOL is a single viewing and multi-spectral prototype polarimeter that provides accurate polarized measurements in f i ve spectral bands.The accuracy and stability of retrieval algorithms associated with the single view angle operation of MICROPOL are limited by assuming aerosol microphysical models.In China,Cheng et al.(2011)used a Directional Polarimetric Camera(DPC)to retrieve aerosol optical properties over cities.A DPC detects the same target twice,at 200 m and 4000 m to estimate the land surface and atmosphere,respectively.However,more monitoring is needed over the vast regions in China.

To better monitor aerosols in China,the Anhui Institute of Optics and Fine Mechanics,Chinese Academy of Science(AIOFM-CAS)has developed the Advanced Atmosphere Multi-angle Polarization Radiometer(AMPR)with high polarized detection accuracy.It can obtain multi-angular,multispectral and polarized data.A more detailed discussion of the AMPR is found in section 2.

This paper focuses on the retrieval algorithm of aerosol properties using multi-angular,multi-spectral,and polarized data from the AMPR over vegetation without a priori knowledge of land surface.We introduced an iterative method to quantify the polarized ref l ectance of the land surface and aerosol.The polarized ref l ectance at 1640 nm was selected as the initial estimate of surface.Then,an iterative method was adopted to obtain the ref l ectance of the land surface and atmosphere step by step.Retrievals of the ground-based sun photometer(CE318)were used to validate the algorithm(Dubovik and King,2000;Li et al.,2006).The algorithm,based on RSP measurements(RSP algorithm)and developed by Waquet et al.(2009a),was also used for the validation of the AMPR algorithm.

2.Instrument,calibration and data selection

The AMPR sensor is currently a sub-orbital based prototype that will eventually become space-based.The main goal of the AMPR is to retrieve a suite of aerosol and cloud optical thicknesses and parameters over land.The spectral and spatial responses of radiance measurements are complex,so the AMPR has six polarized spectral channels with center wavelengths at 490,555,665,865,960,and 1640 nm.Measurements at 665,865,and 1640 nm were used to retrieve aerosol properties in this paper where the molecular and aerosol optical thicknesses were small.Accordingly,the multiple scattering inf l uences were less and the single scattering approximation is realistic(Br′eon et al.,1997).The 960 nm measurements were used to estimate the water vapor column amounts.

The AMPR scan starts from?55°to+55°of nadir with a sampling interval of 1°.Thus,each scan contains 111 instantaneous samples at every wavelength.Two Wollaston prisms were used to detect the polarization in the 0°,90°,45°,and 135°directions.The subastral point spatial resolution was about 52 m at an altitude of 3 km with a scan period of 0.8630 s.About one-third of the time was used to scan the target and the remaining time was reserved for calibration purposes.

Calibration of the AMPR was based on Song et al.(2012)in order to eliminate the instrumental polarization effects and to derive the calibrated polarization parameters.Before and after the f l ight experiment,the AMPR was calibrated twice in the laboratory.There were 15 days between these two calibrations.The accuracy of the degree of linear polarization(DoLP)was better than 0.01.The calibration lamp attached to the AMPR instrument functioned while measurements were being taken.Data show that the AMPR was in normal operating conditions.

The measured polarized ref l ectance of a scan period was the input of the algorithm.The relative polarized radiance of the vegetation surface had less spectral dependence(Waquet et al.,2007,2009b).All three wavelengths shared the same surface polarized ref l ectance.We assumed that the atmospheric conditions did not change during the scanning period.Hence,we can obtain the multi-angular scattering information of aerosol.When the scattering angle was larger than 160°,the spectral responses of polarized ref l ectance were negligible and thus,the data were omitted.

3.Aerosol retrieval algorithm of AMPR

3.1.Polarized ref l ectance at viewing level

The normalized total radiance,R,and normalized polarized radiance,Rp,are def i ned from Stokes’vector:

where E0is the f l ux density at the top of atmosphere andμsis the cosine of the solar zenith angle.Thus,the upward polarized ref l ectance at the viewing altitude can be simulated in the following form(Cairns et al.,1997;Deuz′e et al.,2001):

Fig.1.Complex refractive index of aerosol at the moment of plane transit on 10 August 2012 over target f i eld(TF).

whereθvis the viewing zenith angle,θsis the solar zenith angle,and?is the relative azimuth angle.λis the wavelength.The letter“a”represents the atmospheric optical properties,which include the molecular and aerosol optical properties.Rp,atmois the contribution of the atmosphere polarized refl ectance and Rp,surfis the bidirectional polarized re fl ectance factor.T1and T2are the downward and upward transmission terms,respectively:

Here,ψand?areestimatedbyLafrance(1997),zistheviewing altitude andμvis the cosine of the viewing zenith angle.The molecule and aerosol optical thickness below level z can be calculated as

whereτmoleandτaeroare the whole optical thicknesses of the molecule and aerosol,respectively.Haand Hmare set to approximately 2 km and 8 km,respectively(Waquet et al.,2007,2009b).

3.2.Aerosol retrieval scheme

We report the results of the CE318 retrievals on 10 August 2012.These results demonstrate the temporal variability of the aerosol optical properties over the target f i eld.Figure 1 shows the complex refractive index of the aerosols at the moment of plane transit,including the real and imaginary parts.Both the real and imaginary parts are nearly steady at the different wavelengths but change after a period of time.Figure 2 shows the size distribution of the aerosols at the same time with the refractive index.The amount of small aerosol particles decreased and the center of the f i rst peak moved left in the afternoon.As polarization is more sensitive to small particles than large ones(Deuz′e et al.,2001),these changes are ref l ected by the aerosol optical thickness(AOT)and the ?Angstr¨om exponent.The aerosol distribution is time-varying and is considered in the retrieval.

A lookup table(LUT)approach was adopted in the algorithm to retrieve the AOT and the?Angstr¨om exponent.The?Angstr¨om exponent was introduced as an incarnation of the aerosol size distribution and the refractive index.Low?Angstr¨om exponent values correspond to large particles,while large values correspond to small particles,as the ?Angstr¨om exponent is a function of the size distribution and refractive index.The aerosol scattering phase function is related to these two parameters as well and is therefore related to the?Angstr¨om exponent.The?Angstr¨om exponent is calculated from the AOTs at 665 and 865 nm as follows:

Fig.2.Size distribution of aerosol at the moment of plane transit over TF on 10 August 2012.

whereλ1andλ2are two wavelengths,τ1andτ2are the relative AOTs of the wavelengths.

The polarized phase function(P)and single scattering albedo(SSA)are calculated using the Mie theory(Bohren and Huffman,1983;Grainger et al.,2004)by assuming spherical aerosol particles.The LUT structure is summarized in Table 1.The polarized re fl ectance was calculated using a vector radiative transfer model called successive order of scattering(SOS)(Lenoble et al.,2007).In the AMPR method,the LUT is used to store the aerosol polarized refl ectance only,thus the surface polarized re fl ectance is not included.The simulated polarized re fl ectance at the instrument level was calculated using Eq.(3).

Figure 3 shows the aerosol retrieval scheme.A leastsquare fi tting method was used to search for the simulated polarized re fl ectance,Rp,simu,that best matches the measured polarized re fl ectance,Rp,meas.The aerosol properties and surface re fl ectance converge to the actual value.In the condition of M wavelengths and N viewing angles,the residual termcan be def i ned as:

Table 1.Structure of LUT.

whereλiis the ith wavelength. θsj,θvjand ?jare the jth viewing geometry.

The AMPR measured polarized ref l ectance at 1640 nm was used as the initial estimation of the land surface by assuming that the aerosol and atmospheric molecule scattering were small(Wang et al.,2005).The f l ow chart(Fig.3)mechanism can be summarized as follows:

(1)Input the estimation of surface ref l ectance into the al-gorithm;

Fig.3.Flow chart of AMPR retrieval algorithm.

(2)Calculate Rp,simufrom the LUT and the surface refl ectance using Eq.(3);

(3)Calculate the least residual term using Eq.(9)and search the least one.The aerosol properties at 665 and 865 nm were obtained synchronously;

(4)Can the difference between the last steps be ignored?If no,work through steps 5 to 7.If yes,the aerosol properties have been obtained,so go to step 8;

(5)Estimate the aerosol properties at 1640 nm from the properties at 665 and 865 nm using Eq.(8)and calculate the aerosol polarized re fl ectance at 1640 nm from the LUT;

(6)Calculate the surface polarized re fl ectance(Rp,surf)by eliminating the aerosol polarized re fl ectance from the measured polarized re fl ectance at 1640 nm using the derivative of Eq.(3);

(7)The surface polarized re fl ectance from step 6 is used astheinputofthesurfacere fl ectanceandstartoverfromsteps 1 to 3;

(8)Calculate the surface polarized re fl ectance(Rp,surf)using the derivative of Eq.(3).

3.3.Surface contribution to upwelling polarized radiance

The surface polarized re fl ectance was obtained in step 8.At the same time,the surface polarized model[Eq.(10)],developed by Nadal and Bre′on(1999),was used to estimate the surface polarized re fl ectance for validation:

whereγ=(π?Θ)/2 is the incident angle on the re fl ected element,Θ is the scattering angle,and Fpis the Fresnel coeffi cient for polarized light.ρa(bǔ)ndβare empirical coef fi cients that are determined from the normalized difference vegetation index(NDVI)of the observed surface and the groundtype classi fi cation.

The NDVI was computed using the re fl ectance at 665 and 865 nm.Parametersρa(bǔ)ndβare from Nadal and Bre′on(1999).Figure 4 shows the surface polarized re fl ectance calculated by the two methods.It can be found that the angular responses of surface polarization are at the same current.The differences between these two methods are in the order of magnitude of 10?4,less than 1/10 of the total polarized re fl ectance.That is to say,an accurate surface polarized refl ectance can be obtained from the AMPR algorithm.

4.Experiments

To validate the algorithm and monitor the aerosol distribution over the Beijing–Tianjin–Tangshan region,the AMPR experimentwasconductedinTianjinon10August2012.The area covered east of Tianjin,south of Tangshan,and a corner of Bohai Bay as shown in Fig.5.A large area of farmland in Tangshan,primarily covered by plants,was chosen as the target f i eld(TF).Haze was present in this area during the morning but disappeared in the afternoon.

Two experimental f l ights(Y-12 airplane)were performed in the morning and afternoon along the same path.The summaries of the f l ights are presented in Table 2.In the morning,the AMPR scan direction was along the f l ight track while it was across the track in the afternoon.The AMPR scan angle was limited to between ?38°and+29°because of the smaller dimensions of the downward window in the plane.A POS AV(position and attitude recorder made by Applanix Company)was used to record the position and attitude of the AMPR.

Figure 5 shows a map of the local conditions and f l ight track.The red line is the segment of the f l ight track.The white square is the TF,which is mainly covered by plants.The ground-based CE318 is situated in Qichang(39.1773°N,118.3404°W,the white star)south of the TF.It provided the spectral AOT,the aerosol complex refractive index,and par-ticle size distribution.

Fig.4.Surface polarized ref l ectance over vegetation computed using AMPR algorithm and Nadal and Br′eon(NDVI＞0.3,ρ=0.007,andβ=140)method:(a)solar zenith angle=32.0°,in the morning;(b)solar zenith angle=37.2°,in the afternoon.

Fig.5.Map of experimental f i eld and f l ight track segment in Tianjin experiment.

Table 2.Main characteristics of the f l ight on 10 August 2012.

5.Results

A case study of the AMPR observations performed in the TF is presented.The AOT at 665 nm and the?Angstr¨om exponent(computed between AOTs at 665 and 865 nm)were retrieved from the AMPR measurements.The same constraints(AOTat670nmandthe?Angstr¨omexponent,whichwascomputed between AOTs at 670 nm and 870 nm)from the CE318 measurements in the TF were used to validate the AMPR retrievals.

Fig.6.Simulated,measured and surface polarized ref l ectance at the aircraft altitude:(a)solar zenith angle=32.0°in the morning;(b)solar zenith angle=37.2°in the afternoon.Green(665 nm)and red(865 nm)lines are the simulated polarized ref l ectance.Dark gray(665 nm)and dark cyan(865 nm)symbols are the measured polarized ref l ectance.Black line and symbols are the surface polarized ref l ectance.

Table 3.AOT and?Angstr¨om exponent retrieved from AMPR and CE318.

Figure 6 shows the simulated and measured polarized refl ectance at the height of the aircraft.The atmospheric polarized re fl ectance was computed using the SOS model with a black surface.The aerosol microphysical properties were derived from the CE318 measurement.The surface polarized re fl ectance was derived from the AMPR calculation process.The simulated polarized re fl ectance(atmosphere+surface)was calculated from Eq.(3)and has good agreement with the measurements.

Retrievals of CE318(before and after the over fl ight)and AMPR in the morning and afternoon are reported in Table 3.The AMPR retrieved AOT was smaller than the CE318 retrieved AOT while theA?ngstro¨m exponent was larger.One of the reasons is that polarized light stems mainly from small particles(Deuze′ et al.,2001).Another reason may be the overestimation of the land surface albedo by using the refl ectance of 1640 nm as the initial estimate of the land surface.

The RSP algorithm was used to verify the AMPR algorithm.Measurements at 1640 nm were substituted for the 2250 nm wavelength retrievals in the RSP algorithm(Fig.7).Retrievals of CE318 at the moment of the plane transit were used to verify the results.On the whole,the retrievals of both the algorithms were consistent and close to the retrievals of CE318.

Aerosols over the TF appear to be a mixture of different types(e.g.,urban-industrial and oceanic).The fi ne mode,urban-industrial aerosols correspond to largeA?ngstro¨m exponent values,while the coarse mode marine aerosols correspond to smallerA?ngstro¨m exponent values.The marine aerosols and morning haze were dominant during the morn-ing hours,but in the afternoon the dominant mode shifted to the urban-industrial aerosols.The AOT became smaller as the morning haze disappeared.These results suggest a diurnal cycle in aerosol variability.

Fig.7.Retrieved AOT and?Angstr¨om exponent over TF from the AMPR algorithm(line and squares),RSP algorithm(line and triangles),and CE318(the black star).(a)AOT in the morning;(b)AOT in the afternoon;(c)?Angstr¨om exponent in the morning;and(d)?Angstr¨om exponent in the afternoon.Time=0 is the CE318 results.Time=1 is 0827 LST in the morning and 1418 LST in the afternoon.

6.Conclusions

The AMPR is a newly developed instrument.It provides directional,polarized,and multi-spectral measurements.We developed an algorithm to retrieve the aerosol optical thickness and the?Angstr¨om exponent based on the AMPR measurements.The algorithm is based on a LUT and an iterative method.Experiments in Tianjin were conducted to validate the algorithm.

Using the aerosol retrieval algorithm for the AMPR measurement,the temporal variability of the aerosol optical properties over the TF were analyzed.The results in the morning and afternoon were analyzed and compared with CE318 measurements.Agreement between the retrievals of the AMPR and the nearby CE318 is adequate.Along the f l ight line,the retrievals of the AMPR algorithm match the retrievals of the RSP algorithm and CE318 well,which demonstrates the potential of the AMPR algorithm.The preliminary validation is encouraging.

However,limited cases make further comparisons diff icult.More experiments(currently in progress)will yield futurevalidresultsasnewequipmentdesignedtoobtainthesurface bidirectional polarization distribution function(BPDF)of land surfaces is being developed.The degree of overestimation of land surface properties when using the AMPR can be reduced by using this equipment.Combining this new approach with LIDAR will enhance the ability of the AMPR to quantify particle size and will also be addressed in future studies.

Acknowledgements.This research was supported by the Chinese Airborne Remote Sensing System,the Major National Science and Technology Infrastructure Construction Projects,and the Key Programs of the Chinese Academy of Sciences(Grant No.KGFZD-125-13-006).The authors are grateful for the helpful suggestions of researcher LI Zhengqiang and his team from the Institute of Remote Sensing Applications.

Br′eon,F.M.,J.L.Deuz′e,D.Tanr′e,and M.Herman,1997:Validation of spaceborne estimates of aerosol loading from Sun photometer measurements with emphasis on polarization.J.Geophys.Res.,102,17 187–17 195.

Brunekreef,B.,and S.T.Holgate,2002:Air pollution and health.The Lancet,360(9341),1233–1242.

Bohren,C.F.,and D.R.Huffman,1983:Absorption and Scattering of Light by Small Particles.John Wiley,New York,448 pp.

Cairns,B.,L.D.Travis,and E.E.Russell,1997:Polarization:Ground-based upward-looking and aircraft/satellitebased downward-looking measurements.Proc.SPIE,Vol.3220,Satellite Remote Sensing of Clouds and the Atmosphere II,103(January 1,1997),doi:10.1117/12.301140.

Cairns,B.,B.E.Carlson,R.X.Ying,A.A.Lacis,and V.Oinas,2003:Atmospheric correction and its application to an analysis of Hyperion data.IEEE Trans.Geosci.Remote Sens.,41(6),1232–1245.

Cheng,T.H.,X.F.Gu,D.H.Xie,Z.Q.Li,T.Yu,and X.F.Chen,2011:Simultaneous retrieval of aerosol optical properties over the Pearl River Delta,China using multi-angular,multispectral,and polarized measurements.Remote Sens.Environ.,115,1643–1652.

Chowdhary,J.,B.Cairns,and L.Travis,2002:Case studies of aerosol retrievals over the ocean from multiangle,multispectral photopolarimetric remote sensing data.J.Atmos.Sci.,59(3),383–397.

Chowdhary,J.,and Coauthors,2005:Retrieval of aerosol scattering and absorption properties from photopolarimetric observations over the ocean during the CLAMS experiment.J.Atmos.Sci.,62,1093–1117,doi:10.1175/JAS3389.1.

Dai,T.,D.Goto,N.A.J.Schutgens,X.Dong,G.Shi,and T.Nakajima,2014:Simulated aerosol key optical properties over global scale using an aerosol transport model coupled with a new type of dynamic core.Atmos.Environ.,82,71–82.

Deuze′,J.L.,F.M.Bre′on,P.Y.Deschamps,C.Devaux,M.Herman,A.Podaire,and J.L.Roujean,1993:Analysis of POLDER(POLarization and Directionality of Earth’s Refl ectances)airborne instrument observations over land surfaces.Remote Sens.Environ.,45,137–154.

Deuz′e,J.L.,and Coauthors,2001:Remote sensing of aerosols over land surfaces from POLDER-ADEOS-1 polarized measurements.J.Geophys.Res.,106,4913–4926.

Diner,D.J.,J.V.Martonchik,R.A.Kahn,B.Pinty,N.Gobron,D.L.Nelson,and B.N.Holben,2005:Using angular and spectral shape similarity constraints to improve MISR aerosol and surface retrievals over land.Remote Sens.Environ.,94,155–171.

Duan,A.M.,G.X.Wu,Y.M.Liu,Y.M.Ma,and P.Zhao,2012:Weather and climate effects of the Tibetan Plateau.Adv.Atmos.Sci.,29(5),978–992,doi:10.1007/s00376-012-1220-y.

Dubovik,O.,and M.D.King,2000:A f l exible inversion algorithm for retrieval of aerosol optical properties from Sun and sky radiance measurements.J.Geophys.Res.,105,20 673–20 696.

Dubovik,O.,and Coauthors,2011:Statistically optimized inversion algorithm for enhanced retrieval of aerosol properties from spectral multi-angle polarimetric satellite observations.Atmospheric Measurement Techniques Discussions,3,4967–5077.

Grainger,R.G.,J.Lucas,G.E.Thomas,and G.B.Ewen,2004:Calculation of Mie derivatives.Appl.Opt.,43,5386–5393.

Hansen,J.E.,M.Sato,A.Lacis,R.Ruedy,I.Tegen,and E.Matthews,1998:Climate forcings in the industrial era.Proceedings of the National Academy of Sciences of the United States of America,95,12 753–12 758.

Hasekamp,O.,P.Litvinov,and B.Andr′e,2011:Aerosol properties over the ocean from PARASOL multiangle photopolarimetric measurements.J.Geophys.Res.,116,D14204,doi:10.1029/2010JD015469.

Hauser,A.,D.Oesch,N.Foppa,and S.Wunderle,2005:NOAA AVHRR derived aerosol optical depth over land.J.Geophys.Res.,110,D08204.doi:10.1029/2004JD005439.

Haywood,J.and O.Boucher,2000:Estimates of the direct and indirect radiative forcing due to tropospheric aerosols:a review.Rev.Geophys.,38,513–544.

Knobelspiesse,K.D.,B.Cairns,B.Schmid,M.O.Rom′an,and C.B.Schaaf,2008:Surface BRDF estimation from an aircraft compared to MODIS and ground estimates at the Southern Great Plains site.J.Geophys.Res.,113,D20105,doi:10.1029/2008JD010062.

Lafrance,B.,1997:Simplif i ed model of the polarized light emerging from the atmosphere.Correction of the stratospheric aerosol impact on POLDER measurements.Ph.D.dissertation,Universit′e des Sciences et Techniques de Lille,France.(in French)

Lenoble,J.,M.Herman,J.L.Deuz′e,B.Lafrance,R.Santer,and D.Tanr′e,2007:A successive order of scattering code for solving the vector equation of transfer in the earth’s atmosphere with aerosols.Journal of Quantitative Spectroscopy and Radiative Transfer,107(3),479–507.

Li,Z.Q,P.Golouba,C.Devaux,X.F.Gu,J.L.Deuz′e,Y.L.Qiao,and F.S.Zhao,2006:Retrieval of aerosol optical and physical properties from ground-based spectral,multi-angular,and polarized sun-photometer measurements.Remote Sens.Environ.,101(4),519–533.

Mishchenko,M.I.,and L.D.Travis,1997:Satellite retrieval of aerosol properties over the ocean using polarization as well as intensity of ref l ected sunlight.J.Geophys.Res.,102,16 989–17 013.

Nadal,F.,and F.M.Br′eon,1999:Parameterization of surface polarized ref l ectance derived from POLDER spaceborne measurements.IEEE Trans.Geosci.Remote Sens.,37,1709–1718.

Remer,L.A.,and Coauthors,2005:The MODIS aerosol algorithm,products,and validation.J.Atmos.Sci.,62(4),947–973.

Song,M.X.,B.Sun,X.B.Sun,and J.Hong,2012:Polarization calibration of airborne multi-angle polarimetric radiometer.Optics and Precision Engineering,20(6),1153–1158.(in Chinese)

Wang,X.H.,J.Hong,Y.L.Qiao,F.G.Meng,Y.J.Zhang,and P.Gong,2005:Retrieval of aerosol optical thickness with dualangle multispectral means.High Technology Letters,15(9),101–105.(in Chinese)

Waquet,F.,J.F.L′eon,P.Goloub,J.Pelon,D.Tanr′e,and J.L.Deuz′e,2005:Maritime and dust aerosol retrieval from polarized and multispectral active and passive sensors.J.Geophys.Res.,110,D10S10,doi:10.1029/2004JD004839.

Waquet,F.,P.Goloub,J.L.Deuz′e,J.F.L′eon,F.Auriol,C.Verwaerde,J.Y.Balois,and P.Franc?ois,2007:Aerosol retrieval over land using a multiband polarimeter and comparison with path radiance method.J.Geophys.Res,112,D11214,doi:10.1029/2006JD008029.

Waquet,F.,B.Cairns,K.Knobelspiesse,J.Chowdhary,L.D.Travis,B.Schmid,and M.I.Mishchenko,2009a:Polarimetric remote sensing of aerosols over land.J.Geophys.Res.,114,D01206,doi:10.1029/2008JD010619.

Waquet,F.,J.F.L′eon,B.Cairns,P.Goloub,J.L.Deuz′e,andF.Auriol,2009b:Analysis of the spectral and angular response of the vegetated surface polarization for the purpose of aerosol remote sensing over land.Appl.Opt.,48(6),1228–1236.