30SATELLITE-BASED MEASUREMENT OFATMOSPHERIC AEROSOLS

RUDOLF B. HUSAR

Department of Energy, Environment and Chemical Engineering, Washington University, Saint Louis, Missouri

30.1 Introduction 667

30.2 Background on Satellite Remote Sensing 668

30.3 Aerosol Physical and Optical Properties 669

30.4 Principles of Satellite Aerosol Detection andMeasurements 672

30.5 Challenges of Satellite Aerosol Measuring Systems 675

30.6 Satellite Data and Information Systems 675

30.7 Applications 676

30.8 Future Developments 677

30.9 Acknowledgments 678

30.10 List of Acronyms 678

30.11 References 678

30.1 INTRODUCTION

Radiation sensors on earth-orbiting satellites offer a way toobserve atmospheric aerosols at about kilometer resolution,near real time over much of the globe. The aerosol back-scattering of solar radiation allows us to observe the patternof atmospheric aerosols from major pollution events, duststorms, forest fires, and other aerosol events. In fact, thepatches of bluish smoke-haze and yellow dust plumes are pro-minent features of the earth as viewed from space. Satelliteaerosol measurements have also shown great potential tocontribute to earth science, to the assessment of air quality,and to the management of some disasters. Global-scaleaerosol monitoring has the further potential to improve ourunderstanding of aerosol-induced climate effects. Thus, it isexpected that satellite aerosol observing and measuringsystems will continue to expand their utility.

Unfortunately, the quantitative measurement of aerosolproperties has proven to be a challenge since the beginningof the satellite era in the 1960s. This is no surprise, sincesatellite-based measurement is among the most complexaerosol measuring techniques. In principle, atmospheric aero-sols are well suited for satellite remote sensing. The sun

provides a stable light source, the aerosol scattering coversthe entire solar spectrum, and the radiation is easily detectableby high resolution sensors. The difficulties for quantitativeaerosol measurements include: (1) Aerosol scattering is justone of the four components of the detectable radiation(Fig. 30-1a) and separation of the weak aerosol signal fromthe other signal components is error-prone; (2) Associatingthe angular back-scattering by aerosols with intrinsicaerosol properties is burdened with ambiguity; and (3) Thedownward-looking satellite sensors measure the total back-scattering, which is the integral over the entire atmosphericcolumn.

This chapter is a brief summary of satellite aerosol detec-tion and measurement. The chapter begins with a backgroundon aerosol remote sensing followed by a summary of aerosoloptical parameters relevant to satellite detection. The bulk ofthis chapter is devoted to the physical principles and the chal-lenges of satellite aerosol detection. Following a section onsatellite data dissemination, illustrative application examplesare given for earth science and air quality management. Thechapter is concluded with a list of desirable developmentsin satellite aerosol measurements. For more details, thereader is guided to the excellent textbooks on atmospheric

Aerosol Measurement: Principles, Techniques, and Applications, Third Edition. Edited by P. Kulkarni, P. A. Baron, and K. Willeke# 2011 John Wiley & Sons, Inc. Published 2011 by John Wiley & Sons, Inc.

667

aerosols (Seinfeld and Pandis 1998; Friedlander 2000), thereview of atmospheric aerosol measurements by McMurry(2000), and atmospheric visibility measurements and scienceby Watson (2002). For more details on satellite aerosolmeasurement systems see the authoritative article byKing et al. (1999), the recent broad review by Hoff andChristopher (2009) and the comments by Hidy et al. (2009).

30.2 BACKGROUND ON SATELLITEREMOTE SENSING

Atmospheric aerosol characterization using optical tech-niques has a rich history, a vigorous present, and bright pro-spects for the future. The abundance of atmospheric opticaldata prompted early investigators to seek information about

aerosol properties via the optical data. Following theKrakatoa eruption, Kiessling (1884) concluded that he volca-nic dust with characteristic size between 1 and 2 mm causedthe spectacular optical phenomena worldwide. Wegener(1911) used a simple but most elegant reasoning to inferthat the characteristic size of the atmospheric haze particlesis comparable to the wavelength of visible light. With theadvent of computers, numerical inversion procedures weredeveloped to infer aerosol size distributions from optical,that is, spectral and angular aerosol-scattering measurements.For example, based on the inversion of spectral extinctiondata, Penndorf (1957) concluded that the peculiar blue sunand blue moon phenomena in Europe were due to agedCanadian forest fire smoke with the characteristic size of0.6–0.8 mm. Recently, a more sophisticated inversionscheme was offered by Dubovik and King (2000) for the

Figure 30-1 (a)

Q15

Radiation components detected by a satellite aerosol sensor. (b) Polar and geostationary satellite orbits.

668 SATELLITE-BASED MEASUREMENT OF ATMOSPHERIC AEROSOLS

extraction of columnar aerosol size distribution data from theAERONET1 (Holben et al. 1998) sun-sky photometer net-work data.

Satellite aerosol detection begun in the mid-1960s with theobservation of “anomalous gray shades” in DefenseMeteorological Satellite Program (DMSP) satellite images.By the 1970s, these gray shades were identified as forestfire smoke over Central America (Parmenter 1971), Saharadust (Carlson and Prospero 1972), and other major aerosolevents. The application to air pollution was pioneered byRanderson (1968). Quantitative aerosol detection begun inthe early 1970s with the pioneering work of Griggs (1975),who observed that over dark water the excess radiancedetected by a satellite was correlated with the aerosol opticaldepth (AOD) of the atmospheric aerosol column. Later Fraser(1976) pointed out that the relationship between radiance andAOD depends on the type of aerosol, such as windblown dustor industrial haze. The potential of geostationary satellites toaugment surface-based particulate air pollution measure-ments was illustrated by Lyons and Husar (1976). By the1980s, a significant number of satellites were deployed thatyielded increasingly better aerosol characterization. Inanother seminal paper, Fraser et al. (1984) laid out a pro-cedure for satellite-based measurement of regional sulfatehaze over the eastern United States, including the estimationof the aerosol mass transport rates out of the region.

The remote sensing of the earth’s surface and of atmos-pheric composition is significantly perturbed by the presenceof atmospheric aerosols. Much of the early work was there-fore focused on devising aerosol correction algorithms forremoving the spurious aerosol signal. The earliest quantitat-ive global-scale aerosol measurements were obtained as by-products arising from aerosol correction of advanced veryhigh resolutionQ1 radiometer (AVHRR) sea surface temperaturereadings (Stowe 1991) and yielded a multi-year global clima-tology of aerosol optical depth over the oceans (Husar et al.1997). The columnar ozone detection by the total ozone map-ping spectrometer (TOMS) sensor was also distorted byabsorbing aerosols and the subsequently extracted aerosoldata yielded the global absorbing aerosol maps over land aswell over oceans (Herman et al. 1997). The ocean color andchlorophyll detection by the sea-viewing wide field-of-viewsensor (SeaWiFS; Gordon and Wang 1994) also contributedsignificant advances through aerosol corrections algorithms.By the 1990s a wide array of sensors were deployed for themeasurement of atmospheric aerosols. These sensors useback-scattered solar radiation in multiple narrow wavelengthchannels for moderate resolution imaging spectroradiometer(MODIS; King et al. 1992), SeaWiFS (Gordon and Wang1994), and polarization of back-scattered radiation(POLDER; Deuze et al. 2001), and combined angular andspectral scattering for multi-angle imagingQ1 spectroradiometer

(MISR; Diner et al. 1998). It is fair to say, however, that noneof the sensors can by itself fully characterize the complex,dynamic, multidimensional aspects of the atmospheric aero-sol system. For an extensive review of recent satellite aerosolmeasurements, see Hoff and Christopher (2009).

Current earth-observing satellites are circling the eartheither in geostationary or polar orbit, representing twodifferent aerosol-sensing and monitoring opportunities(Fig. 30-1b). The fixed geostationary perspective and dataevery 30 min allows the monitoring of dynamic meteorologi-cal phenomena such as major dust clouds, smoke plumes, orregional pollution events as they evolve throughout the day.Geostationary meteorological satellite data are routinely dis-tributed through the Internet minutes after they are takenand are used in forecasting and other decision support appli-cations. Polar orbiting platforms at about 100-km elevationscan the earth once a day observing always the “sunnyside.” The low orbit permits polar platform sensors to havehigher spatial resolution and geo-location accuracy as wellas sensors at multiple narrow wavelengths. Most quantitativeaerosol data sets and climatologies are derived from polarorbiting satellites carrying a variety of sensors.

30.3 AEROSOL PHYSICAL AND OPTICALPROPERTIES

This section reviews key characteristics of atmospheric aero-sols that are relevant to satellite remote sensing. Emphasis isplaced on the observed regularities in size distribution, light-scattering efficiency, angular scattering pattern, and verticaldistribution. A distinguishing characteristic of atmosphericaerosols is the high dimensionality of the “data space.” Atany location and time (x, y, z, t), aerosols are distributedover particle diameter, D, which spans at least Q2two ordersof magnitude (0.1 and 10 mm) in the optically active sub-range. Atmospheric particles are also composed of discretechemical species with composition, C. Each size class for agiven chemical species may have a unique shape, S. Asubtle, but for aerosol optics significant, dimension is thenature of aerosol mixing, X. Some of the aerosol species aremixed externally, while other species are mixed internallyin a single particle and their radiative effects are additive,while the optical effect of internally mixed particles ismuch less predictable (White 1986). For a formal treatmentof the aerosol size-composition distribution functions, seeFriedlander (1970) and Mc Murry (2000).

The full characterization of an aerosol system requireseight orthogonal dimensions (x, y, z, t, D, C, S, X ). Satellitesensors have high spatial resolution (x, y) but they detectthe optical effect of a vertical column, which is an integralover five dimensions (z, D, C, S, X ). The de-convolution ofthis integral is a notoriously ill-posed mathematical problemthat does not possess a unique solution. In other words, the1See list of acronyms at the end of the chapter.

30.3 AEROSOL PHYSICAL AND OPTICAL PROPERTIES 669

number of unknown aerosol parameters is much larger thanthe known entities from optical measurements. This inversionproblem is made extra uncertain by the fact that the aerosoloptical signal at the top of the atmosphere is weak comparedto the noise from the surface reflectance, interference fromclouds, and from the complications of a meta-stable aerosolsystem.

A strategy toward the solution of the inversion problem isto reduce the dimensionality of the aerosol system. Suchreduction can be achieved by observing the regularities ofatmospheric aerosol behavior and applying those regularitiesto satellite aerosol measurements. The key aerosol microphy-sical parameters relevant to the optical detection of aerosolsare shown in Figure 30-2a.

The aerosol volume (or mass) is typically distributed(Fig. 30-2a) between the fine particle mode in the diameterrange 0.1 mm to about 2 mm range and the coarse particlemode above 2 mm. In a nutshell, the model introduced byWhitby and co-workers (Whitby et al. 1972; Whitby 1978)states that each aerosol mode has a characteristic size distri-bution, chemical composition, and optical properties. Moredetails on size distribution characteristics of atmosphericaerosols are presented in Chapter 4. The multi-modal concepthas been widely adopted as the standard size distributionmodel by the aerosol remote-sensing community (Remerand Kaufman 1998; Kahn et al. 2001; Dubovik et al. 2002).

Fine mode particles (below 1–2 mm) are mostly formedby combustion and condensation processes, resulting inspherical liquid droplets that are frequently meta-stable.Hence, the size distribution and optical properties of fine par-ticles is highly dynamic and responsive to environmentalconditions such as gas particle conversion, relative humidity,and chemical interactions among particles. Over urban indus-trial regions the fine particle mode is composed of sulfate,organic, and nitrate species, each having somewhat differentsize distribution (Seinfeld and Pandis 1998). Fine particlestend to be internally mixed, but may exhibit multiple massmodes. Sulfate particles, for instance, occur in two distinctlydifferent mass modes depending on whether the formationmechanism is photochemical gas particle conversion (massmode � 0.2 mm) or through liquid-phase reactions (massmode � 0.5 mm) (McMurry et al. 1996). Fine particles exhi-bit another regularity of increased mass median diameter withincreasing concentration. This is significant because it allowsthe formulation of dynamic aerosol models that directly relatethe spectral and angular scattering characteristics to aerosolconcentration (Remer and Kaufman 1998; Dubovik andKing 2000). Coarse-mode atmospheric particles tend to orig-inate from primary emissions, for example, windblown dust,sea salt, fly-ash, and road dust, and they are irregular in shape.The radiative impact of coarse particle mixtures are additive.An exception to this rule is when ambient air is ingested into

Figure 30-2 Aerosol physical properties relevant to optical detection. (a) Aerosol mass distribution. (b) Light-scattering and absorption perunit aerosol volume (or mass). (c) Change in the phase function as a function of extinction coefficient. Q13

670 SATELLITE-BASED MEASUREMENT OF ATMOSPHERIC AEROSOLS

cloud droplets and the result is internal mixing of all particlesin the cloud droplets. Sea salt particles may also grow at highhumidity.

The most critical aerosol optical property for satellite aero-sol retrieval is the angular scattering since satellites detect thecumulative scattering of sunlight in the view-path of thesensor (Fig. 30-1a). The directional dependence of scatteringis defined by the phase function P(u) which is the ratio of theenergy scattered into direction, u, to the average energyscattered into all directions. See Chapter 13 for detailed dis-cussion on fundamentals of light scattering by aerosol par-ticles. The scattering angle, u, is the angle between the sun,the aerosol volume, and the sensor (Fig. 30-1a). Unlike thesymmetric Rayleigh scattering by pure air, the scatteringphase function for aerosols is elongated in the direction ofthe incoming light. Barteneva (1960), observed a remarkabledependency of the phase function shape on the aerosol extinc-tion coefficient in the former Soviet Union. With increasingextinction coefficient, the “cigar”-shaped aerosol phase func-tion becomes increasingly shifted toward the forward direc-tion (Fig. 30-2c). This pattern is consistent with the notionthat at higher extinction coefficients, the characteristic par-ticle size increases and the scattering is more and more for-ward directed. The broader applicability of Barteneva’sobservations was confirmed through airborne polar nephel-ometer measurements in the United States by Johnson(1981). However, the nature of the phase function varies con-siderably (Sakunov et al. 1996). A pixel-level comparison ofaerosol optical depths (AODs) derived from MODIS andMISR satellite aerosol instruments sensors found a significantdeviation as a function of the scattering angle between the twosatellite aerosol measuring systems (Mishchenko et al. 2009).This further highlights the need for increased attention to theaerosol phase function.

Another feature of atmospheric aerosols is that at anygiven region, there is a high correlation between the totallight-scattering coefficient and the mass concentration offine particles, below 2.5 mm in size. The slope of the corre-lation corresponds to scattering efficiency, a � 3–5 m2/g(Malm and Hand 2007), which is generally applicable for sul-fates, nitrates, and organics in the fine particle mode with apeak in extinction efficiency in the 0.3–0.8 mm size range(Fig. 30-2b). For atmospheric dust particles, the correspond-ing extinction efficiency per unit mass is much less (a �0.6 m2/g), as documented by White (1986). The observedregularity and ubiquity of this relationship allows the estima-tion of fine particle columnar mass concentration Mfc[g/m2]from measured optical thickness, t through Mfc ¼ t/a.

The vertical layering of the atmospheric aerosols alsorequires special consideration. It is known that aerosols tendto reside in distinct layers of the atmosphere, as illustratedschematically in Figure 30-3. The vertical distribution canbe subdivided into three layers, each layer having differentaerosol types, atmospheric residence time, and mixingcharacteristics. The stratospheric or Junge layer (Fig. 30-3a)is composed primarily of volcanic sulfuric acid aerosolsthat circle the earth at 10–15 km elevation for a year ortwo after volcanic eruptions. Below the stratosphere is thetropospheric layer, which is the carrier of dust, smoke, oroccasional industrial haze ejected from the boundary layer.Tropospheric aerosols are thin pancake-like layers that maycover 1000-km size areas and tend to circle the earth duringtheir 1–2 week residence time. The individual layers oftropospheric dust, smoke, and haze do not interact, hencetheir optical effects are additive. The bottom layer is theplanetary boundary layer (PBL) that contains a largevariety of aerosols from anthropogenic and natural sources(Fig. 30-3). Precipitation and other removal mechanisms

Figure 30-3 (a) Schematic representation of the aerosol vertical distribution. (b) Astronaut photo of stratospheric aerosol layers. (c) Complexphysical and radiative interaction between aerosols and clouds.

30.3 AEROSOL PHYSICAL AND OPTICAL PROPERTIES 671

restrict the aerosol residence time in the PBL to 3–5 days.Each aerosol type has its own sources and transport andoptical characteristics and therefore deserves to be treated sep-arately with specific aerosol physical and optical parameteri-zation. In Figure 30-3a, the aerosol types that are in separateboxes are externally mixed and their respective radiativeeffects are additive. A group of aerosol types in the boundarylayer that are in adjacent boxes represent aerosol types that aregenerally internally mixed and their physical, chemical, andoptical properties and behaviors need to be considered jointly.

The above discussion is a simplified description of thecharacteristics of atmospheric aerosols. A challenge of satel-lite aerosol measurements is to clearly delineate whichaspects of aerosol characterization can be supported by satel-lite sensors and their retrieval algorithms.

30.4 PRINCIPLES OF SATELLITE AEROSOLDETECTION AND MEASUREMENTS

The science of satellite aerosol measurements is based onwell-understood fundamental physical principles: Mietheory of light scattering and the theory of radiative energytransfer through scattering and absorbing media. The verysame physical laws that govern satellite aerosol detectionare applied to atmospheric visibility studies. In fact, satelliteaerosol detection and daytime vision by the human eye oper-ate on the same general principle of passive remote sensing(Fig. 30-1a): The radiation source is the sun; the target isthe earth’s surface; the detector is an array of color-sensitivesensors, that may be the retina in the human eye or the satelliteradiation sensor. In between the reflecting surface and thelight sensor is the radiatively active atmosphere consistingof aerosols, air, and clouds. However, the two fields havedifferent objectives and evolved rather independently fromeach other. The objective of visibility studies is to measureand model the ambient aerosol microphysics and chemistryand to evaluate the resulting effects on the optical environ-ment. Satellite aerosol remote sensing on the other handintends to solve the much more difficult inverse problem ofderiving the properties of atmospheric aerosols from themeasured optical effects at the top of the atmosphere. Theinherent problem is that the retrieval of the unknown aerosolproperties requires knowledge of those same properties. Thecurrent practice for resolving the inversion paradox is to pre-pare a discrete set of “aerosol models,” and to choose themodel that best fits the satellite optical data in the form ofspectral or angular scattering. The aerosol models and the sur-face reflectance models are typically tailored to the character-istics of the particular sensor.

The radiative signal received by a satellite is the combinedcontribution of surface reflectance and aerosol reflectance.For the sake of illustration, a single-scatter radiative transfertheory is used here to show the aerosol interactions with

solar radiation. The separation of the aerosol signal fromthe surface reflectance can be achieved by the solution ofthe appropriate radiation transfer equation as derived in(Husar and White 1976). The resulting relationship betweenthe surface reflectance and aerosol optical parameters isshown in Figure 30-4.

The aerosol perturbation of the upwelling radiation con-sists of two competing effects. On one hand, light-scatteringparticles tend to add reflectance to bright surfaces. Thisexcess reflectance depends both on the aerosol optical thick-ness, t, as well as the aerosol scattering phase function, P. Theother role of aerosols is to act as a filter that exponentiallydiminishes the upwelling radiation from a reflecting surfaceand from aerosol scattering itself. The net result of thesetwo competing aerosol effects is illustrated in Figure 30-5d,where the apparent reflectance is plotted against the aerosoloptical thickness. The bundle of curves corresponds to differ-ent values of surface reflectance R0. The simplest case iswhen the surface is black, R0 ¼ 0 and the received radiationis purely due to aerosol scattering along the path, that is,the path radiance PR ¼ (1 2 e2t )P (Kaufman 1993). Foran optically thin aerosol layer, t , 1, PR is proportional tothe optical thickness, since self-extinction of the aerosollayer is minimal. As the aerosol layer approaches t . 2,self-extinction becomes more significant and by t . 4, theapparent reflectance asymptotically approaches the value ofP, which is the case for radiative equilibrium. Good examplesfor the strong path radiance are dark surfaces such as ocean orvegetation at blue wavelength where R0 is small (,0.05),compared to P � 0.3 (Fig. 30-5b,c).

The opposite case occurs when the surface reflectance R0

is high compared to P. In this case, the addition of aerosoloptical thickness actually diminishes the initial high surfacereflectance, since the filter term dominates over the aerosolreflectance term. As the aerosol layer above the bright surfacebecomes optically thick, t . 4, the apparent reflectanceasymptotically approaches the value of aerosol reflectance,P. For example, an aerosol layer on top of bright cloudswill tend to decrease the brightness of clouds. Figure 30-5a

Figure 30-4 Illustrative radiative model for aerosol remotesensing.

672 SATELLITE-BASED MEASUREMENT OF ATMOSPHERIC AEROSOLS

is an example of a layer of yellow dust on top of white cloudsreducing the cloud reflectance. The next interesting case iswhen the aerosol phase function has the same value as the sur-face reflectance function, P ¼ R0. in the direction of sensor’sline of sight. In this case, adding aerosol will not change theapparent reflectance. At P � R0 (e.g., soil and vegetation at0.8 mm), the reflectance is unchanged by typical haze aero-sols and aerosol detection is not possible.

The critical measure, whether aerosol will increase ordecrease the apparent reflectance, is the ratio of the aerosoland surface reflectances, P/R0. Figure 30-5d also showsthat with increasing aerosol optical thickness the apparentreflectance of bright and dark surfaces converge toward thevalue of the aerosol scattering function, P. In other words,

with increasing AOT, the contrast between dark Q3and brightsurfaces diminishes and they become indistinguishable as t

exceeds 4. In the case of an optically thick aerosol layer,the filtering and scattering effects of aerosols are balanced,that is, the aerosol system reaches radiative equilibrium.The implication is that the retrieval of t for optically thickaerosol layers (t . 4) is not possible. Aerosol retrieval isalso limited when P � R0, which occurs over the bright sur-faces of deserts and rocky terrains.

The optical thickness, t, of the aerosol column is derivedfrom the excess reflectance, that is, the difference betweenthe apparent reflectance of the top of the atmosphere.Through the simple radiative transfer theory it can be shownthat the excess reflectance and the aerosol optical thickness

Figure 30-5 (a) Spectral reflectance with and without aerosol clouds. (b) Vegetation. (c) Ocean. (d) Apparent surface reflectance as a functionof P/R0 and AOT.

30.4 PRINCIPLES OF SATELLITE AEROSOL DETECTION AND MEASUREMENTS 673

are related by the following simple expression, t ¼ 2ln(R 2

P)/(R0 2 P). The value of P is obtained from fitting theobserved and retrieved surface reflectance spectra. Forexample, in light haze at 412 nm, it is found that P ¼ 0.38.

The color Figure 30-6 provides a graphic illustration of thekey steps in satellite aerosol retrieval. The hazy SeaWiFSimage (Fig. 30-6a) shows the composite signal received bythe sensor on July 16, 2009. The image was synthesizedfrom the blue (0.412 mm), green (0.555 mm), and red(0.67 mm) channels. The aerosol-free surface reflectance(Fig. 30-6b) was derived from time series data. The aerosoloptical depth (Fig. 30-6c) was calculated from the excessradiance, that is, the difference between the hazy and haze-free reflectances at the blue (412 nm) wavelength. The result-ing aerosol pattern shows qualitatively the hazy patches (red)over the ocean as well as in the vicinity of cloud systems. Theblack areas are blocked out by the cloud mask.

Figures 30-6a,b,c show the received signal at threewavelengths. At blue wavelength (412 nm) Figure 30-6d,the apparent reflectance, R, reaching the satellite sensor isdominated by the haze since the surface reflectance is lowover the ocean and vegetation. The aerosol patterns are clearlydiscernible, while all the surface features are obscured byhaze. Hence, the blue wavelength is well suited for aerosoldetection over land (Hsu et al. 2006) but surface characteriz-ation is difficult (see Fig. 30-5c). At red wavelength(670 nm), Figure 30-6e, the apparent reflectance is contribu-ted to significantly by both surface reflectance and aerosolreflectance. However, the magnitude of the surface reflec-tance differs greatly for vegetated and solid surfaces, suchas urban centers. Hence, at the red wavelength aerosol detec-tion over land is difficult because it is very sensitive to thetype of underlying surfaces. Over the ocean the aerosol reflec-tance is high compared to the water reflectance, and therefore

Figure 30-6 (a) True-color composite of SeaWiFS satellite data for July 15, 1999. (b) Surface reflectance in the absence of aerosols andclouds. (c) Derived aerosol optical depth, AOD. (d–f) Measured apparent reflectance at 412-, 670-, and 870-nm wavelength, respectively,showing increasing atmospheric transparency at higher wavelengths. (See color insert.)

674 SATELLITE-BASED MEASUREMENT OF ATMOSPHERIC AEROSOLS

quantitative aerosol detection is possible. The near-IR wave-length (865 nm), Figure 30-6f, is uniquely suitable for aerosoldetection over the ocean since the ocean reflectance isbelow 1%. Hence, virtually all the apparent reflectance isdue to aerosol scattering. Over land, the reflectance of bothvegetated and solid surfaces is high and comparable to theaerosol reflectance P, that is, P/R0 � 1 and aerosol detectionis limited.

30.5 CHALLENGES OF SATELLITE AEROSOLMEASURING SYSTEMS

Practical satellite aerosol measurements face a number ofadditional challenges beyond the dynamic and multidimen-sional aerosol system. Clouds obscure the detection of bothunderlying surfaces and atmospheric aerosols. The areas deli-neated by a “cloud mask” are unavailable for aerosol detec-tion. Since large portions of the tropics and northernlatitudes are cloudy throughout the year, cloudiness is amajor limitation of satellite remote sensing of both aerosolsand surfaces. The cloud edges are not easily discernible, par-ticularly during humid conditions and hygroscopic (e.g., sul-fate) haze. Thin cirrus clouds are also difficult to identify.Consequently, improper cloud masks are a source of errorin satellite-derived aerosol measurements. Cloud shadowson the earth’s surface also constitute a complication foraerosol retrieval.

Air or Rayleigh scattering of blue light always contributesto the radiation detected by the satellite sensor. The magni-tude of air scattering can be corrected by rigorous calculationsbased on the known optical properties of air molecules,the elevation of the surfaces, and by application of well-established radiative transfer code, such as that offered byVermote et al. (2002). Absorption by stratospheric ozoneand tropospheric water vapor also attenuates selected bandsin the solar spectrum passing through the atmosphere. Thestratospheric ozone layer has an interfering absorption bandin the visible range (0.52–0.74 mm). Whenever possible,the satellite sensor wavelength bands are located in trans-mission windows, that is, at wavelengths away from majormolecular absorption bands.

From the perspective of aerosol detection, the highly tex-tured and colorful reflection from the earth’s surface constitu-tes a significant background noise that needs to be dealt with.The surface reflectance, that is, the fraction of the incomingsolar radiation reflected by a surface, depends on many par-ameters most notably on the nature of the surface itself andthe geometric and spectral distribution of incoming andreflected radiation. Ocean surfaces reflect the incoming radi-ation, mostly by mirror-like, specular reflection. The areas of“sun glint” can be identified from the sun-surface-sensorgeometry and can be masked out just like clouds (Khanet al. 2007). Ocean waves tend to broaden the angular

spread of the specular reflection and may also add spuriousreflectance from foaming whitecaps. Solid land surfaces aremostly diffuse Lambertian reflectors, with minor specularcomponents. Vegetation is a nearly diffuse Lambertian reflec-tor but exhibits an extra reflection “hot spot” back toward thesun. Over land, the surface reflection is defined by the bidir-ectional diffuse reflectance function (BDRF; Roujean et al.1992). It depends on the angle of the incoming radiationand the angle of reflected radiation as well as on wavelength.Vegetation reflectance has a peak in the green (0.5 mm) andrises sharply to over 40% in the near IR. Typical soil orrock reflectance gradually increases from about 10–15% inthe blue to over 40% in the near IR. The spectral reflectanceof water is low in the visible range and vanishes in the nearIR, making it practically black. The spectral reflectance ofmany surfaces varies with season. Unfortunately, the apriori quantification of the spectral BDRF is not availablefor the ever-changing earth surfaces at all locations and alltimes. This necessitates the extraction of key features ofBDRFs from the “dirty” satellite images using spectralreflectance values for the same pixel on different observationconditions (e.g., Raffuse 2003).

The performance of satellite aerosol sensors and theirrespective retrieval algorithms are evaluated primarily againstobservations of optical thickness data provided by theAERONET (Holben et al. 1998) sun photometer network.This federated network has strategically located sites thatcharacterize typical aerosol regimes such as dusty desertlocations, biomass smoke areas, marine background, aswell as urban industrial sites with complex aerosol mixtures.The validation results appear to confirm the anticipatedretrieval accuracy of MODIS and MISR products (Abdouet al. 2005; Kahn et al. 2005, 2007; Remer et al. 2008).However, a direct comparison of MODIS and MISR aerosoldata show significant disagreements at the pixel level aswell as between long-term and spatially averaged aerosolproperties, particularly over land (Liu and Mishchenko2008).

30.6 SATELLITE DATA AND INFORMATIONSYSTEMS

Earth observing satellites are by far the most prolific sourcesof environmental monitoring data. Literally terabytes ofdigital data are collected, transmitted, and processed daily.An integral part of earth observing satellite systems is thedata flow and processing network, that is, the informationsystem that distributes the relevant processed data to theend users. Virtually all the raw and processed satelliteproducts are publicly accessible from their respectiveagencies. NASA, NOAA, and other agencies have a richvariety of data products relevant to atmospheric aerosols. Infact, atmospheric particulate matter, that is, dust, smoke, and

30.6 SATELLITE DATA AND INFORMATION SYSTEMS 675

haze are represented prominently (about 15%) in the NASAproduct showcases, like the Earth Observatory. Most ofthese aerosol products are images from the MODIS sensoron the TERRA and AQUA satellites. Other relevant datainclude AERONET federated sun photometer data, spaceLIDAR data from the PICASSO active LIDAR sensor,and the OMI data from the joint NASA-European SpaceAgency program.

Recent developments in surface and satellite sensing alongwith new information technologies now allow real-time,“just-in-time” data analysis for the characterization and par-tial explanation of major air pollution events as well asmore in-depth postanalysis (Husar and Poirot 2005).However, for the users of satellite data, the new developmentsintroduced a new set of problems. The “data deluge” problemis especially acute for analysts interested in aerosol pollution,since the aerosol processes are inherently complex, thenumerous relevant data range from detailed surface-basedchemical measurements to extensive satellite remote sensingand the integration of these requires the use of sophis-ticated data integration and modeling science methods.Unfortunately, neither the current science nor the tools andmethods are adequate for such sophisticated data integration.As a consequence satellite-derived earth observations areseverely underutilized in making societal decisions. Aremedy is anticipated from the Global Earth ObservationSystem of Systems (GEOSS), (GEOSS 2005) an emergingpublic information infrastructure for finding, accessing, andapplying diverse earth observations that are useful for scienceand decision makers.

30.7 APPLICATIONS

There is increasing demand for satellite aerosol measure-ments in earth science, air quality management, and disastermanagement. Satellite-derived global aerosol climatologieshave improved our general understanding of global biogeo-chemical processes of sulfur, organics, mineral dust, andother substances that are vital to the earth system. Verticalaerosol column observations are key to the characterizationof the atmospheric aerosols, since most of the aerosol massas well as the atmospheric processes occur aloft. Surface-based measurements characterize only a thin horizontal aero-sol layer. Satellite observations made it possible to identifyand derive global-scale aerosol source regions (e.g.,Prospero et al. 2002), global-scale transport patterns (e.g.,Husar et al. 1997), and also crude estimates of atmosphericaerosol residence times. These observation-based emissionrates and atmospheric residence times, in turn, are improvingthe performance of global and regional chemical transportand forecast models.

With increasing concern about human-induced climatechange, quantitative global-scale monitoring of radiatively

active atmospheric aerosols has become a major goal.While aerosol back-scattering to space tends to reduce theincoming solar radiation, aerosol absorption contributes tothe heating of the atmosphere. However, the magnitude ofthese perturbations and the relative contribution of thehuman-induced aerosol burden to heating and cooling effectsis not well understood. Unfortunately, the current satellite-derived aerosol measurements are too uncertain to authorita-tively resolve these subtle aerosol effects (Mishchenko et al.2009).

Air quality management is a recent application area of sat-ellite aerosol measurements. In the past, air quality regu-lations for aerosols relied on surface-based monitoringnetworks to enforce the air quality standards. More recentlythe U.S. Environmental Protection Agency’s (USEPA)Exceptional Event (EE) Rule (Federal Register 2008) expli-citly encourages the use of satellites for the documentationof aerosol contributions that originate outside the jurisdictionof air quality control agencies. In fact, compelling satelliteobservations of smoke and dust events prompted the intro-duction of the EE Rule. Figure 30-7 is an example of anexceptional smoke event where biomass smoke from forestand agricultural fires in southern Mexico and Guatemalawas transported across the entire eastern United States andcaused record surface aerosol concentrations through mostof the transport path. The spatial pattern of the bluishsmoke plume and white clouds for May 15–16, 1998, wasderived from the SeaWiFS satellite data. The semiquantitativeabsorbing aerosol index measured by the TOMS satellitesensor (green overlay) indicates that the aerosol is light-absorbing smoke, not sulfate haze. The aerosol extinctioncoefficient (red contour lines), derived from surface visibilityobservations, indicates that the smoke was reaching theground and impacts human health. Such combination ofmulti-sensory satellite observations along with surface moni-toring data and diagnostic transport models provides theevidence that the pollution event was due to uncontrollable,extra-jurisdictional causes and not from urban-industrialsources. Similar integration of satellite and surface obser-vations and models have elucidated the intercontinentaltransport of windblown dust from the East Asian GobiDesert and its impact on the west coast of North America(Husar et al. 2001).

Satellite aerosol observations are expected to continueto contribute to several aspects of air quality management,including:(1) providing direct observational evidence ofregional and long-range aerosol transport; (2) helpingimprove emission inventories and tracking emission trends;(3) evaluating air quality models; and (4) complementingsurface networks through filling of spatial-temporal gaps.However, before satellite aerosol observations can reachthe “promised land” (Hoff and Christopher 2009), it willbe necessary to better understand their measurementlimitations (Hidy et al. 2009) and to develop science-based

676 SATELLITE-BASED MEASUREMENT OF ATMOSPHERIC AEROSOLS

methods of integrating satellite data with surface obser-vations, emission inventories, and chemical transportmodels (NRC 2010).

30.8 FUTURE DEVELOPMENTS

Quantitative satellite measurements of aerosol properties areconstrained by inherent physical and mathematical limit-ations. However, both the scientific understanding and thetools and methods of satellite aerosol sensing are maturing.Therefore, it may be appropriate to express desirable develop-ments as perceived by this writer and the need for moreinterdisciplinary collaboration. Much of the knowledge on“aerosol models,” including aerosol size distributiondynamics, chemical composition, and optical properties, hasbeen generated through atmospheric aerosol and visibilityresearch and is directly applicable to the improvement of satel-lite aerosol detection. Conversely, satellite aerosol monitoringcan significantly enhance the understanding of atmosphericaerosols, including emissions, transport, and spatio-temporalpatterns. Thus, dynamic models for different aerosol typescould be co-developed by aerosol scientists and remote-sensing specialists. More robust and continually updatedsurface reflectance models could help aerosol retrieval as

well as surface characterization. Iterative co-retrieval of aero-sol and surface properties could be one possible approach.

Future aerosol retrievals could utilize complementary andredundant information from multiple sensors, that is, throughpixel-level fusion of multiple sensors’ data. Feature-level datafusion, for example, TOMS/OMI data with MODIS/MISR,could strengthen aerosol type detection. Fusion of geosta-tionary and polar satellite data could yield a better chara-cterization of aerosol dynamics throughout the day. Anopportunity for the integration of multiple sensors is beingpursued by the “A-Train,” a synchronized constellation ofeight satellites that fly on near-identical orbits as a pack(Stephens et al. 2002).

A more difficult issue is: What comprehensive, integratedsystem is needed for improving air quality management usingsatellite and ground-based observations along with evolvingmodeling and emission inventories (Hidy et al. 2009).Lastly, the concept of “aerosol retrieval” could well beexpanded beyond the explicit use of the optical aerosol sig-nal processing arising from single instruments. A more com-plete effort that is directed toward the full characterization ofthe entire eight-dimensional aerosol system could producedramatic improvements in our understanding and also providebenefits to many application areas, such as air quality andhuman health, climate change, and disaster management.

Figure 30-7 Mexican forest fire smoke plume transport over eastern United States observed through the SeaWiFS and TOMS satellite sensorsand surface visibility. (a) May 15, 1998; (b) May 16, 1998. (See color insert.)

30.8 FUTURE DEVELOPMENTS 677

30.9 ACKNOWLEDGMENTSQ4

This work was partially supported by the NASA GrantsNNL05AA00A and NNX08BA33G as well as USEPAGrant XA83228301. Erin Robinson and Janja Husar wereinstrumental in the preparation of the manuscript and theirhelp is greatly appreciated.

30.10 LIST OF ACRONYMS

AERONET Aerosol robotic network

AOD Aerosol optical depth

AQUA Earth-observing NASA satellite

AVHRR Advanced very high resolution radiometer

BDRF Bidirectional diffuse reflectance function

USEPA U.S. Environmental Protection Agency

GEOSS Global Earth Observation System of Systems

IR Infrared

MODIS Moderate resolution imaging spectroradi-ometer

MISR Multiangle imaging spectroradiometer

NASA National Aeronautics and Space adminis-tration

NOAA National Oceanic and AtmosphericAdministration

OMI Ozone monitoring instrument

PICASSO Pathfinder instrument for cloud and aerosolspaceborne observations

TERRA Earth-observing NASA satellite

TOMS Total ozone mapping spectrometer

SeaWiFS Sea-viewing wide field-of-view sensor

30.11 REFERENCES

Abdou, W. A., D. J. Diner, J. V. Martonchik, C. J. Bruegge,R. A. Kahn, B. J. Gaitley, K. A. Crean, L. A. Remer, andB. Holben. 2005. Comparison of coincident MultiangleImaging Spectroradiometer and Moderate Resolution ImagingSpectroradiometer aerosol optical depths over land and oceanscenes containing Aerosol Robotic Network sites. J. Geophys.Res. 110: D10S07.

Barteneva, O. D. 1960. Scattering functions of light in theatmospheric boundary layer. Izv. Akad. Nauk SSR, Ser. Geofiz.:1852–1865.Q5

Carlson, T. N., and J. M. Prospero. 1972. The large-scale movementof Saharan air outbreaks over the northern equatorial Atlantic.J. Appl. Meteorol. 11:283–297.

Deuze, J. L., F. M. Breon, C. Devaux, P. Goloub, M. Herman,B. Lafrance, F. Maignan, A. Marchand, F. Nadal, and G. Perry.2001. Remote sensing of aerosols over land surfaces from

POLDER-ADEOS-1 polarized measurements. J. Geophys. Res.106(D5):4913–4926.

Diner, D. J., J. C. Beckert, T. H. Reilly, C. J. Bruegge, J. E. Conel,R. A. Kahn, and J. V. Martonchik. 1998. Multi-angle ImagingSpectroRadiometer (MISR) instrument description and exper-iment overview. IEEE Trans. Geosci. Remote Sens. 36:1072–1087.

Dubovik, O., and M. D. King. 2000. A flexible inversion algorithmfor retrieval of aerosol optical properties from sun and sky radi-ance measurements. J. Geophys. Res. 105(D16):20673–20696.

Dubovik, O., B. Holben, T. F. Eck, A. Smirnov, Y. J. Kaufman,M. D. King, D. Tanre, and I. Slutsker. 2002. Variability ofabsorption and optical properties of key aerosol types observedin worldwide locations. J. Atmos. Sci. 59:590–608.

Federal Register. 2008. Rules: Revised Exceptional Event DataFlagging Submittal and Documentation Schedule to FederalRegister [FR Doc E8-29747] [40 CFR Part 50]. http://www.thefederalregister.com/d.p/2008-12-16-E8-29747

Fraser, R. S. 1976. Satellite measurement of mass of Sahara dust inthe atmosphere. Appl. Opt. 15:2471–2479.

Fraser, R. S., Y. J. Kaufman, and R. L. Mahoney. 1984. Satellitemeasurements of aerosol mass and transport. Atmos. Environ.18:2577–2584.

Friedlander, S. K. 1970. The characterization of aerosols distributedwith respect to size and chemical composition. J. Aerosol Sci.1:295–307.

Friedlander, S. K. 2000. Smoke, Dust, and Haze: Fundamentals ofAerosol Dynamics. Oxford University Press, New York.

GEOSS 2005. Group on earth Observations j Home. http://www.earthobservations.org/ Q14

Gordon, H. R., and M. Wang. 1994. Retrieval of water-leaving radi-ance and aerosol optical thickness over the oceans withSeaWiFS: a preliminary algorithm. Appl. Opt. 33:443–452.

Griggs, M. 1975. Measurements of atmospheric aerosol opticalthickness over water using ERTS-1 data. J. Air Pollut. Contr.Assoc. 25:622.

Hidy, G. M., J. R. Brook, J. C. Chow, M. Green, R. B. Husar, C. Lee,R. D. Scheffe, A. Swanson, and J. G. Watson. 2009. Remotesensing of particulate pollution from space: Have we reachedthe promised land? J. Air Waste Manage. Assoc. 59:1130–1139. Q6

Herman, J. R., P. K. Bhartia, O. Torres, C. Hsu, C. Seftor, andE. Celarier. 1997. Global distribution of UV-absorbingaerosols from Nimbus 7/TOMS data. J. Geophys. Res.102(D14):16911–16922.

Hoff, R. M., and S. A. Christopher. 2009. Remote sensing of par-ticulate pollution from space: Have we reached the promisedland? J. Air Waste Manage. Assoc. 59:645–675.

Holben, B. N., T. F. Eck, I. Slutsker, D. Tanre, J. P. Buis, A. Setzer,E. Vermote, J. A. Reagan, Y. J. Kaufman, and T. Nakajima.1998. AERONET-A Federated Instrument Network and DataArchive for Aerosol Characterization – an overview. RemoteSensing of Environment 66:1–16.

Hsu, N. C., S. Tsay, M. D. King, and J. R. Herman. 2006. Deep blueretrievals of Asian aerosol properties during ACE-Asia. IEEETrans. Geosci. Remote Sens. 44:3180.

678 SATELLITE-BASED MEASUREMENT OF ATMOSPHERIC AEROSOLS

Husar, R. B., and W. H. White. 1976. On the color of the LosAngeles smog. Atmos. Environ. 10:199–204.

Husar, R. B., J. M. Prospero, and L. L. Stowe. 1997.Characterization of tropospheric aerosols over the oceans withthe NOAA advanced very high resolution radiometer opticalthickness operational product. J. Geophys. Res. 102(D14):16889–16909.

Husar, R. B., D. M. Tratt, B. A. Schichtel, S. R. Falke, F. Li, D. Jaffe,S. Gasso, T. Gill, N. S. Laulainen, F. Lu, M. C. Reheis, Y. Chun,D. Westphal, B. N. Holben, C. Gueymard,1Q7 I. McKendry,N. Kuring, G. C. Feldman, C. McClain, R. J. Frouin,J. Merrill, D. DuBois, F. Vignola, T. Murayama, S. Nickovic,W. E. Wilson, K. Sassen, N. Sugimoto, and W. C. Malm..2001. Asian dust events of April 1998. J. Geophys. Res.106(D16):18317–18330.

Husar, R. B., and R. L. Poirot. 2005. DataFed and FASTNET:Tools for agile air quality analysis. Environ. Manag.(September):39–41.Q8

Johnson, R. W. 1981. Daytime visibility and nephelometer measure-ments related to its determination. Atmos. Environ.15:1835–1845.

Kahn, R., P. Banerjee, and D. McDonald. 2001. Sensitivity ofmultiangle imaging to natural mixtures of aerosols over ocean.J. Geophys. Res. 106(D16):18219–18238.

Kahn, R. A., B. J. Gaitley, J. V. Martonchik, D. J. Diner, K.A. Crean, and B. Holben. 2005. Multiangle ImagingSpectroradiometer (MISR) global aerosol optical depth vali-dation based on 2 years of coincident Aerosol RoboticNetwork (AERONET) observations. J. Geophys. Res.110(D10):S04.Q9

Kahn, R. A., M. J. Garay, D. L. Nelson, K. K. Yau, M. A. Bull,B. J. Gaitley, J. V. Martonchik, and R. C. Levy. 2007.Satellite-derived aerosol optical depth over dark water fromMISR and MODIS: Comparisons with AERONET and impli-cations for climatological studies. J. Geophys. Res.—Atmospheres 112(D18): 205.

Kaufman, Y. J. 1993. Aerosol optical thickness and atmosphericpath radiance. J. Geophys. Res. 98(D2):2677–2692.

Kiessling, J. 1884. Uber den Einfluß kunstlich erzeugter Nebel aufdirektes Sonnenlich. Meteorol. Zeitschr. 117(1):117–126.

King, M. D., Y. J. Kaufman, W. P. Menzel, D. Tanre, NGSF Center,and M. D. Greenbelt. 1992. Remote sensing of cloud, aerosol,and water vapor properties from the moderate resolution imagingspectrometer (MODIS). IEEE Trans. Geosci. Remote Sens.30:2–27.

King, M. D., Y. J. Kaufman, D. Tanre, and T. Nakajima. 1999.Remote sensing of tropospheric aerosols from space: Past,present, and future. Bull. Am. Meteorol. Soc. 80:2229–2259.

Liu, L., and M. I. Mishchenko. 2008. Toward unified satellite clima-tology of aerosol properties: Direct comparisons of advancedlevel 2 aerosol products. J. Quant. Spectrosc. RadiativeTransfer 109:2376–2385.

Lyons, W. A., and R. B. Husar. 1976. SMS/GOES Visible imagesdetect a synoptic-scale air pollution episode. Monthly WeatherReview 104:1623–1626.

Malm, W. C., and J. L. Hand. 2007. An examination of the physicaland optical properties of aerosols collected in the IMPROVEprogram. Atmos. Environ. 41:3407–3427.

McMurry, P. H., X. Zhang, and C. T. Lee. 1996. Issues in aerosolmeasurement for optics assessments. J. Geophys. Res.101(D14):19189–19197.

McMurry, P. H. 2000. A review of atmospheric aerosol measure-ments. Atmos. Environ. 34:1959–1999.

Mishchenko, M. I., I. V. Geogdzhayev, L. Liu, A. A. Lacis,B. Cairns, and L. D. Travis. 2009. Toward unified satelliteclimatology of aerosol properties: What do fully compatibleMODIS and MISR aerosol pixels tell us? J. Quant. Spectrosc.Radiative Transfer 110:402–408.

NRC. 2010. Global Sources of Local Pollution: An Assessment ofLong-Range Transport of Key Air Pollutants To and From theUnited States. National Research Council, National AcademyPress.

Parmenter, F. C. 1971. Picture of the month: Smoke from slashburning operations. Monthly Weather Review 99:684–685.

Penndorf, R. B. 1957. New Tables of total Mie scatteringCoefficients for spherical particles of real refractive indexes(1.33� 1.50). J. Opt. Soc. Am, 47:1010–1014.

Prospero, J. M., P. Ginoux, O. Torres, S. E. Nicholson, and T.E. Gill. 2002. Environmental characterization of global sourcesof atmospheric soil dust identified with the NIMBUS 7 TotalOzone Mapping Spectrometer (TOMS) absorbing aerosolproduct. Rev. Geophysics 40(1):2–1. Q10

Raffuse, S. M. 2003 Estimation of daily surface reflectance over theUnited States from the SeaWiFS sensor. Masters Thesis,Washington University, St. Louis, MO.

Randerson, D. 1968. A study of air pollution sources as viewed byearth satellites. J. Air Pollut. Control Assoc. 18:249–254.

Remer, L. A., and Y. J. Kaufman. 1998. Dynamic aerosolmodel: Urban/industrial aerosol. J. Geophys. Res. 103(D12):13859–13872.

Remer, L. A., R. G. Kleidman, R. C. Levy, Y. J. Kaufman, D. Tanre,S. Mattoo, J. V. Martins, C. Ichoku, I. Koren, and H. Yu. 2008.Global aerosol climatology from the MODIS satellite sensors.J. Geophys. Res. 113(D14):S07. Q9

Roujean, J. L., M. Leroy, and P. Y. Deschanps. 1992. A bidirectionalreflectance model of the Earth’s surface for the correction ofremote sensing data. J. Geophys. Res. 97(D18):20455–20468.

Sakunov, G. G., D. Barteneva, and V. F. Radionov. 1996.Experimental studies of light scattering phase functions ofthe atmospheric surface layer. Atmos. Oceanic Physics31:750–758.

Seinfeld, J. H., and S. N. Pandis. 1998. Atmospheric Chemistry andPhysics: From Air Pollution To Climate Change. Wiley-Interscience, New York.

Stephens, G. L., D. G. Vane, R. J. Boain, G. G. Mace, K. Sassen,Z. Wang, A. J. Illingworth, E. J. O’Connor, W. B. Rossow,and S. L. Durden. 2002. The Cloudsat Mission and the A-Train. Bull. Am. Meteorol. Soc. 83:1771–1790.

Stowe, L. L. 1991. Cloud and aerosol products at NOAA/NESDIS.Global and Planetary Change GPCHE 4, 4, no. 1/3. Q11

30.11 REFERENCES 679

Vermote, E. F., N. Z. El Saleous, and C. O. Justice.2002. Atmospheric correction of MODIS data in thevisible to middle infrared: first results. Rem. Sens. Environ.83:97–111.

Watson, J. G. 2002. Visibility: Science and regulation. J. Air WasteManag. Assoc. 52:628–713.

Wegener, A. L. 1911. Thermodynamic der Atmosphere. JohannAmbrosius Barth, Leipzig.

Whitby, K. T., R. B. Husar, and B. Y. H. Liu. 1972. The aerosol sizedistribution of Los Angeles smog. J. Coll. Interf. Sci.39:177–204.

Whitby, K. T. 1978. The physical characteristics of sulfur aerosols.Atmos. Environ. (1967) 12:135–159. Q12

White, W. H. 1986. On the theoretical and empirical basis forapportioning extinction by aerosols: A critical review. Atmos.Environ. (1967) 20:1659–1672. Q12

680 SATELLITE-BASED MEASUREMENT OF ATMOSPHERIC AEROSOLS