A workshop on the "Passive Remote Sensing of Tropospheric Aerosols and Atmospheric Correction for the Aerosol Effect" was conducted in Washington, DC, April 15-19, 1996. The interest in tropospheric aerosols (liquid and solid particles suspended in the air) was resurrected recently when climate modelers indicated that tropospheric aerosols generate the main uncertainty in anthropogenic forcing considered in predicting climate change, twice as large as the uncertainty in the greenhouse warming. Aerosols may counteract a large part (or most) of the present, globally-averaged, greenhouse forcing; but their regional rather than global scale is expected to introduce even more important climate effects, from cooling in the North Atlantic region to possible reduction of atmospheric mixing in the tropics. Aerosols are also increasingly important in understanding atmospheric chemistry, being sinks to atmospheric species and surfaces for fast chemical reactions. Aerosols are considered the main long-range transport mechanism for nutrients between continents and between continents and oceans. They fertilize the Amazon Basin, generate the red top soil in Bermuda, and are the main source of iron for oceanic phytoplankton. Aerosols serve as an indicator of the presence of air pollution, reflect the magnitude of biomass burning, and are involved in acid depositions. Aerosol interaction with solar radiation inhibits observations of the Earth's surface, including oceanic and land productivity.

The workshop brought together, in a highly scientifically stimulating but socially relaxing atmosphere, most of the U.S. and international experts on remote sensing of aerosols and of atmospheric corrections, that are presently responsible for the development of algorithms for the new satellite systems: EOS (MODIS, MISR, and EOSP), ADEOS (POLDER, OCTS, and GLI), MERIS, the new AVHRR, and others. The workshop, combining 30-minutes presentations and 2 hours of discussions each day, reviewed and intercompared the physical principles used by the different algorithms, in order to stimulate critical discussions aimed at understanding the differences between the algorithms. Such discussions can foster the generation of improved algorithms for the individual satellite sensors and generate collaborations on algorithms that will use the data from several sensors simultaneously. In order to share the results of the workshop with a wider community, a special issue in the Journal of Geophysical Research (JGR) will be devoted to up to 26 papers. A discussion paper that summarizes the 5 discussions in the workshop is planned for the special issue. An introduction was written by Prof. Jacqueline Lenoble, a long-time expert in the field and founder of the Department for Atmospheric Optics in Lille that fostered many of the scientists who participated in the workshop.

The workshop was organized by Yoram Kaufman (NASA/GSFC) from the MODIS Science Team (presently also the AM project scientist), Didier Tanré (University of Lille) from the MODIS and POLDER Science Teams, Teryuyki Nakajima (University of Tokyo) leader of the GLI Science Team and a close associate of the MODIS team, Howard Gordon (University of Miami) from the MODIS and MISR Teams, and Michael King (NASA/GSFC) MODIS team member and EOS Senior Project Scientist. The workshop was considered to be very successful, probably due to its focused objectives and special format. The workshop was organized by a scientific steering committee, with key scientists representing a broad array of instruments and disciplines. A detailed agenda was written before the prospective attendees were contacted, resulting in a high response rate. The workshop included a mixture of presentations (30 minutes each, mostly in the morning) and discussions (2 hours every day) and was limited to 30 presenters, and 50 total participants. The discussions were led by prominent scientists, most of them with no direct role in the new satellite systems. To achieve a friendly and relaxed social atmosphere we all stayed in one hotel in the best part of the city, had social events, and long 2-4 hour lunch breaks. We had excellent technical support provided by Jorge-Scientific and Applied Research Corporation (ARC).

Presentations invited for the workshop by the steering committee included 5 background papers, 25 papers related to the algorithms being developed, and 2 general talks on the aerosol effect on climate. The first 4 days of the workshop were devoted, respectively, to the 4 main topics: remote sensing of aerosols over land, remote sensing of aerosols over the ocean, atmospheric corrections for the aerosol effect over land and ocean, and evaluation of the remote sensing data using ground-based and airborne measurements. In each day, 4-8 papers on the subject were presented, followed by a two-hour discussion. Friday was devoted to discussions only and a summary. The following is a summary of the activity in the workshop.

MONDAY -- Remote sensing of aerosols over the land

The workshop started with 4 background papers. J. Prospero talked about long-term ground-based monitoring of aerosol physical and chemical properties by the Atmosphere/Ocean Chemistry Experiment (AEROCE) island network. He emphasized the role of dust in the aerosol forcing, including its anthropogenic component due to land use change. Thirty years of AEROCE monitoring shows a systematic increase of the dust deposition in Barbados, associated with expansion of land use in the Sahel and reduction in the rainfall index. The combination of the AEROCE data with the NOAA aerosol product from AVHRR indicates that half of the dust originates from arid regions with land disturbance, rather than from deserts (see articles in Nature April 4, 1996). The ground-based measurements are used to separate the scattering coefficient (an indicator of the aerosol scattering of radiation and backscattering of solar light to space) into the contribution from several components. P. Koepke discussed a global data set of climatology of the aerosol microphysical data and the corresponding optical properties. M. King presented an overview of the new satellite systems that will be used to monitor aerosols. These followed a talk by J. Penner on the aerosol characterization that is used in estimating aerosol effects on the global climate. This includes sources, budgets, chemical transformations, and interaction with radiation. The main difficulty is in estimating the aerosol forcing on climate through aerosol effects on cloud microphysics and albedo (indirect effect). This is difficult due to a feedback effect of cloud maximum saturation to the aerosol concentration and the nonlinear and complex relationship between the aerosol mass and the number of cloud condensation nuclei (CCNs) it can produce, due to variations in the aerosol size distribution. The aerosol indirect effect depends also on the concentration of preexisting aerosol particles that is difficult to estimate. The simplifications used in present climate models were emphasized. It is concluded that the complex aerosol problem can be seriously addressed only by a combination of long-term monitoring from space-based and ground-based platforms accompanied by monitoring the properties of the aerosolvertical profile and by extensive field campaigns.

Six talks on remote sensing of aerosols over the land started in the afternoon with presentations for EOS--MODIS, MISR, and EOSP; and ADEOS--POLDER and TOMS. These satellite systems differ in their spectral, angular, and spatial monitoring of aerosols. In the talks and the discussion that followed, headed by H. Grassl, the advantages and limitations of satellite remote sensing were indicated. Over land, satellites can determine the spatial distribution of the aerosol optical thickness with accuracy and coverage that depend on the sensor characteristics, from an error in the optical thickness of 0.05-0.10 in the solar channels to 0.01 in the IR channels. The aerosol size distribution cannot be retrieved, though some information on the aerosol type is present. Polarization measurements may be useful for determining the aerosol refractive index or shape (sphericity). The satellite data are evaluated and supplemented by measurements of the spectral optical thickness and size distribution from a network of sun/sky radiometers and by chemical measurements in order to distinguish between the forcing of different aerosol species and sources. Comparisons with aerosol transport and evolution models are also being planned. Field experiments with aircraft sampling and lidar systems are needed to evaluate the vertical structure of the aerosol layers. Many problems were discussed, including: the effect of particle non-sphericity on the satellite analysis; the possibility of deriving the aerosol radiative forcing directly from the measured radiances rather than from the derived optical thickness; and the need for reporting on and consistency between the assumptions used in deriving the aerosol optical thickness and the later use of them in climate models. The satellite data can give only partial information on the vertical distribution. The use of the water vapor band at 1.375 um or the use of oxygen absorption bands can distinguish between tropospheric and stratospheric aerosols. There is a need for a satellite lidar system (GLAS planned for EOS) and ground-based monitoring of aerosols with lidars. The only information on absorption given today by satellite data is from the absorption difference in TOMS UV channels. There is a need to develop operational methods for estimation of the aerosol absorption from satellites and from ground-based measurements. Some experimental methods are available.

TUESDAY -- Remote sensing of aerosols over the ocean

Tuesday we started with 7 papers describing techniques for operational remote sensing of aerosols over the ocean from EOS--MODIS and MISR (an overall talk on EOSP was given the previous day); ADEOS--GLI, OCTS, and POLDER; and enhanced AVHRR, SeaWiFS, and MERIS. They were followed by a paper on laboratory measurements of aerosols and a discussion headed by B. Herman of the remote sensing techniques. Over the oceans, the dark and more predictable water surface reflectance allows the determination of additional aerosol parameters. The aerosol optical thickness is derived with higher accuracy (an error of 0.01-0.05). Information on the aerosol model or the particle size can be obtained from most systems, and refractive index from polarization measurements. Many of the problems mentioned in regard to remote sensing over the land were discussed in the context of measurements over the ocean. Even though the satellite aerosol information is much more accurate and informative over the ocean than over the land, there is a need for an island-based network of sun/sky radiance measurements to derive the detailed aerosol properties, in collaboration with aerosol chemical measurements and lidar measurements of the aerosol vertical distribution.

The assumptions used in the inversion of the satellite data are used differently by the algorithm developers. There was a discussion about unifying the assumptions. The conclusion, suggested by H. Grassl, was that since it is not possible before launch to determine what is the optimum type of analysis, and since the need for assumptions varies from sensor to sensor, any unification of the assumptions should be postponed for data reanalysis a few years after launch.

WEDNESDAY -- Atmospheric corrections over land and ocean

Three papers on atmospheric corrections over land were given for MODIS and for POLDER. A discussion of corrections for MISR was given the first day as part of the derivation of the aerosol optical thickness. Derivation of the surface BRDF was stressed in the last talk and in the other talks. Over-the-ocean-correction discussion included 5 papers which were discussed in the framework of MODIS, MISR, MERIS, OCTS, and POLDER. Two parallel discussions followed, one for the ocean and one for the land.

Atmospheric corrections over the land use two pathways. In the first (e.g., MODIS), the aerosol optical thickness is derived and used with the proper aerosol model for the correction, using surface BRDF properties derived from recently corrected satellite data. In the second, preferred by the multi-view MISR instrument, the surface BRDF properties and the atmospheric corrections are derived simultaneously with remote sensing of aerosols. We discussed the problem of generation of aerosol models used in the correction, and the methods for supplementing aerosol data missing from satellite remote sensing with aerosol climatology.

Atmospheric corrections over the ocean are difficult, due to the large (10 times) contamination of the signal by the atmospheric aerosol. The correction works to overcome this difficulty by using the difference in the spectral properties of the ocean surface (almost black in the near IR) and of the aerosol (monotonously decreasing optical thickness with wavelength, with a decrease rate that depends on the particle size distribution). In addition to spectral properties of the water, uncertainty in the prediction of the far glint effects and of foam spectral effects makes the analysis even more difficult. But previous success using validated retrievals from the CZCS and the improved spectral and calibration capability of the new satellite systems makes a workable solution a possibility. Two basic procedures for the inversion of the satellite data were suggested. For typical scanning radiometers (MODIS, SeaWiFS, OCTS, and MERIS) most procedures use the near-IR radiances to find the aerosol loading and model, using look-up tables to extrapolate this information to the visible channels, and then remove the aerosol effect from the satellite data. For sensors with multi-angle capability (POLDER and MISR) this additional information is incorporated in the near IR to better constrain the aerosol model.

Thursday -- Evaluation of the remote sensing methods

The day started with two general talks on the measurement techniques. F. Valero discussed aircraft measurements, emphasizing that the vertical variation of direct and diffuse radiative fluxes, as influenced by the presence of aerosols, can be used directly to relate the aerosol optical thickness, derived from space, with the aerosol direct radiative forcing of climate. J. Reagan followed with discussion of ground-based instrumentation, emphasizing the need and availability of inexpensive pulse lidar systems to study the vertical distribution of aerosols and their optical properties. The validation session included 4 papers: 2 for validation of remote sensing of aerosols and 2 for validation of atmospheric corrections over the ocean. No presentations were made for validation of the atmospheric correction over the land, though plans to do so exist. The validation of the remote sensing of aerosols is based on ground-based sun/sky radiometers, including measurements of polarization. Sky measurements of radiance and polarization are similar in their physical sense to satellite measurements of upwelling radiation. The ground-based measurements have the advantage of observations on a background of black space (not ocean or land reflectance or polarization) and of measurements in a wide range of scattering angles (2-150deg.) instead of the narrower range from satellites (40-180deg. in the best conditions and 100-160deg. in average conditions). The physical similarity and advantages of the ground-based instruments make them the prime source of evaluation of the satellite data. The ground-based instruments can also measure the incident solar radiation, thus allowing the direct derivation of the spectral optical thickness and, by comparison with the analysis of the sky data, the aerosol single scattering albedo (a measure of aerosol absorption).

Validation of the correction over the ocean will also use sun/sky observations to determine the correctness of the aerosol model derived in the correction, but will concentrate on detailed measurements of the spectral water-leaving radiance.

Conclusions

The success of the workshop is probably due to the combination of a focused objective and a special format that promotes plenty of discussion, formal and informal, in a relaxed atmosphere. A special issue in the JGR will summarize the scientific material and discussion from the workshop for the general scientific community. The two main scientific conclusions resulting from the workshop are: (1) the present remote sensing strategy is weak in deriving the aerosol single scattering albedo, a measure of the aerosol absorption and a critical parameter in understanding the effects of aerosols on climate; and (2) the use of algorithms with different types of data and different assumptions is expected to be beneficial in improving aerosol remote sensing techniques.