D. Tanré

AERONET-a federated instrument network and data archive for aerosol Characterization

Abstract The concept and description of a remote sensing aerosol monitoring network initiated by NASA, developed to support NASA, CNES, and NASDA’s Earth satellite systems under the name AERONET and expanded by national and international collaboration, is described. Recent development of weather-resistant automatic sun and sky scanning spectral radiometers enable frequent measurements of atmospheric aerosol optical properties and precipitable water at remote sites. Transmission of automatic measurements via the geostationary satellites GOES and METEOSATS’ Data Collection Systems allows reception and processing in near real-time from approximately 75% of the Earth’s surface and with the expected addition of GMS, the coverage will increase to 90% in 1998. NASA developed a UNIX-based near real-time processing, display and analysis system providing internet access to the emerging global database. Information on the system is available on the project homepage, http://spamer.gsfc.nasa.gov . The philosophy of an open access database, centralized processing and a user-friendly graphical interface has contributed to the growth of international cooperation for ground-based aerosol monitoring and imposes a standardization for these measurements. The system’s automatic data acquisition, transmission, and processing facilitates aerosol characterization on local, regional, and global scales with applications to transport and radiation budget studies, radiative transfer-modeling and validation of satellite aerosol retrievals. This article discusses the operation and philosophy of the monitoring system, the precision and accuracy of the measuring radiometers, a brief description of the processing system, and access to the database.

The MODIS Aerosol Algorithm, Products, and Validation

The Moderate Resolution Imaging Spectroradiometer (MODIS) aboard both NASA’s Terra and Aqua satellites is making near-global daily observations of the earth in a wide spectral range (0.41–15 m). These measurements are used to derive spectral aerosol optical thickness and aerosol size parameters over both land and ocean. The aerosol products available over land include aerosol optical thickness at three visible wavelengths, a measure of the fraction of aerosol optical thickness attributed to the fine mode, and several derived parameters including reflected spectral solar flux at the top of the atmosphere. Over the ocean, the aerosol optical thickness is provided in seven wavelengths from 0.47 to 2.13 m. In addition, quantitative aerosol size information includes effective radius of the aerosol and quantitative fraction of optical thickness attributed to the fine mode. Spectral irradiance contributed by the aerosol, mass concentration, and number of cloud condensation nuclei round out the list of available aerosol products over the ocean. The spectral optical thickness and effective radius of the aerosol over the ocean are validated by comparison with two years of Aerosol Robotic Network (AERONET) data gleaned from 132 AERONET stations. Eight thousand MODIS aerosol retrievals collocated with AERONET measurements confirm that one standard deviation of MODIS optical thickness retrievals fall within the predicted uncertainty of 0.03 0.05 over ocean and 0.05 0.15 over land. Two hundred and seventy-one MODIS aerosol retrievals collocated with AERONET inversions at island and coastal sites suggest that one standard deviation of MODIS effective radius retrievals falls within reff 0.11 m. The accuracy of the MODIS retrievals suggests that the product can be used to help narrow the uncertainties associated with aerosol radiative forcing of global climate.

Operational remote sensing of tropospheric aerosol over land from EOS moderate resolution imaging spectroradiometer

Daily distribution of the aerosol optical thickness and columnar mass concentration will be derived over the continents, from the EOS moderate resolution imaging spectroradiometer (MODIS) using dark land targets. Dark land covers are mainly vegetated areas and dark soils observed in the red and blue channels; therefore the method will be limited to the moist parts of the continents (excluding water and ice cover). After the launch of MODIS the distribution of elevated aerosol concentrations, for example, biomass burning in the tropics or urban industrial aerosol in the midlatitudes, will be continuously monitored. The algorithm takes advantage of the MODIS wide spectral range and high spatial resolution and the strong spectral dependence of the aerosol opacity for most aerosol types that result in low optical thickness in the mid-IR (2.1 and 3.8 pm). The main steps of the algorithm are (1) identification of dark pixels in the mid-IR; (2) estimation of their reflectance at 0.47 and 0.66 pm; and (3) derivation of the optical thickness and mass concentration of the accumulation mode from the detected radiance. To differentiate between dust and aerosol dominated by accumulation mode particles, for example, smoke or sulfates, ratios of the aerosol path radiance at 0.47 and 0.66 pm are used. New dynamic aerosol models for biomass burning aerosol, dust and aerosol from industrial/urban origin, are used to determine the aerosol optical properties used in the algorithm. The error in the retrieved aerosol optical thicknesses, r,, is expected to be AT, = 0.05 5 0.27,. Daily values are stored on a resolution of 10 X 10 pixels (1 km nadir resolution). Weighted and gridded 8-day and monthly composites of the optical thickness, the aerosol mass concentration and spectral radiative forcing are generated for selected scattering angles to increase the accuracy. The daily aerosol information over land and oceans (Tunr& et al., this issue), combined with continuous aerosol remote sensing from the ground, will be used to study aerosol climatology, to monitor the sources and sinks of specific aerosol types, and to study the interaction of aerosol with water vapor and clouds and their radiative forcing of climate. The aerosol information will also be used for atmospheric corrections of remotely sensed surface reflectance. In this paper, examples of applications and validations are provided.

Remote sensing of aerosol properties over oceans using the MODIS/EOS spectral radiances

Spectral radiances measured at the top of the atmosphere in a wide spectral range (0.55–2.13 μm) are used to monitor the aerosol optical thickness and the aerosol size distribution (integrated on the vertical column) of the ambient (undisturbed) aerosol over the oceans. Even for the moderate resolution imaging spectrometer (MODIS) wide spectral range, only three parameters that describe the aerosol loading and size distribution can be retrieved. These three parameters are not always unique. For instance, the spectral radiance of an aerosol with a bilognormal size distribution can be simulated very well with a single lognormal aerosol with an appropriate mean radius and width of distribution. Preassumptions on the general structure of the size distribution are therefore required in the inversion of MODIS data. The retrieval of the aerosol properties is performed using lookup table computations. The volume size distribution in the lookup table is described with two lognormal modes: a single mode to describe the accumulation mode particles (radius 1.0 μm). Note that two accumulation modes may be present, one dominated by gas phase processes and a second dominated by cloud phase processes. The coarse mode can also be split into several partially overlapping modes describing maritime salt particles and dust. The aerosol parameters we expect to retrieve are η, the fractional contribution of the accumulation mode to scattering; τ, the spectral optical thickness; and rm, the mean particle size of the dominant mode. Additional radiative quantities such as asymmetry parameter and effective radius are derived subsequently. The impact of the surface conditions, wind speed and chlorophyll content on the retrieval is estimated, the impact of potential sources of error like the calibration of the instrument is also tested. The algorithm has been applied successfully to actual data sets provided by the Thematic Mapper on Landsat 5 and by the MODIS airborne simulator on the ER-2 and tested against ground and airborne measurements. A first estimate of the general accuracy is Δτ = ±0.05±0.05τ (at 550 nm), Δrm = 0.3rm, Δη = ±0.25.

Cloud and aerosol properties, precipitable water, and profiles of temperature and water vapor from MODIS

The Moderate Resolution Imaging Spectroradiometer (MODIS) is an Earth-viewing sensor that flies on the Earth Observing System Terra and Aqua satellites, launched in 1999 and 2002, respectively. MODIS scans a swath width of 2330 km that is sufficiently wide to provide nearly complete global coverage every two days from a polar-orbiting, Sun-synchronous, platform at an altitude of 705 km. MODIS provides images in 36 spectral bands between 0.415 and 14.235 /spl mu/m with spatial resolutions of 250 m (two bands), 500 m (five bands), and 1000 m (29 bands). These bands have been carefully selected to enable advanced studies of land, ocean, and atmospheric properties. Twenty-six bands are used to derive atmospheric properties such as cloud mask, atmospheric profiles, aerosol properties, total precipitable water, and cloud properties. We describe each of these atmospheric data products, including characteristics of each of these products such as file size, spatial resolution used in producing the product, and data availability.

Remote sensing of aerosols over land surfaces from POLDER‐ADEOS‐1 polarized measurements

The polarization measurements achieved by the POLDER instrument on ADEOS-1 are used for the remote sensing of aerosols over land surfaces. The key advantage of using polarized observations is their ability to systematically correct for the ground contribution whereas the classical approach using natural light fails. The estimation of land surface polarizing properties from POLDER has been examined in a previous paper. Here we consider how the optical thickness, δ 0 , and Angstrom exponent, α , of aerosols are derived from the polarized light backscattered by the particles. The inversion scheme is detailed and illustrative results are presented. Maps of the retrieved optical thickness allow for detection of large aerosol features and, in the case of small aerosols, the δ 0 and α retrievals are consistent with correlative ground-based measurements. However, because polarized light stems mainly from small particles, the results are biased for aerosol distributions containing coarser modes of particles. To overcome this limitation, an aerosol index defined as the product AI = δα 0 is proposed. Theoretical analysis and comparison with ground-based measurements suggest that AI is approximately the same when using δ 0 and α related to the entire aerosol size distribution or derived from the polarized light originating from the small polarizing particles alone. This invariance is specially assessed by testing the continuity of AI across coastlines, given the unbiased properties of aerosol retrieval over ocean. Although reducing the information concerning the aerosols, this single parameter allows a link between the POLDER aerosol surveys over land and ocean. POLDER aerosol index global maps enable the monitoring of major aerosol sources over continental areas.

Comparisons of the TOMS aerosol index with Sun‐photometer aerosol optical thickness: Results and applications

A nearly 20-year global data set (1979–1994 and 1996 to the present) of tropospheric absorbing aerosols has been developed from total ozone mapping spectrometer (TOMS) backscattered radiance measurements in the range from 331 to 380 nm. The occurrence of aerosols is derived directly from measured backscattered radiances and is represented by a quantity known as the aerosol index. Previous theoretical model simulations have demonstrated that the aerosol index depends on aerosol optical thickness (AOT), single scattering albedo, and aerosol height and that the AOT can be determined provided that the microphysical properties and height of aerosols are known. In this paper we show that the TOMS aerosol index measurements are linearly proportional to the AOT derived independently from ground-based Sun-photometer instruments over regions of biomass burning and regions covered by African dust. We also show how this linear relationship can be used to directly convert the aerosol index into AOT for smoke and dust aerosols for the regions near the Sun-photometer sites and how information about aerosol height can be inferred from the results. Finally, we apply this method to the TOMS data over the last two decades and find a significant increase in the amount of biomass burning smoke in the African savanna regions during the 1990s in addition to the more obvious increase in South America.

Climatology of dust aerosol size distribution and optical properties derived from remotely sensed data in the solar spectrum

Simultaneous spectral remote observations of dust properties from space and from the ground create a powerful tool for the determination of ambient dust properties integrated on the entire atmospheric column. The two measurement methods have a complementary sensitivity to variety of dust properties. The methodology is demonstrated using spectral measurements (0.47-2.21 mm) from Landsat TM over the bright Senegalian coast and dark ocean, and Aerosol Robotic Network (AERONET) radiances measured in several locations. We derive (1) the dust size distribution, showing a dominant coarse mode at 1-5 mm and a secondary mode around 0.5 mm effective radius; (2) dust absorption, which is found to be substantially smaller than reported from previous measurements; (3) the real part of the refractive index which varies within the range 1.53- 1.46; and we show that (4) the effect of the dust nonspherical shape on its optical properties is not significant for scattering angles ,1208.

Remote sensing of aerosols over land surfaces including polarization measurements and application to POLDER measurements

Ground-based measurements of the diffuse skylight and airborne measurements of the light reflected by land surfaces are examined, especially with regard to their polarization properties. The reported land surface reflections correspond to multidirectional polarized measurements performed by the Polarization and Directionality of Earth Reflectances (POLDER) airborne version on very clear days. These observations are analyzed for retrieving the polarization properties of scattering by terrestrial aerosols and reflection by ground targets, respectively. The results suggest that the polarized light is much more sensitive to atmospheric scattering than to reflection by natural surfaces, especially by vegetative cover. Theoretical modeling supports this hypothesis. Finally, application of these results to aerosol remote sensing over land surfaces from POLDER measurements is discussed.

Information on aerosol size distribution contained in solar reflected spectral radiances

Information on the aerosol size distribution contained in the reflected solar spectral radiances detected over the oceans can be reduced into two quantities. These quantities have been determined in an unbiased way with the use of the principal components. Consequently, only one to two parameters of the size distribution can be retrieved. For a single-mode distribution these parameters are the effective radius of the particles and the width of the size distribution. The accuracy of the retrieval depends on the view and illumination directions. Accurate knowledge of the refractive index, real and imaginary parts, is not critically important for the retrieval as long as the retrieved particles are smaller than 1.0 μm. An error budget shows that very clean conditions are not suitable for getting any information on the aerosol size distribution. A surprising result of this investigation is that the spectral reflectance of a bimodal-lognormal distribution can be simulated very well with spectral reflectance of a single lognormal with an appropriate radius and width of distribution, σ, that do not necessarily correspond to an average of the bimodal values. The present results change drastically our philosophy regarding the retrieval scheme. Additional studies are needed to confirm the present results for nonspherical particles.