A14A-01
An investigation of cloud and aerosol properties using CALIOP and MODIS
The launch of CALIPSO in June of 2006 provides over a year of global active remote sensed measurements of clouds and aerosol. The CALIOP measurements on CALIPSO offer a unique perspective of the earth's atmosphere with its capability to measure aerosols and clouds with very high vertical resolution and sensitivity. The near coincident orbit with the EOS AQUA suit of passive remote sensed cloud properties offers an opportunity to make direct comparisons between the passive and active measurements. Using collocated MODIS and CALIOP measurements this presentation will investigate the sensitivity of CALIOP and MODIS at detecting and distinguishing clouds and aerosols with the goal of charactering the sensitivity differences between the active and passive retrievals. The analysis will include both global and region comparisons.
A14A-02
Initial comparisons of MODIS and CALIPSO level 2 aerosol products and their cloud masking techniques
The CALIOP (Cloud-Aerosol Lidar with Orthogonal Polarization) instrument aboard the CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation) satellite has been acquiring data as part of the A-Train constellation of satellites since June of 2006. A complete level 1B data set and a limited set of level 2 data products have been publicly available since December 2006. As of December 2007, the level 2 data products have been augmented to include aerosol optical depths and extinction profiles. In conjunction, the MODIS-Aqua level-2 data set, MYD04, is readily available for essentially all days in orbit since June of 2002 and provides column integrated aerosol observations at a spatial resolution of 10x10km at nadir. In this paper we present comparisons of the standard MODIS-Aqua level-2 aerosol data to the initial release of the CALIPSO level-2 aerosol data set. As a first step towards investigating the cloud masking techniques for the two different instruments, we will show comparisons of the operational 500m-resolution cloud mask used in the MODIS-Aqua aerosol algorithm to single-shot CALIOP observations and the standard MODIS cloud mask product, MYD35, for the same scenes. Additionally, for a selected set of coincident and collocated measurements of several different aerosol types, we compare the optical depths retrieved by the two sensors.
A14A-03
Impacts of 3-D radiative effects on satellite-based cloud screening and aerosol optical depth (AOD) retrieval and their consequences on comparing AOD form different satellite sensors
Several recent studies comparing global and regional aerosol optical depths (AODs) derived from space-based sensors have revealed large differences amongst the different sensors. These studies all point to differences in cloud screening strategies as one of the main culprits behind these differences, but differences in strategy on how the remaining clear pixel radiances are sampled for reporting AODs over a domain (typically 10 to 100 km in scale) should also be recognized. Cloud climatologies reveal that these domains often contain clouds, especially over ocean (i.e., > 95% of the time for 10 km scale domain), implying that clear pixels (~ 1 km scale) identified within the domain over which the AOD is to be reported is often near (i.e., a few km) cloud. This proximity of clear pixels to clouds implies the need to examine the impact of 3-D radiative interaction between clear and cloudy regions on AOD retrievals. We present the first detailed examination on how 3-D radiative transfer impacts satellite cloud screening and the subsequent AOD retrieval applied to a sample of the remaining clear pixels. The 3-D radiative transfer through predefined heterogeneous boundary-layer cloud fields embedded in a range of horizontally homogeneous aerosol fields have been carried out to produce synthetic satellite images. Our simulations are numerous, but they are restricted to a wavelength of 0.67 μm, and the procedures of cloud screening and AOD retrieval are applied to these single channel synthetic images. While not representative of any real operational algorithm, physical insight on the connection between cloud screening and AOD retrievals caused by 3-D radiative effects was gained. We show that significant overlap between the radiance distribution of clear and cloudy sky exists, the degree to which depends on many factors (resolution, solar zenith angle, surface reflectance, aerosol optical depth, cloud top variability, etc.). The 3-D radiative pathways that lead to this overlap were examined, revealing that the darkening of clouds by the shadow and leakage pathways can cause clouds of high optical depths (up to 5 in our simulations) to fall within the overlap region, and that the surface-cloud interaction pathway plays an important role in the brightening of clear regions. Large (up to 100's %) systematic errors in AOD retrievals were observed that depended on the details of the cloud mask and the factors that influence the clear/cloud radiance overlap, especially the solar zenith angle. Different sampling strategies commonly employed by modern sensors in producing domain-averaged AOD were performed showing that AOD retrievals on the domain-averaged radiances from all clear pixels produced the smallest AOD biases with the weakest (but still large) dependence on solar zenith angle. AOD biases tended to peak in the solar zenith angle range of 30° to 50°, and remained large even when perfect cloud screening was specified. Our results suggest that the dependence of AOD bias with solar zenith angle may, in part, explain some of the differences observed in AOD climatologies derived from sensors in different orbits, as does the different sampling strategies employed by the different sensors in deriving domain-averaged AOD.
A14A-04
Examination of cumulus cloud contamination on aerosol retrieval from Terra instruments over ocean
Large quantitative discrepancy in aerosol climatologies derived from the passive satellite sensors has long been recognized. One of the major factors causing this discrepancy is hypothesized to be small cumuli, because cumuli tend to partially fill satellite fields-of-view (pixels) and these partially cloudy pixels may or may not be used for aerosol retrieval depending on the quality of the cloud screening algorithm for each satellite instrument. However, examining the quality of a cloud screening algorithm requires the knowledge of the true cloud fields, which can only be provided from coincident high resolution satellite images. Fortunately, with the coincident 15 m observation from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), cumulus cloud contamination on aerosol retrieval can now be examined for both the Moderate Resolution Imaging SpectroRadiometer (MODIS) and the Multi-angle Imaging SpectroRadiometer (MISR) aboard Terra satellite. From Sep. to Dec. 2004, 124 ASTER scenes containing only cumuli over the tropical western Atlantic were collected with overlapped MODIS and MISR observations. The average cloud fraction of all the ASTER scenes examined is 8.6%, half of which is contributed from clouds less than 2 km in diameter. With reference to ASTER, 9.6% of MISR 1.1 km pixels used to retrieve aerosol properties, which are reported at 17.6 x 17.6 km2 regions, contain some clouds, but 86% of the cloud-contaminated pixels have a cloud fraction less than 1% of a pixel area. The information on which 1 km pixels were used in the MODIS aerosol retrieval is not available, making the assessment of cloud contamination on the MODIS aerosol retrieval difficult. However, when comparing to MODIS at pixel level, MISR aerosol optical depth is only 0.02 larger than MODIS on average with a standard deviation of 0.015. Although our results reveal that cumulus cloud contamination is not a major factor that causes the large discrepancy in aerosol climatologies, it is noticed that both aerosol optical depth and Angstrom exponent do have a correlation with the spatial distribution of cumuli.
A14A-05
Intercomparison of cloud properties derived from MODIS and METEOSAT-SEVIRI
Cloud properties inferred from meteorological satellite imagers provide an important reference for the evaluation of cloud processes and cloud-radiation interactions in climate and weather prediction models. To date, the MODIS instrument on board the polar-orbiting TERRA and AQUA satellites offers a unique data source, due to the large number of spectral channels, the high spatial resolution, and the accurate on-board calibration. In contrast to polar orbiters, geostationary satellite provide regular and frequent sampling, and are thus better suited to monitor the development of clouds and resolve their diurnal cycle. EUMETSAT operates the METEOSAT series of geostationary satellites, which provide full-disk images covering Europe and Africa at 15 minute intervals. However, users have to sacrifice some accuracy in comparison to MODIS, due to fewer and wider spectral channels, coarser resolution, and a lower calibration accuracy. To stimulate the use of METEOSAT data in the scientific community, several satellite application facilities have been initiated at European national weather services. Within the Satellite Application Facility on Climate Monitoring, the Royal Netherlands Meteorological Institute (KNMI) has developed retrievals of cloud thermodynamic phase, optical thickness, effective radius and water path. As part of the validation activities, a detailed comparison of these cloud products to corresponding MODIS products has been carried out and is presented here. Special attention is given to the various mechanisms responsible for the observed differences, and their individual contributions to the combined deviations. These include the algorithms themselves, differing spatial resolution and viewing geometry, and discrepancies in sensor calibration and spectral response. In particular, the dependence on sensor resolution and the effects of partially filled pixels on derived cloud properties are studied. A thorough understanding of these mechanisms is of central importance for the interpretation of observed discrepancies between different satellite-based cloud datasets, and their relevance for model validation studies.
A14A-06
Global Distribution and Vertical Structure of Clouds Revealed by CALIPSO
Understanding the effects of clouds on Earth's radiation balance, especially on longwave fluxes within the atmosphere, depends on having accurate knowledge of cloud vertical location within the atmosphere. The Cloud- Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) satellite mission provides the opportunity to measure the vertical distribution of clouds at a greater detail than ever before possible. The CALIPSO cloud layer products from June 2006 to June 2007 are analyzed to determine the occurrence frequency and thickness of clouds as functions of time, latitude, and altitude. In particular, the latitude-longitude and vertical distributions of single- and multi-layer clouds and the latitudinal movement of cloud cover with the changing seasons are examined. The seasonal variablities of cloud frequency and geometric thickness are also analyzed and compared with similar quantities derived from the Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) using the Clouds and the Earth's Radiant Energy System (CERES) cloud retrieval algorithms. The comparisons provide an estimate of the errors in cloud fraction, top height, and thickness incurred by passive algorithms.
A14A-07
Influence of cloud top properties on liquid water retrievals from MODIS and AMSR- E
Satellite observations in the near Infrared (NIR) and Microwave (MW) are often used to retrieve integrated atmospheric cloud water paths. Comparisons abound, some of them concluding that the NIR returns higher cloud water paths than the MW and others obtaining the opposite result. The NIR cloud water paths from the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument on AQUA are compared to those from the Advanced Microwave Scanning Radiometer (AMSR-E) to show that the cloud top properties may be used to discriminate cases where the NIR estimates higher cloud water paths than the MW and vice versa. Conflicting results in earlier studies can then be traced back to the filtering criteria used in their comparisons. An analysis is presented for a spatial resolution of 0.25 x 0.25 degree using the equinox and solstice days of 2003-2006. A climatology of cloud water paths, and the relative abundance of ice and liquid phases during these days will be shown. The cloud top pressures and temperatures where MW and NIR differences change sign are presented. Implications for the conclusions of scientific analyses of cloud water paths based on either MODIS or AMSR-E data are discussed.
A14A-08
Joint distributions of ice clouds and relative humidity: What can we learn from A-train measurements?
Using recent advances in cloud optical and microphysical properties from the Atmospheric Infrared Sounder (AIRS) and the Moderate Resolution Imaging Spectroradiometer (MODIS), along with vertical cloud profiles of ice water content (IWC) and cloud-type from CloudSat and the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO), and temperature and specific humidity retrievals from AIRS/AMSU, we investigate joint distributions of clouds, relative humidity (RH), and cloudy/clear sky RH anomalies on the regional and global scales. The distributions are partitioned by cloud-type, latitude, formation/maintenance mechanism, and other geophysical variables. Composite cases are formed from thousands of observations to investigate the differences and similarities between the joint distributions among the partitioned categories. Regional and global scale variability is explored using continuous data along the CloudSat/CALIPSO ground track, and the results are compared to previous modeling, surface-based, in situ, and satellite studies.
A14A-09
Comparisons of CloudSat cloud features and TRMM precipitation features
The Cloud Features (CFs) are defined by vertically grouping the pixels with cloud mask and cloud radar reflectivity greater than -28 dBZ from CloudSat GeoProf-2B products. The horizontal size, maximum reflectivity profile and other properties for each CF are summarized. For comparison, precipitation features (PFs) are defined by vertically grouping the nadir only pixels with TRMM precipitation radar reflectivity greater than 20 dBZ using 9 years of TRMM observations. Since CloudSat has sun synchronizing orbit and only observes at two local times, the global distribution of cloud occurrence from CFs and precipitation occurrence from PFs over tropics are compared at different altitudes near 1:30 AM and 1:30 PM local times. The regional difference of ratios from PF precipitation area to areas with cloud radar reflectivity greater than -10 dBZ, 0dBZ, 10dBZ and with cloud mask from CFs are demonstrated. To help evaluating the representativeness of the cloud climatology from CloudSat as a whole instead of at the two local times, the statistics of 20 dBZ area and area of TRMM VIRs 11 micron brightness temperature < 235 K from subset of PFs near 1:30 AM and 1:30 PM local time are shown along with those from all PFs. The regional differences between the two are shown.
A14A-10
Precipitation estimation from CloudSat
CloudSat, a satellite in the A-Train constellation containing a nadir-pointing 94 GHz cloud radar, is well suited for the detection of precipitation and the quantification of precipitation intensity, being especially sensitive to lighter rainfall. A new algorithm for retrieving precipitation intensity from CloudSat observations is described. Considering path integrated attenuation (PIA) by hydrometeors as a measurement rather than noise, it is possible to estimate rainfall intensity using observations of the decrease in the surface backscatter under precipitating columns. Multiple scattering within the precipitating column can be significant for rainfall exceeding a few millimeter per hour, so Monte Carlo modeling is used to simulate the relationship between rainfall and observed PIA for various vertical profiles of precipitation. A model of the melting layer is also incorporated into the Monte Carlo calculations to better represent the transition from snow to rain. Results of a year of analysis of precipitation incidence and accumulation using the retrieval are presented, including comparisons with estimates from other sensors. The regional variation of precipitation efficiency, in terms of the fraction of clouds that produce precipitation, is also discussed.