The 18th Clouds and the Earth's Radiant Energy System (CERES) Science Team meeting was held at the State University of New York at Stony Brook on September 15-17, 1998. The CERES Science Team reached a unanimous decision to archive the ERBE-like data products from CERES on the Tropical Rainfall Measuring Mission (TRMM). The ES8 (Level 2), ES9 (Level 3 monthly averages with the individual hourly observations), and ES4 (monthly averages only) will be fully available after October 1998.
Bruce Wielicki of the NASA Langley Research Center (LaRC), CERES Co-Principal Investigator, opened the meeting with an EOS program status report. The Earth Observing System morning satellite (EOS AM) launch is delayed until mid-to-late 1999 due to spacecraft control software problems. EOS PM is still on schedule for a December 2000 launch. For the EOS Follow-on, CERES is being proposed as part of three mission concepts: aerosol radiative forcing, cloud feedback, and climate variability. The climate variability mission would fill the time gap between EOS PM and the National Polar-orbiting Operational Environmental Satellite System (NPOESS) to ensure a long time series of radiation budget measurements.
Jack Cooper (LaRC) presented the instrument status report. A Tiger Team is investigating the cause of a voltage anomaly experienced by the Proto-Flight Model aboard the TRMM. The team has verified that the voltage fluctuations are a result of an anomalous voltage converter and not an improperly operating converter voltage monitor. Radiation damage testing performed on spare voltage converters has confirmed that radiation did not singularly cause the anomaly. Further tests are being conducted on spare voltage converters in order to determine the cause of this anomaly. The final outcome will include a recommendation for the operation of the instrument on TRMM and an assessment of the impact to the EOS AM-1 and EOS PM-1 instruments.
Robert B. Lee III (LaRC) showed that the CERES/TRMM calibration stability is excellent. The internal calibration module (ICM) ground-to-flight radiometric stability was ±0.2% for both the total and shortwave (SW) channels and ±0.4% for the window (WN) channel. The Instrument Team recommended using the ground count conversion coefficients to reduce CERES TRMM Earth radiance measurements.
Kory Priestley (LaRC) showed ground calibration results for the EOS AM/PM flight models. Flight Model 3 and 4 offset measurements are similar to those of the first three CERES instruments and are not expected to be a substantial concern in data analysis.
The ERBE-like data will be validated and archived first. These data do not rely on coincident observations from the cloud imager. Richard Green (LaRC) showed large differences seen between the CERES and Earth Radiation Budget Experiment (ERBE) top-of-atmosphere (TOA) fluxes over the tropics (20S to 20N). CERES TOA longwave (LW) fluxes for the tropics range from 4 to 8 Wm-2 higher than ERBE. The TOA SW fluxes range from 0 to 3 Wm-2 lower than for ERBE. The large 1998 ENSO may help to explain the variation in LW flux differences from a peak of 8 Wm-2 in February to a minimum of 4 Wm-2 in July. The changes between ERBE and CERES are primarily in the total-sky fields (not clear-sky). Only about 1 Wm-2 of the LW change in July appears to be explainable by calibration differences. Comparisons of the long time series of Earth Radiation Budget Satellite (ERBS) nonscanner data which overlaps both ERBS and CERES scanner data support the change in LW flux between ERBE and CERES time periods (late 1980s to late 1990s). An increase of 3 to 4 Wm-2 in tropical mean LW flux also occurred in early 1991.
David Young (LaRC) showed early validation results for ERBE-like monthly-averaged data. He compared tropical means of the first 7 months of CERES data with ERBS results. CERES had 11% more scenes identified as clear than ERBS observed for the same month from 1985-1989. This difference decreased as the El Niño event came to an end. CERES total-sky LW is 3% higher than ERBE, and the difference decreased as El Niño ended. The CERES values are consistent with ERBS non-scanner measurements for the same time period. Small clear-sky LW anomalies are consistent with model calculations. Total-sky SW is 2.7% lower than ERBE, of which up to 2% is likely due to temporal sampling noise. The clear-sky SW is 2.9% lower than ERBE on average, but the deficit ranges from 1.8% for ocean to 6.1% for land and 8.7% for deserts. Young concluded that all CERES/ERBE differences are within the ERBE-like uncertainty limits and recommended archival of the ERBE-like monthly products.
Lou Smith (Virginia Tech) compared CERES scanner results with ERBE wide-field-of-view (WFOV) data. CERES total channel estimates of nighttime ERBE WFOV fluxes are biased less than 1%, comparable to the ERBS scanner-to-nonscanner comparisons. Daytime LW biases were less than 2%.
There has been considerable progress on validating the CERES WN channel. Richard Green reported that both the Angular Distribution Models (ADM)s and Tropical Mean results show that the WN has more limb darkening than the broadband LW radiance. Fred Rose of Analytical Services and Materials, Inc. (AS&M) used MODTRAN to validate spectral correction coefficients for unfiltering the WN measurements. David Kratz (LaRC) used line-by-line models to establish the relationship between total and WN channel radiances for the case of high, cold clouds. Kory Priestley then performed a 3-channel intercomparison which implied an inconsistency between the SW channel and SW portion of the total channel at the 0.65% level.
Patrick Minnis (LaRC) briefed the team on new methods and algorithms for deriving CERES cloud products from the Visible-Infrared Scanner (VIRS) imager data. Cloud patterns from the current algorithm appear quite reasonable, but phase determination may overestimate water (liquid) clouds; heavy aerosols are missed; and a correction for viewing zenith and solar zenith angle dependence is needed. CERES-retrieved cirrus and daytime stratus cloud heights and temperatures agree well with surface measurements, but nighttime comparisons are not as close. Cloud emissivities and fractions are generally close to those from surface observations, but CERES-derived liquid water paths and effective droplet radii are, in general, larger than those from the surface. The optical depths from CERES agree well with surface observations during the daytime, but are much smaller at night. Future plans are to incorporate other Co-Investigator's algorithms, implement a tri-spectral nighttime retrieval algorithm, and use VIRS channel 2 for cloud mask, phase, and particle size calculations.
Norman Loeb (Hampton University) reported on the development of ADMs. He developed new CERES LW and WN ADMs for the ERBE scene types and noted significant differences from the ERBE ADMs. Details are given in the ADM Working Group report. He examined POLDER (Polarization and Directionality of the Earth's Reflectances) data to determine the influence of cloud properties on ADMs and albedo. Loeb determined that when all observations at all POLDER angles are considered, absolute ADM albedo bias errors are less than 0.03% compared to 0.3% for albedos derived using the independent pixel assumption.
Tom Charlock (LaRC) reported that the Surface and Atmospheric Radiation Budget (SARB) derived albedos for VIRS clouds are 0.07 higher than the CERES observations. The pre-launch albedos for Advanced Very High Resolution Radiometer (AVHRR) clouds had compared well to ERBE. CERES/ARM Validation Experiment (CAVE) results show that more SW absorption is needed in the atmosphere. The Fu-Liou radiative transfer code was modified in the LW to include explicit TRMM WN output and other spectral improvements. SARB has added the WN channel to the flux constrainment algorithm. SARB is considering switching from the Goddard Data Assimilation Office (DAO) to European Center for Medium-Range Weather Forecasts (ECMWF) data for their radiative transfer calculations.
Don Cahoon (LaRC) described the CERES helicopter field experiment conducted at the Southern Great Plains (SGP) Atmospheric Radiation Measurement (ARM) site in August 1998. Spectral bidirectional reflectance distribution functions (BRDFs) were measured from helicopter-borne instruments over many types of cropland. The results of this experiment will be presented as CERES/ARM/GEWEX Experiment (CAGEX-3) and will be widely disseminated through a web site.
Robert Seals (LaRC) of the Langley Distributed Active Archive Center (DAAC) discussed the wide range of DAAC services for producing, archiving, and distributing Earth science data. Bruce Barkstrom (LaRC) cited a need for improving introductory information on CERES data products in the LaRC DAAC web ordering tool for CERES data. He showed an example of the ES-8 data product description and outlined the data quality summary pages needed for the ERBE-like Level 2 and Level 3 data products.
Jim Kibler (LaRC) presented the CERES Data Management System (DMS) status. All DMS subsystem deliveries to the DAAC are on schedule. Instrument and
ERBE-like subsystems are running at the DAAC using the LaRC TRMM Information System (LaTIS).
Lin Chambers (LaRC) gave an update on the CERES Students' Cloud Observations On-Line (S'COOL) Project. The S'COOL project successfully participated in a joint French/U.S. demonstration of science education cooperation during a May visit to France by Mrs. Clinton. Dan Goldin and Bruce Barkstrom were at a school in the Washington area. Schools in New York and France also participated. Currently S'COOL has 142 registered participants from 16 countries on 5 continents.
Instrument Working Group: Robert B. Lee III led the Instrument Working Group meeting in discussions of the accuracy of the CERES instrument on TRMM. Measurement accuracy and precision goals have been satisfied. The group is examining the instrument ground calibration data in an attempt to understand the 1% inconsistency in the ERBE-like 3-channel checks performed by Priestley for deep convective clouds (0.8%), and Richard Green on the Tropical Mean day/night check (1.2%). This inconsistency is within the CERES goal of 1% SW absolute calibration accuracy; calibration changes are not currently planned.
Cloud Working Group: Patrick Minnis led the discussion of cloud retrieval, archival, data dissemination, and validation issues. The team explored the possibility of using ECMWF data rather than DAO for constructing the Meteorology, Ozone, and Aerosol (MOA) data set. Man-Li Wu (GSFC) presented information about current and planned changes in DAO products that will address some of the concerns about DAO. The Langley cloud team will provide a summary of their work on differences between MOA skin temperatures and International Satellite Cloud Climatology Project (ISCCP) TOA temperatures for use in DAO validation.
There is an apparent overestimate of effective droplet radius in the water cloud retrievals. Accuracy of the 3.7 micron channel calibration and knowledge of the solar constant in this spectral interval were discussed as possible sources of errors. Also, team members discussed the inaccuracies inherent in using a bidirectional model as a possible source of the VIRS albedos being lower than other narrowband instruments.
Minnis presented a summary of both short- and long-term improvements planned for the cloud algorithm, including multi-layer retrievals. Larry Stowe (NOAA) noted that the Clouds from AVHRR (CLAVR) algorithm has a multi-layer flag that can be used for improving identification of multi-layer pixels. Ron Welch (U. Alabama Huntsville) will tune his cloud mask to VIRS and develop improved masking techniques and multi-layer retrievals specific to VIRS. Stowe will investigate thresholds for application to the visible and near-infrared channels for the purpose of flagging aerosols in pixel data in the cloud algorithm. Team members were encouraged to visit the cloud working group's web site at Langley: http://lposun.larc.nasa.gov/~cwg/ for updates on algorithm changes and validation.
Welch showed recent results of classification studies using Moderate Resolution Imaging Spectroradiometer (MODIS) Airborne Simulator (MAS) data from Alaska and smoke detection results from AVHRR data. He also presented results of his global detection of contrails from AVHRR imagery from October 1992 and discussed the impact of smoke aerosols on clear-sky SW fluxes in ERBE data using TOMS UV data for aerosol detection. He suggested visiting www.atmos.uah.edu/~sundar for further information.
Bing Lin (Hampton University) presented a study of the collocation of TRMM Microwave Imager (TMI) data with VIRS. After mapping VIRS data into the TMI fields of view, he identified a clear-sky temperature bias that implies that TRMM's recalibration was necessary.
Minnis presented a summary of contrail detection over the U.S. using NOAA-11 and NOAA-12 during April 1994. He also showed results from a LW anisotropy study that showed differences in clear-sky temperatures from GOES-8 and GOES-9 of as much as ±4 C when viewing the same area. Diurnal difference cycles were presented for various scene types and viewing conditions. Recent helicopter studies from the ARM SGP site confirmed the existence of such anisotropy.
Surface and Atmospheric Radiation Budget (SARB) Working Group: A joint meeting of the SARB and Surface-only Working Groups co-chaired by Tom Charlock (LaRC) and David Kratz addressed algorithm and validation issues.
Fred Rose discussed recent modifications to the Fu-Liou radiative transfer code made in cooperation with Qiang Fu (Dalhousie University) and David Kratz. The CKD (Clough, Kneizys, and Davies) continuum, which was limited to the window region in the original code, was extended to the 5-2200 cm-1 region. A new set of correlated-k's which accounts for the absorption of CFCs and weak bands of carbon dioxide was incorporated into the code, bringing the Fu-Liou results into better agreement with line-by-line computations.
Shashi Gupta (AS&M) presented new scene-dependent global surface emissivity maps for the 12 spectral intervals of the Fu-Liou code, the CERES WN channel, and the broadband LW region. Use of the broadband emissivity map instead of the assumption of a black surface in net LW flux computations yielded differences of up to 6 Wm-2 over some regions. Emissivity maps are available to the science community at the following web address: http://tanalo.larc.nasa.gov:8080/surf_htmls/SARB_surf_html.
Shi-Keng Yang (Research and Data Systems Corporation) discussed the status of the Stratospheric Monitoring-group Ozone Blended Analysis (SMOBA) product which is the primary source of ozone data for CERES. SMOBA provides assimilated 3-D ozone fields based primarily on Solar Backscatter UltraViolet (SBUV/2) data from NOAA-14. Due to a malfunction of the grating on SBUV/2, the data stream was interrupted during parts of May and June 1998. The data stream was revived by shifting to a different wavelength, but the new data showed inconsistencies with prior data. A recalibration of the instrument on August 12 resulted in a decrease of 10-12 dobson units in the column ozone amount.
Man-Li Wu discussed DAO efforts toward improving the Goddard Earth Observing System (GEOS-2) meteorological fields. She compared clear-sky OLR fields from GEOS-2 and ECMWF. The differences over land areas were attributed to corresponding differences in ground temperature, and those over ocean areas to differences in moisture fields. The problems with the upper tropospheric humidity (UTH) in GEOS-2 were attributed to deficiencies in the model convection scheme.
Yaping Zhou (State University of New York at Stony Brook) compared results obtained from radiation modules of two versions of the Community Climate Model (CCM2 and CCM3) with data sets available from the CAGEX-1 web page. All models overestimated outgoing LW and underestimated downward LW flux; the Fu-Liou code provided the best agreement with observations and CCM2 provided the worst.
CERES ADM Working Group: Norman Loeb organized and led the first official ADM working group meeting in discussions of critical ADM/inversion research issues. These include spectral correction, definition of ADM scene types, and ways of consistently identifying scene types at all angles. The formation of this new working group is a reflection of the importance and complexity of this area of research.
Lin Chambers presented 2-D model results showing how incorrect scene identification due to 1D-cloud property retrieval errors can lead to ADM-derived albedo errors. Overall mean bias errors using the ADM approach were smaller than 1D approach at least by a factor of 2. She also showed the advantage of using multiple-view satellites (CERES rotating azimuth plane scanner; POLDER) to estimate albedo. Instantaneous rms errors in albedo were reduced by 50% when albedos inferred at multiple-view angles were averaged together and compared with truth as opposed to treating albedos in each view angle individually.
Richard Green led a discussion on how ADMs vary with height above the Earth's surface. He provided an example of an ADM for a Lambertian Earth and showed that neglect of the ADM dependence with height can lead to an error of about 1% in the derived TOA net flux.
Lou Smith showed recent LW and WN channel along-track measurements from ES-8. Overall, limb darkening in the WN channel was stronger than in the LW channel. The difference did not depend on scene type.
Norman Loeb compared new LW and WN ADMs derived from CERES SSF for ERBE scene types ("VIRS_12") with LW ERBE ADMs (Suttles et al.). VIRS_12 LW ADMs showed much more limb darkening than the Suttles LW models. He also found stronger limb darkening in the WN ADMs compared to the LW ADMs. Based on simulations from MODTRAN, Loeb argued that under clear-sky conditions, high tropical column water vapor and surface temperatures are the likely causes for the stronger limb darkening in the WN channel. However, in contrast to Smith's results, VIRS_12 LW and WN limb darkening showed a strong dependence on scene typeit tended to increase with cloud cover, particularly in the WN channel. The reason for the apparent discrepancy between the Smith and Loeb results may be because of differences between ERBE-like MLE (used in Smith's study) and VIRS imager scene identification (used in Loeb's study).
Yong Hu (Hampton University) showed some early statistics of cloud optical depth retrievals stratified by viewing geometry. In the fixed azimuth scan mode, VIRS mean optical depths show a 10% (relative) decrease between nadir and the most oblique VIRS viewing zenith angle (45 degrees). When VIRS mean optical depths are stratified by CERES viewing zenith in the rotating azimuth plane scan mode, no viewing zenith angle dependence is observed (since CERES viewing zenith angles are independent of VIRS viewing zenith angles). He also showed a systematic increase in cloud optical depth retrievals with solar zenith angle, confirming Loeb's earlier ERBS and AVHRR studies.
Time Interpolation and Spatial Averaging (TISA) Working Group: David Young led discussions of software development, current temporal and spatial averaging studies, and ongoing CERES ERBE-like validation efforts. Stephanie Weckmann (Virginia Tech) analyzed ERBE scanner and nonscanner data to show how the ERBE nonscanner data can be used to validate CERES results. Comparisons with Scanner for Radiation Budget (ScaRaB) data are also continuing. The team discussed several science issues that must be resolved. Can we define monthly mean errors for products that use geostationary data for temporal interpolation? Should we develop standard products which calculate monthly mean relationships between variables such as albedo and cloud particle size? Are there variables such as aerosols that require separate algorithms? How do we most effectively use cloud information from the geostationary satellites?
Bruce Barkstrom (LaRC) gave a preview of the studies which he plans to undertake while on a 6-month sabbatical at the University of Illinois Urbana-Champaign. He will develop pattern recognition techniques for use in the characterization of cloud spatial structures.
Robert Cess (State University of New York at Stony Brook) analyzed measurements of direct, diffuse, and global SW radiation from eight stations in Canada, and the SGP/ARM site in Oklahoma. He noticed a very unusual feature in the data from one Canadian station and all data from the ARM site. Most of the afternoon data from these sites obtained during fall and winter showed diffuse radiation values less than the Rayleigh limit. The cause remains unexplained.
Jim Coakley (Oregon State University) presented results of a pixel-level cloud retrieval algorithm that incorporates information from adjacent pixels and regions. The algorithm retrieves information about pixels containing broken clouds rather than assuming that all non-clear pixels are totally filled with clouds.
Leo Donner (GFDL) presented results of a study of radiation and microphysics in the clouds of the Tropical Ocean Global Atmosphere Combined Ocean Atmosphere Response Experiment (TOGA-COARE) region. The objective of the study was to improve the cloud system model using observations from the TOGA-COARE region. Comparisons of modeled and observed surface and TOA radiation data showed that the ice sedimentation process in clouds has a significant impact on surface radiation.
Qingyuan Han (University of Alabama Huntsville, representing Ron Welch) presented results on monitoring the indirect radiative forcing of aerosols using satellite data. Increased concentration of aerosols in the atmosphere leads to increased concentrations of cloud condensation nuclei (CCN) and cloud droplets, decreased droplet radius, and increased cloud optical thickness and albedo. The optical thickness of thin clouds was substantially affected by increased aerosol concentration while thick clouds were largely unaffected. The magnitude of the radiative forcing was estimated to be about 0.25 Wm-2 for the Southern Hemisphere (considered pristine), and about 0.77 Wm-2 for the Northern Hemisphere (considered polluted).
Patrick Minnis (LaRC) showed that azimuthal and viewing zenith angle dependence of skin temperature and surface emissivity were evident in both satellite and helicopter data sets. The magnitude of the variation is related to terrain roughness and possibly surface type. The anisotropic effects may be corrected for by using visible-channel BRDFs. The phenomenon will be examined more closely using GOES, VIRS, helicopter, and aircraft data sets.
V. Ramanathan (Scripps) presented observational data valuable for CERES related studies obtained during the Indian Ocean Experiment (INDOEX). These data consist of surface radiometric measurements, and retrievals of optical properties, vertical distributions, and chemical composition of ambient aerosols. Aerosol optical properties were derived from comparisons of site-measured and model-derived diffuse and global SW fluxes, and indicated that ambient aerosols were highly absorbing. These results can be used for developing models of aerosol radiative forcing for CERES.
David Randall (Colorado State University) presented the results of a numerical experiment on stratus cloud feedbacks in a coupled ocean-atmosphere model. It was primarily a study of the feedbacks between stratus clouds and the sea surface temperature (SST). The experiment was initiated by modifying the stratus parameterization in the model so as to increase the cloud optical depth (brighter clouds). Randall showed that while the response of an atmosphere GCM was a straight forward increase in the SW cloud radiative forcing (SWCRF) at the ocean surface, the response of the coupled ocean atmosphere GCM was far more complex because of the feedbacks. In the coupled model, increased cloud optical depth led to increased SWCRF, to lower SST, to increased atmospheric stability, to more stratus clouds, and to the whole cycle all over again, indicating a positive feedback.
Larry Stowe (NOAA) summarized his dark count corrections to VIRS 1.6 µm data to account for the 5 µm leak. He showed VIRS aerosol retrievals that utilized 1.6 µm data corrected with a 1.6 µm regression between night time 1.6 µm data and 10.8 µm temperatures and differences between 10.8 and 12 µm data. This correction is 0.6% to 1.2%. Stowe also found small errors of 0.1% to 0.2% in the visible data but did not determine the cause. He will do a similar regression for the full range of temperatures rather than just for the clear-sky temperature ranges that he used when checking for aerosols. He will check his regressions periodically now that the night time radiances are available.
Si-Chee Tsay (GSFC) summarized NASA's involvement in the South African Reginal Science Initiative (SAFARI), a project aimed at studying and establishing linkages between the physical, chemical, and biological processes in South African ecosystems. The approach is to develop synergies between remote sensing, modeling, and airborne and surface-based observational capabilities of the participating organizations. GSFC provides the surface-measured radiation data. Toward that end, SRB measurements are already being made, mostly using commercially available instruments. A limited number of specialized instruments, e.g., micropulse lidars, cimel sun photometers, MFRSR, NIP, and microwave radiometers will also be deployed. Pre-SAFARI experiments will be conducted in 1999 during February-March (the wet season) and August-September (the dry season). Several full-fledged coupled experiments will follow during the years 2000 onwards.
Shi-Keng Yang (representing Jim Miller, NOAA/NCEP) reviewed the status of the newest version of the radiation module of National Centers for Environmental Prediction (NCEP) data assimilation model. This model is being used for the second phase of the NCEP reanalysis and for the processing for the Atmospheric Model Intercomparison Project (AMIP II). An examination of the global mean energy budget of the model shows an imbalance. Comparisons of the TOA radiation fields with ERBE results, and of surface radiation fields with the Langley SRB data set show large differences.
David Young (LaRC) compared time-averaged fluxes from CERES crosstrack and rotating azimuth plane (RAP) scanners. He showed that the RAP and crosstrack-instantaneous SW fluxes agree to within 1%. The 6-month mean LW and SW fluxes are not biased by the inclusion of RAP data. He recommended that both RAP and crosstrack data be included in the CERES ERBE-like monthly mean products.
The next CERES Science Team meeting is scheduled for April 27-29, 1999 at the NASA Langley Research Center. The focus will be twofold: the progress of validation for clouds, ADMs, SARB, and TISA, and new science results from the science team.