Clouds and the Earth's Radiant Energy System (CERES) Science Team Meeting
--Bruce R. Barkstrom (brb@ceres.larc.nasa.gov), Principal Investigator, NASA Langley Research Center
--Gary G. Gibson (g.g.gibson@larc.nasa.gov), NASA Langely Research Center
The 13th Clouds and the Earth`s Radiant Energy System (CERES) Science Team meeting was held March 13-15, 1996 at the NASA Goddard Space Flight Center (GSFC) in Greenbelt, MD. The focus of the meeting was CERES instrument status, algorithm development, and validation plans. The CERES instrument is designed to provide a climate data set suitable for examining the role of clouds in the radiative heat balance of the climate system. The CERES Science Team blends expertise in broadband radiometry, cloud and radiation remote sensing, and climate modeling. The Science Team guides the definition of the CERES instrument and science studies.
Michael King, EOS Senior Project Scientist, hosted the meeting and opened with an EOS science overview. He identified several key areas of EOS oversight and responsibilities related to CERES. The EOS Project Science Office will conduct peer reviews of instrument calibration and validation plans and will support field experiments and correlative measurement programs for validation. A second review of CERES Algorithm Theoretical Basis Documents, which should closely correspond to the flight version of the algorithms, is scheduled for the fall of 1996.
Leonard Kopia, Robert B. Lee III, and G. Louis Smith presented the instrument status report. The CERES instrument was delivered to GSFC in October 1995 and fully integrated on the Tropical Rainfall Measuring Mission (TRMM) spacecraft on February 14, 1996. Instrument weight and power are slightly below the TRMM allocations. Except for discrete Electro-Magnetic Compatibility (EMC) exceedances, the Proto-Flight Model (PFM) met all physical, electrical, and thermal performance requirements for TRMM. A waiver was approved for radiated emission and radiated susceptibility levels. Threshold levels in CERES (worst case) show a 60 dB margin between spacecraft allowable EMC limits for emissions and instrument sensitivity. The instrument noise is 2-3 counts with less than 1 count microstrain offset and no azimuthal variation of the offset. By comparison, the Earth Radiation Budget Experiment (ERBE) had 2.5 counts noise, 10-20 counts offset, and 30-40 counts offset due to azimuthal variation. Sensor sensitivity (Noise Equivalent Power; NEP) is five times better than ERBE (36 nW vs. 190 nW). Two calibration anomalies occurred: sensor gain variation with temperature (~0.4%), and existence of a second detector time constant. Both anomalies are being considered in the data reduction algorithms. The first full comprehensive spacecraft level testing is scheduled for April 22. EMI/EMC testing begins June 1, thermal/vacuum tests on June 27, and vibration testing on October 18. The spacecraft will be shipped to Japan on March 29, 1997, and will be ready for launch on August 1, 1997.
Several improvements were made to ensure compatibility with EOS spacecraft EMC requirements. Sensor changes to reduce susceptibility include the removal of a feedback capacitor in detector output signal lines, addition of signal buffers for better line impedance balance, and addition of filter networks to the high- voltage lines, high-voltage monitor lines, and bridge- balance control lines. These changes have resulted in a 20-30 times improvement in the radiated susceptibility response of the sensor electronics. A flexible ground strap was added between the rotating azimuth assembly and the pedestal to eliminate residual radiated emissions. Pedestal harnesses and cable layout were reworked to provide better shielding and minimize cable lengths. Connectors were also reworked to improve shielding. Sensor units for Flight Model 1 (FM1) were aligned and characterized and are in the process of spectral characterization using the Fourier Transform Spectrometer. FM2 sensors are being aligned on the mounting plate. Sensor scan assembly is in progress and will be completed April 15 for FM1 and May 17 for FM2. The schedule shows 3.7 weeks of slack for FM1 delivery to EOS AM-1 on November 1, 1996, and 5.3 weeks of slack for FM2 delivery on January 10, 1997.
Jim Kibler presented the CERES Data Management System development schedule and a Release 1 software integration and test schedule. He also gave a comprehensive report on software development responsibilities and status, data products, and Release 2 issues for each Working Group. The ERBE-like Release 1 software, supporting data files, and test plan were delivered to the LaRC Distributed Active Archive Center (DAAC) on February 15, 1996. This is the first EOS code to be delivered to the DAAC. The remaining CERES Release 1 software has now been delivered to the Langley DAAC for testing and integration. This represents over 200,000 lines of code, and is another major delivery milestone on the way to launch. All CERES subsystems were delivered on or ahead of schedule, and early indications are that the integration of the software at the DAAC is going very well.
The Instrument Working Group, led by Robert B. Lee III, focused on the validation plan, a review of Release 1 software, and discussion of Release 2 plans and issues.
The instrument validation and consistency plan was completed. Instrument data products include the filtered broadband shortwave (0.3 - 5.0 um), total (0.3 - >100 um), and window (8 - 12 um) radiances. Validation plans encompass flight calibration analyses (internal calibration module and Mirror Attenuator Mosaic [MAM]), multi-channel comparisons (Inversion Working Group), multi-satellite intercomparisons (TISA Working Group+), single spacecraft cross-track and rotating azimuth plane (RAP) instrument comparisons, and geolocation/coastline detection studies.
The Release 1 algorithm was delivered to the DAAC in February 1996. This code converts radiometric and housekeeping parameters and performs the geolocation calculations. The Release 2 (flight) version is being developed as a refinement of Release 1 plus solar and internal calibration collections for in-flight radiometric and geolocation calibration/validation. Dominique Crommelynck briefed the Team on the European Geostationary ERB (GERB) Experiment. He reviewed the instrument design and problem areas involving the detector, thermal isolation of array cells, and precision of the rotating scan mechanism. The GERB data should be especially useful in conducting diurnal studies, tracking cloud systems, and process/feedback studies.
The Working Group was led by Bruce Wielicki in discussions of validation, cloud analysis methods, and software development. He showed a timetable through 1999 of software development goals for Release 1, Release 2, DAAC deliveries, optimizations, and pre- and post-launch validation of VIRS and MODIS.
Validation plans for all cloud and inversion subsystems were presented. Wielicki stressed the need for objective validation of the cloud mask. Pat Minnis offered McIdas as a way to get the surface observer cloud archives in real time for validation. He has October 86 data ready for use. Michael King suggested that ARM micropulse lidar data are available on the "web." The Group agreed, at Bryan Baum's suggestion, that the CERES cloud algorithm should also be able to work with current NOAA satellites so that we can validate with ARM data and other validation opportunities. In fact, the cloud algorithm should be decoupled from the rest of CERES so that we can validate with field experiment data, e.g., lidar, radar, Experimental Cloud Lidar Pilot Study (ECLIPS) taken during non-ERBE months.
Jim Coakley suggested a consistency check by computing zonal averages of clear-sky temperature over sections of the oceans and examining them to see if they fall into other than normal distributions. Coakley also suggested using LITE and NOAA-11 coincidences for cloud height validation; using NOAA-12 data would be difficult because it is in a near-terminator orbit. Ron Welch focused on the need to check consistency of retrieved cloud parameters as you cross surface-type and cloud-type "boundaries." In late 1997, Surface HEat Budget of the Arctic Ocean (SHEBA) will provide good polar retrievals for validation. Wielicki stressed that vertical profilers were needed and discussed the feasibility of ground-based lidar profilers. Future field experiments were identified and prioritized for use in validating various retrievals. Several validation studies were identified and will be reported at future Science Team meetings.
Richard Green presented a plan to validate CERES data against the historical ERBS data by using Earth validation targets. The average nighttime longwave radiance over tropical ocean for a month was found to vary by only 0.5%. This quantity will provide an early validation of the CERES radiances. Green also showed that the average limb darkening can be established in the same manner, which is useful in validating the scanner offsets. He showed the limb-darkening function for both Nimbus-7 and ERBS and suggested that the difference could cause an inconsistency in the CERES ERBE-like product since the Angular Distribution Models (ADMs) are constructed with Nimbus-7 data and applied to ERBS radiances.
The Release 1 cloud analysis code was delivered to the DAAC. Baum stressed the need for consistent treatment and feedback in all algorithms regarding bad data, missing data, partially processed data, etc. He summarized the lessons learned from Release 1. The EOSDIS Toolkit has initially represented a major overhead for resources, but it is expected that this approach will result in long-term savings. The team reconfirmed a past decision to exempt the Science Team from using the Toolkit for contributed code and instead delegate algorithm I/O issues to the software implementation staff. There was insufficient manpower to verify all contributed Release 1 algorithms, but for Release 2 it is mandatory that all algorithms be carefully checked out before being sent to the DAAC.
Pat Minnis showed global Visible Infrared Near-infrared Techniques (VINT) retrievals of cloud properties for one day of data. Minnis emphasized the need for the Co-Is to get all of the input before we can confidently verify that the CERES code gives results consistent with the off-line code. Steve Platnick showed water droplet radii retrievals and suggested that they were qualitatively the same as those derived from VINT. Minnis showed some preliminary results from his nighttime (10.7, 12, and 3.9 um channels) cloud property retrieval algorithm, which is being developed for Release 2. Minnis' technique of using a lapse rate of 7 k/km for clouds lower than 2 km was adopted for the cloud algorithm. This will allow for the effects of inversions that are not picked up in the relatively coarse Meteorological, Ozone, and Aerosol (MOA) temperature profiles.
Ron Welch expressed concern about the cloud mask results from the first day of CERES cloud results. His code was run with a "global" algorithm specification rather than more specific types of algorithm specifiers (smoke, polar, etc.) that limit the geographical coverage. Larry Stowe showed Pathfinder results and gave a status report on Pathfinder. They have processed 15 months of AVHRR data for Pathfinder and will soon have 18 months. Baum presented some recent results of remote sensing of low-level clouds in marine polar air masses using AVHRR multispectral imagery and concluded that radiometric data from these clouds cannot be fully interpreted using a single cloud particle distribution.
Jim Coakley compared radiative fluxes from calibrated AVHRR and plane-parallel theory with observations. He applied the spatial coherence technique to identify similar, uniform, single-layer cloud regimes over the ocean for a month and formed a "global" composite of plane-parallel regions to extend the effective spatial scale. His study, along with an extension of his work by Norman Loeb, showed that when 1-D theory is used to infer cloud optical depth directly from observations at nadir, a systematic increase with solar zenith angle is observed. This increase is most pronounced at solar zenith angles greater than 63 degrees. On average, differences between observed and plane-parallel reflectances are more sensitive to changes in solar zenith angle than to viewing and azimuth angles. The differences are likely due to cloud 3-D effects.
Dave Randall compared Colorado State University (CSU) GCM results with ISCCP data and concluded that thin cirrus clouds are missed by current cloud retrieval algorithms. Thin cirrus clouds are at least as important radiatively as upper tropospheric water vapor. Changes in thin cirrus amount could strongly influence the climate. The CSU GCM is a good estimator of cloud occurrence with respect to ISCCP for optically thick ([[tau]] > 22) clouds. For thin ([[tau]] < 3.6) clouds, ISCCP detects significantly fewer clouds than the model. He suggested that, due to multiple cloud layers, ISCCP tends to overestimate optical thickness of high clouds.
Richard Green showed preliminary validation results of a new set of ERBE-like ADMs constructed from the Nimbus-7 data and the Radiance Pairs Method (RPM). The ERBE production ADMs show an erroneous 10% albedo growth from nadir to the limb, which was eliminated by the new ADMs except for viewing zenith angles greater than 70 degrees. Applying these new ADMs to ERBS data increased the global shortwave (SW) flux by 5 Wm- 2 for a terminator orbit and resulted in a -5 Wm- 2 change for a noon orbit. The longwave (LW) flux changes were within 1 Wm- 2. Validation of the new ADMs is continuing.
Alexander Ignatov, UCAR visiting scientist (representing Larry Stowe), showed aerosol products using NOAA AVHRR as a TRMM/VIRS prototype. He suggested that independent retrievals at 0.63 and 1.6 um are desired to improve the overall aerosol dataset.
Tom Charlock led SARB Working Group discussions on validation, initial Release 1 algorithm results, and Release 2 plans. For SARB, cloud physical and narrowband radiative properties data derived from cloud imagers such as VIRS or MODIS are used along with atmospheric temperature and humidity data in a radiative transfer model to calculate broadband radiative fluxes at the surface of the Earth, through the atmosphere, and up to the top of the atmosphere (TOA). Charlock reviewed the activities of the Atmospheric Radiation Measurement (ARM) Science Team and pointed out the synergies which exist between SARB activities and the ARM program.
Charlock and Charlie Whitlock presented a SARB validation plan which uses the CERES/ARM/GEWEX [Global Energy and Water Cycle Experiment] Experiment (CAGEX) and several surface networks. Whitlock briefly reviewed the first year of surface observations from the CERES Walker Tower validation site. A near-constant broadband surface albedo in combination with a false-color Landsat image dramatically illustrated the potential of the site for validating the ERBE-like SW and LW net surface products. The current focus is to finalize a near-term strategy for future operations. Michael King indicated that some surface sites will likely be enhanced with new instruments. Crommelynck described a balloon experiment in which radiation measurements were made for validating model calculations. Lessons learned from problems encountered in such experiments will be valuable in future measurement programs for validation.
The SARB Release 1 code was delivered to the DAAC. Fred Rose presented results from several applications of the "flux constraint algorithm," which will be used to constrain the SARB radiative transfer computations with the TOA flux measurements obtained from CERES instruments. Comparison of model calculations with ERBE data showed that the application of the constraint algorithm improved the agreement significantly even though some differences persist. Application of the constraint algorithm to CAGEX data over the Oklahoma ARM/Cloud and Radiation Testbed (CART) site yielded similar comparisons. Tim Alberta reviewed the CAGEX datasets and presented comparisons of the radiative transfer computations from the Fu-Liou code with ERBE data. Significant biases were noticed in some of these comparisons for both clear-sky and all-sky SW fluxes. CAGEX datasets are available to the science community for use and evaluation on the web.
Shashi Gupta presented the first results from Release 1 processing of surface-only, cloudy-sky LW fluxes for an hourly swath on October 1, 1986. The results were examined in terms of the meteorological inputs to the algorithms, and the sensitivities of the LW fluxes to these inputs. The results were consistent with the physical relationships, and agreed well with the climatological distribution of LW fluxes. Anand Inamdar (representing V. Ramanathan) reviewed the surface-only, clear-sky LW algorithm. He indicated that the group may use climatological data on cloud thicknesses from WMO atlases in developing a cloudy-sky surface LW algorithm.
Dave Kratz reviewed the state of the SW radiative transfer in clear atmosphere in light of the disagreements between model calculations and observations. He identified several possible causes for these disagreements including differences between the SW parameterizations used in the models and the line-by-line computations, the use of Lorentz line shapes in many models, and neglecting the far wings of lines and continua. He also reviewed the current status of the "surface only" SW and LW algorithms in Release 1 processing and the results from those algorithms.
Bob Cess presented a summary of recent studies that support his theory of anomalously high atmospheric SW absorption. Using the results of several researchers, Cess concluded that (1) plane-parallel models overestimate cloud albedo, (2) models overestimate surface SW absorption by overestimating surface insolation, (3) satellite-surface and stacked-aircraft measurements both indicate excess cloud SW absorption, (4) evidence indicates that the cause is macrophysical rather than microphysical, and (5) it is highly unlikely that observational evidence of excess SW absorption is due to satellite sampling errors.
Dave Randall presented results of studies with a Single Column Model (SCM). Meteorological observations from the Southern Great Plans (SGP) ARM/CART site were used to force the advective processes at the SCM boundaries. Results from the SCM were compared with observations of cloud formation and precipitation etc. from the ARM/CART site. The model is still in an early stage of development. Shi-Keng Yang (representing Jim Miller) presented results from simulations from the National Centers for Environmental Prediction (NCEP) model with a new SW parameterization based on Chou's work. Comparison with ERBE data showed improvement over earlier results, but significant differences still remained. Comparison of NCEP cloud amount with ISCCP showed significant biases off the west coast of South America, north of India, and off the west coast of South Africa. Some dynamics-related problems in the Tropics were suggested. Maurice Blackmon showed several new developments/improvements to the physics in an NCAR GCM (CCM3). A new deep convection scheme and reformulations of the planetary boundary layer diagnostic, clouds, and radiation have resulted in closer comparisons between the CCM3 and observations. Long-term CERES measurements are needed to validate the models and provide a sound basis for developing new physical parameterizations.
Takmeng Wong led the TISA Working Group where the agenda encompassed validation, satellite sampling, temporal interpolation, algorithm development, and Release 2 issues.
Wong presented the TISA validation plans. Validation is required for LW and SW TOA total-sky and clear-sky fluxes, window radiance, LW surface flux, atmospheric flux, cloud amount (total and levels), cloud particle size, cloud liquid and ice water path, cloud emittance and optical depth (daytime only), and cloud height and thickness. Pre-launch validation involves applying CERES algorithms to ERBE data and comparing the results to "truth" data from other satellite and surface observations. When CERES data are available, qualitative evaluations will be made based on comparisons to previous ERBE data, ERBE wide-field-of-view data (if available), Scanner for Radiation Budget (ScaRaB) data (if available), GOES cloud properties, ARM data, CAGEX, and operational ground station observations. EOS AM results will also be validated by comparison with TRMM data.
Release 1 TISA algorithms have been delivered to the DAAC on schedule. Maria Mitchum summarized the outstanding TISA Release 2 issues: (1) improve ERBE-like clear-sky/cloudy-sky threshold, (2) derive technique for normalizing the geostationary cloud products to CERES observations, (3) use local time vs. GMT in binning the data, (4) derive technique for calculating vertically integrated cloud properties, (5) incorporate a surface emittance map, (6) evaluate image enhancement techniques for spatially averaging flux and cloud data, (7) explore RAP scanner sampling errors, and (8) develop a technique to incorporate ISCCP B1 rather than B3 data.
Wong used Minnis' Layer Bispectral Threshold Method (LBTM) "truth" cloud dataset over the TOGA region in the western Pacific Ocean for December 1992 to determine temporal interpolation errors for cloud amount and visible optical depth. Results showed that temporal interpolation errors are large for the CERES single satellite product and decrease as the number of satellites increases. The use of cloud products from geostationary satellites reduces CERES temporal interpolation errors for cloud amount and optical depth by up to 50%. He concluded that geostationary cloud data should be incorporated into the CERES cloud temporal interpolation scheme. The improved CERES cloud and TOA flux data should reduce errors in both the surface and atmospheric radiative flux products.
Dave Young discussed options for acquiring geostationary data. He showed data provided by Mathew Schwaller (GSFC) on spatial and temporal resolution, data volume, contents, and expected time delay for ISCCP data products. A tentative decision was made to use ISCCP B1 data for CERES temporal interpolation. A sample B1 dataset has been acquired for testing. Young also discussed the problem of clear-sky data gaps. For ERBE, 10-50% of land regions lacked monthly mean clear-sky data. He concluded that the primary cause of the data gaps is inadequate LW thresholds and proposed development of regional, monthly LW thresholds from ERBE data for use on CERES.
Dave Doelling presented some LW results of ongoing CERES spatial sampling studies. Spatial sampling rms errors for the RAP scanner were about twice that of a crosstrack scanner for all three satellites. A "truth" radiation and cloud data set is being developed using a full month of GOES 8-km, 1-hr data and LBTM clouds for testing interpolation methods and determining spatial sampling errors. This dataset, in conjunction with the CERES Point Spread Function, will provide a tool for estimating crosstrack and RAP scanner sampling errors, evaluating arithmetic and resolution enhancement techniques for spatial averaging of radiation and clouds, e.g., image sharpening, and determining spatial sampling errors for cloud parameters.
The Science Team toured the GSFC Spacecraft Integration Facility to observe the CERES instrument mounted on the TRMM spacecraft. Chris Kummerow, TRMM Deputy Project Scientist, gave an overview of the TRMM mission, spacecraft, and instrument status. While the TRMM precipitation data provide information on latent heating within the tropical atmosphere, CERES provides information on radiative heating. These two components dominate the tropical energy budget. Of particular interest to CERES, the VIRS instrument has overcome problems with the cooling system and is now scheduled for delivery in April for integration on the TRMM spacecraft. There is currently speculation about a TRMM follow-on mission around 2002 in a high-inclination orbit. VIRS has been mentioned as a likely instrument for this mission, but no decision has been made concerning the CERES instrument. King reiterated that CERES measurement of radiative fluxes is a primary research goal of the U.S. Global Change Research Program. The Science Team strongly recommended that a CERES instrument be included on any TRMM follow-on mission. Bruce Wielicki will contact the appropriate scientists in Japan to make the case for CERES.
Validation plans are due by March 31, 1996. The next CERES Science Team meeting is scheduled for September 18-20, 1996 at the Langley Research Center. Major topics will include instrument flight qualification tests, approval of the CERES validation plan, Release 1 algorithm testing, and Release 2 status.