Using CALIPSO and CloudSat Data to Verify Cloud Analyses at Air Force Weather Agency

For decades, the Air Force Weather Agency (AFWA) and its predecessor organizations have performed global cloud analyses of meteorological satellite data for the Department of Defense. Recent improvements to the analysis process permit single-layer cloud base and top estimates for each satellite data pixel analyzed as cloudy. These analyses are aggregated into polar stereographic grids for each hemisphere at 24 km resolution, generating a product known as the World Wide Merged Cloud Analysis (WWMCA).

Verification of the accuracy of these pixel-level analyses has been somewhat subjective, requiring a human analyst to compare satellite imagery to the resulting analysis to identify over- or under-analysis of cloud cover. These defects are then examined in more detail to determine if the cloud analysis algorithm can be parametrically “tuned” to reduce cloud-detection errors in future analyses.

CloudSat and CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation) data offer independent sources of atmospheric opacity measurements that can be used to validate AFWA's cloud-mask and cloud-height analysis results. The Cloud Profiling Radar (CPR) on CloudSat penetrates dense cloud cover and precipitation that provides cloud-base observations in virtually all weather conditions, but does not detect thinner cloud very well. The Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) instrument aboard CALIPSO is more sensitive to the smaller ice particles of optically thinner cirrus cover, but cannot fully penetrate optically dense clouds to reliably characterize their bases. These two observation data sets are therefore highly complementary. This presentation will show examples of these strengths and weaknesses for our verification purposes.

During our initial comparison of CloudSat and CALIPSO data to AFWA's cloud analysis, we found that in general the CALIPSO Level 2 cloud product detected less cloud cover than AFWA's analysis. Because AFWA's analysis is tuned for operational customers based upon their feedback, we chose to create a simple alternative Level-2-like product from the Level 1B 532-nm total attenuated backscatter data. The CloudSat Level 2 cloud product was used as-is; its resolution of about 2.5 km was comparable to the typical 3-4 km resolution of the AFWA pixel-level cloud analysis.

In addition to specific case studies, we will present some aggregate statistics resulting from automated matchups of the AFWA pixel-level cloud detections to the CloudSat and CALIPSO detections.