Comparison of Cloud Properties from CALIPSO-CloudSat and Geostationary Satellite Data

Cloud properties are being derived in near-real time from geostationary satellite imager data for a variety of weather and climate applications and research. Assessment of the uncertainties in each of the derived cloud parameters is essential for confident use of the products. Determination of cloud amount, cloud top height, and cloud layering is especially important for using these real -time products for applications such as aircraft icing condition diagnosis and numerical weather prediction model assimilation. Furthermore, the distribution of clouds as a function of altitude has become a central component of efforts to evaluate climate model cloud simulations. Validation of those parameters has been difficult except over limited areas where ground-based active sensors, such as cloud radars or lidars, have been available on a regular basis. Retrievals of cloud properties are sensitive to the surface background, time of day, and the clouds themselves. Thus, it is essential to assess the geostationary satellite retrievals over a variety of locations. The availability of cloud radar data from CloudSat and lidar data from CALIPSO make it possible to perform those assessments over each geostationary domain at 0130 and 1330 LT. In this paper, CloudSat and CALIPSO data are matched with contemporaneous Geostationary Operational Environmental Satellite (GOES), Multi-functional Transport Satellite (MTSAT), and Meteosat-8 data. Unlike comparisons with cloud products derived from A-Train imagers, this study considers comparisons of nadir active sensor data with off-nadir retrievals. These matched data are used to determine the uncertainties in cloud-top heights and cloud amounts derived from the geostationary satellite data using the Clouds and the Earth s Radiant Energy System (CERES) cloud retrieval algorithms. The CERES multi-layer cloud detection method is also evaluated to determine its accuracy and limitations in the off-nadir mode. The results will be useful for constraining the use of the passive retrieval data in models and for improving the accuracy of the retrievals.