Assessing the accuracy of MISR and MISR‐simulated cloud top heights using CloudSat‐ and CALIPSO‐retrieved hydrometeor profiles

Abstract

AbstractSatellite retrievals of cloud properties are often used in the evaluation of global climate models, and in recent years satellite instrument simulators have been used to account for known retrieval biases in order to make more consistent comparisons between models and retrievals. Many of these simulators have seen little critical evaluation. Here we evaluate the Multiangle Imaging Spectroradiometer (MISR) simulator by using visible extinction profiles retrieved from a combination of CloudSat, CALIPSO, MODIS, and AMSR‐E observations as inputs to the MISR simulator and comparing cloud top height statistics from the MISR simulator with those retrieved by MISR. Overall, we find that the occurrence of middle‐ and high‐altitude topped clouds agrees well between MISR retrievals and the MISR‐simulated output, with distributions of middle‐ and high‐topped cloud cover typically agreeing to better than 5% in both zonal and regional averages. However, there are significant differences in the occurrence of low‐topped clouds between MISR retrievals and MISR‐simulated output that are due to differences in the detection of low‐level clouds between MISR and the combined retrievals used to drive the MISR simulator, rather than due to errors in the MISR simulator cloud top height adjustment. This difference highlights the importance of sensor resolution and boundary layer cloud spatial structure in determining low‐altitude cloud cover. The MISR‐simulated and MISR‐retrieved cloud optical depth also show systematic differences, which are also likely due in part to cloud spatial structure.