Comparison of marine stratocumulus cloud top heights in the southeastern Pacific retrieved from satellites with coincident ship‐based observations

In order to better understand the general problem of satellite cloud top height retrievals for low clouds, observations made by NOAA research vessels in the stratocumulus region in the southeastern Pacific during cruises in 2001 and 2003 to 2006 were matched with near‐coincident retrievals from the Moderate Resolution Imaging Spectroradiometer (MODIS) and Multiangle Imaging SpectroRadiometer (MISR) instruments on the Terra satellite, along with a limited set of ISCCP 30‐km (DX) retrievals. The ISCCP cloud top heights, determined from the cloud top pressures, were found to be biased high by between 1400 and 2000 m within the limited comparison data set. Like the International Satellite Cloud Climatology Project (ISCCP) results, the MODIS retrievals were biased high by more than 2000 m, while the MISR retrievals had errors on the order of 230 to 420 m, with the wind corrected heights having almost no bias. The extremely large bias in the ISCCP and MODIS retrievals was traced to their reliance on low‐resolution observations or models of the atmospheric temperature structure. Cloud top height retrievals based on satellite cloud top temperatures and a constant atmospheric lapse rate agreed substantially better with the ship‐based measurements.