Impact of Finite Resolution Measurement on Low Cloud Climatology Derived from Satellite Data

It has been previously demonstrated that the cloud cover derived from meteorological satellites could be biased because clouds may partially fill the finite fields-of-view of the satellite instrument. However, to date, robust statistics from observations to show this bias do not exist. Examining this bias requires temporal and spatial coincident measurements from both high-resolution and low-resolution satellites, which can now been achieved using data from three satellite instruments onboard EOS Terra: the Moderate Resolution Imaging Spectroradiometer (MODIS), the Multiangle Imaging Spectral Radiometer (MISR), and the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER).

From September 2004 to March 2005, 157 ASTER scenes that contained only low clouds were collected with coincident MISR and MODIS observations. The time and location coincided with the Rain In Cumulus over the Ocean (RICO) experiment. For each ASTER scene, a cloud mask was generated at 15m resolution and treated as truth. In comparison to the truth, MODIS and MISR were overestimating low cloud cover by 22% and 36%, respectively, which are remarkably large when considering the climate sensitivity to low cloud cover. When looked at in term of a cloud mask's ability to detect fully and partially cloud filled pixels, MISR has an agreement rate of 84% relative to ASTER, while MODIS has only 37%. The implication of this in deriving low cloud properties and its dependence on the spatial distribution of the clouds will be demonstrated in conjunction with its potential impact on climate change research.