Abstract
ASTER stereo and ASTER infrared (IR) retrieved cloud top heights (CTHs) at 90 m spatial resolution are compared to operational Multi-Angle Imaging SpectroRadiometer (MISR) stereo and MODerate Resolution Imaging Spectroradiometer (MODIS) thermal IR techniques at 1100 m and 5000 m spatial resolution, respectively. ASTER data availability limits this study to trade wind cumulus clouds only. ASTER IR, MISR stereo and MODIS IR cloud height frequency distributions were derived for 41 trade wind cumulus cloud scenes with no cirrus contamination. All three retrievals produce a bimodal CTH distribution with maxima at 650 and 1500 m.
Several scenes are investigated in further details utilizing the ASTER stereo heights. The ASTER stereo-derived CTH frequency distribution is also bimodal and coincides best with a surface-based ceilometer-derived cloud base height distribution, in the sense that it produces the fewest CTH retrievals below the typical lidar-observed cloud base height. Sensitivity to spatial resolution was investigated by degrading ASTER stereo and IR CTH retrievals to the MISR spatial resolution, and also by deriving ASTER stereo heights from degraded radiances. The degraded ASTER stereo CTH retrievals remain bimodal, but showed less high clouds and more lower clouds, whereas the ASTER stereo heights from degraded radiances lose the bimodal structure similarly to the ASTER degraded resolution IR technique.
Overall, the stereo technique appears more suitable for retrieving CTH for trade wind cumulus clouds than IR techniques, because it shows less sensitivity to resolution. With either technique CTHs should be produced at the high spatial resolution (finer than 1 km) and with subsampling preferred over spatial averaging, if necessary. Nonetheless, even at a spatial resolution of 1 km (which often exceeds the horizontal size of many trade wind cumulus clouds) both stereo and IR techniques reflect the CTH bimodality characterizing the scenes we studied.
