Cumulus cloud base height estimation from high spatial resolution Landsat data: a Hough transform approach

This study develops a semiautomated methodology for estimating cumulus cloud base heights using high-spatial-resolution Landsat multispectral scanner data. The approach employs a variety of image processing techniques to match cloud edges with their corresponding shadow edges. Cloud base height is then estimated by computing the separation between the corresponding generalized Hough transform reference points. Sixteen subregions, each 30 km×30 km in size, are selected for four Landsat scenes. Standard deviations of cloud base height within each of the subregion range from about 100 m to 150 m. Differences between cloud base heights computed using the Hough transform and a manual verification technique are small. It is estimated that cloud base height accuracies of 50-70 m may be possible using HIRIS and ASTER instruments in the Earth Observation Satellite (EOS) Global Climate Change program