Base height estimation for low and high-level clouds from MODIS data

Abstract

The algorithm’s approbation results for estimating the cloud base height from passive remote sensing data from space are presented. We apply the technology of artificial neural networks. The algorithm combines two existing approaches in this area: the use of statistical relationships between the cloud base height and other cloud features, and the use of the "donor-recipient" concept. We apply the Kohonen self-organizing map as a classifier. CALIOP data (CALIPSO satellite) and MODIS data (Aqua satellite) are used at the training stage of the selected neural network. Retrieving of the cloud base height by a tuned classifier is already carried out only on the basis of the passive remote sensing results from space. The algorithm makes it possible to estimate the studied parameter for low and high-level clouds at  15 . We discuss the results of retrieving the cloud base height from MODIS satellite images obtained over the territory of Western Siberia in 2013.