Cloud-Base Height Retrieval from MODIS Satellite Data Based on Self-Organizing Neural Networks

Abstract

An algorithm for retrieval of cloud-base height (CBH) from passive remote sensing data based on artificial intelligence methods is presented. Determining the CBH is considered as a special case of the classification problem. The algorithm is trained by comparing the results of active measurements of the CBH for single-layer clouds by the ground-based ceilometers (ASOS network), CALIOP lidar (CALIPSO satellite), and CPR radar (CloudSat satellite) with the cloud parameters obtained from the MODIS spectroradiometer (Aqua satellite). The results of estimating the capabilities of active tools to determine the CBH depending on the optical thickness of clouds are presented. The CBH retrieval algorithm is based on the use of three independent Kohonen neural networks trained on the data of the above devices. The results of determining the CBH for single-layer clouds by the developed classifier based on daytime MODIS images of the territory of Western Siberia obtained in summer are discussed. It is established that the algorithm generally underestimates the CBH. The average bias of the resulting scores from the ASOS/CALIOP/CPR reference data is −0.2 km with a standard deviation of 1.2 km.