Detection of cloud cover using dynamic thresholds and radiative transfer models from the polarization satellite image

Abstract

The detection of cloud using satellite observations is important for retrieval algorithms, image visualization and climate applications. Identifying the presence of cloud and establishing an accurate cloud cover results is a challenging task. Here, we use data from the PARASOL satellite data. In this paper, we describe a new algorithm to detect cloud based on multi-spectrum and polarization characteristics of the polarization images. This uses the new dynamic thresholds obtained by statistics for different atmosphere models and underlying surfaces in different time and areas. Aiming at improving the accuracy of the final cloud detection, especially for those special scenes, such as bright land surface and severe haze. Also, an ice/snow detection algorithm has been adapted from the cloud detection algorithm.A validation for the new cloud detection results has been carried out. In comparisons with other satellite instruments, such as the Moderate Resolution Imaging Spectroradiometer (MODIS), the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite (CALIPSO) and CloudSat. Experiment results show the new dynamic cloud detection algorithm has obvious advantages over the official cloud detection algorithm to the whole track results in accuracy, especially when in some areas covered by the bright cloud-free objects. The overall accuracy compared with CloudSat/CALIPSO is 95.07%. The new algorithm improves the accuracy of cloud detection results and constructs a new theoretical model for the cloud detection algorithm of the multi-angle polarization satellite image, such as the Directional Polarimetric Camera (DPC) launched onboard the GF-5 (GaoFen-5) Satellite.