An Objective Regional Cloud Mask Algorithm for GOES Infrared Imager with Pixel-Dependent Thresholds

A cloud mask (CM) algorithm is required for identifying cloud-free pixels for direct assimilation of infrared radiance observations from Geostationary Operational Environmental Satellites (GOESs) in mesoscale forecast modeling systems. In this study, a regional, pixel-dependent CM algorithm is developed in which the threshold for each CM test for a target pixel is determined by a one-dimensional optimization approach based on probability distribution functions of the nearby cloudy and clear-sky pixels within a 10ox10o box centered at the target pixel. and tested for GOES-12 imager infrared channels. The purpose of the proposed CM algorithm is to isolate cloud-free pixels from cloudy pixels for data assimilation using mesoscale forecast models. In this CM algorithm, the implicit dynamic thresholds are determined at pixel-resolutions. An innovative aspect of this algorithm is that the thresholds for each CM test are determined by a one-dimensional optimization approach considering the local distribution of clouds. It is observed that the distribution of optimized thresholds over land displays more variation than over ocean in addition to larger average threshold values. The performance of the new developed CM algorithm is evaluated by comparing with Moderate Resolution Imaging Spectroradiometer (MODIS) CM during the period from May 19 to May 23 in 2008. Totally 5,616,090 GOES-12 Infrared Imager pixels are tested and the average Probability of Correct Typing (PCT) based on MODIS CM is 92.94% over land, and 91.50% over ocean respectively.

Index Terms—quality control, cloud mask, GOES, optimization.