A procedure for the detection and removal of cloud shadow from AVHRR data over land

Although the accurate detection of cloud shadow in AVHRR scenes is important for many atmospheric and terrestrial applications, relatively little work in this area has appeared in the literature. This paper presents a new multispectral algorithm for cloud shadow detection and removal in daytime AVHRR scenes over land. It uses a combination of geometric and optical constraints, derived from the pixel-by-pixel cross-track geometry of the scene and image analysis methods to detect cloud shadow. The procedure works well in tropical and midlatitude regions under varying atmospheric conditions (wet-dry) and with different types of terrain. Results also show that underdetected cloud shadow ran produce errors of 30-40% in observed reflectances for affected pixels. Moreover, radiative transfer calculations show that the effects of cloud shadow are comparable to or exceed those of aerosol contamination for affected pixels. The procedure is computationally efficient and hence could be used to produce improved weather forecast, land cover, and land analysis products. The method is not intended for use under conditions of poor solar illumination and/or poor viewing geometry