Filling satellite image cloud gaps to create complete images of evapotranspiration

摘要

Satellite images often have clouds in portions of the images. When estimating vegetation consumptive water use using the surface energy balance method METRIC, the evapotranspiration, expressed as ET_rF = ET / ET_r, where ET_r is tall reference evapotranspiration computed from weather data, ET_rF cannot be directly estimated for these areas because cloud temperature masks surface temperature and cloud albedo masks surface albedo. ET_rF for clouded areas must be filled in before application of further integration processes so that those processes can be uniformly applied to an entire image. A linear interpolation is used to fill in ET_rF for clouded portions of images. The linear interpolation is used rather than curvilinear interpolation, such as the spline introduced later to interpolate between cloud-corrected images, because some periods between cloud-free pixel locations can be as long as several months. Often, the change in crop vegetation amount, and thus ET_rF, is uncertain during that period. Thus the use of curvilinear interpolation can become speculative.