|
Wide availability of videodata recorded with the AVHRR instrument used onboard NOAA satellites and frequent recurrence of images of examined fragments of the Earth's surface stimulate attempts at solving problems of resource- ecological monitoring based on these data. Unfortunately, poor spatial resolution and significant distortions of geometrical and radiobrightness characteristics of these images make their analysis by algorithms of thematic processing difficult. Because resource-ecological monitoring includes first of all temporal analysis of adjusted videodata, a problem of automatic mutual superimposition of images of an examined region arises, because these image are recorded from different turns of the satellite orbit. In its turn, this problem raises a problem of obtaining equal resolutions of images. An integrated image correction procedure includes the normalization of illumination of images of the Earth's underlying surface recorded in the daytime with consideration of the solar zenith angle and the calibration of video data with the determination of the albedo (in channels 1 and 2) and thermodynamic temperatures (in channels 3, 4, and 5). The elimination of geometrical distortions takes into account variations in the projection of a scanning spot of the AVHRR instrument on a cylinder and includes the recalculation of radiobrightness for a fixed spot diameter. Cartographic fixing of the data and mutual adjustment of images recorded from different turns of the satellite orbit are carried out by minimization of a quadratic criterion of the discrepancy of contour gradients of these images. The integrated image correction allows ensembles of images affixed to the coordinates to be accumulated and their temporal variations to be analyzed to detect forest fires in real time and to forecast the weather against the background of natural phenological changes in portraits of the Earth's underlying surface (EUS).
|
