地球観測衛星画像解析における表面形状データの空間分解能についての検討

抄録

Currently, many land observation satellites can be used in order to monitor natural environment. For satellite data analysis, reference data is a very important. For instance, DSM (Digital Surface Model) is one of the important data, that can make correct shading and shadow effects in the satellite imagery. Therefore, suitable ground sampling distance of DSM should be understand.
In this study, high resolution DSM by aerial LiDAR was used, that ground sampling distance showed 0.2m. A simple satellite image simulation was carried out by computing shading and shadow with the DSM and other field data including spectral reflectance of trees. The simulated satellite image was compared with real satellite image. The target satellite image was ALOS AVNIR-2. The ground sampling distance of ALOS AVNIR-2 is 10m. The standard deviation of the simulation showed less than 3.0 (W/m²/sr/micro m) in infra-red band.
The ground sampling distance of DSM was changed from 0.2m to 2.0m. Then, the standard deviation of the simulation showed over 15.0 (W/m²/sr/micro m). A permissive value of the accuracy in infra-red band was 5.1 (W/m²/sr/micro m). The value was computed from standard deviation of real satellite data in simple land cover area. Then, 0.2m ground sampling distance showed same standard deviation with real satellite data.