| Summary: | 碩士 === 國立彰化師範大學 === 地理學系 === 104 === Remote sensing data are widely used in land cover classification, crop productivity prediction, natural resources investigation, and hazard detection. There are some differences between the remote sensing image and the surface caused by the terrain relief and the atmospheric condition. Apart from the complicated light reaction in the air, some radiometric reflection from the surface may be influenced by shadows and other topographic effects. Topographic correction is one of the most important data preprocessing steps and choosing a suitable image calibration method is essential to research.
Topographic correction models (TCMs) are based on empirical, physical or semi-empirical theories. The Minnaert, cosine, SCS, C, and SCS+C were selected as the TMCs in this study, and some statistics data were used to assess the effectiveness in the Lienhuachih Experimental Forest. Results show that the DEM resolution not only affect the cosine value of the illumination angle, but also the parameter calculated from those TCMs. With the coarse DEM resolution data, parameter c decline and cosi increase, while parameter k seems to have no linear relationship from it. Topographic corrections make little difference from the original ones in the NDVI values, and they decrease in steep slopes. With the effectiveness assessment consequences, the SCS+C is the most appropriate method in the study area, and C model comes the next. Meanwhile, the Minnaert model cause some missing pixels in the corrected image.
|
|---|
