Topographic variations affect the reflectance properties of the Earth's surface and are often removed in remote sensing studies, especially when significant terrain variations exist. In this study, however, we show that shading effects assessed by Landsat can be treated as a signal that stores important topographic information, especially when the spectral characteristics of a surface are homogenous. The coastal transverse dunes of the Ashdod area, and the desert linear dunes of Nizzana (both located in Israel), were selected to investigate the abovementioned idea. The dune heights in these areas are 10 m on average (relative to their surroundings) and have maximum slopes of 33°. An innovative method for extracting slope, aspect, and height data for sand dunes using Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+) images was developed, based on the regularity and periodicity of dunes' landscapes. Using two Landsat images representing different sun zenith and azimuth angles, reflectance values of each image were converted to cos( i) values ( i=incident angle between the surface normal and the solar beam radiation), applying histogram matching methods. The slope and aspect of each pixel were determined as those that give the best prediction of the observed value of cos( i). Height profiles were then extracted, using simple trigonometric relationships. The accuracies of heights and slopes along selected profile lines were to the order of 1 m and 3°, respectively (at a spatial resolution of 15 m). Best results were obtained when the images included one from the summer and the other from the winter, corresponding to maximum difference in solar zenith and azimuth angles. Errors in heights were attributed to surface heterogeneity (e.g., presence of biogenic soil crusts in the rainy season), geometric correction errors, cast shadows, and Bidirectional Reflectance Distribution Function (BRDF) effects. Comparison to Advanced Thermal Emission and Reflection Radiometer (ASTER) 3D information showed that the proposed method is better in representing the topographic variation of the area than the digital elevation model (DEM) produced by ASTER.
Read full abstract