UTILIZATION OF TANDEM-X DEM FOR TOPOGRAPHIC CORRECTION OF SENTINEL-2 SATELLITE IMAGE

Observing a Changing Earth; Science for Decisions— Monitoring, Assessment, and Projection

Dericks Praise Shukla., & Sharad Kumar Gupta.

2017-12-13

Topographic shadows of irregular mountains obstruct the analysis of satellite images in hilly areas. Topographic shading creates ambiguity among different elements of scene having similar hue and saturation but different intensities, which leads to misinterpretation of surface features. Due to this effect, there is high variability in the reflectance response of similar vegetation types i.e. sunny areas show more than actual reflectance, whereas shaded areas show less than expected reflectance. In this paper, the effects of digital elevation model (DEM) has been assessed on topographic correction of Sentinel-2 satellite images. TanDEM-X digital elevation model with spatial resolution of 12 meters is used in the study. DEMs was resampled at 10 meters to match the spatial resolution of satellite image. In this research work, six known methods of corrections are compared: the cosine, semi-empirical C, Minnaert, CHuang Wei, SCS + C and VECA correction depending on the solar incidence angle and exitance angle. Correlations and linear regressions between reflectance values of satellite image and topographic parameters (slope, aspect derived from DEM) are investigated. Ideally, the slope of regression line between illumination condition and image band should be zero, as topography should not control the solar illumination over terrain. However, in hilly regions, topography plays a critical role by controlling the amount of sunlight falling on north and south faces of mountains. Hence, after application of topographic correction methods, this slope of regression line decreases. Coefficient of variation, mean and standard deviation was used for the statistical analysis of the results obtained after topographic correction. The results of the study shows that classical cosine method overcorrect the image. As seen from the statistical analysis, VECA correction method provides good correction of topography. However, the C-Huang Wei, Minnaert, C and SCS + C-correction methods yielded an impressive reduction in the visual topography effects.