Topographic correction of NDVI in mountainous regions

The Normalized Difference Vegetation Index (NDVI) is based on the normalized difference of reflectivity between red and NIR spectra. But even the same land cover will exist significant variation in apparent reflectivity because the difference of phototropic and apheliotropic, especially in mountainous regions. Since vegetation has stronger reflectance within the NIR, the influence is obvious than other bands. The objective of this study is to rectify the topographic effect on NIR spectral band based on Non-Lambertain Assumption, which considers the surface bi-directional reflectivity. Minnaert factor is used to correct the discrepancy. The incident angle, exitance angle and the terrain topography were parameters considered in this method. The study area is Shimen reservoir watershed in Taiwan, the major land cover here is forest. SPOT 5 images, 40m DEM and a land use map were used in this research. The study procedure could be divided into five parts: First is converting the image data from digital number (DN) to radiance for normalization. Secondly, solar angle and terrain aspect were used to define the phototropic and apheliotropic sides. In this step, shadow and non-vegetation area were masked in the same time. Third, different vegetations with solar phases could be classified using a land use map. After that, Effective Incidence Angle (EIA) and Effective View Angle (EVA) could be calculated according to the Minnaert correction. Finally, the effective normal radiometric response could be rectified after the Minnaert factor. The result shows the NDVI variation of forest canopy between phototropic and apheliotropic areas can be well corrected. The topographic correction proposed in this study is efficient in correcting NDVI values in mountainous regions.