A Correction Method of NDVI Topographic Shadow Effect for Rugged Terrain

Abstract

The influence of solar altitude and azimuth angles makes shadows prevalent in remote sensing images of rugged terrains. Consequently, the normalized difference vegetation index (NDVI) of the shadow areas is much lower than that of sunlit areas - a phenomenon known as the NDVI topographic shadow effect. In this study, we developed an NDVI topographic shadow effect correction (NTSEC) model. The NTSEC is based on the difference in solar radiation between the sunlit and shadow areas, and introduces a variable factor that indicates the intensity of shadows to simulate the reflectance for direct solar light not received in the shadow areas. The simulated reflectance for direct solar light in the shadow areas is used as a compensation value to be summed with the original reflectance for calculating the corrected NDVI. Landsat 8 Operational Land Imager images from two different regions were used to test the NTSEC method and multiple strategies were employed to provide objective evaluative results. The model performance was compared with four commonly used topographic correction (TC) models: C-correction, Minnaert, variable empirical coefficient algorithm, and statistical empirical. The results show that NTSEC can suppress the overcorrection of the self shadow areas that are difficult to eliminate in the TC methods, while simultaneously improving the undercorrection of the cast shadow areas. The corrected NDVI of the self shadow and cast shadow areas were almost the same. The NDVI differences between shady and sunny slopes were significantly reduced after correction. In addition, the NTSEC method did not produce outliers, and the NDVI of different land cover types retained interclass stability. In summary, the NTSEC model can be used as a simple and robust method for correcting the NDVI in shadow areas.