Assessment of different topographic corrections in AWiFS satellite imagery of Himalaya terrain

Abstract

The influence of topographic effects in optical satellite imagery is not investigated very extensively in the Himalayan terrain. The topographic variability causes a problem of differential illumination due to steep and varying slopes in rugged Himalayan terrain. Therefore, topographic corrections are essential for qualitative and quantitative analysis of snow cover applications. The present paper discusses the implementation of different topographic correction models on AWiFS sensor onboard IRS P6 satellite images and the qualitative and quantitative comparative analysis in detail. Both the Lambertian and non-Lambertian assumptions have been considered in the present analysis with the aim to explore best suitable empirical model for rugged terrain. The main topographic methods implemented are: Lambertian assumptions are found to be very unrealistic over Himalayan terrain as these lead to either underestimation or overestimation of physical parameters significantly both on sunlit slopes as well as the slopes away from the Sun. This problem is overcome by considering non-Lambertian assumption. Minneart constant and C-correction coefficients for all AWiFS satellite bands are estimated using regression analysis. All the results due to topographic effects are investigated qualitatively and quantitatively using four criteria namely visual analysis, validation with field measurements (in-situ observations), spectral reflectance of training samples of snow on the south and north aspects and graphically. The visual analysis confirms the minimization of three dimensional relief effects in two-stage normalization and slope matching methods and retrieves some of the information under mountain shadow. Due to the very bright surface of snow fields there is likely to be more diffuse reflected light in these areas than over darker vegetated surfaces. The qualitative analysis in other methods does not extract any information on shady slopes. The quantitative validation of topographic results in satellite imagery with in-situ observations shows underestimation of spectral reflectance of snow significantly except for slope matching technique. It is also apparent that although all the topographic methods correct the reflectance of training snow samples on the south and north aspects but most acceptable values are achieved using slope matching. The results obtained from graphical analysis reveal that mean reflectance after all topographic corrections are independent of illumination. This study also suggests that the suitability of topographic models can not be concluded as successful based on single criterion. Slope matching technique is the only technique which satisfies all the four criteria successfully and produces the best result for Himalayan terrain.