PARENT SESSION
Poster Session #18: Remote Sensing and GIS.
Tuesday, August 7, 2001. Presentation from 3:30 PM to 5:00 PM. Exhibition Hall
87
Correlation of NDVI and topographic indices in a rugged mountain area of Korea.
Chang, HyunJeong1, Kang, Sinkyu1, Lee, Dowon1, 1
ABSTRACT- In this study, three image pre-processing techniques were compared to reduce the image distortion caused by rugged topography. Several topographic indices were also tested for explaining vegetation patterns of a rugged mountain area at both regional and landscape scales. The Minnaert's methods without atmospheric correction, with band ratio correction, and with bulk correction were applied for image pre-processing. NDVI from a Landsat TM image was used as a surrogate for vegetation patterns. Elevation, slope, skyview, and surface area index were used as topographic indices. Effective range of semivariogram, fractal dimension, and standard deviation of the topographic indices were also suggested as indices for topographic variability. Study area was divided into 154 sub-regions of 3km by 3km. All the indices were calculated in every sub-region. Minnaert's method with band ratio correction was the most effective to minimize the image distortion to 59.6%, compared with others, 47.9% and 55.8%. The elevation and standard deviation of slope were well correlated with mean NDVIs. Elevation and skyview were well correlated with standard deviation of NDVI. The results indicate that NDVI shows significant correlation with different topographic indices depending on the topography of sub-regions. Local topography has an influence on the spatial patterns of NDVI in rugged mountain areas
KEY WORDS: NDVI, Correlation analysis, Topographic index, Remote Sensing