Satellite Image Analysis for the Geologic Structure and Land Surface Environments in the Chugaryung Rift Valley Near Cholwon

The surface environmental characteristics around the lava plateau of Cholwon plain in Chugaryung Rift Zone are studied comparing the geological research result, the topographical digital elevation distribution and the multispectral data of the Landsat TM. According to the digital terrain distribution, the last lava eruption is verified to have originated at the Mt. Ori, 2 km southwest away from Pyeonggang, and spreaded to outskirts, especially flowing southward along the Hantan River up to Jeonkok. The satellite multispectral data analysis could not present the detail geological distribution, but only the scheme of the boundary between the lava plateau and the bedrock, which is distinctive in topography. The general landcover is easily classified into vegetation, water and dried soil, showing their characteristics in band ratios such as 2 to 5 of B4/B3 for vegetation, 10 to 25 of B1/B7 for water, and 2 to 3 of B5/B2 for soil or cement cover in cities. NDVI analysis and forest survey in the mountains near Cholwon have shown around 0.15 for needle leaf forest indicating the highest value in April scene, whereas larger than 0.5 in September but less than 0.05 in April for deciduous forest. April's NDVI image suggests that the persistent needle leaf forest is very active in the mountains far west and near northeast from Pyeonggang in North Korea, and around Kalmalup in South Korea, but not active in the Demilitarized Zone. September's NDVI image shows that the vegetation is vigorous in DMZ and mountain areas rather than in plain areas cultivating for agriculture. It is remarkable that there are relatively many medium sized reservoirs in North Korea just north side of DMZ and the north and south boundaries of DMZ are distinguishable. Considering the pixel's resolution (30 m), the military guard route seems to be strong in spectral reflectance comparing with the environs in spite of its small size. The unsupervised landwver classification presents different patterns according to the year and season and then the classification is not so effective and need to be supported by any new approach.