ANALYSIS OF LIGHT INTENSITY DATA BY THE DMSP/OLS SATELLITE IMAGE USING EXISTING SPATIAL DATA FOR MONITORING HUMAN ACTIVITY IN JAPAN
Keywords: GIS, Land Use, Urban, Classification, Data mining, Comparison, Building, Vector
Abstract. The main objective of this research is to show how much can be monitored various human activities using night light images by the DMSP/OLS from NOAA/NGDC. In Japan, various human activities can be monitored easily without satellite images because we can use many kinds of detailed spatial dataset and statistics. On the other hand detailed spatial data are not developed adequately especially in developing countries. Night light images by the DMSP/OLS can help to monitor them in such countries.
Therefore we discuss how to use night light images of the DMSP/OLS for this objective in Tohoku region, Japan. Human activates were explained by 3 factors, i.e. road distribution, accumulation of buildings and dynamic population. These data and light images of the DMSP/OLS were resampled into the same aggregate unit and compared with a light intensity of the DMSP/OLS. In addition it is shown which factor of human activates explains the light intensity more clear than other factors by multiple regression analyses using all factors. Results of multiple regression analyses show that impacts by road distribution are strong in urban and suburban areas and impacts by building are strong in rural areas. Impacts by dynamic population are weak in all areas. Finally estimated images of light intensities were developed using results of multiple regression analysis and they were compared with the actual image of light intensity. The compared result shows that tendency of spatial distribution of the light intensity by the estimated result agrees rather well with tendency by the DMSP/OLS.