Abstract
By applying the regression analysis, this paper investigated the relationship between the ground surface temperature observed by NOAA-AVHRR and various environmental factors such as elevation, albedo of ch. 1, albedo of ch. 2, vegetation index, cosy and latitude. y is the incident angle of the direct sun light to the ground surface. The test area was specified to Tohoku district, which is located in the most northern part of Honshu, Japan, and has the size of about 200 km × 600 km. Three mid-afternoon AVHRR images collected on April 17, 1988, June 12, 1986 and October 8, 1987 by NOAA-9 were used in the analysis. The sample points were selected at every 5 km distance, but points in the areas of water, snow and cloud, or with large amounts of data variation in their neighbor-hoods were excluded.
The elevation and the vegetation index appeared to be significant to all the three images, but other explanation variables were not consistent. The April image had the smallest data range of the ground surface temperature and the correlation coefficient remained to be 0.53. In the June image, the most effective variable was the vegetation index, but the two albedos were comparable to this. Its highest correlation coefficient was 0.85. The October image showed the highest correlation coefficient as 0.84, and the latitude was the most important variable. Large positive residues appeared in inland basins where the vegetation index are not so high, and large negative residues appeared in coastal regions which would be affected by sea wind.
