Analysis of Nocturnal Thermal Environments in Meso-Scale Region in Winter using Satellite Derived Thermal-IR Image

Abstract

Nocturnal thermal environments in Kanto and Koshin-etsu region of Japan in winter were analyzed using LANDSAT TM thermal-IR images. Four scenes in winter (14th Dec. 1984, 25th Feb. 1986, 26th Dec. 1986, 27th Jan. 1987) and one scene in summer (4th Aug. 1984) were investigated. Digital terrain model data, air temperature data at 68 AMeDAS stations and aerological observation data at Tateno were used simultaneously.
The LANDSAT thermal images were geometrically corrected to UTM coordinates by affin transformation. An atmospheric correction was performed on these images using LOWTRAN 7. Digital numbers of the thermal images were transformed into radiation brightness values using NASA model equation. Radiation brightness values were converted into surface temperatures according to the transmittanse and the pass radiance calculated from LOWTRAN 7.
Changes in surface temperature with elevations were closely related with the vertical profile of atmospheric temperature. The inversion layer due to radiative coolings appeared in the relation between surface temperature and elevation as well as in the relation between aerological data and elevation. The inversion layer did not appear in air temperature profile in the third scene.
Surface temperatures depended on the terrain structures. Scenes in winter showed the cold air lakes in valleys, low surface temperatures due to radiative coolings at plains and the warm layer at ravine hillsides. The cold air lake also existed in summer.
The range of area where the surface temperature affects the air temperature was about from 100 to 300 m. In winter, correlation coefficients between mean surface temperature in the range and air temperature showed the second maximum peak at larger range of several kilometers. Especially, the third scene showed the highest correlation coefficient in the scenes.
Surface temperatures were highly correlated with air temperatures, and the correlation coefficients were about 0.9. Spatial variations in surface temperatures were greater than those in air temperatures.