- 著者
-
北村 彩子
泉 岳樹
松山 洋
- 出版者
- 公益社団法人 東京地学協会
- 雑誌
- 地学雑誌 (ISSN:0022135X)
- 巻号頁・発行日
- vol.113, no.4, pp.495-511, 2004-08-25 (Released:2009-11-12)
- 参考文献数
- 29
- 被引用文献数
-
2
The thermal infrared images observed by satellites represent integral of radiations from both surface and atmosphere. This has been pointed out qualitatively, however, it has not been clarified quantitatively. Using Landsat-5 TM images (Kanto scene, Path107, Row35), this study quantitatively investigated the ratio of the radiant flux densities of surface temperature and those of air temperature. A multiple regression analysis was applied in this investigation. Four daytime scenes of the thermal infrared images of Landsat-5 TM (28 Feb1992, 25 Feb 1997, 13 Dec 1998, 30 Jan 1999, all were fine), and meteorological data in meteorological observatories, AMeDAS stations and Terrestrial Environment Research Center, University of Tsukuba were used for the analysis. Generally, surface temperature around10 : 00 JST is not observed when Landsat passes the study area, so the diurnal variation of thesurface temperature and energy budget at each site was calculated by the method of Kondo (1992) who set exchange coefficient constant throughout a day.It was clarified that the radiant flux densities of surface temperature and those of air temperature equally contribute to the radiant flux densities of brightness temperature observed by Landsat-5 TM, except for a case of strong wind since the constant value of exchange coefficient was not appropriate in this case. In the case of 13 Dec 1998, correlation between brightness temperature and air temperature, obtained in this study (r=0.71) was better than that of Yan and Mikami (2002) (r=0.53) who analyzed the same thermal infrared images. This was due to the difference of the area studied. In this case, correlation betweenradiant flux densities of brightness temperature and those of air temperature were also 0.71. Moreover, the multiple correlation coefficient among brightness temperature, surface temperature and air temperature (r=0.76), and radiant flux densities of brightness temperature, that of both surface temperature and air temperature (r=0.76) was better than the single correlation coefficients between brightness temperature and air temperature, and radiant flux densities of them. Since AIC (Akaike's Information Criterion) of the multiple correlation analysis was smaller than that of the single correlation analysis, this study statistically showed that the radiant flux densities of brightness temperature observed byLandsat-5 TM represented equal contribution of both surface temperature and air temperature.