Application of satellite data to the studies of agricultural meteorology: Relationship between ground temperature from GMS IR data and AMeDAS air temperature

The purpose of the present study is to estimate air temperature in areas where there is no meteorological observation site, using satellite thermal IR data. Surface temperature from GMS IR data derived by eq. (1) was compared with AMeDAS (meteorological observation site) air temperature. The results are summarized as follows: 1) The maximum correlation coefficients between AMeDAS air temperature and surface temperature from GMS IR data is 0.90, the minimum is 0.30 and the mean is 0.60±0.15. 2) The correlation coefficients are affected by the precipitable water and decrease with increasing precipitable Water as shown in Fig. 2. 3) The correlation coefficients for each GMS observed time are better at night and in the morning than during the day (Table 2). 4) Also, the small values of the regression coefficients appear during the day and the large values at night and in the morning (Table 2). 5) The standard deviations which indicated scattering around the regression line are large at 12:00 and 15:00, but small at 06:00 and 09:00 (Table 2). The reason that correlation coefficients, regression coefficients and standard deviations between AMeDAS air temperature and surface temperature from GMS IR data are less during the day than at night and in the morning, is caused by ground conditions because the effects of solar radiation on surface temperature depend on ground surface conditions: plant cover, incline of slope etc. The hourly mean deviation from the regression line for surface temperature was calculated to investigate the characteristic of ground surface conditions for each AMeDAS observation site. AMeDAS observation sites were classified into four types according to the patterns of the hourly mean deviation as shown in Fig. 5. Most of type I were distributed in the plain regions: Ishikari, Konsen and Tokachi. Type II appears in the basin regions and type III on the coast of the Pacific Ocean and the Sea of Okhotsuk. The remaining areas are type IV. The standard deviations for all data, over 2.0°C and under 1.3°C were plotted on a map. Most sites with large standard deviations correspond to those of type III, while those with the small standard deviations correspond to sites of type I. The hourly changes of standard deviations are shown in Fig. 7. The large standard deviations appears on the coast of the Pacific Ocean and the Sea of Okhotsuk from 09:00 to 18:00. The sites of the small standard deviations appear more in the morning at 06:00 than during the day and at night. (author)