Abstract
The air temperature near the ground is very important to the growth of the rice plant becuase of its tropical origin. Therefore, information of the air temperature is very important for the cultivation of the rice plant. The purpose of the present study is to estimate the 1.5-m height air temperature in region where there is no meteorological observation site. The temperature estimate will be determined using satellite thermal IR data.
First, field measurements were made to determine the correlation between the air temperature and the surface temperature. The air temperature closely compared with the surface temperature, and the correlation coefficient for all the data was over 0.90. Further analysis was done to relate the air temperature to other meteorological parameters ; net radiation, humidity and wind speed. However, the best correlation coefficient was obtained using surface temperature. This suggests that the air temperature can be accurately estimated by the surface temperature from the satellite IR data.
Secondly, the surface temperature from the GMS IR data derived using equations (1) and (2) was compared with the AMeDAS air temperature in Hokkaido. The maximum correlation coefficient between the AMeDAS air temperature and the surface temperature determined using the GMS IR
data was 0.90, the minimum was 0.30 and the mean was 0.65±0.15.
Thirdly, the air temperature distributions in Hokkaido were estimated using the regression equation which relates the AMeDAS air temperature and the surface temperature determined using the GMS IR data (Fig. 3). The air temperature distributions in Ningxia, China were also estimated by using the GMS IR data.
Fourthly, the hourly mean deviation was calculated for the regression line obtained using the surface temperature from the GMS IR data and the AMeDAS air temperature in order to investigate the characteristic of the ground surface condition for each of the AMeDAS sites. The AMeDAS sites were classified according to their patterns for the hourly mean deviation. Using the results of this classification, the temperature distributions of the five areas in Hokkaido were estimated. The accuracy of the estimation using this method was 1.0±0.1 (°C).
