基于地面红外检测系统验证的灌区地表温度遥感反演.

It is an important development trend in modern agriculture to utilize the remote sensing data and real-time field monitoring data for irrigation management, and to realize the agriculture informatization by using precision information technology. In this paper, in order to validate land surface temperature by remote sensing inversion, we designed and installed 4 sets of monitoring systems to collect field data on line, including crop canopy temperature, air temperature, air humidity, wind speed, solar radiation, soil moisture/temperature, and so on. The Jiefangzha Irrigation Region was selected as one of the research area, situated in the western part of the Hetao Irrigation District (4025N, 10709E). The other one was in the Daxing Experimental Station, Beijing (3937N, 11625E). The instruments were installed in the main agriculture crop fields (maize, spring wheat and sunflower) in Jiefangzha Irrigation Region, Inner Mongolia and in the rotation field of winter wheat-summer maize (Daxing Experimental Station, Beijing). The land surface temperature in the survey area was obtained by the infrared remote sensing inversion of Landsat7 and Landsat 8 in 2015. The land surface emissivity was determined by 2 methods, a simple estimation by Sobrino method and the Qin Zhihao method. Five pixels with 30 m×30 m each was selected around the monitoring system. The observed data at 11:00 and 12:00 by the instrument in the field was compared with the inversion results from remote sensing data. The results showed that the land surface temperature by the remote sensing inversion could agree well with the field crop canopy temperature. The monitoring data in situ could be the representative of the surrounding condition, which was about 90 m×90 m (5 pixels). The calculation of land surface emissivity based on Qin Zhihao method was suitable for different crops. The statistics parameters based on the Qin Zhihao method made a good performance in the sunflower field in 2015 with the coefficient of determination (R2), root mean square error (RMSE), relative error (RE) and Willmott index of 0.85, 1.97℃, 6.5% and 0.94, respectively. In the maize field, it was suitable in using the Sobrino method, with the R2, RMSE, RE and Willmott index of 0.76, 2.32℃, 7.8% and 0.92, respectively. The 2 methods had no significant difference in Daxing Station, Beijing. But the Sobrino method was better for the spring wheat in Jiefangzha Irrigation Region. The layout scheme and reasonable numbers of the monitoring systems, the drought diagnosis and irrigation management using multiple source data and the optimization and improvement of the monitoring system would be the key points to be studied in the future. [ABSTRACT FROM AUTHOR]