| Remotely sensed land surface temperature(LST) is an important parameter description of the land surface and atmosphere energy exchange. It was often used in soil evapotranspiration, crop yield estimation, environmental monitoring and other research areas. The accuracy of the LST directly or indirectly affects the reliability of the research results. But limited by the small observing area, the in situ measured data had a deficient representation. Because of the scale effect, the validation of the heterogeneous LST will face a lot of uncertainty. In addition, it will be affected by the method which was used on processing the in situ measured data. The accuracy of the in situ measured data will directly affect the LST validation. Therefore, the LST validation process was composed of several links. Both the scale effect and the accuracy of the in situ measured data should be considered.In this paper, the downstream of Heihe River basin in northeastern Ejin Banner which was covered by sparse vegetation was selected as the study area. This paper used the thermal infrared radiation data which was measured in the area, the four component outgoing longwave radiation data, the emissivity data, and satellite remote sensing data as study data. The methods of converting the in situ measured data to the LST, extracting the component temperature and improving the spatial representation of the in situ measured data were studied.The influence of atmospheric factors was mainly considered when studying the method of converting the in situ measured data to the LST. In this paper, the MODIS(Moderate Resolution Imaging Spectroradiometer) atmospheric profiles product was used to simulate the transmission, the upward longwave radiation and the downward longwave radiation of atmosphere in different zenith angle and height. An estimation method which can obtain high time resolution downward radiation in 8-14μm was established by using downward long wave radiation data and simulation of atmospheric parameters. Through the simulation results, we can find that the upward radiation measured by the four component net radiation sensor was strongly influenced by the atmospheric transmittance and the upward radiation of the atmosphere. At 6m height, the upward radiation of atmosphere is 40-80W/m2, and atmospheric transmittance is about 80%. At 24 m height, the upward radiation of atmosphere is about 110W/m2, and atmospheric transmittance is about 70%. It will bring over 5 K deviations if ignoring the influence of the two atmospheric parameters.The portable thermal imager can obtain the thermal infrared temperature image quickly. It was often used to measure LST. In this paper, a method of extracting component temperature from thermal imager was established. The results show that the component temperature extracted by this method present a good consistency with stationary thermodetector data.This paper established relationship between the LST and both the multiband reflectance data and coefficient of variation by using fitting method. The results show that the fitting relationship can well describe the LST. For two daytime ASTER images, the square of correlation coefficient is above 0.7 and RMSEs(root mean square error) are about 1.9 K and 3.2 K respectively. For two daytime Landsat 8 images, the square of correlation coefficient is above 0.88, and RMSEs are 1.2 K and 1.5 K respectively. At the same time, it shows similar calculation accuracy when the fitting relationship was applied to different spatial resolution data. It means that the fitting relationship was scale invariance. Use the component temperature and fitting relationship can improve the scale space representation of the in situ measured data. The low spatial resolution remote sensing LST data can be verified. |
PDF Full Text Request