The correlation of multi-channel thermal infrared remote sensing information and its effect on retrieval of land surface temperature

The informational correlation among seven thermal infrared channels of MODIS is simulated. Results show that they have high correlation coefficient, which becomes one of the main obstacles to raise accuracy of LST retrieved by the remote sensing method. By means of principal component analysis we have proven 14 independent equations for retrieval of LST at most having five independent informative components, if seven bands are chosen for retrieval of LST, only three independent informative components exist. Therefore the property of retrieval of LST belongs to the underdetermined inverse problem. We cannot expect to get a stable and high accurate LST retrieved by various currently used remote sensing methods based on determined equations. From now on it is a main research topic to retrieve LST by underdetermined equations.