Synchronous retrieval of land surface temperature and emissivity

This is an old topic for more than ten years to retrieve land surface temperature (LST) from satellite data, but it has not been solved yet. At first, people tried to transplant traditional split window method of sea surface temperature (SST) to the retrieval of LST, but it was found that the emissivities of land surface (epsilon(i)) must be involved in atmospheric correction. Then many different formulas appeared with assumption of emissivities known. In fact, emissivities of land surface with pixel size cannot be known beforehand because of various reasons, so in recent years the focus of attention has been transferred to retrieving emissivities (epsilon(i)) and LST at the same time. Therefore. we have to solve missing equations problem. For this same people try to introduce middle infrared information, but new problems will be brought in which means that it is very difficult to describe middle infrared BRDF of targets with high accuracy and the scattering of atmospheric aerosol cannot be ignored. Therefore a different way is offered to solve this problem only using two thermo-infrared bands data based on three assumptions, constant emissivities in two measurements, and the same atmospheric parameters for neighbouring pixels and the difference of emissivity (Delta e) of two channels can be known beforehand. Results of digital simulations show that it is possible to retrieve LST with its root mean square (RMS) of errors less than I K and RMS of relative error of ground radiance at 7 parts per thousand if the error of atmospheric temperature at +/-2 degrees C and the relative error of atmospheric water vapor at +/- 10 % can be satisfied. Results have been confirmed by initial field test.