Simulation of split-window algorithm performance

Various methods for correction of the atmospheric influence in the thermal IR-windows 3-5 μm and 10-13 μm, have been proposed. One such method is the split-window technique for correction of AVHRR data. From earlier work, there are significant variations in the suggested coefficients of this algorithm. Over land surfaces, variations in emissivity have also to be compensated for. In this paper, the optimum choice of coefficients is investigated against a set of standard atmospheres using a prediction model including the emissivity effects of a land surface. It is obvious from the results that instead of using fixed standard coefficients estimated over a wide spread of atmospheric conditions, the optimum coefficients should be computed based on the best guess of the present atmospheric path parameters and on estimated emissivity values of the individual pixels