Retrieval of surface temperature in boreal forest zone from SSM/I data

Novel inversion methods for the retrieval of surface temperature (air temperature at ground level) in forested areas using space-borne multi-channel microwave radiometer data are presented and analyzed. The first technique is an inversion method based on the use of a constrained least squares algorithm for the inversion of a semi-empirical emission model. The other methods discussed are empirical approaches: multiple linear regression and polarization difference formulas. The validity of the inversion method, as well as the feasibility of the empirical approaches, are evaluated in the case of Finnish boreal forests employing SSM/I data. The results show that for conifer dominated boreal forests the surface temperature can be estimated reliably from SSM/I measurements during snow-free conditions. The highest test site-wise determined correlation coefficients (r) between the ground-based reference values (near-surface air temperature) and the SSM/I-based estimates are above 0.97 and the corresponding unbiased rms errors are smaller than 1.3/spl deg/C. These values were obtained using morning time overpasses of the SSM/I without any data rejection.