Abstract
An algorithm to simultaneously retrieve land surface temperatures and spectral emissivities in the microwave spectrum for applications with SSM/I measurements is developed and tested over the HAPEX-Sahel region. A closed but overspecified system of radiative transfer equations, written in terms of unknown surface temperatures and unpolarized spectral emissivities at 19, 37, and 85 GHz, is solved with a non-linear least-squares optimization technique. Clear-sky multispectral SSM/I brightness temperature measurements at full resolution pixel scale, obtained at least twice during a diurnal cycle (corresponding to the ascending and descending nodes of a DMSP satellite orbit), are used as input over a target area. It is assumed that the spectral emissivities remain constant during the time period under consideration. High-resolution temperature-moisture profiles obtained from the Météo France-CNRM balloon-launch station sited at Hamdallay are used to help account for atmospheric effects in the radiative calculations. Retrieved surface temperatures are consistent with ground-based radiometer skin temperature measurements and in situ soil temperature measurements obtained during the 1992 Intensive Observing Period (IOP) at the West-Central supersite for the 8 October golden day case. Qualitative agreement is also found with split window surface temperature estimates retrieved from AVHRR satellite measurements, although this cannot be considered a rigorous verification procedure because of uncertainties in the IR technique and dissimilar overpass times between DMSP and NOAA satellites. The microwave emissivity estimates are consistent with results reported in the published literature. The algorithm is applied over an extended North African region surrounding the HAPEX-Sahel study area for an ensemble of golden days to illustrate its potential for large-scale applications. As an independent test of its utility, output from the algorithm which requires representative estimates of surface temperature and emissivity in the forward radiative transfer model calculations. This test is successful in that rain events are correctly identified and their intensity is accurately retrieved in the course of the 1992 IOP, demonstrating the value of a simultaneous temperature-emissivity algorithm in aiding other types of terrestrial remote sensing schemes.