An Improved Method to Retrieve Surface Emissivity in a Canary Pine Forest Using Aster Data

Resumen

The widely-used Temperature and Emissivity Separation (TES) algorithm has been shown to provide reliable temperature and emissivity estimates from land-leaving thermal data. Nevertheless, TES has some important limitations mainly related to its inability to correct important inaccuracies for gray bodies, such as vegetation. In this study, this problem is addressed through an improvement to the current NEM (Normalized Emissivity Method) module included in the TES algorithm. The proposed method, called Modified NEM (ModNEM), has been specifically designed to retrieve an accurate surface emissivity for bodies with a spectral behavior typical of gray-bodies, i. e., flat and near spectrally invariant. ModNEM selects two different approximations of surface temperature instead of the usual maximum temperature, with the aim of accounting for the peculiar spectral behavior of brightness temperature and emissivity for vegetation. The NEM and TES, as well as the ModNEM, have been used to recover surface emissivity in a pine forest located in Tenerife Island (Canary Islands, Spain) using data from the Advanced Spaceborne Thermal Emission Reflection Radiometer (ASTER). Results have been compared and validated using reference emissivity values obtained from the pines by means of the box method. This validation study showed that high uncertainties are associated with the standard techniques (up to 0.034), whereas ModNEM results in lower uncertainties in emissivity estimates (< 0.005 for all ASTER channels). The results suggest that the current approximations in the existing NEM module are not suitable for surfaces with a spectral behavior similar to a grey body. This work was supported by MICINN under Grant CGL2010-22189-C02.