Estimating Forest Surface Temperature: a Case Study in Canary Islands (spain)

The forest temperature is one key parameter in the models developed for providing early fire risk alerts. The satellite remote sensing data are the best way to obtain this parameter both regionally and globally. This study proposes an approach to obtain Forest Surface Temperature (FST) by means of Terra-ASTER (Advanced Spaceborne Thermal Emission and Reflectance Radiometer) multiband algorithms. We have proposed a multiband (MB) linear model and a Generalized Additive Model (GAM). The algorithms coefficients are obtained from simulated atmospheric data using the radiative transfer code MODTRAN 4.0 and 524 (381 day-time and 143 night-time) atmospheric soundings launched in Tenerife, Canary Island, Spain. These algorithms have been applied to a fully vegetated forest covered with Pinus canariensis. GAM algorithm showed improved residual statistics with respect to the MB both for night and day. Differences between ASTER Surface Kinetic Temperature Product (AST-08) and GAM are in the order of the standard performance of ASTER TES algorithm.