Abstract
The land surface temperature (LST) and emissivity (LSE) derived from Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data were evaluated in a low spectral contrast volcanic site at an altitude of 2000 m on the island of Tenerife, Spain. The test site is almost flat, thermally homogeneous, and without vegetation cover or variation in its surface composition. ASTER data correspond to six scenes, under both day- and nighttime conditions during 2008. This case study analyzes the impacts of the sources of inaccuracies using the temperature–emissivity separation (TES) algorithm. Uncertainties associated with inaccurate atmospheric correction were minimized by means of local soundings and the climate advantages of the area.
Concurrent ground-based radiometric measurements were performed for LST, and laboratory and field measurements for LSE, to obtain reference values. The TES evaluation showed a good level of agreement in the emissivity derived for ASTER bands 13 and 14 [root-mean-square difference (RMSD) lower than 0.002] and discrepancies in ASTER bands 10 and 11 that were within the expected performance of the algorithm (±0.015). However, out-of-threshold errors were retrieved in band 12, producing an artificial increase in spectral contrast. The underestimated TES LSE spectra point to the presence of a roughness effect at measurement scales that may increase the laboratory band emissivity values. TES LST comparison with ground data showed an RMSD value of 0.5 K. ASTER standard products AST08 (LST) and AST05 (LSE) atmospherically corrected by means of Naval Research Laboratory (NRL) data were also tested, showing a similar level of performance for the TES implemented with local soundings, but failed in high-humidity atmospheric conditions.
Corresponding author address: África Barreto, Dept. de Física FEES, Universidad de La Laguna, 38206 La Laguna, Spain. Email: afriba@ull.es
