LST retrieval algorithm adapted to the Amazon evergreen forests using MODIS data

Abstract Amazonian tropical forests play a significant role in global water, carbon and energy cycles. Considering the importance of this biome and climate change projections, the monitoring of vegetation status of these rainforests becomes of significant importance. In this context vegetation temperature is presented as a key variable linked with plant physiology. In particular some studies showed the relationship between this variable and the CO2 absorption capacity and biomass loss of these tropical forests proving the potential use of vegetation temperature in the monitoring of the vegetation status. Nevertheless, the use of thermal remote sensing data over tropical forests still has some limitations being of special importance the atmospheric correction under very humid conditions and the possible high occurrence of cloudy pixels. In order to mitigate these limitations over the Amazon region, we present in this paper a new processing methodology to derive a LST product from Moderate Resolution Imaging Spectroradiometer (MODIS) data. The LST product was generated using a tuned split-window equation and cloud information derived from the Multi-Angle Implementation of Atmospheric Correction (MAIAC) algorithm. This LST product was validated using simulated and in situ data, and intercompared to the MODIS LST standard product (MOD11). Validation analysis shows that the new LST product reduces the RMSE by 0.6 to 1 K when compared to the MODIS standard LST product, mainly because of a reduction of the bias. We also show a preliminary intercomparison between MODIS and Visible Infrared Imaging Radiometer Suite (VIIRS) LST spatial patterns to illustrate the feasibility of VIIRS to extend forward the MODIS LST temporal series.