Estimativa da produtividade primária bruta em uma floresta tropical sazonalmente seca

Abstract

Gross Primary Productivity (GPP) is characterized by the rate of carbon absorption during photosynthesis, providing crucial information about seasonal variations in the carbon cycle. GPP has been monitored worldwide through flux towers, using the Eddy Covariance (EC) technique, and by integrating clustered models with remote sensing data. This study aimed to evaluate different methods of estimating GPP from remote sensing, derived from Landsat 8 and MODIS (Moderate Resolution Imaging Spectroradiometer) data. The following models were tested: Vegetation Photosynthesis Model (VPM), Temperature and Greenness Model (TG), Vegetation Index Model (VI), Light Use Efficiency (LUE), and MOD17A2H product, at the Seridó Ecological Station (ESEC-Seridó), in Serra Negra do Norte municipality, RN, from January 1, 2014, to December 31, 2015. Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Land Surface Water Index (LSWI) were derived from Landsat 8, available on Google Earth Engine (GEE), and used as inputs to the models, integrated with micrometeorological variables. The models were compared with GPP measured by the Eddy Covariance technique (GPPEC). Overall, all models underestimated GPPEC, especially in the dry season, with MODIS model overestimating in both seasons, notably in the dry season, by approximately 95.5%. The results indicated that the TG model estimated using Landsat 8 data showed the best performance for the study area (seasonally dry tropical forest). The second-best performance was from the LUE models, also estimated from Landsat 8, considering the variability in light use efficiency. Therefore, the validation and comparison of models conducted in this study will provide valuable information for the development of future gross primary productivity estimation models, given the need for model calibration using observed data to reduce uncertainties arising from parameter usage.