Estimativa da produtividade primária bruta no nordeste do Brasil com produtos MODIS e influência da base de dados meteorológicos.

Abstract

Gross primary production (GPP) is one of the fundamental components of the global carbon cycle, in this sense it has been the object of studies around the world, which have intensified in the context of climate change. In this research, the GPP estimates were analyzed for the Northeastern of Brazil biomes, in order to evaluate the MOD17A2H product (GPPmodis), and two more models recalculated based on meteorological data from the Instituto Nacional de Meteorologia (INMET) being 1: GPPmet combined with the fraction of photosynthetically active radiation of the MOD15A2H product (FPAR); and 2: GPPevi combined with the enhanced vegetation index from the MOD09A1 product (EVI09), for the years 2012 to 2018, covering the oceanic phenomena: El Niño, La Niña, positive gradient of the Atlantic in 2012 and years of neutral conditions and their relationships with the environmental and land use and change variables. The estimates were compared with the GPP measured in a micrometeorological tower (GPPtorre) at the Serra Talhada site inserted in the Caatinga biome for June 2014 to July 2015 period, from Ameriflux database. The results show that the interannual variation of GPPmodis was slightly influenced by the total annual precipitation conditioned by climatic events, and there were no influences from changes in land cover. All estimates presented a well-defined seasonality, due to this, there were correlations between them and with the environmental variables. Regarding the comparison with GPPtorre, the biggest errors were observed in GPPmodis and GPPevi, while GPPmet did not show significant differences (p-value ≤ 0.01) which may have been a reflection of the local calibration of photosynthetically active radiation (PAR). Therefore, GPPmet proves to be the best model, as long as the PAR percentage is adjusted, it can be applied to Caatinga biome, mainly due to the availability of these data without additional financial costs. However, the applicability to other biomes in the Northeast still needs to be further studied. With these results, it is suggested the refinement of the PAR data in MOD17A2H, including percentages closer to the reality of the biomes, mainly in the tropics, but in the absence of data, GPPmodis is the best alternative to obtain the GPP.