Obtaining carbon feedbacks in the big data era: a perspective to Central-north of Brazil

• Ano de defesa: 2020
• Autor(a) principal: Santos, Gustavo André de Araújo [UNESP]
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Carbon Cycle; Climate Change; MODIS; NASA POWER; OCO-2; SIF; Ciclo do carbono; Mudança Climática
• Link de acesso: http://hdl.handle.net/11449/204390

With the emergence of the era of earth observation in high resolution, remote sensing data's exponential growth has occurred in recent years. Thus, it is possible to carry out studies to monitor the carbon cycle on a global and regional scale, thus generating carbon feedback that can contribute to climate governance and decision-making to mitigate the effects that cause climate change. Therefore, this study was designed with the purpose of printing feedbacks related to the carbon cycle using Big Data in earth observation using remote sensing and climatic variables obtained by simulation models for the Central-North of Brazil, an important region due to the presence of important biomes such as the Amazonia, Cerrado, Pantanal and Atlantic Forest, besides being one of the most representative regions in the advancement of Brazilian agribusiness. Three studies were carried separately for the following sub-regions: Mato Grosso, Mato Grosso do Sul, and Eastern Amazonia. For each region were discussed different problems. A time series was analyzed from January 2015 to December 2018. The variables XCO2, Solar-induced fluorescence at 757nm, and 771nm were extracted from Orbiting Carbon Observatory-2 -OCO-2. The NDVI (MOD13A1), EVI (MOD13A1), and evapotranspiration (MOD16A2), data from Moderate Resolution Imaging Spectroradiometer - MODIS and climate variables (precipitation, wind speed, air temperature and relative humidity) as of Prediction of Worldwide Energy Resources - NASA POWER. The data were submitted to descriptive statistics, regression, correlation, temporal analysis and spatial interpolation with the kriging method and hotspots. It was observed that both in biomes and forest areas in Mato Grosso do Sul, the temporal variation of atmospheric CO2 concentration is mainly governed by photosynthesis (SIFR²adj. = 0.07-0.55; p<0.05, NDVIR²adj. = 0.18-0.63; p<0.05, and EVIR²adj. = 0.20-0.49; p<0.05) and that photosynthesis is positively related to evapotranspiration (R²adj. = 0.20-0.44; p<0.001) and air temperature (R²adj. = 0.26-0.44; p<0.001). In Mato Grosso and Eastern Amazonia, SIF was also an important variable in explaining the temporal variability of XCO2 (r= -0.84; p<0.01). However, in these regions, this relationship is also observed spatially. In general, the time variations of XCO2 in the north-central region of Brazil varies between the dry and rainy periods, this was clear in all studies. As for the spatial variation of XCO2, it varies according to the type of land use and the time of year. Given the results presented in this work, it is clear that the use of big data from remote sensing observations are valuable tools in understanding the carbon cycle since the relationships observed in the intersection of these data generate results that are clearly explained by the physical and biological processes around the soil-plant-atmosphere system.