Mapeamento da severidade de incêndios florestais, perdas de carbono e análise das variáveis da paisagem que influencia a sua ocorrência em uma área da Amazônia Central

Abstract

There are estimates that forest fires in the Amazon will become more intense and frequent, with the increase in severe droughts and the existence of different ignition sources, such as deforestation and agricultural fire management. The central Amazon is known for having relatively preserved forest areas, compared to those near the “deforestation arc”. However, in 2015, a year of intense ‘El Niño’, it was observed that large areas were burned in this region. The main objective of this study was to analyze the spatial and temporal dynamics of forest fires and their impacts on the forest in an area of the Negro-Solimões interfluve. To this end, the decision thresholds of the ∆NBR spectral index were adjusted to map the severities of these fires. The “Normalized Burn Ratio” (NBR) spectral index, which uses the near-infrared and short-wave infrared bands, was created for mapping burned areas and is one of the most widely used indices worldwide, including in the Amazon. The adjustment was made with data collected in the field from areas with different severities, hemispherical photographs of the forest canopy, and Landsat images. Radiometric normalization was performed before calculating the ∆NBR. PRODES data were used to help construct the deforestation masks. Forest fires were mapped from 1995 to 2021. To assess the influence of environmental variables, land and river access routes, land cover, and land tenure on the spread of fire across the landscape, the Weight of Evidence Coefficients statistical method was used in the Dinamica EGO software. The Maximum Cumulative Water Deficit (MCWD) was also calculated. To assess changes in forest structure, a forest inventory was carried out with 5 plots (0.25 ha each) in each of the four severity classes, ‘Unchanged’, ‘Low’ (LS), ‘Moderate’ (MS) and ‘High Severity’ (HS). Tree individuals were divided into forest compartments such as live trees, dead trees, trees without canopy, live palms, dead palms and woody residues. Canopy openness data were also collected. The agreement between the field severity classification and the classification based on the thresholds adjusted for the ∆NBR spectral index was satisfactory (overall accuracy = 74.2%; Kappa coefficient = 0.635). If the classic Key & Benson thresholds were used, the fire severity classification would be underestimated. For the temporal analysis, the years with the most fires were those with El Niño events, 1997 (4,179 ha), 2009 (1,843 ha) and 2015 (17,475 ha). There was little repetition of burned areas, with 95%, 4% and 0.08% burning once, twice and three times, respectively. For 2015, the year with the largest burned area, 60.7%, 19% and 20% were classified as MS, HS and LS, respectively. The MS class has approximately 33% of canopy openness. Although 2015 had the largest burned area, its MCWD (-181 mm) was smallerr than that of 1997 (-235 mm) and 2009 (-224 mm), indicating that, among the variables analyzed in addition to drought, factors such as slope, proximity to roads and watercourses, indigenous lands and conservation units also influence the occurrence of fires. The severity classes respond similarly to the influence of the variables. The total dry biomass stock for Unchanged was 242 (± 60) Mg ha-1. The mortality rate was 29.4, 49 and 88.6% for LS, MS and HS. The largest committed CO2 emission occurred in 2015, 3,966 Gg CO2. Most of this emission occurred in MS (2,241 Gg CO2), the class that had the largest burned area. The results of the threshold adjustment corroborate the importance of this adjustment for each study area to classify fire severity using ∆NBR. Areas of the central Amazon that are considered relatively well preserved are subject to large losses due to forest fires during a severe drought event, even with relatively constant deforestation rates. The classification of fire severity contributes to improving estimates of carbon emissions and the impacts of fire on the structure of the Amazon forest.