Tropical forests concentrate the largest diversity of species on the planet and play a key role in maintaining environmental processes. Due to the importance of those forests, there is growing interest in mapping their components and getting information at an individual tree level to conduct reliable satellite-based forest inventory for biomass and species distribution qualification. Individual tree crown information could be manually gathered from high resolution satellite images; however, to achieve this task at large-scale, an algorithm to identify and delineate each tree crown individually, with high accuracy, is a prerequisite. In this study, we propose the application of a convolutional neural network—Mask R-CNN algorithm—to perform the tree crown detection and delineation. The algorithm uses very high-resolution satellite images from tropical forests. The results obtained are promising—the $Recall$ , $Precision$ , and $F_1$ score values obtained were were $0.81$ , $0.91$ , and $0.86$ , respectively. In the study site, the total of tree crowns delineated was $\mathrm{59,062}$ . These results suggest that this algorithm can be used to assist the planning and conduction of forest inventories. As the algorithm is based on a Deep Learning approach, it can be systematically trained and used for other regions.

Extreme droughts have been recurrent in the Amazon over the past decades, causing socio-economic and environmental impacts. Here, we investigate the vulnerability of Amazonian forests, both undisturbed and human-modified, to repeated droughts. We defined vulnerability as a measure of (i) exposure, which is the degree to which these ecosystems were exposed to droughts, and (ii) its sensitivity, measured as the degree to which the drought has affected remote sensing-derived forest greenness. The exposure was calculated by assessing the meteorological drought, using the standardized precipitation index (SPI) and the maximum cumulative water deficit (MCWD), which is related to vegetation water stress, from 1981 to 2016. The sensitivity was assessed based on the enhanced vegetation index anomalies (AEVI), derived from the newly available Moderate Resolution Imaging Spectroradiometer (MODIS)/Multi-Angle Implementation of Atmospheric Correction algorithm (MAIAC) product, from 2003 to 2016, which is indicative of forest's photosynthetic capacity. We estimated that 46% of the Brazilian Amazon biome was under severe to extreme drought in 2015/2016 as measured by the SPI, compared with 16% and 8% for the 2009/2010 and 2004/2005 droughts, respectively. The most recent drought (2015/2016) affected the largest area since the drought of 1981. Droughts tend to increase the variance of the photosynthetic capacity of Amazonian forests as based on the minimum and maximum AEVI analysis. However, the area showing a reduction in photosynthetic capacity prevails in the signal, reaching more than 400 000 km ² of forests, four orders of magnitude larger than areas with AEVI enhancement. Moreover, the intensity of the negative AEVI steadily increased from 2005 to 2016. These results indicate that during the analysed period drought impacts were being exacerbated through time. Forests in the twenty-first century are becoming more vulnerable to droughts, with larger areas intensively and negatively responding to water shortage in the region. This article is part of a discussion meeting issue ‘The impact of the 2015/2016 El Niño on the terrestrial tropical carbon cycle: patterns, mechanisms and implications’.

We report large-scale estimates of Amazonian gap dynamics using a novel approach with large datasets of airborne light detection and ranging (lidar), including five multi-temporal and 610 single-date lidar datasets. Specifically, we (1) compared the fixed height and relative height methods for gap delineation and established a relationship between static and dynamic gaps (newly created gaps); (2) explored potential environmental/climate drivers explaining gap occurrence using generalized linear models; and (3) cross-related our findings to mortality estimates from 181 field plots. Our findings suggest that static gaps are significantly correlated to dynamic gaps and can inform about structural changes in the forest canopy. Moreover, the relative height outperformed the fixed height method for gap delineation. Well-defined and consistent spatial patterns of dynamic gaps were found over the Amazon, while also revealing the dynamics of areas never sampled in the field. The predominant pattern indicates 20–35% higher gap dynamics at the west and southeast than at the central-east and north. These estimates were notably consistent with field mortality patterns, but they showed 60% lower magnitude likely due to the predominant detection of the broken/uprooted mode of death. While topographic predictors did not explain gap occurrence, the water deficit, soil fertility, forest flooding and degradation were key drivers of gap variability at the regional scale. These findings highlight the importance of lidar in providing opportunities for large-scale gap dynamics and tree mortality monitoring over the Amazon.

In January 2020, an extreme precipitation event occurred over southeast Brazil, with the epicentre in Minas Gerais state. Although extreme rainfall frequently occurs in this region during the wet season, this event led to the death of 56 people, drove thousands of residents into homelessness, and incurred millions of Brazilian Reais (BRL) in financial loss through the cascading effects of flooding and landslides. The main question that arises is: To what extent can we blame climate change? With this question in mind, our aim was to assess the socioeconomic impacts of this event and whether and how much of it can be attributed to human-induced climate change. Our findings suggest that human-induced climate change made this event >70% more likely to occur. We estimate that >90,000 people became temporarily homeless, and at least BRL 1.3 billion (USD 240 million) was lost in public and private sectors, of which 41% can be attributed to human-induced climate change. This assessment brings new insights about the necessity and urgency of taking action on climate change, because it is already effectively impacting our society in the southeast Brazil region. Despite its dreadful impacts on society, an event with this magnitude was assessed to be quite common (return period of $\sim$4 years). This calls for immediate improvements on strategic planning focused on mitigation and adaptation. Public management and policies must evolve from the disaster response modus operandi in order to prevent future disasters.

Extreme droughts in Amazonia cause anomalous increase in fire occurrence, disrupting the stability of environmental, social and economic systems. Thus, understanding how droughts affect fire patterns in this region is essential for anticipating and planning actions for remediation of possible impacts. Focused on the Brazilian Amazon biome, we investigated fire responses to the 2010 and 2015/2016 Amazonian droughts using a remote sensing data. Our results revealed that the 2015/2016 drought surpassed the 2010 drought in intensity and extent. During the 2010 drought, we found a maximum area of 846,800 km2 (24% of the Brazilian Amazon biome) with significant (p<0.05) rainfall decrease in the first trimester, while during the 2015/2016 the maximum area reached 1,702,800 km2 (47% of the Brazilian Amazon biome) in the last trimester of 2016. On the other hand, the 2010 drought had a maximum area of 840,400 km2 (23% of the Brazilian Amazon biome) with significant (p<0.05) land surface temperature increase in the first trimester, while during the 2015/2016 drought the maximum area was 2,188,800 km2 (61% of the Brazilian Amazon biome) in the last trimester of 2015. Unlike the 2010 drought, during the 2015/2016 drought, significant positive anomalies of active fire and CO2 emissions occurred mainly during the wet season, between October 2015 and March 2016. During the 2010 drought positive active fire anomalies resulted from the simultaneous increase of burned forest, non-forest vegetation and productive lands. During the 2015/2016 drought, however, this increase was dominated by burned forests. The two analysed droughts emitted together 475 Tg CO2, with 233 Tg CO2 in 2010, 147 Tg CO2 in 2015 and 95 Tg CO2 in 2016, which represented respectively 209%, 136%, 82% of the Brazil’s national targets for reducing carbon emissions from deforestation by 2020 (1288 Tg CO2 from 1996-2005 average deforestation or a rate of 0.11 Pg CO2 year-1). Finally, we anticipate that the increase of fires during the droughts showed here may intensify and can become more frequent in Amazonia due to changes in climatic variability if no regulations on fire use are implemented.

Monitoring changes in tree cover for assessment of deforestation is a premise for policies to reduce carbon emission in the tropics. Here, a U-net deep learning model was used to map monthly tropical tree cover in the Brazilian state of Mato Grosso between 2015 and 2021 using 5 m spatial resolution Planet NICFI satellite images. The accuracy of the tree cover model was extremely high, with an F1-score >0.98, further confirmed by an independent LiDAR validation showing that 95% of tree cover pixels had a height >5 m while 98% of non-tree cover pixels had a height <5 m. The biannual map of deforestation was then built from the monthly tree cover map. The deforestation map showed relatively consistent agreement with the official deforestation map from Brazil (67.2%) but deviated significantly from Global Forest Change (GFC)’s year of forest loss, showing that our product is closest to the product made by visual interpretation. Finally, we estimated that 14.8% of Mato Grosso’s total area had undergone clear-cut logging between 2015 and 2021, and that deforestation was increasing, with December 2021, the last date, being the highest. High-resolution imagery from Planet NICFI in conjunction with deep learning techniques can significantly improve the mapping of deforestation extent in tropical regions.