Wildfires are a critical driver of landscape transformation on Earth, representing a dynamic and ephemeral process that poses challenges for accurate early detection. To address this challenge, researchers have increasingly turned to deep learning techniques, which have demonstrated remarkable potential in enhancing the performance of wildfire detection systems. This paper provides a comprehensive review of fire detection using deep learning, spanning from 1990 to 2023. This study employed a comprehensive approach, combining bibliometric analysis, qualitative and quantitative methods, and systematic review techniques to examine the advancements in fire detection using deep learning in remote sensing. It unveils key trends in publication patterns, author collaborations, and thematic focuses, emphasizing the remarkable growth in fire detection using deep learning in remote sensing (FDDL) research, especially from the 2010s onward, fueled by advancements in computational power and remote sensing technologies. The review identifies “Remote Sensing” as the primary platform for FDDL research dissemination and highlights the field’s collaborative nature, with an average of 5.02 authors per paper. The co-occurrence network analysis reveals diverse research themes, spanning technical approaches and practical applications, with significant contributions from China, the United States, South Korea, Brazil, and Australia. Highly cited papers are explored, revealing their substantial influence on the field’s research focus. The analysis underscores the practical implications of integrating high-quality input data and advanced deep-learning techniques with remote sensing for effective fire detection. It provides actionable recommendations for future research, emphasizing interdisciplinary and international collaboration to propel FDDL technologies and applications. The study’s conclusions highlight the growing significance of FDDL technologies and the necessity for ongoing advancements in computational and remote sensing methodologies. The practical takeaway is clear: future research should prioritize enhancing the synergy between deep learning techniques and remote sensing technologies to develop more efficient and accurate fire detection systems, ultimately fostering groundbreaking innovations.

Dryland regions around the world are facing intricate challenges due to climate change and human activities. The Caatinga biome in Brazil, an exceptional dryland ecosystem covering approximately 86.3 million hectares, is particularly impacted by human influence. We conducted an extensive study analyzing changes in land use and land cover within the Caatinga region over a span of 35 years, from 1985 to 2019. This study leverages collective knowledge and collaborative effort with the MapBiomas project to provide valuable insights into the biome’s landscape. It maps eight principal land cover classes using Landsat Collection 1 Tier 1 data normalized to top-of-atmosphere reflectance. All data processing was carried out within the Google Earth Engine platform, and the graphics were generate using R version 3.6.2. This study achieved an impressive 80% global accuracy in the time series of Caatinga land use and land cover (LULC) changes, with allocation and area discrepancies of 11.6% and 8.5%, respectively. The extensive 35-year LULC dataset reveals a substantial 11% reduction in natural vegetation in the Caatinga biome, translating to a loss of 6.57 million hectares. This decline is primarily attributed to the expansion of cattle ranching and agriculture; all types of natural vegetation have experienced decreases, with Savanna Formation (SF) areas declining by 11% and Forest Formation (FF) areas declining by 8%. In contrast, pasturelands expanded by 62% and agricultural land expanded by 284% during this period. With their urgent and significant real-world for informing social, economic, and environmental policy decisions within the Caatinga and other dryland regions globally, these findings underscore the importance and immediacy of our research.

This study provides a comprehensive overview of Phase I of the deforestation dryland alert system. It focuses on its operation and outcomes from 2020 to 2022 in the Caatinga biome, a unique Brazilian dryland ecosystem. The primary objectives were to analyze deforestation dynamics, identify areas with highest deforestation rates, and determine regions that require prioritization for anti-deforestation efforts and conservation actions. The research methodology involved utilizing remote sensing data, including Landsat imagery, processed through the Google Earth Engine platform. The data were analyzed using spectral unmixing, adjusted Normalized Difference Fraction Index, and harmonic time series models to generate monthly deforestation alerts. The findings reveal a significant increase in deforestation alerts and deforested areas over the study period, with a 148% rise in alerts from 2020 to 2022. The Caatinga biome was identified as the second highest in detected deforestation alerts in Brazil in 2022, accounting for 18.4% of total alerts. Hexagonal assessments illustrate diverse vegetation cover and alert distribution, enabling targeted conservation efforts. The Bivariate Choropleth Map demonstrates the nuanced relationship between alert and vegetation cover, guiding prioritization for deforestation control and native vegetation restoration. The analysis also highlighted the spatial heterogeneity of deforestation, with most deforestation events occurring in small patches, averaging 10.9 ha. The study concludes that while the dryland alert system (SAD-Caatinga—Phase I) has effectively detected deforestation, ongoing challenges such as cloud cover, seasonality, and more frequent and precise monitoring persist. The implementation of DDAS plays a critical role in sustainable forestry by enabling the prompt detection of deforestation, which supports targeted interventions, helps contain the process, and provides decision makers with early insights to distinguish between legal and illegal practices. These capabilities inform decision-making processes and promote sustainable forest management in dryland ecosystems. Future improvements, including using higher-resolution imagery and artificial intelligence for validation, are essential to detect smaller deforestation alerts, reduce manual efforts, and support sustainable dryland management in the Caatinga biome.

Wildfires significantly impact ecosystems, economies, and biodiversity, particularly in fire-prone regions like the Caatinga biome in Northeastern Brazil. This study integrates machine learning with climate and land use data to model current and future fire dynamics in the Caatinga. Using MaxEnt, fire probability maps were generated based on historical fire scars from Landsat imagery and environmental predictors, including bioclimatic variables and human influences. Future projections under SSP1-2.6 (low-emission) and SSP5-8.5 (high-emission) scenarios were also analyzed. The baseline model achieved an AUC of 0.825, indicating a strong predictive performance. Key drivers of fire risk included the mean temperature of the driest quarter (with an importance of 14.1%) and isothermality (12.5%). Temperature-related factors were more influential than precipitation, which played a secondary role in shaping fire dynamics. Anthropogenic factors, such as proximity to farming and urban areas, also contributed to fire susceptibility. Under the optimistic scenario, low-fire-probability areas expanded to 29.129 Mha, suggesting a reduced fire risk with climate mitigation. However, high-risk zones persisted in the Western Caatinga. The pessimistic scenario projected an alarming expansion of very-high-risk areas to 12.448 Mha, emphasizing the vulnerability of the region under severe climate conditions. These findings underline the importance of temperature dynamics and human activities in shaping fire regimes. Future research should incorporate additional variables, such as vegetation recovery and socio-economic factors, to refine predictions. This study provides critical insights for targeted fire management and land use planning, promoting the sustainable conservation of the Caatinga under changing climatic conditions.

The semi-arid Caatinga biome is particularly susceptible to fire dynamics. Periodic droughts amplify fire risks, while anthropogenic activities such as agriculture, pasture expansion, and land-clearing significantly contribute to the prevalence of fires. This research aims to evaluate the effectiveness of a fire detection model and analyze the spatial and temporal patterns of burned areas, providing essential insights for fire management and prevention strategies. Utilizing deep neural network (DNN) models, we mapped burned areas across the Caatinga biome from 1985 to 2023, based on Landsat-derived annual quality mosaics and minimum NBR values. Over the 38-year period, the model classified 10.9 Mha (12.7% of the Caatinga) as burned, with an average annual burned area of approximately 0.5 Mha (0.56%). The peak burned area reached 0.89 Mha in 2021. Fire scars varied significantly, ranging from 0.18 Mha in 1985 to substantial fluctuations in subsequent years. The most affected vegetation type was savanna, with 9.8 Mha burned, while forests experienced only 0.28 Mha of burning. October emerged as the month with the highest fire activity, accounting for 7266 hectares. These findings underscore the complex interplay of climatic and anthropogenic factors, highlighting the urgent need for effective fire management strategies.