Forest fire probability zonation using dNBR and machine learning models: a case study at the Similipal Biosphere Reserve (SBR), Odisha, India
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.Forest fire risk is high in 40.85% of Similipal Biosphere Reserve, with 2021 being the peak year. Machine learning models identified land use and vegetation indices as key factors for targeted fire management.
Area Of Science
- Environmental science, focusing on forest ecosystems and fire dynamics.
- Remote sensing and geospatial analysis for environmental monitoring.
- Machine learning applications in ecological risk assessment.
Background
- Forest fires pose a significant threat to biodiversity and environmental balance.
- Effective forest fire probability (FFP) identification is crucial for mitigation strategies.
- The Similipal Biosphere Reserve (SBR) faces recurring fire challenges.
Purpose Of The Study
- To assess forest fire trends and susceptibility in the SBR from 2012 to 2023.
- To compare the performance of four machine learning models in predicting forest fire probability.
- To identify key conditioning factors influencing forest fire susceptibility.
Main Methods
- Utilized four machine learning models: XGBTree, AdaBag, Random Forest (RF), and Gradient Boosting Machine (GBM).
- Created a forest fire inventory using the delta normalized burn ratio (dNBR) index and incorporated 19 conditioning factors.
- Generated FFP maps and evaluated model performance using ROC-AUC, MAE, MSE, and RMSE metrics; frequency ratio (FR) model used for variable importance.
Main Results
- Approximately 40.85% of the SBR is classified as high to very high susceptibility to forest fires.
- The Random Forest model demonstrated the highest accuracy with an AUC of 0.965.
- Land use/land cover (LULC), NDVI, and NDMI were identified as the most influential factors for fire susceptibility, with 2021 being the peak fire year.
Conclusions
- Machine learning models, particularly RF, effectively predict forest fire probability in the SBR.
- Precise FFP maps generated can guide targeted interventions and enhance fire management strategies.
- Findings support policymakers and conservationists in mitigating the impact of forest fires through data-driven insights.
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
Related Concept Videos
Water plays a significant role in the life cycle of plants. However, insufficient or excess of water can be detrimental and pose a serious threat to plants.
Under normal conditions, water taken up by the plant evaporates from leaves and other parts in a process called transpiration. In times of drought stress, water that evaporates by transpiration far exceeds the water absorbed from the soil, causing plants to wilt. The general plant response to drought stress is the synthesis of hormone...
There have been five major extinction events throughout geological history, resulting in the elimination of biodiversity, followed by a rebound of species that adapted to the new conditions. In the current geological epoch, the Holocene, there is a sixth extinction event in progress. This mass extinction has been attributed to human activities and is thus provisionally called the Anthropocene. In 2019 the human population reached 7.7 billion people and is projected to comprise 10 billion by...
In the ever-evolving field of public health, statistical analysis serves as a cornerstone for understanding and managing disease outbreaks. By leveraging various statistical tools, health professionals can predict potential outbreaks, analyze ongoing situations, and devise effective responses to mitigate impact. For that to happen, there are a few possible stages of the analysis:
Predicting Outbreaks
Predictive analytics, a branch of statistics, uses historical data, algorithmic models, and...