Abstract:Bushfire is one of the major natural disasters that cause huge losses to livelihoods and the environment. Understanding and analyzing the severity of bushfires is crucial for effective management and mitigation strategies, helping to prevent the extensive damage and loss caused by these natural disasters. This study presents an in-depth analysis of bushfire severity in Australia over the last twelve years, combining remote sensing data and machine learning techniques to predict future fire trends. By utilizing Landsat imagery and integrating spectral indices like NDVI, NBR, and Burn Index, along with topographical and climatic factors, we developed a robust predictive model using XGBoost. The model achieved high accuracy, 86.13%, demonstrating its effectiveness in predicting fire severity across diverse Australian ecosystems. By analyzing historical trends and integrating factors such as population density and vegetation cover, we identify areas at high risk of future severe bushfires. Additionally, this research identifies key regions at risk, providing data-driven recommendations for targeted firefighting efforts. The findings contribute valuable insights into fire management strategies, enhancing resilience to future fire events in Australia. Also, we propose future work on developing a UAV-based swarm coordination model to enhance fire prediction in real-time and firefighting capabilities in the most vulnerable regions.
What problem does this paper attempt to address?
### Problems the Paper Aims to Solve
This paper aims to conduct an in-depth analysis of the severity of wildfires in Australia over the past 12 years by combining remote sensing technology and machine learning methods, and to predict future wildfire trends. Specifically, the main objectives of the study include:
1. **Mapping Fire Severity**: Using NASA's FIRMS (Fire Information for Resource Management System) satellite data to map the severity of wildfires in Australia over the past 12 years.
2. **Identifying and Analyzing Trends and Patterns in Fire Occurrence and Severity**: By analyzing historical data, identifying temporal and spatial patterns of fire occurrence, and trends in fire severity.
3. **Developing a Model to Predict Future Fire Trends**: Based on historical data, developing a predictive model that can forecast future fire trends.
4. **Proposing Firefighting Recommendations for High-Risk Areas**: Considering factors such as population density and vegetation type, providing targeted firefighting recommendations for high-risk areas.
### Background and Significance
Wildfires are one of the major natural disasters faced by Australia, causing significant losses to human life and the environment. Understanding and analyzing the severity of wildfires is crucial for developing effective management and mitigation strategies, helping to prevent the widespread damage caused by these natural disasters. In recent years, the frequency and severity of wildfires in Australia have increased significantly, particularly during the 2019-2020 "Black Summer" wildfire event, which burned over 18.6 million hectares of land, resulted in 33 deaths, destroyed more than 3,500 homes, and an estimated 1 billion animals died. These events highlight the immediate danger of wildfires and the necessity of developing comprehensive management strategies.
### Methods and Data
1. **Data Collection**:
- Utilizing MODIS (Moderate Resolution Imaging Spectroradiometer) satellite data, covering multiple time frames, providing data at different resolutions for monitoring fire events and their severity.
- Using NASA's FIRMS data, covering fire events from February 2012 to January 2024, focusing on understanding fire severity and predicting future fire trends.
2. **Data Preprocessing**:
- Cleaning the data to remove duplicates, ensuring each fire event is counted only once.
- Normalizing the data, standardizing attributes of different scales to between 0 and 1 to improve the convergence speed of learning algorithms.
- Splitting the dataset by year to analyze trends and patterns in fire severity over the past 12 years in detail.
- Handling missing values to ensure data integrity and accuracy.
3. **Feature Extraction**:
- Calculating spectral indices from Landsat imagery, such as NDVI (Normalized Difference Vegetation Index), NBR (Normalized Burn Ratio), EVI (Enhanced Vegetation Index), NDWI (Normalized Difference Water Index), BI (Burn Index), SWIR1 and SWIR2, and dNBR (Differenced Normalized Burn Ratio).
- Combining topographic and climatic features, such as elevation, slope, Topographic Position Index (TPI), average temperature, total precipitation, and Soil Moisture Index (SMI).
4. **Model Training and Evaluation**:
- Using the XGBoost regression model to predict fire severity, with dNBR as the target variable.
- Determining the optimal hyperparameter configuration through manual tuning, including 500 boosting rounds, a learning rate of 0.1, a maximum tree depth of 5, an 80% training data sampling rate, and an 80% feature sampling rate per tree.
- Using 80% of the data for training, 20% for testing, and performing 5-fold cross-validation to evaluate the model's generalization ability.
### Results and Recommendations
Through the above methods, the research team successfully developed a high-accuracy predictive model with an accuracy rate of 86.13%, capable of effectively predicting fire severity across different Australian ecosystems. Additionally, the study identified future high-risk areas and provided data-driven recommendations to support targeted firefighting efforts. These findings offer valuable insights for Australia's future fire management strategies, enhancing the ability to respond to fire events.
### Future Work
The research team also proposed future research directions, including