Project Overview
Wildfires significantly alter watershed characteristics, creating conditions that can lead to devastating debris flows during subsequent rainfall events. This project employs high-resolution topographic data and machine learning approaches to better understand, predict, and mitigate post-fire debris flow hazards in vulnerable communities across fire-prone regions.
Research Motivation
Climate change has intensified wildfire activity across the western United States, creating unprecedented challenges for debris flow hazard management. Traditional empirical models often fail to capture the complex interactions between fire effects, topography, and hydrology that control debris flow initiation and runout. Our research addresses this critical gap through innovative data-driven approaches.
Research Objectives
- Develop machine learning models for debris flow susceptibility mapping
- Quantify fire effects on watershed hydrology and geomorphology
- Create high-resolution hazard maps for at-risk communities
- Improve early warning systems for post-fire debris flows
- Optimize mitigation strategies and infrastructure placement
Methodology & Innovation
High-Resolution Data Integration:
- Pre- and post-fire LiDAR for topographic change detection
- Multi-temporal satellite imagery for burn severity assessment
- Ground-based monitoring of soil properties and vegetation recovery
- Rainfall radar and gauge network data integration
Machine Learning Approaches:
- Random Forest and Gradient Boosting for susceptibility modeling
- Convolutional Neural Networks for spatial pattern recognition
- Ensemble methods for uncertainty quantification
- Transfer learning across different fire-affected regions
Physical Process Integration:
- Hydrologic modeling of altered watershed response
- Slope stability analysis in fire-affected terrain
- Sediment supply and transport capacity modeling
- Physics-informed machine learning architectures
Study Areas
Southern California:
- Thomas Fire (2017) - Montecito debris flows
- Woolsey Fire (2018) - Malibu watershed impacts
- Apple Fire (2020) - San Bernardino National Forest
Northern California:
- Camp Fire (2018) - Paradise area recovery
- North Complex Fire (2020) - Sierra Nevada foothills
- Dixie Fire (2021) - Multi-county impacts
Community Applications
Our research directly supports vulnerable communities through:
- Real-time hazard assessment during storm events
- Land-use planning in fire-prone watersheds
- Emergency evacuation route optimization
- Infrastructure hardening and protection strategies
- Insurance risk assessment and pricing models
Expected Outcomes
- Improved debris flow prediction accuracy by 30-40%
- Reduced false alarm rates in early warning systems
- Cost-effective mitigation strategy recommendations
- Enhanced community resilience to compound hazards
- Transferable methodologies for global application
Collaboration & Impact
This project involves close collaboration with Cal Fire, USGS Landslide Hazards Program, local emergency management agencies, and international research partners. Results are being integrated into operational hazard assessment tools and emergency response protocols.