AI Modeling & Data Analytics for Wildfire Preparedness, Mitigation, and Response (PNNL, January 12, 2024)
Summary
This is a presentation by Andre Coleman, Ph.D., Chief Scientist of the Earth Systems Predictability & Resiliency Group, to the Washington State House Innovation, Community & Economic Development, and Veterans Committee on January 12, 2024. It states that wildfires in Washington are occurring more frequently and with increasing burn area, and contrasts the challenges of standard analytics with advanced analytics that integrate multiple sources of data. It describes RADR, a multi-risk assessment, situational awareness and forecasting system, and models for prevention and mitigation and for response to active events. The conclusion states that Washington could leverage federally funded R&D and AI models to improve wildfire analytic tools, and that PNNL welcomes partnerships.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
- Issuer
- Congressional materials
- Document type
- 2024 01 12 A31582 D264208 Pnnl Ai Modeling Data Analystics For Wildfire Preparedness M
- Date
- 2024-01-12
- Case
- 2024 01 12 A31582 D264208 Pnnl Ai Modeling Data Analystics For Wildfire Preparedness M
Cite this document
“AI Modeling & Data Analytics for Wildfire Preparedness, Mitigation, and Response (PNNL, January 12, 2024).”
Original at app.leg.wa.gov, dated 2024-01-12. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/2024-01-12-2024-01-12-a31582-d264208-pnnl-ai-modeling-data-analystics-for-wildfire-preparedness-m/
, SHA-256 f743ada2a66e23625c6a4cf880cdb659dfeedc7d72f6a43d7ed735c48a024540.Full text
AI Modeling & Data
Analytics for Wildfire
Preparedness,
Mitigation, and
Response
Andre Coleman, Ph.D.
Chief Scientist
Earth Systems Predictability & Resiliency Group
January 12, 2024
.
Washington State House Innovation, Community &
Economic Development, and Veterans Committee
UNCLASSIFIED
Wildfires in Washington occurring more
frequently and with increasing burn area
WA Burn Area (5-year Groups) 2,403 Wildland fires since 2000
8.6 Number of acres burned in
Washington since 2000
MILLION
19% Portion of state in burn area
256k Largest burn area
(Carlton Complex)
ACRES
Fires greater than 100,000 acres
17 (13 since 2014)
190k Burn area average
ACRES 2000-2009
Fire is prevalent in both forest and rangelands
386k Burn area average
2010-2019
ACRES
UNCLASSIFIED 2
AI-powered tools increase utility of existing data
and ability to solve complex problems
Challenges of standard analytics Advanced analytics
• Outdated datasets used in models to • Integrate multiple sources of data to
inform today’s conditions consistently inform on hazard risk
• Monitoring frequency isn’t keeping • Automated data retrieval and AI models
pace with highly dynamic events help provide frequent data and analysis
updates
• Inconsistent modeling and analytics
approaches across organizations • Enable downstream consequence
analysis beyond the hazard itself
• Lack of common data and tools
across jurisdictions and entities • Provide accessible, intuitive, and
actionable products to a variety of end-
users
UNCLASSIFIED 3
Rapid Analytics for Disaster Response
RADR: A multi-risk assessment, situational awareness
and forecasting system
• Flood Detection
• Structural Damage
• Rubble/Debris Detection
• Building Damage
• Vegetation Damage
• Transportation Barriers
• Infrastructure Damage
• Change Analysis
• Pre-Event Wildfire Risk
• Thermal Anomaly Detection
• Active Wildfire Monitoring
• Wildfire Spread Forecasting
UNCLASSIFIED 4
Different types of models enable
triple-line defense for wildfire resiliency
RECOVERY &
PREVENTION & MITIGATION RESPONSE LONG-TERM
PREPAREDNESS
Monthly/ Future
Pre-Season Forecasted Active Event
Weekly Post-Event Wildfire
Fire Risk Fire Spread Monitoring
Fire Risk Risk
UNCLASSIFIED 5
Models for prevention and mitigation
focus on pre-season and ongoing fire risk
PREVENTION & MITIGATION
Pre-Season Fire Risk Monthly/Weekly Fire Risk
• Historic meteorology • Current meteorology
• Fire occurrence, size, • Latest fuelscape
Types of and duration
• Fuel moisture state
Data • Seasonal analogs
• Soil moisture state
• Latest fuelscape
• Wildfire risk + spread
• Probabilistic risk
modeling
• Corridor-specific risk
Applications
Vegetation Infrastructure Community Consequence Planning, Infrastructure
Management Hardening Resiliency Analysis Preparedness, & Site
Monitoring, & Operations
UNCLASSIFIED Policy Actions 6
Models for response include forecasts
and monitoring of active events
RESPONSE
Forecasted Fire Spread Active Event Monitoring
• Current meteorology • High-resolution satellite
• Forecasted meteorology imagery (sees through
Types of • Vegetation state smoke)
Data • Fuel moisture • 3-4x/day coverage
• Ignition point scenarios • Automated mapping
and delivery
• Active fire front(s)
• Community and asset
• Spot fires exposure risk
Applications
Resourcing Community Evacuation Infrastructure & Situational Resourcing & Infrastructure &
& Response Protection / Site Operations Awareness Response Site Operations
Targeted UNCLASSIFIED 7
Maintenance
Models for recovery and preparedness
consider post-event data and future risk
RECOVERY & LONG-TERM PREPAREDNESS
Post-Event Future Wildfire Risk
• Damage inventory • Climate projections
• Burn severity • Vegetation dynamics
Types of • Vegetation state • Land management
Data • Soils & topography • Wildfire risk modeling
• Projected/design storms
• Downstream asset risk
Applications
Flood Risk Landslide Recovery Monitoring Long-Term Policy
Risk Planning & Development
UNCLASSIFIED Mitigation 8
RADR weighs hazard consequences to inform
highest priority actions and investments
Health & Safety
Community Risk
Environment
Wildfire Risk
Demographics (Landscape) Asset Risk
Infrastructure Wildfire Risk Environmental
(Corridor) Risk
Critical Facilities
{Entity Defined}
Economics
User-Defined Fire Modeling Topic-Level Compound Consequence
Priorities Scenarios Consequence Risk Risk Score
UNCLASSIFIED 9
Conclusion
• Washington could leverage federally funded R&D and AI models to improve
analytic tools for wildfire risk assessment, mitigation, response, and recovery.
• Development of validated, trusted data, models, and tools can help entities
build a common understanding and coordinate actions based on current
information.
• PNNL welcomes partnerships that continue the path of innovation to evaluate,
validate, or integrate new hazard analytics products into decision-making or
workflows.
UNCLASSIFIED 10
Thank you
Thank You
Andre.Coleman@pnnl.gov
11
File and source
- File
- 2024-01-12_a31582_d264208_pnnl-ai-modeling-data-analystics-for-wildfire-preparedness-m.pdf
- Size
- 1,974,719 bytes
- SHA-256
- f743ada2a66e23625c6a4cf880cdb659dfeedc7d72f6a43d7ed735c48a024540
- Original
- app.leg.wa.gov