In the Wake of the Northern California Fires, ME Professor Tarek I. Zohdi Looks to Fire Hazard Mitigation Research

Friday, October 13, 2017

Fire Prediction, Response, and Suppression in the 21st Century

Originally published on October 9, 2017 in the PEER News Digest

by T.I. Zohdi and K.M. Mosalam

 

Today on October 9, 2017, the numerous burning wild fires in Napa and Sonoma Counties in Northern California are a blunt reminder of how vulnerable our communities are to natural hazards. The state of California has recently experienced extended periods of drought, and where there is uninhibited urban growth, fires will thrive. Without a strategy that efficiently marshals available modern technology, California residents are faced with recurring episodes of fire-related tragedy and loss. Listed below are several approaches to address this hazard.

 

Ubiquitous technology now exists for continuous monitoring of statewide fire hazard, with seamless coordination between satellite networks, drones, computerized prediction systems, and a fully-engaged social network to enable rapid response, feedback and strategic planning. There are many fire-fighting strategies and tools that are being continuously proposed by private industry, municipal, state and federal agencies, academia, and the military. PEER’s mission towards a resilient California includes supporting research activities that will evaluate and pursue approaches that can be aggregated to provide the best possible policies, strategies, and technologies for natural hazards. This includes fire hazard mitigation control (which involves prediction, response, and suppression) for both fire fighters and California residents. In particular:


• Evaluate and help develop the deployment of massive numbers of cost-effective drones for:

- Personalized (“tethered”) safety of fire-fighters

- High-risk delivery of fire retardants
- High-risk delivery of emergency supplies

• Evaluate and help develop coordinated cost-effective satellite imaging for California
• Evaluate and help develop modern, predictive simulation tools
• Develop a social network comprised of non-passive citizens with a large array of smart phones and social networking tools to catalogue and report local conditions where they live and work in order to develop a high-resolution mosaic of the conditions in California as a whole in real time.

 

On average, wildfires burn 4.3 million acres in the United States annually. In recent years the federal government has spent $1 billion a year on fire suppression. California, Colorado, Oregon, and Arizona experience conflagrations each year.

 

The risk of major wildfires can be reduced partly by a reduction or alteration of fuel. In wilderness areas, reduction can be accomplished by either conducting controlled burns, deliberately setting areas ablaze in less dangerous weather when conditions are less volatile, or by physical removal of trees, which is an existing strategy in many U.S. forests. Alteration of fuels, which involves reducing the structure of fuel ladders, can be accomplished by hand crews with chain saws or by large mastication equipment that shreds trees and vegetation to a mulch. Such techniques are best used within the wild land/urban interface where communities connect with wild open space. Prescribed burns in the back country, away from human habitations, are not particularly effective in preventing large fires. All large catastrophic fires in the U.S. have been wind-driven events where the amount of fuel (trees, shrubs, etc.) has not been the most important factor in fire spread. In Northern California alone, wildfires burned an estimated 1,000,000 acres every summer, and the range continues to increase. With a state-wide drought and harsh terrain, over 2000 fires burned across the region and firefighters have been spread thin, leaving some fires to burn with little to no attention. Classical wildfires, also known as a wildland fires or forest fires, are defined as an uncontrolled fire often occurring in wildland areas, but which can also consume houses or agricultural resources. Common causes include lightning strikes, human carelessness, arson, heat waves, droughts, and cyclical climate changes.

 

The evolution of personalized drones has been rapid. Recently there have been significant efforts to create automatic solutions for early wildfire detection. An integrated approach is best, based on a practical combination of different detection systems depending on wildfire risk and the size of the area. Satellite and aero-monitoring can provide a wider view and may be capable of monitoring very large and low-risk areas. Limitations include the long distance to satellites in geostationary orbits and the short window of observation time for satellites in polar orbits. Another area with a great deal of promise is to harness advances in computational and applied mathematics. In such approaches, forest-fire modeling frequently involves a dynamic system of cells with nearest neighbor interaction. The essential features are:


• A partitioning of the domain into cells
• Each cell having an index on combustibility (how much is available to burn)
• Simple rules of propagation (nearest neighbor interaction)

 

Early versions of this approach go back to Henley (1989) and Drossel and Schwabel (1992), who used cellular automaton on a grid of cells. A cell can be empty, occupied by a tree, or burning. For example, Drossel and Schwabel’s model (1992) is defined by rules that are executed simultaneously involving: (1) A burning cell turns into an empty cell; (2) A tree will burn if at least one neighbor is burning; and (3) A tree ignites with a certainty if the neighbor is burning.

 

The linked simulation developed by Zohdi is based on coding developed with PEER funding and depicts the following:

 

• A city block
• The size and color of each structure indicate the amount of combustible material, which are randomly generated
• The wind is blowing towards the far-right corner
• The black airborne material is equal to the amount of mass burned below it
• Each cell has the potential to catch on fire, which is based on the probabilistic measure of the chance to burn based on the state of its neighbors and the wind current
• In this specific simulation, the fire starts in the middle of the city block, and the buildings downwind from the initial fire burn