Science Policy
day one project

Next-Generation Fire and Vegetation Modeling for a Hot and Dry Future

06.20.23 | 9 min read | Text by Matthew Hurteau


Wildfires are burning in ways that surprise even seasoned firefighters. Our current models cannot predict this extreme fire behavior—nor can they reproduce recent catastrophic wildfires, making them likely to fail at predicting future wildfires or determining when it is safe to light prescribed fires. 

To better prepare the fire management community to operate in a new climate, Congress should establish and fund five regional centers of excellence (CoE) to develop, maintain, and operate next-generation fire and vegetation models to support wildland fire planning and management. Developing five regional CoEs (Southeast, Southwest, California, Pacific Northwest, Northern/Central Rockies) will ensure that researchers pursue a range of approaches that will ultimately lead to better models for predicting future wildfire behavior, improving our ability to safeguard human lives, communities, and ecosystems.

Challenge and Opportunity

In the decade ending in 2021, total federal wildfire suppression expenditures surpassed $23 billion, which is a fraction of the total costs of damages from wildfire over that period. For example, the 2018 wildfires in California are estimated to have amounted to $148.5 billion in economic costs for the state. The costs of suppressing fire, and the societal and natural resources costs of extreme wildfire, will continue to increase with increasing temperatures. 

Fewer than 2% of ignitions become large wildfires, but it is this 2% that cause most of the damage because they are burning under extreme conditions. The area of forests burned by wildfire annually in the western United States has been increasing exponentially since 1984. While the number of ignitions remains relatively constant from year to year, climate change is drying fuels and making forests more flammable. As a result, no matter how much money we spend on wildfire suppression, we will not be able to stop increasingly extreme wildfires. Thus, we need to better understand where the risks lie on our landscapes and work proactively to reduce them. 

When vegetation—especially dead vegetation—is subjected to high temperatures, any moisture absorbed during the winter months quickly evaporates. As a result, increasingly hot summers are making our forests more flammable. Live vegetation moisture content does not react as quickly as dead vegetation, but sharp increases in air temperature when conditions are dry can make live plants more flammable as well. While this relationship between temperature and ecosystem flammability has remained consistent over time, until the past decade we had not reached a level of warming that dried ecosystems sufficiently to allow for consistent extreme fire behavior. This is in part because large dead fuels, such as dead trees and logs, did not dry sufficiently to become flammable for the majority of the fire season until recently. 

Our current operational models for simulating wildfire and vegetation are incapable of reproducing the extreme fire behavior and rapid ecosystem change that we are now experiencing. Forest growth-and-yield models, such as the Forest Vegetation Simulator, used by managers have served them well for decades. However, because they are built using statistical relationships between past tree growth and climate, they are incapable of capturing the effects of changing climate, especially extreme events, on tree growth and mortality. Similarly, our operational fire models, such as FARSITE, that are used for both management planning and simulating fire spread to plan fire suppression activities are not designed to deal with the substantial ecosystem changes that are occurring from climate change. These fire models have served us well in the past, but increasing temperature and a drying atmosphere are causing conditions that far exceed the data used to build these models. 

For example, our current operational fire models do not account for large dead trees and logs and how they contribute to fire spread or for the way fire behaves in the wildland–urban interface. Yet wildfires are increasingly burning through communities, and the number of dead trees and logs is increasing because of drought- and insect-induced tree mortality and is increasingly available to burn because of high temperatures. The 2020 Creek Fire in the Sierra Nevada, California, burned through an area of extensive tree mortality from prolonged drought and insect outbreaks. The operational fire spread model ELMFIRE, which is used to predict fire spread of active wildfires, was unable to predict the mass fire behavior created by the massive number of dead trees.

Managing wildfire risk both prior to and during wildfires requires advanced models that are able to account for changing climatic conditions. We need new wildfire models that account for the increasing fuel dryness that facilitates extreme fire behavior, and we need new vegetation models that account for the effects of extreme drought and temperature on vegetation mortality. The research and development necessary to prepare us for our increasingly flammable world requires both fundamental and applied research, neither of which is sufficient on its own. 

Further, we need to ensure that we commit to maintaining these models as the climate continues to change so that we do not create another tool that fails to serve us well within a decade or two. As the climate continues to change, these next-generation fire and vegetation models will be challenged with novel conditions that require continuous efforts to ensure they are capable of capturing the dynamics of the system. In addition, we must ensure that the mechanistic understanding of the system that develops is applied to supporting fire and vegetation management decision-making. This will require ongoing experimentation and observations of actual wildfire behavior, along with extensive data collection to characterize how quickly the flammability of the system changes as a function of vegetation type and weather conditions. 

Developing these next-generation models is necessary for both fire suppression and management planning. Incident command teams rely on fire spread models to help plan suppression efforts for active wildfires, and thus having better predictions of fire spread is essential for effective operations and firefighter safety. Likewise, planning forest treatments that are effective for reducing the risk of high-severity wildfire under extreme weather conditions requires better vegetation and fire models that can capture the influence of changing climate on the probability that high-severity wildfire occurs. 

Plan of Action

Developing and future-proofing next-generation fire and vegetation models will require new and sustained investment. Further, we must accept that these advanced models will require a level of expertise to operate that we cannot expect from a land manager trained in natural resource management, requiring that we fund expert model users to support management planning and suppression efforts. 

As with all research and development, there are many possible pathways. Regional differences in weather, vegetation, and management history will alter climate effects on vegetation growth, mortality, and flammability. Similar to the Manhattan Project approach of simultaneously pursuing two different ignition systems when there was more than one potential viable alternative, we lack the necessary understanding to pick a “winning” model at this point. 

To account for regional differences in vegetation and the research momentum that is developing in different nascent modeling approaches, an effective and robust federal investment would entail the following actions. 

Recommendation 1. Congress should establish and fund five centers of excellence housed at academic institutions in the Southeast, Southwest, California, Pacific Northwest, and Northern/Central Rockies to develop and maintain next-generation fire and vegetation models that are capable of modeling extreme fire behavior and can be operationalized to support planning for wildfire and vegetation management and to support wildfire suppression. 

Establishing five centers with this geographic distribution will allow for investigation into the forest types where the majority of wildfire area occurs and will capture the range of climatic conditions under which wildfires are occurring. It will also take advantage of past and ongoing regional research efforts that will form the information foundation for each center. While these centers should have largely independent research programs, it will be necessary to coordinate some large-scale experimentation and to ensure that research findings and advances are shared rapidly. To achieve these objectives, one center should be selected to act as the coordinating center for the network. 

Recommendation 2. Congress should require institutional partnerships between the host institutions and federal research institutions (e.g., U.S. Forest Service Research and Development, Department of Energy National Labs, U.S. Geological Survey, etc.). 

We are currently in an all-hands-on-deck situation in the fire and fuels research community, and we need to operate in a collaborative and regionally coordinated manner. Requiring partnerships between the academic centers of excellence and federal research facilities within each region will ensure that effort is not duplicated and a wider range of expertise. For example, efforts are under way at federal research facilities that could be integrated within the regional fire centers. The integration will ensure collaboration between academic and federal partners and allow for the overall research effort to draw on the strengths of these different types of institutions. 

Recommendation 3. Congress should mandate and fund the centers to operate these next-generation models and support wildfire and vegetation management planning and operations. 

To date, we have relied on fire and vegetation models developed by the research community to use data collected by fire and forest managers and packaged so that natural resource professionals can operate the models. Both of these constraints have contributed to the limitations of our current suite of models. We can no longer afford the limitations imposed by expectations on the research community to develop models that a natural resource professional can run on a desktop computer. Accounting for a range of factors, such as how changing climatic conditions will directly change the amount of fuel on the landscape and also for how short-term changes in weather will interact with longer-term changes in climate and influence fuel moisture, requires a more sophisticated approach to simulating the system than is necessarily accessible to a non-expert user. Expecting a natural resource professional to use an advanced coupled atmosphere-biosphere fire model would be like teaching someone how to balance their checkbook and then expecting them to calculate exactly how much they need to save every week for retirement. Further, important feedback to model improvement will come from repeated application by expert model users. To deploy next-generation fire and vegetation models in a manner that will effectively support fire and natural resource management decision-making, each center will employ experts who will work collaboratively with managers in response to their requests to run simulations for pre-fire management and suppression operations planning.

Recommendation 4. Congress should mandate the creation of strategic plans to support implementation and coordination across centers. 

Each center will develop a five-year strategic plan to guide its research and development efforts. Following strategic plan development, representatives from the five centers will convene to determine necessary coordinated experimentation and implementation plans to facilitate coordinated efforts. The coordinating center will hold biannual leadership meetings to ensure data and information flow and identify additional opportunities for collaboration among individual centers. 


Establishing five centers of excellence to develop, maintain, and operate next-generation models will cost approximately $26 million per year, which is less than 1% of the 2021 federal wildfire suppression expenditure. This level of funding would provide $5 million per year per center (plus an additional $1 million per year for the coordinating center). The annual budgets would fund staff scientist and research assistant positions, provide support for the experiments necessary to develop and parameterize new models, provide computing resources for computationally sophisticated models, and fund staff analysts to run the models in support of managers. Initially, the majority of the annual appropriation would be focused on model development, transitioning to maintaining and operating the models to support land management as the technology matures. 

The centers could be supported through National Science Foundation (NSF) funding. NSF could provide financial support for five university host institutions (one in each region) selected through a competitive bidding process. In turn, these university host institutions can manage the required federal partnerships. Selection of university host institutions could be based in part on demonstrated capacity to manage successful partnerships with federal institutions. 

It is imperative that we invest in new models that will support more effective mitigation to reduce wildfire severity, otherwise spending on suppression will continue to balloon despite improved fire intelligence.

Frequently Asked Questions
Are there universities with demonstrated capacity to perform the kind of work required to make this centers of excellence program successful?

Yes. Just a few examples include the colocation of the University of Georgia with fire researchers in the U.S. Forest Service (USFS) Southern Research Station; the University of New Mexico’s existing relationships with Los Alamos National Lab, Sandia National Lab, and the U.S. Geological Survey; and the University of Washington’s long-standing relationship with the USFS Pacific Northwest Fire and Environmental Applications research group.

Why might NSF be the right agency to fund the proposed centers?

NSF is in wildland fire research and, jointly with the National Institute of Standards and Technology, already funds research on fire in the wildland–urban interface. While much of the research needed to develop next-generation fire and vegetation models is basic, all wildland fire research is inherently applicable. NSF hosted a five-day Wildfire and the Biosphere Innovation Lab, and the findings included the assertion that “support for applied research will be most effective by aiming at both short- and long-term applications and solutions,” acknowledging that the application of research findings is an important part of the research enterprise.

Large investments in hazardous fuels management are being made now. Will models developed through this research have an impact in the near term?

Yes. These centers will bring together and build from ongoing efforts. There are already efforts under way to develop optimal treatment strategies that account for changing climatic conditions using advanced forest landscape models. This approach, with some refinement and validation, will be useful for informing treatment placement within the next two years.

Why do we need five centers of excellence? Wouldn’t it be more efficient to have one center that tracks and evaluates regional modeling efforts and creates best practices for management application?

This is functionally the system we have now. The Fire Research Management and Exchange System (FRAMES) provides a clearinghouse of models developed for fire and vegetation modeling to inform management. FRAMES may be a good interface to help increase manager awareness of the models the five centers will develop, but it is not a mechanism for facilitating the research and development needed to tackle the wildfire problem. We need five centers because there are already a number of efforts under way to develop new fire and vegetation models. None of the models will be perfect because they all take different approaches and there are tradeoffs inherent in any given approach. With simultaneous investment, we will be able to capitalize on the aspects of each model that best simulate a part of the fire spread or vegetation growth process and then develop a system that incorporates the best of each model. Competition within the U.S. scientific enterprise has helped our country achieve high global standing. Funding five centers will shift that competition away from researchers spending much of their time competing for funding and focus it on competing with their best ideas in a way that prepares us for managing wildfire in the future.