Computational modeling of extreme wildland fire events: A synthesis of scientific understanding with applications to forecasting, land management, and firefighter safety

https://doi.org/10.1016/j.jocs.2020.101152Get rights and content

Highlights

  • Key requirements are accurate simulation of fine-scale atmospheric winds and fire thermodynamic feedbacks upon the atmospheric circulations.

  • Drought and fuel accumulation enhancements are concentrated on "plume-driven" events climbing slopes.

  • Extreme wind-driven events contain microscale peak winds that exceed surface weather station measurements and mesoscale model predictions.

  • Complex events may contain wind-driven and plume-driven events at different places and times throughout their lifetimes.

Abstract

The understanding and prediction of large wildland fire events around the world is a growing interdisciplinary research area advanced rapidly by development and use of computational models. Recent models bidirectionally couple computational fluid dynamics models including weather prediction models with modules containing algorithms representing fire spread and heat release, simulating fire-atmosphere interactions across scales spanning three orders of magnitude. Integrated with weather data and airborne and satellite remote sensing data on wildland fuels and active fire detection, modern coupled weather-fire modeling systems are being used to solve current science problems. Compared to legacy tools, these dynamic computational modeling systems increase cost and complexity but have produced breakthrough insights notably into the mechanisms underlying extreme wildfire events such as fine-scale extreme winds associated with interruptions of the electricity grid and have been configured to forecast a fire's growth, expanding our ability to anticipate how they will unfold. We synthesize case studies of recent extreme events, expanding applications, and the challenges and limitations in our remote sensing systems, fire prediction tools, and meteorological models that add to wildfires' mystery and apparent unpredictability.

Introduction

Wildfire occurrence is punctuated by extreme events - only 3–5 % exceed 100 ha in size [1] - yet the largest 1 % of fires account for 80–96 % of area burned [2,3]. In particular, communities around the world are at risk from large, severe wildfires that threaten people directly and create life-threatening impacts on air quality, water availability and quality, and soil integrity. A new term ‘megafire’ [4] was coined following the devastating 2000 U.S. wildfire season to reflect a perception that some wildfires in the U.S., and perhaps worldwide, were reaching new levels in either size, impact on society, or in severity at least in part due to a century of aggressive fire suppression. Indeed, the number of large fires has been increasing in western states [5]. Statistical analyses over broad regions indicate relationships between regional acres burned and conditions intuitively favorable for fire [5,6]. These perceptions have led to widespread attribution of large fires in total and -- more speculatively -- individual fires to climate change and fuel accumulation.

Meanwhile, physically based, mechanistic studies investigating the dynamics producing large, intense fires, confirming such speculations, have been lacking. Much progress has been made in the broad, interdisciplinary research area of wildland fire science since the 2000 wildfires, particularly in the ecological precursors to and impacts of wildfires on living systems, though fewer resources have been allocated to study the physics of fire itself. That fire behavior research has been concentrated on a sequence of instrumented small, low intensity prescribed fire field experiments the results of which have had limited applicability to large, high intensity events. Two-dimensional fire models that are based on kinematic relationships between fire behavior and environmental parameters are still broadly used for research, operations, and planning. Though these simpler tools have had wide use and utility in modeling fire spread in simple conditions, often through ad hoc calibration of inputs to obtain observed growth rates, they have been unable to represent complex mountain airflows, transient phenomena, and wildfire events that evolve through dynamic fire-atmosphere interactions - characteristics of the most destructive, highest impact events.

More specifically, the most extreme events frequently lie at an end of a spectrum. At one end, fire growth is driven by strong ambient winds, characterized by practitioners as a "wind-driven" event, while at the other end of the spectrum where ambient winds are generally weak, fire growth amplifies due to an internal feedback loop in which winds sometimes ten times stronger than ambient are generated by the heat released by the fire itself, commonly referred to as a "plume-driven" event. In such events, media commonly quote fire managers who describe fire behavior as no longer being simple or intuitive but beyond what simple operational models can predict. Meanwhile, forests themselves are being shaped by drought, human settlement, silviculture, previous fires, blowdowns, and other disturbances. Fires in this environment are forged by intricate mountain winds, including winds amplified by the fire itself. Thus, the complex fire environment, particularly in forested mountainous areas, exceeds current understanding and operational tools, making fires appear unpredictable, increasing uncertainty, and inhibiting the use of prescribed fires that might mitigate the hazard but carry their own uncertainties and risks.

Scientific advances such as coupled numerical weather prediction-wildland fire computational models that capture fire-atmosphere interactions and remote sensing of fires and fuels including quantitative measurements of vegetation structure are transforming fire science. These coupled models have led to better understanding of fire events and phenomena, have reproduced key features during the unfolding of large fire events (e.g., [7]), begun painting a more nuanced picture of how fire environment factors such as fuel amount and condition affect fire behavior (e.g. [8],), and, integrated with satellite active fire detection data, serve as a paradigm for next-generation fire growth forecasting systems (e.g. [9],). Coupled model simulations are broaching operational forecast use but introduce their own uncertainties associated with numerical weather prediction, namely, the decrease in their skill with the length of predicted time, limits to predictability that are particularly severe at finer scales, and ambiguities in the interpretation of fire detection and weather data that might be used to assess simulation fidelity.

Here, we synthesize research using computational models that showed how complex atmospheric flows interact with wildland fire behavior and how recent case studies of large wildfires (discussed fires are listed in Table 1) have improved understanding of transient, hazardous fire behavior and illuminated some under-recognized, difficult to understand weather-fire behavior combinations that are not well represented by current two-dimensional, kinematic models. Examples of these factors include the impact of gust fronts and shifting winds on fire behavior, fire-induced winds, large fire whirls, transient plume behavior, coastal airflows, and complex topographic airflow effects. In addition, we discuss challenges and limitations in our computational models, remote sensing systems, and underlying meteorological models that add to wildfires' mystery and apparent unpredictability. As presented, Section 2 describes the models used in wildland fire simulations, including the considerations associated with developing and applying models that shift land management from widely used laptop-based empirical projections to high performance computing-based expert systems. It also describes the expanding pool of fire detection and mapping data available to initiate and evaluate simulations and surface weather station networks tasked with revealing complex mountain airflows. Section 3 describes application of the new generation of models to scientific investigations of the mechanisms driving extreme fires, practical uses in fire management and mitigation, and distilling knowledge about complex, transient fire phenomena to improve wildland firefighter safety. Differences in the predictions from what should be similar systems are noted. Section 4 draws together the knowledge, technical, and safety advances brought forth by the infusion of computational modeling into an area traditionally relying on empirical learning and calls out areas needing further advancement.

Section snippets

Tools and methods

As the complexities of wildfire behavior became more recognized, fire modeling tools evolved from simple algorithms to computational models. This section discusses new modeling tools, data sources that may be used to initialize and assess them, and the constraints and limits that accompany these more complex models.

Applications

In this section, we synthesize some recent studies to give insight on some overarching fire research issues and the new generation of computational weather - wildland fire models being used to address them, such as what causes distinct types of outlier fires. We also discuss applications, such as identifying situations with potential for large fire growth, fire growth forecasting, and how studies of transient fire behavior are being used to improve firefighter safety.

In typical use, a coupled

Conclusions

The recognition that much of the current complexity surrounding wildland fires arises from their interaction with the fluid surrounding them has led to an expansion in the use of computational models. These models' uses range from basic understanding of fire behavior, investigation into the causes of extreme events, more sophisticated forecasting systems based on perceptions that current kinematic models no longer suffice, to developing training material on this new understanding for

CRediT authorship contribution statement

Janice L. Coen: Conceptualization, Methodology, Visualization, Software, Formal analysis, Investigation, Writing - original draft, Writing - review & editing. W. Schroeder: Conceptualization, Methodology, Software, Formal analysis, Investigation, Writing - original draft, Writing - review & editing. S. Conway: Conceptualization, Methodology, Formal analysis, Investigation, Writing - review & editing. L. Tarnay: Conceptualization, Methodology, Formal analysis, Investigation, Writing - review &

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

NCAR is sponsored by the National Science Foundation (NSF). This material is based upon work supported by FEMA under Award EMW-2015-FP-00888 and NSF under Award 1561093. Any opinions, findings, and conclusions or recommendations expressed in this material are the authors' and do not reflect the views of NSF.

Dr. Janice Coen is a Project Scientist at the National Center for Atmospheric Research in Boulder, Colorado. She studies fire behavior and its interaction with weather using coupled weather-fire computer simulation models and by analyzing infrared imagery of wildland fires. She received a B.S. in Engineering Physics from Grove City College and an M.S. and Ph.D. from the Department of Geophysical Sciences at the University of Chicago. She has served on the Board of Directors of the International

References (70)

  • P.E. Dennison et al.

    Large wildfire trends in the western United States, 1984–2011

    Geophys. Res. Lett.

    (2014)
  • A.L. Westerling et al.

    Warming and earlier spring increase western US forest wildfire activity

    Science

    (2006)
  • J.L. Coen et al.

    Simulation and thermal imaging of the 2006 Esperanza wildfire in southern California: application of a coupled weather-wildland fire model

    Int. J. Wildland Fire

    (2014)
  • J.L. Coen et al.

    Deconstructing the king megafire

    Ecol. Appl.

    (2018)
  • J.L. Coen et al.

    Use of spatially refined remote sensing fire detection data to initialize and evaluate coupled weather-wildfire growth model simulations

    Geophys. Res. Lett.

    (2013)
  • E.K. Noonan-Wright et al.

    Developing the US wildland fire decision support system

    J. Combust.

    (2011)
  • M.A. Finney

    FARSITE: Fire Area Simulator—Model Development and Evaluation

    (1998)
  • J. Ramirez et al.

    New approaches in fire simulations with wildfire analyst

  • R.D. Stratton

    Guidance on spatial wildland fire analysis: models, tools, and techniques

    General Technical Report RMRS-GTR-183

    (2006)
  • R.R. Linn

    A Transport Model for Prediction of Wildfire Behavior

    (1997)
  • W.E. Mell et al.

    A physics-based approach to modeling grassland fires

    Int. J. Wildland Fire

    (2007)
  • J.L. Coen

    Simulation of the Big Elk Fire using coupled atmosphere-fire modeling

    Int. J. Wildland Fire

    (2005)
  • J.L. Coen

    Modeling Wildland Fires: A Description of the Coupled Atmosphere-Wildland Fire Environment Model (CAWFE); NCAR Technical Note NCAR/TN-500+STR

    (2013)
  • J.L. Coen et al.

    WRF-fire: coupled weather-wildland fire modeling with the weather research and forecasting model

    J. Appl. Meteor. Climatol.

    (2013)
  • J. Toivanen et al.

    Coupled atmosphere-fire simulations of the black saturday kilmore east wildfires with the unified model

    J. Adv. Model. Earth Syst.

    (2019)
  • J.L. Coen

    Some requirements for simulating wildland fire behavior using insight from coupled weather-wildland fire models

    Fire

    (2018)
  • J.L. Coen et al.

    Coupled weather-fire modeling: from research to operational forecasting

    Fire Manag. Today

    (2017)
  • W.C. Skamarock

    Evaluating mesoscale NWP models using kinetic energy spectra

    Mon. Weather Rev.

    (2004)
  • J.L. Coen et al.

    The generation and forecast of extreme winds during the origin and progression of the 2017 Tubbs Fire

    Atmosphere

    (2018)
  • J.L. Coen et al.

    The High Park Fire: coupled weather-wildland fire model simulation of a windstorm-driven wildfire in Colorado’s Front Range

    J. Geophys. Res. Atmos.

    (2015)
  • J. Coen et al.

    Extreme wildfire events: understanding and prediction

    AGU Fall Meeting Abstracts

    (2018)
  • C.C. Simpson et al.

    Resolving vorticity-driven lateral fire spread using the WRF-Fire coupled atmosphere-fire numerical model

    Nat. Hazards Earth Syst. Sci. Discuss.

    (2014)
  • T.L. Clark et al.

    Source code documentation for the Clark-Hall Cloud- scale model code version G3CH01

    NCAR Technical Note NCAR/TN-426+STR

    (1996)
  • T.L. Clark et al.

    Terrain-induced Turbulence over Lantau Island: 7 June 1994 Tropical Storm Russ Case Study

    J. Atmos. Sci.

    (1997)
  • R.C. Rothermel

    A mathematical model for predicting fire spread in wildl and fuels

    Research Paper INT-115

    (1992)
  • Cited by (19)

    • Fire return intervals explain different vegetation cover responses to wildfire restoration in two Sierra Nevada basins

      2022, Forest Ecology and Management
      Citation Excerpt :

      For example, detailed explorations of how fire, land cover change and forest growth have altered the hydrological dynamics of ICB (and the sensitivity of these changes to climate change) have previously been conducted using the process-based ecohydrological model RHESSys (Boisramé et al., 2019; Rakhmatulina et al., 2021). Similar process-based models are available to describe fire ignition and spread (Mann et al., 2016; Coen et al., 2020), as well as forest disturbance and demography (Butler and Dickinson, 2010; Seidl et al., 2011). However, highly resolved process-based models have well-known limitations, including large data requirements to support calibration and validation, computational expense, uncertainty associated with typically high levels of parameterization, and challenges in interpreting cause-and-effect amidst complex interactions between modeled processes.

    • UFS-Net: A unified flame and smoke detection method for early detection of fire in video surveillance applications using CNNs

      2022, Journal of Computational Science
      Citation Excerpt :

      Many fire detection methods have been proposed so far, which are usually based on physical characteristics of fire such as heat, pressure, gas, smoke, etc. [3,4], and can be detected by specific sensors. These sensors need to be close to the fire source and have acceptable performance in small environments; however, they lose their effectiveness in large and open areas [5–7]. Nowadays, visual surveillance systems are widely used in different environments [8–11]; In the meantime, computer vision technology is growing rapidly and is used in a variety of applications [12–17].

    View all citing articles on Scopus

    Dr. Janice Coen is a Project Scientist at the National Center for Atmospheric Research in Boulder, Colorado. She studies fire behavior and its interaction with weather using coupled weather-fire computer simulation models and by analyzing infrared imagery of wildland fires. She received a B.S. in Engineering Physics from Grove City College and an M.S. and Ph.D. from the Department of Geophysical Sciences at the University of Chicago. She has served on the Board of Directors of the International Association of Wildland Fire and as an Associate Editor for the International Journal of Wildland Fire.

    Dr. Wilfrid Schroeder received his M.Sc. degree in Environmental Engineering in 2001 from the Federal University of Rio de Janeiro/Brazil, and his Ph.D. degree in Geography in 2008 from the University of Maryland/USA. He is a physical scientist with NOAA where he leads the development of satellite-based biomass burning detection algorithms and related applications.

    Scott Conway graduated from Colorado State University with a Bachelor of Science degree in Natural Resource Management and GIS. He worked as Forest Ecologist for the USDA Forest Service in the Sierra Nevada Mountains from 2000 to 2019, at the Pacific Southwest Region’s Remote Sensing Lab where he pioneered geospatial dataset application solutions for managers and decision makers, and District Ranger for the Truckee Ranger District on the Tahoe National Forest. He currently consults as Forest Ecologist with Spatial Informatics Group on forest project assessment, analysis, implementation, and monitoring toward defensible communities surrounded by resilient forests.

    Dr. Leland Tarnay is an ecologist working for the USDA Forest Service Region 5 Remote Sensing Lab. Lee received his Bachelor of Science in Biology from the University of California, Davis (1995), and his Ph.D. from the University of Nevada, Reno (2001). He spent 10 years as Yosemite National Park’s air resource specialist before joining the Forest Service in 2015. While he is interested in all interactions between land and the atmosphere, his current core expertise is in smoke monitoring, emissions estimation, dispersion modeling, and mapping of forest fuels and structure.

    View full text