Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “wildfires”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 235 records · Page 13
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Machine learning and artificial intelligence for wildfire prediction

Wildfire ignition, intensity, and spread rates are tightly linked with water cycle extremes. The science of wildfire prediction has traditionally encompassed the use of physical and empirical models to quantify the direction and speed of fire spread, plume injection and fire-aerosol impacts on atmospheric composition, predictions of fire season severity on subseasonal-to-seasonal (S2S) time scales, and assessment of the spatial and temporal patterns of fire risk across landscapes. Together with expanding observation networks, machine learning and artificial intelligence (AI) have the potential to revolutionize the application of such models for fire science, saving lives, protecting critical infrastructure, and providing more accurate estimates of wildfire-climate feedbacks.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Improving North American Wildfire Prediction by Integrating a Machine-Learning Fire Model in a Land Surface Model

Wildfires have shown increasing trends in both frequency and severity across the Contiguous United States (CONUS). However, process-based fire models have difficulties in accurately simulating the burned area over the CONUS due to a simplification of the physical process and cannot capture the interplay among fire, ignition, climate, and human activities. The deficiency of burned area simulation deteriorates the description of fire impact on energy balance, water budget, and carbon fluxes in the Earth System Models (ESMs). Alternatively, machine learning (ML) based fire models, which capture statistical relationships between the burned area and environmental factors, have shown promising burned area predictions and corresponding fire impact simulation. We develop a hybrid framework (ML4Fire-XGB) that integrates a pretrained eXtreme Gradient Boosting (XGBoost) wildfire model with the Energy Exascale Earth System Model (E3SM) land model (ELM) version 2.1. A Fortran-C-Python deep learning bridge is adapted to support online communication between ELM and the ML fire model. Specifically, the burned area predicted by the ML-based wildfire model is directly passed to ELM to adjust the carbon pool and vegetation dynamics after disturbance, which are then used as predictors in the ML-based fire model in the next time step. Evaluated against the historical burned area from Global Fire Emissions Database 5 from 2001-2020, the ML4Fire-XGB model outperforms process-based fire models in terms of spatial distribution and seasonal variations. Sensitivity analysis confirms that the ML4Fire-XGB well captures the responses of the burned area to rising temperatures. The ML4Fire-XGB model has proved to be a new tool for studying vegetation-fire interactions, and more importantly, enables seamless exploration of climate-fire feedback, working as an active component in E3SM.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Simulated wildfire burned area over the CONUS during 2001-2020

Wildfires have shown increasing trends in both frequency and severity across the Contiguous United States (CONUS). However, process-based fire models have difficulties in accurately simulating the burned area over the CONUS due to a simplification of the physical process and cannot capture the interplay among fire, ignition, climate, and human activities. The deficiency of burned area simulation deteriorates the description of fire impact on energy balance, water budget, and carbon fluxes in the Earth System Models (ESMs). Alternatively, machine learning (ML) based fire models, which capture statistical relationships between the burned area and environmental factors, have shown promising burned area predictions and corresponding fire impact simulation. We develop a hybrid framework (ML4Fire-XGB) that integrates a pretrained eXtreme Gradient Boosting (XGBoost) wildfire model with the Energy Exascale Earth System Model (E3SM) land model (ELM). A Fortran-C-Python deep learning bridge is adapted to support online communication between ELM and the ML fire model. Specifically, the burned area predicted by the ML-based wildfire model is directly passed to ELM to adjust the carbon pool and vegetation dynamics after disturbance, which are then used as predictors in the ML-based fire model in the next time step. Evaluated against the historical burned area from Global Fire Emissions Database 5 from 2001-2020, the ML4Fire-XGB model outperforms process-based fire models in terms of spatial distribution and seasonal variations. Sensitivity analysis confirms that the ML4Fire-XGB well captures the responses of the burned area to rising temperatures. The ML4Fire-XGB model has proved to be a new tool for studying vegetation-fire interactions, and more importantly, enables seamless exploration of climate-fire feedback, working as an active component in E3SM.

Liu, Ye↗
DOE OSTI2022

Evaluating Basin-Scale Forest Adaptation Scenarios: Wildfire, Streamflow, Biomass, and Economic Recovery Synergies and Trade-Offs

Active forest management is applied in many parts of the western United States to reduce wildfire severity, mitigate vulnerability to drought and bark beetle mortality, and more recently, to increase snow retention and late-season streamflow. A rapidly warming climate accelerates the need for these restorative treatments, but the treatment priority among forest patches varies considerably. We simulated four treatment scenarios across the 3,450 km 2 Wenatchee River basin in eastern Washington, United States. We used a decision support tool (DST) to assess trade-offs and synergies within and among treatments on wildfire risk and smoke emissions, water yield and snow retention, biomass production, and economic return. Treatment scenarios emphasized prescribed burning ( BurnOnly ), biomass production ( MaxBiomass ), gap-based thinning to optimize water yield ( IdealWater ), and a principle-based restoration scenario ( RA1 ). Fire hazard, smoke emissions, and biomass production metrics were evaluated across scenarios using the Forest Vegetation Simulator, and water yields were modeled using the Distributed Hydrology Soil Vegetation Model. Simulations were summarized to both patch- (10 1 –10 2 ha) and subwatershed- (10 3 –10 4 ha) scales, and treatment effects were evaluated against an untreated baseline landscape. We used logic models to rank effect sizes by scenario across metrics along a continuum between −1 (no or weak effect) to +1 (large effect). All treatments produced benefits across one or more ecosystem services and led to synergistic benefits to water yield and wildfire hazard reduction. Tradeoffs among resource benefits were clear in wilderness where reliance on prescribed burning without mechanical treatment increased costs and eliminated the potential for biomass recovery. The BurnOnly scenario improved fire risk metrics and streamflow, but effect sizes were lower compared to other treatments. IdealWater showed the strongest benefits overall, demonstrating the ability to capture multiple resource benefits through spatially explicit thinning. Our study provides a framework for integrating strategic and tactical models that evaluate tradeoffs and synergies gained through varied management approaches. We demonstrate the utility of decision support modeling to enhance management synergies across large landscapes.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Efficient First-Order Algorithms for Large-Scale, Non-Smooth Maximum Entropy Models with Application to Wildfire Science

Maximum entropy (MaxEnt) models are a class of statistical models that use the maximum entropy principle to estimate probability distributions from data. Due to the size of modern data sets, MaxEnt models need efficient optimization algorithms to scale well for big data applications. State-of-the-art algorithms for MaxEnt models, however, were not originally designed to handle big data sets; these algorithms either rely on technical devices that may yield unreliable numerical results, scale poorly, or require smoothness assumptions that many practical MaxEnt models lack. In this paper, we present novel optimization algorithms that overcome the shortcomings of state-of-the-art algorithms for training large-scale, non-smooth MaxEnt models. Our proposed first-order algorithms leverage the Kullback–Leibler divergence to train large-scale and non-smooth MaxEnt models efficiently. For MaxEnt models with discrete probability distribution of n elements built from samples, each containing m features, the stepsize parameter estimation and iterations in our algorithms scale on the order of O(mn) operations and can be trivially parallelized. Moreover, the strong ℓ1 convexity of the Kullback–Leibler divergence allows for larger stepsize parameters, thereby speeding up the convergence rate of our algorithms. To illustrate the efficiency of our novel algorithms, we consider the problem of estimating probabilities of fire occurrences as a function of ecological features in the Western US MTBS-Interagency wildfire data set. Our numerical results show that our algorithms outperform the state of the art by one order of magnitude and yield results that agree with physical models of wildfire occurrence and previous statistical analyses of wildfire drivers.

Physics↗
DOE OSTI2020

Rapid evolution of aerosol particles and their optical properties downwind of wildfires in the western US

Abstract. During the first phase of the Biomass Burn Operational Project (BBOP) field campaign, conducted in the Pacific Northwest, the DOE G-1 aircraft was used to follow the time evolution of wildfire smoke from near the point of emission to locations 2–3.5 h downwind. In nine flights we made repeated transects of wildfire plumes at varying downwind distances and could thereby follow the plume's time evolution. On average there was little change in dilution-normalized aerosol mass concentration as a function of downwind distance. This consistency hides a dynamic system in which primary aerosol particles are evaporating and secondary ones condensing. Organic aerosol is oxidized as a result. On all transects more than 90 % of aerosol is organic. In freshly emitted smoke aerosol, NH4+ is approximately equivalent to NO3. After 2 h of daytime aging, NH4+ increased and is approximately equivalent to the sum of Cl, SO42, and NO3. Particle size increased with downwind distance, causing particles to be more efficient scatters. Averaged over nine flights, mass scattering efficiency (MSE) increased in ∼ 2 h by 56 % and doubled in one flight. Mechanisms for redistributing mass from small to large particles are discussed. Coagulation is effective at moving aerosol from the Aitken to accumulation modes but yields only a minor increase in MSE. As absorption remained nearly constant with age, the time evolution of single scatter albedo was controlled by age-dependent scattering. Near-fire aerosol had a single scatter albedo (SSA) of 0.8–0.9. After 1 to 2 h of aging SSAs were typically 0.9 and greater. Assuming global-average surface and atmospheric conditions, the observed age dependence in SSA would change the direct radiative effect of a wildfire plume from near zero near the fire to a cooling effect downwind.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Projected increases in wildfires may challenge regulatory curtailment of PM 2.5 over the eastern US by 2050

Anthropogenic contribution to the overall fine particulate matter (PM 2.5 ) concentrations has been declining sharply in North America. In contrast, a steep rise in wildfire-induced air pollution events with recent warming is evident in the region. Here, based on coupled fire–climate–ecosystem model simulations, summertime wildfire-induced PM 2.5 concentrations are projected to nearly double in North America by the mid-21st century compared to the present. More strikingly, the projected enhancement in fire-induced PM 2.5 (~1–2 µg m -3 ) and its contribution (~15 %–20 %) to the total PM 2.5 are distinctively significant in the eastern US. This can be attributed to downwind transport of smoke from future enhancement of wildfires in North America to the eastern US and associated positive climatic feedback on PM 2.5 , i.e., perturbations in circulation, atmospheric stability, and precipitation. Therefore, the anticipated reductions in PM 2.5 from regulatory controls on anthropogenic emissions could be significantly compromised in the future in the densely populated eastern US.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Ecosystem leaf area, gross primary production, and evapotranspiration responses to wildfire in the Columbia River basin

Wildfires impact vegetation mortality and productivity and are increasing in intensity, frequency, and spatial area in the western United States. The rates of vegetation recovery after fires play a major role in the reestablishment of biomass and ecosystem functioning (e.g., structure, resilience, and productivity), but such recovery rates are poorly understood. Here we use remotely sensed data products from the Moderate Resolution Imaging Spectroradiometer (MODIS) to quantify the resistance and resilience of leaf area index (LAI), gross primary production (GPP), and evapotranspiration (ET) to 138 wildfires of various burn severity across the Columbia River basin (CRB) of the Pacific Northwest in 2015. Increasing burn severity caused lower resistance and resilience for all three variables. Resistance and resilience are highest in grasslands, intermediate in savanna, and lowest in needleleaf evergreen forests, consistent with the adaptation of these vegetation types to fire. LAI has consistently lower resistance and resilience than GPP and ET, which is consistent with physical and physiological mechanisms that compensate for reduced LAI. Resilience is influenced by precipitation, vapor pressure deficit (VPD), and burn severity across all three vegetation types; however, burn severity plays a more minor role in grasslands. Increasing wildfire severity will reduce the resistance and resilience and lengthen the recovery time of vegetation structure and fluxes with climate change, with significant consequences for the provision of ecosystem functioning and implications for model predictions.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Short communication: Cosmogenic noble gas depletion in soils by wildfire heating

Measurements of cosmic-ray-produced beryllium-10, neon-21, and helium-3 in quartz in a soil profile from a forested landscape in the Oregon Coast Range show that the cosmogenic noble gases 21 Ne and 3 He are depleted relative to 10Be in the shallow subsurface. The noble gases are mobile in mineral grains via thermally activated diffusion and 10 Be is not, implying that noble gas depletion is the result of surface heating by wildfires and subsequent mixing of partially degassed quartz downward into the soil. Cosmogenic noble gas depletion by wildfire heating of soils is a potential means of estimating wildfire intensity and/or frequency over pre-observational timescales.

58 GEOSCIENCES↗
DOE OSTI2025

ELM2.1-XGBfire1.0: improving wildfire prediction by integrating a machine learning fire model in a land surface model

Wildfires have shown increasing trends in both frequency and severity across the contiguous United States (CONUS). However, process-based fire models have difficulties in accurately simulating the burned area over the CONUS due to a simplification of the physical process and cannot capture the interplay among fire, ignition, climate, and human activities. The deficiency of burned area simulation deteriorates the description of fire impact on energy balance, water budget, and carbon fluxes in the Earth system models (ESMs). Alternatively, fire models based on machine learning (ML), which capture statistical relationships between the burned area and environmental factors, have shown promising burned area predictions and corresponding fire impact simulation. We develop a hybrid framework (ELM2.1-XGBFire1.0) that integrates an eXtreme Gradient Boosting (XGBoost) wildfire model with the Energy Exascale Earth System Model (E3SM) land model (ELM) version 2.1. A Fortran–C–Python deep learning bridge is adapted to support online communication between ELM and the ML fire model. Specifically, the burned area predicted by the ML-based wildfire model is directly passed to ELM to adjust the carbon pool and vegetation dynamics after disturbance, which are then used as predictors in the ML-based fire model in the next time step. Evaluated against the historical burned area from Global Fire Emissions Database 5 from 2001–2019, the ELM2.1-XGBFire1.0 outperforms process-based fire models in terms of spatial distribution and seasonal variations. The ELM2.1-XGBFire1.0 has proven to be a new tool for studying vegetation–fire interactions and, more importantly, enables seamless exploration of climate–fire feedback, working as an active component of E3SM.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS1974

An analysis of wildfire prevention

A model of the production of wildfire ignitions and damages is developed and used to determine wildland activity-regulation decisions, which minimize total expected cost-plus-loss due to wildfires. In this context, the implications of various policy decisions are considered. The resulting decision rules take a form that makes it possible for existing wildfire management agencies to readily adopt them upon collection of the required data.

Heineke, J. M.↗
NASA NTRS1988

Thermal analysis of wildfires and effects on global ecosystem cycling

Biomass combustion plays an important role in the earth's biogeochemical cycling. The monitoring of wildfires and their associated variables at global scales is feasible and can lead to predictions of the influence of combustion on biogeochemical cycling and tropospheric chemistry. Remote sensing data collected during the 1985 California wildfire season indicate that the information content of key thermal and infrared/thermal wave band channels centered at 11.5 microns, 3.8 microns, and 2.25 microns are invaluable for discriminating and calculating fire related variables. These variables include fire intensity, rate-of-spread, soil cooling recovery behind the fire front, and plume structure. Coinciding Advanced Very High Resolution Radiometer (AVHRR) data provided information regarding temperature estimations and the movement of the smoke plume from one wildfire into the Los Angeles basin.

Ambrosia, Vincent G.↗
NASA NTRS1998

Mapping the Distribution of Wildfire Fuels Using AVIRIS in the Santa Monica Mountains

Catastrophic wildfires, such as the 1990 Painted Cave Fire in Santa Barbara or Oakland fire of 1991, attest to the destructive potential of fire in the wildland/urban interface. For example, during the Painted Cave Fire, 673 structures were consumed over a period of only six hours at an estimated cost of 250 million dollars (Gomes et al., 1993). One of the primary sources of fuels is chaparral, which consists of plant species that are adapted to frequent fires and may actually promote its ignition and spread of through volatile organic compounds in foliage. As one of the most widely distributed plant communities in Southern California, and one of the most common vegetation types along the wildland urban interface, chaparral represents one of the greatest sources of wildfire hazard in the region. An ongoing NASA funded research project was initiated in 1994 to study the potential of AVIRIS for mapping wildfire fuel properties in Southern California chaparral. The project was initiated in the Santa Monica Mountains, an east-west trending range in western Los Angeles County that has experienced extremely high fire frequencies over the past 70 years. The Santa Monica Mountains were selected because they exemplify many of the problems facing the southwest, forming a complex mosaic of land ownership intermixed with a diversity of chaparral age classes and fuel loads. Furthermore, the area has a wide diversity of chaparral community types and a rich background in supporting geographic information including fire history, soils and topography. Recent fires in the Santa Monica Mountains, including several in 1993 and the Calabasas fire of 1996 attest to the active fire regime present in the area. The long term objectives of this project are to improve existing maps of wildland fuel properties in the area, link AVIRIS derived products to fuel models under development for the region, then predict fire hazard through models that simulate fire spread. In this paper, we describe the AVIRIS derived products we are developing to map wildland fuels.

Roberts, Dar↗
NASA NTRS2010

Satellite Observation Highlights of the 2010 Russian Wildfires

From late-July through mid-August 2010, wildfires raged in western Russia. The resulting thick smoke and biomass burning products were transported over the highly populated Moscow city and surrounding regions, seriously impairing visibility and affecting human health. We demonstrate the uniqueness of the 2010 Russian wildfires by using satellite observations from NASA's Earth Observing System (EOS) platforms. Over Moscow and the region of major fire activity to the southeast, we calculate unprecedented increases in the MODIS fire count record of 178 %, an order of magnitude increase in the MODIS fire radiative power (308%) and OMI absorbing aerosols (255%), and a 58% increase in AIRS total carbon monoxide (CO). The exceptionally high levels of CO are shown to be of comparable strength to the 2006 El Nino wildfires over Indonesia. Both events record CO values exceeding 30x10(exp 7) molec/ square cm.

Witte, Jacquelyn C.↗
NASA NTRS2018

Using NASA's Remote Sensing Datasets and Land Information System to Characterize Lightning Initiated Wildfires

Can we use modeled information of the land surface and characteristics of lightning beyond flash occurrence to increase the identification and prediction of wildfires? The goals of this study are to: combine observed cloud-to-ground (CG) flashes with real-time land surface model output; and compare data with areas where lightning did not start a wildfire to determine what land surface conditions, rainfall observations, and lightning characteristics were responsible for causing wildfires.

White, Kristopher↗
NASA NTRS2020

Gila Water Resources II: Using Earth Observations to Identify Wildfire Impacts on Hydrologic Functions and Recovery in the Gila National Forest

Wildfires have the potential to cause devastating and long-lasting impacts on ecological systems. In the Gila NF, wildfire events have occurred with increasing frequency and severity over recent years. These disturbances such as the historic Whitewater-Baldy Complex Fire (2012) and Silver Fire (2013) have raised concerns over post-fire soil erosion, flooding, debris flows, and vegetation recovery. Understanding the connection between burn events and ecological functions is crucial for developing effective land management practices within the Gila NF that ensure the conservation of the watershed. Our Gila Water Resources II team worked in partnership with the US Department of Agriculture (USDA) US Forest Service’s (USFS) Gila National Forest and Region 3. This project was designed to provide insight into the influence of wildfires on increased flooding events and to determine if restoration efforts in the Gila NF are having a beneficial and noticeable impact on recovery. The goals of this project included generating data-supported end products to inform land management decisions, including the prioritization of specific regions and time scales for post-fire restoration efforts. To understand recovery trends and hydrologic impact in the Gila NF between 2000 and 2020, this project used NASA EO and ancillary data, including, but not limited to, Landsat 5 Thematic Mapper (TM), Landsat 7 Enhanced Thematic Mapper Plus (ETM+), Landsat 8 Operational Land Imager (OLI), USGS stream gauge data, and data provided by USDA USFS’s Gila National Forest and Region 3.

Water Resources↗
NASA NTRS2020

Eastern Washington Disasters: Integrating NASA Earth Observations to Analyze Spatiotemporal Distributions of Lightning-Caused Wildfires in Eastern Washington

According to the Washington Department of Natural Resources, roughly 36% of large fires in the state since 2010 were caused by lightning. General trends also show a greater increase in the number of lightning-ignited fires over the last three decades. The NASA DEVELOP Eastern Washington Disasters team partnered with The Nature Conservancy’s Washington Chapter to analyze the relationship between lightning strikes and wildfire events in Eastern Washington, with an emphasis on Kittitas and Yakima Counties. Using the International Space Station Lightning Imaging Sensor, the Landsat 5 Thematic Mapper and Landsat 8 Operational Land Imager vegetation moisture index, and Washington Department of Natural Resources historical fire data, the team generated a lightning-caused fire vulnerability index for 2001-2019. Climatology maps of lightning, wildfire, and vegetation moisture of the study area, along with an Esri ArcGIS StoryMap, further communicated project findings. The project results demonstrated that spatiotemporal patterns of lightning-ignited wildfires in Eastern Washington can be useful to inform land management practices and better predict areas that may be more vulnerable to these events.

Disasters↗
NASA NTRS2020

Eastern Washington Disasters: Integrating NASA Earth Observations to Analyze Spatiotemporal Distributions of Lightning-Caused Wildfires in Eastern Washington

According to the Washington Department of Natural Resources, roughly 36% of large fires in the statesince 2010 were caused by lightning. General trends also show a greater increase in the number of lightning-ignited fires over the last three decades. The NASA DEVELOP Eastern Washington Disasters team partnered with The Nature Conservancy’s Washington Chapter to analyze the relationship between lightning strikes and wildfire events in Eastern Washington, with an emphasis on Kittitas and Yakima Counties. Using the International Space Station Lightning Imaging Sensor, the Landsat 5 Thematic Mapper and Landsat 8 Operational Land Imager vegetation moisture index, and Washington Department of Natural Resources historical fire data, the team generated a lightning-caused fire vulnerability index for 2001-2019. Climatology maps of lightning, wildfire, and vegetation moisture of the study area, along with an Esri ArcGIS StoryMap, further communicated project findings. The project results demonstrated that spatiotemporal patterns of lightning-ignited wildfires in Eastern Washington can be useful to inform land management practices and better predict areas that may be more vulnerable to these events.

Disasters↗