Abstract. The annual area burned due to wildfires in the western United States (WUS) increased bymore than 300 % between 1984 and 2020. However, accounting for the nonlinear, spatially heterogeneous interactions between climate, vegetation, and human predictors driving the trends in fire frequency and sizes at different spatial scales remains a challenging problem for statistical fire models. Here we introduce a novel stochastic machine learning (SML) framework, SMLFire1.0, to model observed fire frequencies and sizes in 12 km × 12 km grid cells across the WUS. This framework is implemented using mixture density networks trained on a wide suite of input predictors. The modeled WUS fire frequency matches observations at both monthly (r=0.94) and annual (r=0.85) timescales, as do the monthly (r=0.90) and annual (r=0.88) area burned. Moreover, the modeled annual time series of both fire variables exhibit strong correlations (r≥0.6) with observations in 16 out of 18 ecoregions. Our ML model captures the interannual variability and the distinct multidecade increases in annual area burned for both forested and non-forested ecoregions. Evaluating predictor importance with Shapley additive explanations, we find that fire-month vapor pressure deficit (VPD) is the dominant driver of fire frequencies and sizes across the WUS, followed by 1000 h dead fuel moisture (FM1000), total monthly precipitation (Prec), mean daily maximum temperature (Tmax), and fraction of grassland cover in a grid cell. Our findings serve as a promising use case of ML techniques for wildfire prediction in particular and extreme event modeling more broadly. They also highlight the power of ML-driven parameterizations for potential implementation in fire modules of dynamic global vegetation models (DGVMs) and earth system models (ESMs).
more »
« less
Does hot and dry equal more wildfire? Contrasting short‐ and long‐term climate effects on fire in the Sierra Nevada, CA
Climate and wildfire are closely linked. Climate regulates wildfire directly over short timescales through its effect on fuel aridity and indirectly over long timescales through vegetation productivity and the structure and abundance of fuels. Prediction of future wildfire regimes in a changing climate often uses empirical studies that presume current relationships between short‐term climate variables and wildfire activity will be stationary in the future. This is problematic because landscape‐scale wildfire dynamics exhibit non‐stationarity, with both positive and negative feedback loops that operate at different temporal and spatial scales. This requires that such feedbacks are accommodated in a model framework from which wildfire dynamics are emergent rather than pre‐specified. We use a new model, RHESSys‐WMFire, that integrates ecohydrology with fire spread and effects to simulate a 60‐yr time series of vegetation, fuel development, and wildfire in a 6572‐ha watershed in the Southern Sierra Nevada, USA, with a factorial design of increased temperature and severe drought. All climate scenarios had an initial pulse of elevated area burned associated with high temperature, low precipitation, and high fine fuel loading. There were positive correlations between annual area burned and mean annual maximum temperature and negative correlations with annual precipitation, consistent with understood direct effects of climate on wildfire in this system. Decreased vegetation productivity and increased fine fuel decomposition were predicted with increased temperature, resulting in long‐term reduced fine fuels and area burned relative to baseline. Repeated extreme drought increased area burned relative to baseline and over the long‐term had substantially reduced overstory biomass. Overstory biomass was resilient to repeat wildfire under baseline climate. The model system predicts that the short‐term direct effects of climate on wildfire can differ from long‐term indirect effects such that the simple maxim hotter/drier equals more wildfire can be both true and false, depending on scale.
more » « less- NSF-PAR ID:
- 10449221
- Publisher / Repository:
- Wiley Blackwell (John Wiley & Sons)
- Date Published:
- Journal Name:
- Ecosphere
- Volume:
- 12
- Issue:
- 7
- ISSN:
- 2150-8925
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
more » « less
Abstract Fuel and restoration treatments seeking to mitigate the likelihood of uncharacteristic high‐severity wildfires in forests with historically frequent, low‐severity fire regimes are increasingly common, but long‐term treatment effects on fuels, aboveground carbon, plant community structure, ecosystem resilience, and other ecosystem attributes are understudied. We present 20‐year responses to thinning and prescribed burning treatments commonly used in dry, low‐elevation forests of the western United States from a long‐term study site in the Northern Rockies that is part of the National Fire and Fire Surrogate Study. We provide a comprehensive synthesis of short‐term (<4 years) and mid‐term (<14 years) results from previous findings. We then place these results in the context of a mountain pine beetle (MPB;
Dendroctonus ponderosae ) outbreak that impacted the site 5–10 years post‐treatment and describe 20‐year responses to assess the longevity of restoration and fuel reduction treatments in light of the MPB outbreak. Thinning treatments had persistently lower forest density and higher tree growth, but effects were more pronounced when thinning was combined with prescribed fire. The thinning+prescribed fire treatment had the additional benefit of maintaining the highest proportion of ponderosa pine (Pinus ponderosa ) for overstory and regeneration. No differences in understory native plant cover and richness or exotic species cover remained after 20 years, but exotic species richness, while low relative to native species, was still higher in the thinning+prescribed fire treatment than the control. Aboveground live carbon stocks in thinning treatments recovered to near control and prescribed fire treatment levels by 20 years. The prescribed fire treatment and control had higher fuel loads than thinning treatments due to interactions with the MPB outbreak. The MPB‐induced changes to forest structure and fuels increased the fire hazard 20 years post‐treatment in the control and prescribed fire treatment. Should a wildfire occur now, the thinning+prescribed fire treatment would likely have the lowest intensity fire and highest tree survival and stable carbon stocks. Our findings show broad support that thinning and prescribed fire increase ponderosa pine forest resilience to both wildfire and bark beetles for up to 20 years, but efficacy is waning and additional fuel treatments are needed to maintain resilience. -
more » « less
Abstract Tree rings provide an invaluable long‐term record for understanding how climate and other drivers shape tree growth and forest productivity. However, conventional tree‐ring analysis methods were not designed to simultaneously test effects of climate, tree size, and other drivers on individual growth. This has limited the potential to test ecologically relevant hypotheses on tree growth sensitivity to environmental drivers and their interactions with tree size. Here, we develop and apply a new method to simultaneously model nonlinear effects of primary climate drivers, reconstructed tree diameter at breast height (DBH), and calendar year in generalized least squares models that account for the temporal autocorrelation inherent to each individual tree's growth. We analyze data from 3811 trees representing 40 species at 10 globally distributed sites, showing that precipitation, temperature, DBH, and calendar year have additively, and often interactively, influenced annual growth over the past 120 years. Growth responses were predominantly positive to precipitation (usually over ≥3‐month seasonal windows) and negative to temperature (usually maximum temperature, over ≤3‐month seasonal windows), with concave‐down responses in 63% of relationships. Climate sensitivity commonly varied with DBH (45% of cases tested), with larger trees usually more sensitive. Trends in ring width at small DBH were linked to the light environment under which trees established, but basal area or biomass increments consistently reached maxima at intermediate DBH. Accounting for climate and DBH, growth rate declined over time for 92% of species in secondary or disturbed stands, whereas growth trends were mixed in older forests. These trends were largely attributable to stand dynamics as cohorts and stands age, which remain challenging to disentangle from global change drivers. By providing a parsimonious approach for characterizing multiple interacting drivers of tree growth, our method reveals a more complete picture of the factors influencing growth than has previously been possible.
-
more » « less
Abstract Fire regimes are influenced by both exogenous drivers (e.g., increases in atmospheric CO2and climate change) and endogenous drivers (e.g., vegetation and soil/litter moisture), which constrain fuel loads and fuel aridity. Herein, we identified how exogenous and endogenous drivers can interact to affect fuels and fire regimes in a semiarid watershed in the inland northwestern United States throughout the 21st century. We used a coupled ecohydrologic and fire regime model to examine how climate change and CO2scenarios influence fire regimes. In this semiarid watershed, we found an increase in burned area and burn probability in the mid‐21st century (2040s) as the CO2fertilization effect on vegetation productivity outstripped the effects of climate change‐induced fuel decreases, resulting in greater fuel loading. However, by the late‐21st century (2070s), climatic warming dominated over CO2fertilization, thus reducing fuel loading and burned area. Fire regimes were shown to shift from flammability‐ to fuel‐limited or become increasingly fuel‐limited in response to climate change. We identified a metric to identify when fire regimes shift from flammability‐ to fuel‐limited: the ratio of the change in fuel loading to the change in its aridity. The threshold value for which this metric indicates a flammability versus fuel‐limited regime differed between grasses and woody species but remained stationary over time. Our results suggest that identifying these thresholds in other systems requires narrowing uncertainty in exogenous drivers, such as future precipitation patterns and CO2effects on vegetation.
-
more » « less
Abstract Circum-boreal and -tundra systems are crucial carbon pools that are experiencing amplified warming and are at risk of increasing wildfire activity. Changes in wildfire activity have broad implications for vegetation dynamics, underlying permafrost soils, and ultimately, carbon cycling. However, understanding wildfire effects on biophysical processes across eastern Siberian taiga and tundra remains challenging because of the lack of an easily accessible annual fire perimeter database and underestimation of area burned by MODIS satellite imagery. To better understand wildfire dynamics over the last 20 years in this region, we mapped area burned, generated a fire perimeter database, and characterized fire regimes across eight ecozones spanning 7.8 million km2of eastern Siberian taiga and tundra from ∼61–72.5° N and 100° E–176° W using long-term satellite data from Landsat, processed via Google Earth Engine. We generated composite images for the annual growing season (May–September), which allowed mitigation of missing data from snow-cover, cloud-cover, and the Landsat 7 scan line error. We used annual composites to calculate the difference Normalized Burn Ratio (dNBR) for each year. The annual dNBR images were converted to binary burned or unburned imagery that was used to vectorize fire perimeters. We mapped 22 091 fires burning 152 million hectares (Mha) over 20 years. Although 2003 was the largest fire year on record, 2020 was an exceptional fire year for four of the northeastern ecozones resulting in substantial increases in fire activity above the Arctic Circle. Increases in fire extent, severity, and frequency with continued climate warming will impact vegetation and permafrost dynamics with increased likelihood of irreversible permafrost thaw that leads to increased carbon release and/or conversion of forest to shrublands.
