Wildfires in western U.S. dry forest ecosystems have increased in size and severity during recent decades due primarily to more than a century of fire suppression, exclusion of Indigenous fire, and a rapidly warming climate. Fuel treatments have been employed to restore historical forest conditions and mitigate burn severity. However, their influence on burn severity in the context of other environmental variables and firefighting operations has not been extensively explored. The 2021 Bootleg Fire in south-central Oregon provided an opportunity to evaluate the effectiveness of mechanical thinning (Tx), broadcast burning (Rx), and both treatments combined (TxRx) near the Sycan Marsh Preserve, where pre-fire LiDAR data were also available. We assessed burn severity 1 year after the Bootleg Fire accounting for the local variability of top environmental drivers, fuel treatments, and firefighting operations. We modeled the influence of burn severity drivers using Random Forest and examined mean predictor effects (global scale) and their spatially explicit variability across observations (local scale) using SHapley Additive exPlanations (SHAP) analysis. Within units treated with broadcast burning, the percentage of area burned at low severity was over 80

Wildfire activity has accelerated with climate change, sparking concerns about uncharacteristic impacts on mature and old-growth forests containing large trees. Recent assessments have documented fire-induced losses of large-tree habitats in the US Pacific Northwest, but key uncertainties remain regarding contemporary versus historical fire effects in different forest composition types, specific impacts on large trees within closed versus open canopies, and the role of fuel reduction treatments. Focusing on the 2021 Schneider Springs Fire, which encompassed 43,000 ha in the eastern Cascade Range of Washington and burned during a period of severe drought, this study addresses three interrelated questions: (1) Are burn severity distributions consistent with historical fire regimes in dry, moist, and cold forest types? (2) How does burn severity vary among forest structure classes, particularly large trees with open versus closed canopies? (3) How do fuel reduction treatments influence forest structure and burn severity inside and outside of treated areas? Within each forest type, burn severity proportions were similar to historical estimates, with lower overall severity in dry forests than in moist and cold forests. However, across all forest types combined, high-severity fire affected 30% (4500 ha) of large-tree locations with tree diameters >50 cm. In each forest type, burn severity was lower in locations with large-open structure (<50% canopy cover) than in locations with large-closed structure (>50% canopy cover). Burn severity also was lower inside than outside treated sites in all structure classes, and untreated large-closed forests tended to burn at lower severity closer to treatments. These results highlight the susceptibility of dense, late-successional forests to contemporary fires, even in events with widespread potentially beneficial effects consistent with historical fire regimes. These results also illustrate the effectiveness of treatments that shift large-closed to large-open structures and suggest that treatments may help mitigate fire effects in adjacent large-closed forests. Long-term monitoring and adaptive management will be essential for conserving critical wildlife habitats and fostering ecosystem resilience to climate change, wildfires, and other disturbances.

The boreal biome, the largest terrestrial biome on Earth, is increasingly vulnerable to climate change due to warming twice as rapidly as the global average. Climate change has increased the temperature, frequency, severity, and amount of area burned, which is leading to changes in the spatial extent of forest type and species range. These rapid ecological shifts necessitate fine-scale monitoring of forest type to detect potential type conversions and guide management interventions. In this study, we present a framework for forest type classification combining field plots and high-resolution remote sensing data using machine learning models in the boreal forest of Interior Alaska. For this purpose, we conducted forest type classification at three different levels, including 1. forest and nonforest, 2. hardwood, softwood, and nonforest, and 3. three dominant forest types, including paper birch, black spruce, white spruce, and nonforest. To achieve this goal, we compared the performance of two advanced modeling approaches, the convolutional neural network (CNN) and the XGBoost model. Our datasets included field and high-resolution topographic metrics including elevation, slope, aspect, and solar radiation and canopy height derived from lidar (1 m) and 44 vegetation indices derived from high-resolution (1 m) visible to near infrared (VNIR) hyperspectral data collected by NASA Goddard's Lidar, Hyperspectral and Thermal Imager (G-LiHT) sensor. The remote sensing data were collected under variable sky conditions (clear to overcast) throughout a 1-month growing-season period, and field data collected by United States Department of Agriculture (USDA) Forest Service, Forest Inventory and Analysis program (FIA). In this framework, we also studied the importance of topographic and remote sensing variables for the classification of forest types. We found the CNN model outperformed the XGBoost model in terms of overall accuracy and a macro average F1 score for all three different forest type classifications. The CNN model achieved an overall accuracy of 93.1% for forest or nonforest, 82.6% for hardwood, softwood, and nonforest, and 74.7% for three dominant forest types including paper birch, black spruce, and white spruce along with nonforest. Among the various topographic factors, we found that elevation was the most important factor for discriminating all forest types. In addition, we found that canopy height and vegetation indices including Photochemical Reflectance Index (PRI) (R531 & R570), Pigment Specific Normalized Difference (PSND) (R635 & R800), and Gitelson and Merzlyak (GM1) (R550 & R750) were important for differentiating between hardwood and softwood while Anthocyanin Reflectance Index (ARI1) (R550 & R700) was important for differentiating between forest and nonforest. The high-resolution forest type information can improve our ecological understanding of boreal forest dynamics, estimate above ground biomass, and carbon, and support the national forest inventory and forest managers.

Abstract Background Understanding the impacts of climate change on forest aboveground biomass is a high priority for land managers. High elevation subalpine forests provide many important ecosystem services, including carbon sequestration, and are vulnerable to climate change, which has altered forest structure and disturbance regimes. Although large, regional studies have advanced aboveground biomass mapping with satellite data, typically using a general approach broadly calibrated or trained with available field data, it is unclear how well these models work in less prevalent and highly heterogeneous forest types such as the subalpine. Monitoring biomass using methods that model uncertainty at multiple scales is critical to ensure that local relationships between biomass and input variables are retained. Forest structure metrics from lidar are particularly valuable alongside field data for mapping aboveground biomass, due to their high correlation with biomass. Results We estimated aboveground woody biomass of live and dead trees and uncertainty at 30 m resolution in subalpine forests of the Sierra Nevada, California, from aerial lidar data in combination with a collection of field inventory data, using a Bayesian geostatistical model. The ten-fold cross-validation resulted in excellent model calibration of our subalpine-specific model (94.7% of measured plot biomass within the predicted 95% credible interval). When evaluated against two commonly referenced regional estimates based on Landsat optical imagery, root mean square error, relative standard error, and bias of our estimations were substantially lower, demonstrating the benefits of local modeling for subalpine forests. We mapped AGB over four management units in the Sierra Nevada and found variable biomass density ranging from 92.4 to 199.2 Mg/ha across these management units, highlighting the importance of high quality, local field and remote sensing data. Conclusions By applying a relatively new Bayesian geostatistical modeling method to a novel forest type, our study produced the most accurate and precise aboveground biomass estimates to date for Sierra Nevada subalpine forests at 30 m pixel and management unit scales. Our estimates of total aboveground biomass within the management units had low uncertainty and can be used effectively in carbon accounting and carbon trading markets.

AbstractInterruption of frequent burning in dry forests across western North America and the continued impacts of anthropogenic climate change have resulted in increases in fire size and severity compared to historical fire regimes. Recent legislation, funding, and planning have emphasized increased implementation of mechanical thinning and prescribed burning treatments to decrease the risk of undesirable ecological and social outcomes due to fire. As wildfires and treatments continue to interact, managers require consistent approaches to evaluate treatment effectiveness at moderating burn severity. In this study, we present a repeatable, remote sensing–based, analytical framework for conducting fire‐scale assessments of treatment effectiveness that informs local management while also supporting cross‐fire comparisons. We demonstrate this framework on the 2021 Bootleg Fire in Oregon and the 2021 Schneider Springs Fire in Washington. Our framework used (1) machine learning to identify key bioclimatic, topographic, and fire weather drivers of burn severity in each fire, (2) standardized workflows to statistically sample untreated control units, and (3) spatial regression modeling to evaluate the effects of treatment type and time since treatment on burn severity. The application of our framework showed that, in both fires, recent prescribed burning treatments were the most effective at reducing burn severity relative to untreated controls. In contrast, thinning‐only treatments only produced low/moderate‐severity effects under the more moderate fire weather conditions in the Schneider Springs Fire. Our framework offers a robust approach for evaluating treatment effects on burn severity at the scale of individual fires, which can be scaled up to assess treatment effectiveness across multiple fires. As climate change brings increased uncertainty to dry forest ecosystems of western North America, our framework can support more strategic management actions to reduce wildfire risk and foster resilience.

Tree mortality is rapidly increasing as a result of more frequent and extensive droughts and forest fires across the globe. The increasing pace and scale of disturbance and resulting mortality necessitates the study of tree mor-tality at the scale at which it occurs (individual trees) to develop effective management strategies. Achieving this objective requires high-resolution aerial or satellite remote sensing data supported by field validation and effi-cient modeling. However, the optimal remote sensing data modalities and models for mapping overstory tree mortality remain uncertain and there is a lack of benchmark datasets for reproducible research, comparison, and model improvement. In this research, we propose a framework to generate tree mortality maps at the scale of individual trees and a labeled benchmark dataset using open-source high-resolution remote sensing data. Our benchmark dataset consists of over 15,000 image crops that have been manually labeled via multi-annotator majority for live, dead, mixed, and other classes in National Ecological Observatory Network (NEON) RGB, NEON hyperspectral, and National Agriculture Imagery Program (NAIP) RGBNIR imagery. Specifically, we present a multi-class classification framework that categorizes lidar-derived tree segments as live, dead, mixed, or other. We compared the performance of different machine learning models, remote sensing data modalities and their associated spatial resolution, spectral depth, and orthorectification using our benchmark dataset. Among various models, we found Convolutional Neural Network (CNN) models outperformed all other models with the highest accuracies and F1 scores across all data modalities. Similarly, among remote sensing data modalities, NEON RGB data at 0.1 m resolution performed best due to its high spatial resolution and high quality orthorectification. The overall accuracy of 87.27% with a macro F1 score of 0.76 was achieved using the RGB data with the CNN model. Our framework can be applied to any forest ecosystem in which lidar and raster imagery are available. We applied the framework using the best performing model and data modality to produce an individual tree mortality map over the Teakettle Experimental Forest in the southern Sierra Nevada. The framework and high-resolution mortality map can be valuable resources for forest managers and provide an invaluable basis for studying potential mortality drivers of dominant trees, forest inventory, selective logging, fire disturbance, and succession modeling.

Following a century of fire suppression in western North America, managers use forest restoration treatments to reduce fuel loads and reintroduce key processes like fire. However, annual area burned by wildfire frequently outpaces the application of restoration treatments. As this trend continues under climate change, it is essential that we understand the effects of contemporary wildfires on forest ecosystems and the extent to which post-fire structures are meeting common forest restoration objectives. In this study, we used airborne lidar to evaluate fire effects across yellow pine and mixed conifer (YPMC) forests of California’s Sierra Nevada. We quantified the degree to which forest structures in first-entry burned areas (previously unburned since 1900s) and unburned controls aligned with restoration targets derived from contemporary reference sites. We also identified environmental conditions that contributed to more restorative fire effects. Relative to unburned controls, structural patterns in first-entry burned areas aligned more closely with reference sites. Yet, across all burn severities, first-entry wildfires were only moderately successful at meeting targets for canopy cover (48

Maintaining dense forest habitats for the threatened northern spotted owl (NSO) has proven challenging in seasonally dry, fire-dependent landscapes where low-density conditions were historically dominant and are generally more climate- and disturbance-resilient. To better inform the dual, sometimes-conflicting objectives of species conservation and forest resilience, we developed an approach to evaluate NSO habitat sustainability by: (1) quantifying the structure of high suitability habitat (HSH) associated with NSO using two remotely sensed platforms, (2) estimating current and historical HSH abundance, and (3) identifying HSH locations more likely to persist given current and future forest-zone climate projections and increasing risk of severe wildfire. Tall, closed-canopy conditions effectively comprised the key structural features of HSH, providing a means to map habitat through time. Both historical amounts and contemporary spatial patterns of HSH and other forest and non-forest conditions around occupied NSO sites indicated that HSH and forest resilience goals can be congruent at multiple scales. Independent lines of evidence suggest HSH historically composed similar to 18-24% of the dry and moist mixed-conifer landscape - considerably lower levels than current management goals in many areas. Projected shifts in climate and severe-fire likelihood suggest substantial spatial and temporal shifts where HSH will be sustainable into the future - mainly in currently moist as well as some cold forest types. These findings can inform the potential convergence and trade-offs of species conservation and disturbance resilience goals across local and regional landscapes, based on the inherent capacity of the landscape to support both goals under projected shifts in climate and wildfire.

Contemporary reference sites in California's Sierra Nevada represent areas where a frequent, low-intensity fire regime – an integral ecological process in temperate dry forests – has been reintroduced after several decades of fire suppression. Produced by an intact fire regime, forest structural patterns in these sites are likely more resilient to future disturbances and climate, and thus can provide reference conditions to guide management and ecological research. In this paper, we present a set of 119 delineated contemporary reference sites in the Sierra Nevada yellow pine and mixed-conifer zone along with a suite of key remote sensing-derived forest structure metrics representing conditions within these sites. We also provide a set of summary figures for individual reference sites and sites grouped by dominant climate class. We identified restored frequent-fire landscapes using a combination of fire history, burn severity, management history, and forest type datasets and we delineated individual polygons using catchment basins, fire perimeters, and imagery. Reference sites ranged in size from 101-966 ha with a mean size of 240 ha. Where available (for 59 sites), we used airborne lidar datasets to characterize a suite of key forest structure metrics within reference sites. Across all 119 sites, we provide a set of forest structure metrics produced by the California Forest Observatory. Reference sites were categorized based on their dominant climate class to assist users in identifying the most climatically relevant reference conditions for their project or study area. We encourage the use of the reference sites and associated forest structure datasets for guiding ecologically focused forest management and research in the Sierra Nevada.

Large trees (>= 76.2 cm/>= 30 '' DBH) and especially very large trees (>= 101.6 cm/>= 40 '' DBH) are key structures of Sierra Nevada forests for their ecological function, habitat, and carbon storage. Many of these trees have been lost to historic harvest and more recently to drought and wildfires. Understanding the current frequency and distribution of these large trees is essential to understanding their ecological contribution and management needs. We used airborne lidar to census large trees across three Sierra Nevada landscapes (cumulatively 396 K ha) in lower (dominated by ponderosa pine and mixed conifer) and upper (dominated by red fir) montane forest zones. We used data from a network of Forest Inventory and Analysis (FIA) plots to interpret our lidar-based results for large tree frequency, species, and ages. The lidar data identified > 8 M large and > 2.7 M very large trees, and their mean densities were similar to those from FIA data. Large portions of our study areas had either no or low densities (<20) of large trees per hectare. We found that large and very large tree concentrations were spatially aggregated with most in denser patches containing 20 to 50 + large trees per hectare. Depending on the study area, these often sizable (>1000 ha) patches of dense large trees can cover 20% to 40% of the landscape. (Patches of denser very large trees cover less of the landscape, typically 5% to 10%). However, these large patches are rarely simple blocks. Instead, they typically form complex amorphous matrices interspersed with patches of forests containing shorter trees or non-forest cover. Crucially, almost all large trees were in stands with high canopy cover, suggesting horizontal fuel continuity and low resilience to future wildfires. For lower montane large trees, canopy cover versus large tree density showed almost a unimodal response with canopy cover of 60% to 80% for locations with > 20 large trees per ha. For upper montane large trees, canopy cover versus large tree density showed a more linear relationship for all three study areas. High levels of canopy cover, especially for lower montane forests, suggest settings in which infilling following decades of fire suppression have created overly dense stands with lower resilience to drought and wildfire. Other studies have documented substantial recent losses of these large trees to both factors. The high canopy cover within which almost all large trees exist emphasizes the need for treatment almost everywhere that large trees are present for lower montane forests. This likely will require treatments both within the stands that contain large trees and across the landscapes in which they are found.

Although recent large wildfires in California forests are well publicized in media and scientific literature, their cumulative effects on forest structure and implications for forest resilience remain poorly understood. In this study, we evaluated spatial patterns of burn severity for 18 exceptionally large fires and compared their cumulative impacts to the hundreds of smaller fires that have burned across California forests in recent decades. We used a burn severity atlas for over 1,800 fires that burned in predominantly conifer forests between 1985 and 2020 and calculated landscape metrics to evaluate spatiotemporal patterns of unburned refugia, low-moderate-severity, and high-severity post-fire effects. Total annual area burned, mean annual fire size, and total annual core area burned at high severity all significantly increased across the study period. Exceptionally large fires (i.e., the top 1% by size) were responsible for 58% and 42% of the cumulative area burned at high and low-moderate severities, respectively, across the study period. With their larger patch sizes, our results suggest that exceptionally large fires coarsen the landscape pattern of California’s forests, reducing their fine-scale heterogeneity which supports much of their biodiversity as well as wildfire and climate resilience. Thus far, most modern post-fire management has focused on restoring forest cover and minimizing ecotype conversion in large, high-severity patches. These large fires, however, have also provided extensive areas of low-moderate severity burns where managers could leverage the wildfire’s initial “treatment” with follow-up fuel reduction treatments to help restore finer-scale forest heterogeneity and fire resilience.

Yellow pine and mixed-conifer (YPMC) forests of California’s Sierra Nevada have experienced widespread fire suppression for over a century, resulting in ingrowth and densification of trees, heavy fuel accumulation, and shifts in species composition. Under warmer and drier climates, these forests are primed for stand-replacing fires and severe drought mortality, requiring management interventions to improve their resilience and mitigate future impacts. Characterizations of structural patterns from functioning frequent-fire systems (e.g., contemporary reference sites) provide insights about emergent properties in these systems, which can be used to inform regional restoration efforts. In this study, we used airborne lidar data to quantify and compare multi-scale structures between contemporary reference sites (i.e., forests with repeat low/moderate-severity fire effects) and control sites (i.e., typical fire-suppressed forests) to identify hierarchical patterns that emerge under a modern, frequent, and low-intensity fire regime. We evaluated structures at the neighborhood- (~1 ha), site- (~100-1000 ha), and among-site- (~1000-10000 ha) levels. We consistently observed a nested and hierarchical set of structural patterns across contemporary reference sites, indicating an emergent and self-organizing property of contemporary frequent-fire forests. High proportions of individual trees, small clumps of 2-4 trees, and open space formed mostly open canopy structures at the neighborhood-level, which were consistently arranged in heterogenous spatial patterns across all sites. In fire suppressed control sites, edaphic factors and other non-fire disturbances sometimes produced heterogeneity at the neighborhood- and site-level, but the degree of heterogeneity was not consistent across sites. Structural patterns in contemporary reference sites suggest improved resilience to future fire and drought; thus, we encourage the use of these metrics to guide and inform multi-scale, resilience-focused management in the Sierra Nevada.

Geospatial polygons representing fire-suppressed control sites against which fire-intact reference sites were compared in Chamberlain et al. (2023). Control sites represent areas with 1) no record of fire history, 2) no record of late 20th century or early 21st century timber management, and 3) no "Fast Change" detected by the Landscape Change Monitoring System dataset. All sites are predominantly within the yellow pine and mixed-conifer zone of California's Sierra Nevada, USA. Polygon boundaries were defined using the NHDPlusV2 catchments, and were manually reshaped using aerial imagery to ensure that polygons were > 100 ha, represented primarily forested areas, and excluded major roads, infrastructure, and major rock outcrops. Detailed description of the methods used to produce this dataset provided in: Chamberlain, C.P., Cova, G.R., Cansler, C.A., North, M.P., Meyer, M.D., Jeronimo, S.M.A., Kane, V.R., 2023. Consistently heterogeneous structures observed at multiple spatial scales across fire-intact reference sites. Forest Ecology and Management.

The Composite Burn Index (CBI) is commonly linked to remotely sensed data to understand spatial and temporal patterns of burn severity. However, a comprehensive understanding of the tradeoffs between different methods used to model CBI with remotely sensed data is lacking. To help understand the current state of the science, provide a blueprint towards conducting broad-scale meta-analyses, and identify key decision points and potential rationale, we conducted a review of studies that linked remotely sensed data to continuous estimates of burn severity measured with the CBI and related methods. We provide a roadmap of the different methodologies applied and examine potential rationales used to justify them. Our findings largely reflect methods applied in North America – particularly in the western USA – due to the high number of studies in that region. We find the use of different methods across studies introduces variations that make it difficult to compare outcomes. Additionally, the existing suite of comparative studies focuses on one or few of many possible sources of uncertainty. Thus, compounding error and propagation throughout the many decisions made during analysis is not well understood. Finally, we suggest a broad set of methodological information and key rationales for decision-making that could facilitate future reviews.

Post-fire landscapes are the frontline of forest ecosystem change. As such, they represent opportunities to foster conditions that are better adapted to future climate and wildfires with post-fire management. In western US landscapes, post-fire management has been mostly defined by short-term emergency mitigation measures, salvage harvest to recover economic value, and replanting to achieve full stocking. These approaches are largely incongruent with ecologically based forest management due to their limited scope and objectives. Here, we develop a framework for ecologically based post-fire management. Post-fire management principles are to (i) protect large-diameter trees and fire refugia; (ii) anticipate future fuel accumulation from post-fire tree mortality; (iii) reinitiate and maintain stabilizing fire-vegetation feedbacks; (iv) differentiate between climate- and dispersal-mediated transitions to non-forest; and (v) align species composition and structure with future fire regimes and climate. Stand-scale management strategies to implement these principles include (i) maintain or enhance forest resilience; (ii) restore forest conditions and resist transition to non-forest; and (iii) accept or facilitate transition to non-forest. Determining where and over what extent to deploy these stand-scale strategies in large, burned landscapes is informed by a post-fire landscape evaluation, and expressed with a landscape prescription. A post-fire landscape evaluation is a data-driven characterization of current vegetation conditions, including the immediate changes caused by wildfire, and includes a departure analysis—an evaluation of current conditions against reference conditions. The landscape prescription provides guidance about the distribution of different successional patches and their sizes across the topographic template and identifies priority areas for different post-fire treatments. We develop a geospatial framework to integrate ecological principles with a post-fire landscape evaluation that can be readily applied to management planning after wildfire. We illustrate application of these principles through the development of landscape prescriptions for two watersheds, each burned in a recent large fire, in northeast Washington, USA. Use of ecologically based post-fire management principles and landscape evaluations can help shift often contentious debates over salvage harvesting towards a more productive dialogue around how to best adapt landscapes to future conditions.

We investigated the relative importance of daily fire weather, landscape position, climate, recent forest and fuels management, and fire history to explaining patterns of remotely-sensed burn severity - as measured by the Relativized Burn Ratio - in 150 fires occurring from 2001 to 2019, which burned conifer forests of northeastern Washington State, USA. Daily fire weather, annual precipitation anomalies, and species' fire resistance traits were important predictors of wildfire burn severity. In areas burned within the past two to three decades, prior fire decreased the severity of subsequent burns, particularly for the first 16 postfire years. In areas managed before a wildfire, thinning and prescribed burning treatments lowered burn severity relative to untreated controls. Prescribed burning was the most effective treatment at lowering subsequent burn severity, and prescribed burned areas were usually unburned or burned at low severity in subsequent wildfires. Patches that were harvested and planted <10 years before a wildfire burned with slightly higher severity. In areas managed within 5 years after an initial fire, postfire harvest and planting reduced prevalence of stand-replacing fire in reburns. However, overall, postfire management actions after a first wildfire only weakly influenced the severity of subsequent fires. The importance of fire-fire interactions to moderating burn severity establishes the importance of stabilizing feedbacks in active fire regimes, and our results demonstrate how silvicultural treatments can be combined with prescribed fire and wildfires to maintain resilient landscapes.

We evaluated the effects of postfire management on forest structure in mixed-conifer forests of northeastern Washington, USA. Postfire treatments were harvest-only, harvest combined with planting, planting-only, and postfire prescribed fire. We used aerial light detection and ranging (LiDAR) to measure vertical and horizontal components of postfire forest structure over a period of 2 to 32 years after fires. We compared treated areas to control areas with similar bioclimatic environments and past fire severity. We used niche overlap statistics to quantify distributions of individual forest structure components and PERMANOVA to assess forest structural response to the presence or absence of treatments, past fire severity, time since treatment, and bioclimatic setting. Harvest alone after fire decreased dominant tree height and reduced vertical canopy complexity and the cover of tall trees. Harvest combined with planting increased dominant tree height, vertical complexity, and cover in lower height strata. Planting and prescribed fires showed little difference in forest structure relative to untreated controls. Overall, the burn severity of the initial fire was the strongest influence on postfire structure, and many aspects of vertical and horizontal forest structure showed little difference with increasing time since fire.