Scale-dependent controls on the area burned in the boreal forest of Canada, 1980-2005

Characterization of biophysical contexts leading to severe wildfires in Portugal and their environmental controls

Characterizing the fire regime in regions prone to extreme wildfire behavior is essential for providing comprehensive insights on potential ecosystem response to fire disturbance in the context of global change. We aimed to disentangle the linkage between contemporary damage-related attributes of wildfires as shaped by the environmental controls of fire behavior across mainland Portugal. We selected large wildfires (≥100 ha, n = 292) that occurred during the 2015-2018 period, covering the full spectrum of large fire-size variation. Ward's hierarchical clustering on principal components was used to identify homogeneous wildfire contexts at landscape scale on the basis of fire size, proportion of high fire severity, and fire severity variability, and their bottom-up (pre-fire fuel type fraction, topography) and top-down (fire weather) controls. Piecewise Structural Equation Modeling was used to disentangle the direct and indirect relationships between fire characteristics and fire behavior drivers. Cluster analysis evidenced severe and large wildfires in the central region of Portugal displaying consistent fire severity patterns. Thus, we found a positive relationship between fire size and proportion of high fire severity, which was mediated by distinct fire behavior drivers involving direct and indirect pathways. A high fraction of conifer forest within wildfire perimeters and extreme fire weather were primarily responsible for those interactions. In the context of global change, our results suggest that pre-fire fuel management should be targeted at expanding the fire weather settings in which fire control is feasible and promote less flammable and more resilient forest types.

Future wildfire extent and frequency determined by the longest fire-conducive weather spell

Great efforts have been made to understand the impacts of a changing climate on fire activity; however, a reliable approach with high prediction confidence has yet to be found. By establishing linkages between the longest duration of fire-conducive weather spell and fire activity parameters, this study projected annual area burned (AAB), annual number of fires (ANF), and annual maximum fire size (MFS) into the future. We found that even though the rates of change differ, the spatial pattern of changes for all three parameters are similar by Canadian ecozone. Areas with the lowest fire activity may see higher rates of change in comparison to high fire activity areas. By end of the century, the changes of AAB and MFS for the study area are projected to be about four and five times that of the baseline respectively, while ANF could almost double.

Prediction of regional wildfire activity in the probabilistic Bayesian framework of Firelihood

Modeling wildfire activity is crucial for informing science-based risk management and understanding the spatiotemporal dynamics of fire-prone ecosystems worldwide. Models help disentangle the relative influences of different factors, understand wildfire predictability, and provide insights into specific events. Here, we develop Firelihood, a two-component, Bayesian, hierarchically structured, probabilistic model of daily fire activity, which is modeled as the outcome of a marked point process: individual fires are the points (occurrence component), and fire sizes are the marks (size component). The space-time Poisson model for occurrence is adjusted to gridded fire counts using the integrated nested Laplace approximation (INLA) combined with the stochastic partial differential equation (SPDE) approach. The size model is based on piecewise-estimated Pareto and generalized Pareto distributions, adjusted with INLA. The Fire Weather Index (FWI) and forest area are the main explanatory variables. Temporal and spatial residuals are included to improve the consistency of the relationship between weather and fire occurrence. The posterior distribution of the Bayesian model provided 1,000 replications of fire activity that were compared with observations at various temporal and spatial scales in Mediterranean France. The number of fires larger than 1 ha across the region was coarsely reproduced at the daily scale, and was more accurately predicted on a weekly basis or longer. The regional weekly total number of larger fires (10-100 ha) was predicted as well, but the accuracy degraded with size, as the model uncertainty increased with event rareness. Local predictions of fire numbers or burned areas also required a longer aggregation period to maintain model accuracy. The estimation of fires larger than 1 ha was also consistent with observations during the extreme fire season of the 2003 unprecedented heat wave, but the model systematically underrepresented large fires and burned areas, which suggests that the FWI does not consistently rate the actual danger of large fire occurrence during heat waves. Firelihood enabled a novel analysis of the stochasticity underlying fire hazard, and offers a variety of applications, including fire hazard predictions for management and projections in the context of climate change.

Influence of climatic factor of changes in forest fire danger and fire season length in Turkey

In contrast to the expectations of an increase in annual fire activity and the severity of fire season due to climate change and large fires, which have been occurring in recent years, a downtrend has been identified in fire activity in many studies conducted for the whole of Europe in recent years. Similarly, in Turkey, according to the General Directorate of Forestry statistics, while there is an increase in the number of annual fires, the burnt area has a downtrend pattern. In this study, fire activity and climate data statistics for Turkey were examined along with the fire season length and severity. The results obtained conform with the studies conducted in places from Spain at the westernmost part of Mediterranean Europe to Israel at the easternmost part of the Mediterranean. Considering the changes in temperatures, temperature rise of 2 to 3 °C was detected at all stations in the study area. No decrease was observed in the average temperatures at any of the stations within the study period between 1940 and 2018. On the other hand, the precipitation trend varied according to the stations. Although there have been increases in precipitation in Fethiye, Isparta, and Marmaris since 1960, the decrease in precipitation by 132 mm in Afyon since 1970 and the decrease in precipitation by 137 mm in Bodrum since 1940 are attention-grabbing. These stations are followed by Izmir station with 66 mm and Cesme station with 37 mm of decrease, despite being smaller decreases. In the study, the long-term (1940-2018) data of the meteorological stations discussed within the study, the Canadian Fire Weather Index (FWI) and the Fine Fuel Moisture Code (FFMC) values were calculated. According to the FWI results used in determining the severity and length of fire season on the coastline of Turkey from the northern Aegean to Antalya, the likelihood of large fires decreased by about 52% in 2018 compared to 1970. This decrease in FWI value indicates that the fire severity is reduced. The specified decrease in fire severity also explains the reason of the decrease in the burnt area that occurred over the years in Turkey. No significant change was observed in the FFMC values indicating the possibility of human-induced fires between 1970 and 2018.

Predicting forest fire kernel density at multiple scales with geographically weighted regression in Mexico

Identifying the relative importance of human and environmental drivers on fire occurrence in different regions and scales is critical for a sound fire management. Nevertheless, studies analyzing fire occurrence spatial patterns at multiple scales, covering the regional to national levels at multiple spatial resolutions, both in the fire occurrence drivers and in fire density, are very scarce. Furthermore, there is a scarcity of studies that analyze the spatial stationarity in the relationships of fire occurrence and its drivers at multiple scales. The current study aimed at predicting the spatial patterns of fire occurrence at regional and national levels in Mexico, utilizing geographically weighted regression (GWR) to predict fire density, calculated with two different approaches -regular grid density and kernel density - at spatial resolutions from 5 to 50 km, both in the dependent and in the independent human and environmental candidate variables. A better performance of GWR, both in goodness of fit and residual correlation reduction, was observed for prediction of kernel density as opposed to regular grid density. Our study is, to our best knowledge, the first study utilizing GWR to predict fire kernel density, and the first study to utilize GWR considering multiple scales, both in the dependent and independent variables. GWR models goodness of fit increased with fire kernel density search radius (bandwidths), but saturation in predictive capacity was apparent at 15-20 km for most regions. This suggests that this scale has a good potential for operational use in fire prevention and suppression decision-making as a compromise between predictive capability and spatial detail in fire occurrence predictions. This result might be a consequence of the specific spatial patterns of fire occurrence in Mexico and should be analyzed in future studies replicating this methodology elsewhere.

Forest Understorey Vegetation: Colonization and the Availability and Heterogeneity of Resources

Understorey vegetation comprises a major portion of plant diversity and contributes greatly to nutrient cycling and energy flow. This review examines the mechanisms involved in the response of understorey vegetation to stand development and the overstorey canopy following disturbances. The overall abundance and diversity of the understorey is enhanced with the availability and heterogeneity of light, soil nutrients, soil moisture, and substrates. Vascular plants are positively impacted by the availability and heterogeneity of light and soil nutrients, whereas non-vascular vegetation is more strongly influenced by colonization time, soil moisture, and substrates, and is decreased with a higher proportion of broadleaf overstorey. The availability of resources is a prominent driver toward the abundance and diversity of understorey vegetation, from the stand initiation to stem exclusion stage under a single-species dominated overstorey. However, resource heterogeneity dominates at the later stages of succession under a mixed overstorey. Climate and site conditions modify resource availability and heterogeneity in the understorey layer, but the extent of their influences requires more investigation. Forest management practices (clearcutting and partial harvesting) tend to increase light availability and heterogeneity, which facilitates the abundance and diversity of understorey vascular plants; however, these factors reduce the occurrence of non-vascular plants. Nevertheless, in the landscape context, anthropogenic disturbances homogenize environmental conditions and reduce beta-diversity, as well, the long-term effects of anthropogenic disturbances on understorey vegetation remain unclear, particularly compared with those in primary forests.

Machine learning to predict final fire size at the time of ignition

Fires in boreal forests of Alaska are changing, threatening human health and ecosystems. Given expected increases in fire activity with climate warming, insight into the controls on fire size from the time of ignition is necessary. Such insight may be increasingly useful for fire management, especially in cases where many ignitions occur in a short time period. Here we investigated the controls and predictability of final fire size at the time of ignition. Using decision trees, we show that ignitions can be classified as leading to small, medium or large fires with 50.4 ± 5.2% accuracy. This was accomplished using two variables: vapour pressure deficit and the fraction of spruce cover near the ignition point. The model predicted that 40% of ignitions would lead to large fires, and those ultimately accounted for 75% of the total burned area. Other machine learning classification algorithms, including random forests and multi-layer perceptrons, were tested but did not outperform the simpler decision tree model. Applying the model to areas with intensive human management resulted in overprediction of large fires, as expected. This type of simple classification system could offer insight into optimal resource allocation, helping to maintain a historical fire regime and protect Alaskan ecosystems.

The role of short-term weather conditions in temporal dynamics of fire regime features in mainland Spain

In this paper we investigate spatial-temporal associations of fire weather danger and fire regime features from 1979 to 2013. We analyze monthly time series of fire activity (number of fires and burned area) and fire weather danger rating indices (Fire Weather Index, Burning Index and Forest Fire Danger Index) at two spatial scales: (i) regionally, splitting the Spanish mainland into Northwest, Hinterland and Mediterranean regions; and (ii) locally, using the EMCWF grid. All analyses are based on decomposing time series to retrieve differential indicators of seasonal cycles, temporal evolution and anomalies. At regional scale we apply lagged cross-correlation analysis (4 lags or months before fire) to explore seasonal associations; and trend detection tests on the temporal evolution component. At the local scale, we calculate Pearson correlation coefficients between each individual index and the 18 possible fire-activity subsets according to fire size (all sizes, >1 ha and >100 ha) and source of ignition (natural, unintended and arson); this analysis is applied to both cycles, temporal and anomalies series. Results suggest that weather controls seasonal fire activity although it has limited influence on temporal evolution, i.e. trends. Stronger associations are detected in the number of fires in the Northwest and Hinterland regions compared to the Mediterranean, which has desynchronized from weather since 1994. Cross-correlation analysis revealed significant fire-weather associations in the Hinterland and Mediterranean, extending up to two months prior fire ignition. On the other hand, the association between temporal trends and weather is weaker, being negative along the Mediterranean and even significant in the case of burned area. The spatial disaggregation into grid cells reveals different spatial patterns across fire-activity subsets. Again, the connection at seasonal level is noticeable, especially in natural-caused fires. In turn, human-related wildfires are occasionally found independent from weather in some areas along the northern coast or the Ebro basin. In any case, this effect diminishes as the size of the fire increases. Our work suggests that for some regions of mainland Spain, these fire danger indices could provide useful information about upcoming fire activity up to two months ahead of time and this information could be used to better inform wildland fire prevention and suppression activities.

Large-scale probabilistic identification of boreal peatlands using Google Earth Engine, open-access satellite data, and machine learning

Freely-available satellite data streams and the ability to process these data on cloud-computing platforms such as Google Earth Engine have made frequent, large-scale landcover mapping at high resolution a real possibility. In this paper we apply these technologies, along with machine learning, to the mapping of peatlands–a landcover class that is critical for preserving biodiversity, helping to address climate change impacts, and providing ecosystem services, e.g., carbon storage–in the Boreal Forest Natural Region of Alberta, Canada. We outline a data-driven, scientific framework that: compiles large amounts of Earth observation data sets (radar, optical, and LiDAR); examines the extracted variables for suitability in peatland modelling; optimizes model parameterization; and finally, predicts peatland occurrence across a large boreal area (397, 958 km²) of Alberta at 10 m spatial resolution (equalling 3.9 billion pixels across Alberta). The resulting peatland occurrence model shows an accuracy of 87% and a kappa statistic of 0.57 when compared to our validation data set. Differentiating peatlands from mineral wetlands achieved an accuracy of 69% and kappa statistic of 0.37. This data-driven approach is applicable at large geopolitical scales (e.g., provincial, national) for wetland and landcover inventories that support long-term, responsible resource management.

Estimating Spatial and Temporal Trends in Environmental Indices Based on Satellite Data: A Two-Step Approach

This paper presents a method for employing satellite data to evaluate spatial and temporal patterns in environmental indices of interest. In the first step, linear regression coefficients are extracted for each area in the image. These coefficients are then employed as a response variable in a boosted regression tree with geographic coordinates as explanatory variables. Here, a two-step approach is described in the context of a substantive case study comprising 30 years of satellite derived fractional green vegetation cover for a large region in Queensland, Australia. In addition to analysis of the entire image and timeframe, separate analyses are undertaken over decades and over sub-regions of the study region. The results demonstrate both the utility of the approach and insights into spatio-temporal trends in green vegetation for this site. These findings support the feasibility of using the proposed two-step approach and geographic coordinates in the analysis of satellite derived indices over space and time.

Attribution of the Influence of Human-Induced Climate Change on an Extreme Fire Season

A record 1.2 million ha burned in British Columbia, Canada's extreme wildfire season of 2017. Key factors in this unprecedented event were the extreme warm and dry conditions that prevailed at the time, which are also reflected in extreme fire weather and behavior metrics. Using an event attribution method and a large ensemble of regional climate model simulations, we show that the risk factors affecting the event, and the area burned itself, were made substantially greater by anthropogenic climate change. We show over 95% of the probability for the observed maximum temperature anomalies is due to anthropogenic factors, that the event's high fire weather/behavior metrics were made 2-4 times more likely, and that anthropogenic climate change increased the area burned by a factor of 7-11. This profound influence of climate change on forest fire extremes in British Columbia, which is likely reflected in other regions and expected to intensify in the future, will require increasing attention in forest management, public health, and infrastructure.

Assessing the Minimum Number of Time Since Last Fire Sample-Points Required to Estimate the Fire Cycle: Influences of Fire Rotation Length and Study Area Scale

Boreal forest fire history is typically reconstructed using tree-ring based time since last fire (TSLF) frequency distributions from across the landscape. We employed stochastic landscape fire simulations to assess how large a study area and how many TSLF sample-points are required to estimate the fire cycle (FC) within a given accuracy, and if those requirements change with length of the simulated fire rotation (FRS). FRS is calculated from simulated fire-year maps used to create the TSLF map, and is the “true” measure of fire history that FC estimates should equal. Fire-year maps were created by (i) using a spatially homogenous landscape, (ii) imposing large variations in annual area burned, and (iii) having no age-related change in the hazard of burning. We found that study areas should be ≥3× the size of largest total annual area burned, with smaller-scale areas having a bias that cannot be fixed by employing more samples. For a study area scale of 3×, a FC estimate with an error <10% was obtained with 187 TSLF samples at 0.81 samples per 100 km2. FC estimates were not biased in study area scales that were ≥3×, but smaller-scale areas with a short FRS had an overestimated FC and smaller-scale areas with a long FRS had an underestimated FC. Site specific variations in environmental- and age-related variations in the hazard of burning may require more sample-points; site specific simulations should thus be conducted to determine sample numbers before conducting a TSLF field study.

Direct and indirect climate controls predict heterogeneous early-mid 21st century wildfire burned area across western and boreal North America

Predicting wildfire under future conditions is complicated by complex interrelated drivers operating across large spatial scales. Annual area burned (AAB) is a useful index of global wildfire activity. Current and antecedent seasonal climatic conditions, and the timing of snowpack melt, have been suggested as important drivers of AAB. As climate warms, seasonal climate and snowpack co-vary in intricate ways, influencing fire at continental and sub-continental scales. We used independent records of seasonal climate and snow cover duration (last date of permanent snowpack, LDPS) and cell-based Structural Equation Models (SEM) to separate direct (climatic) and indirect (snow cover) effects on relative changes in AAB under future climatic scenarios across western and boreal North America. To isolate seasonal climate variables with the greatest effect on AAB, we ran multiple regression models of log-transformed AAB on seasonal climate variables and LDPS. We used the results of multiple regressions to project future AAB using GCM ensemble climate variables and LDPS, and validated model predictions with recent AAB trends. Direct influences of spring and winter temperatures on AAB are larger and more widespread than the indirect effect mediated by changes in LDPS in most areas. Despite significant warming trends and reductions in snow cover duration, projected responses of AAB to early-mid 21^st century are heterogeneous across the continent. Changes in AAB range from strongly increasing (one order of magnitude increases in AAB) to moderately decreasing (more than halving of baseline AAB). Annual wildfire area burned in coming decades is likely to be highly geographically heterogeneous, reflecting interacting regional and seasonal climate drivers of fire occurrence and spread.

Attributing extreme fire risk in Western Canada to human emissions

Canada is expected to see an increase in fire risk under future climate projections. Large fires, such as that near Fort McMurray, Alberta in 2016, can be devastating to the communities affected. Understanding the role of human emissions in the occurrence of such extreme fire events can lend insight into how these events might change in the future. An event attribution framework is used to quantify the influence of anthropogenic forcings on extreme fire risk in the current climate of a western Canada region. Fourteen metrics from the Canadian Forest Fire Danger Rating System are used to define the extreme fire seasons. For the majority of these metrics and during the current decade, the combined effect of anthropogenic and natural forcing is estimated to have made extreme fire risk events in the region 1.5 to 6 times as likely compared to a climate that would have been with natural forcings alone.

Predicting environmental mitigation requirements for hydropower projects through the integration of biophysical and socio-political geographies

Uncertainty about environmental mitigation needs at existing and proposed hydropower projects makes it difficult for stakeholders to minimize environmental impacts. Hydropower developers and operators desire tools to better anticipate mitigation requirements, while natural resource managers and regulators need tools to evaluate different mitigation scenarios and order effective mitigation. Here we sought to examine the feasibility of using a suite of multi-faceted explanatory variables within a spatially explicit modeling framework to fit predictive models for future environmental mitigation requirements at hydropower projects across the conterminous U.S. Using a database comprised of mitigation requirements from more than 300 hydropower project licenses, we were able to successfully fit models for nearly 50 types of environmental mitigation and to apply the predictive models to a set of more than 500 non-powered dams identified as having hydropower potential. The results demonstrate that mitigation requirements are functions of a range of factors, from biophysical to socio-political. Project developers can use these models to inform cost projections and design considerations, while regulators can use the models to more quickly identify likely environmental issues and potential solutions, hopefully resulting in more timely and more effective decisions on environmental mitigation.

Tree species richness decreases while species evenness increases with disturbance frequency in a natural boreal forest landscape

Understanding species diversity and disturbance relationships is important for biodiversity conservation in disturbance‐driven boreal forests. Species richness and evenness may respond differently with stand development following fire. Furthermore, few studies have simultaneously accounted for the influences of climate and local site conditions on species diversity. Using forest inventory data, we examined the relationships between species richness, Shannon's index, evenness, and time since last stand‐replacing fire (TSF) in a large landscape of disturbance‐driven boreal forest. TSF has negative effect on species richness and Shannon's index, and a positive effect on species evenness. Path analysis revealed that the environmental variables affect richness and Shannon's index only through their effects on TSF while affecting evenness directly as well as through their effects on TSF. Synthesis and applications. Our results demonstrate that species richness and Shannon's index decrease while species evenness increases with TSF in a boreal forest landscape. Furthermore, we show that disturbance frequency, local site conditions, and climate simultaneously influence tree species diversity through complex direct and indirect effects in the studied boreal forest.

Human and biophysical drivers of fires in Semiarid Chaco mountains of Central Argentina

Fires are a recurrent disturbance in Semiarid Chaco mountains of central Argentina. The interaction of multiple factors generates variable patterns of fire occurrence in space and time. Understanding the dominant fire drivers at different spatial scales is a fundamental goal to minimize the negative impacts of fires. Our aim was to identify the biophysical and human drivers of fires in the Semiarid Chaco mountains of Central Argentina and their individual effects on fire activity, in order to determine the thresholds and/or ranges of the drivers at which fire occurrence is favored or disfavored. We used fire frequency as the response variable and a set of 28 potential predictor variables, which included climatic, human, topographic, biological and hydrological factors. Data were analyzed using Boosted Regression Trees, using data from near 10,500 sampling points. Our model identified the fire drivers accurately (75.6% of deviance explained). Although humans are responsible for most ignitions, climatic variables, such as annual precipitation, annual potential evapotranspiration and temperature seasonality were the most important determiners of fire frequency, followed by human (population density and distance to waste disposals) and biological (NDVI) predictors. In general, fire activity was higher at intermediate levels of precipitation and primary productivity and in the proximity of urban solid waste disposals. Fires were also more prone to occur in areas with greater variability in temperature and productivity. Boosted Regression Trees proved to be a useful and accurate tool to determine fire controls and the ranges at which drivers favor fire activity. Our approach provides a valuable insight into the ecology of fires in our study area and in other landscapes with similar characteristics, and the results will be helpful to develop management policies and predict changes in fire activity in response to different climate changes and development scenarios.

The potential and realized spread of wildfires across Canada

Given that they can burn for weeks or months, wildfires in temperate and boreal forests may become immense (eg., 10(0) - 10(4) km(2) ). However, during the period within which a large fire is 'active', not all days experience weather that is conducive to fire spread; indeed most of the spread occurs on a small proportion (e.g., 1 - 15 days) of not necessarily consecutive days during the active period. This study examines and compares the Canada-wide patterns in fire-conducive weather ('potential' spread) and the spread that occurs on the ground ('realized' spread). Results show substantial variability in distributions of potential and realized spread days across Canada. Both potential and realized spread are higher in western than in eastern Canada; however, whereas potential spread generally decreases from south to north, there is no such pattern with realized spread. The realized-to-potential fire-spread ratio is considerably higher in northern Canada than in the south, indicating that proportionally more fire-conducive days translate into fire progression. An exploration of environmental correlates to spread show that there may be a few factors compensating for the lower potential spread in northern Canada: a greater proportion of coniferous (i.e., more flammable) vegetation, lesser human impacts (i.e., less fragmented landscapes), sufficient fire ignitions, and intense droughts. Because a linear relationship exists between the frequency distributions of potential spread days and realized spread days in a fire zone, it is possible to obtain one from the other using a simple conversion factor. Our methodology thus provides a means to estimate realized fire spread from weather-based data in regions where fire databases are poor, which may improve our ability to predict future fire activity.

Fire activity and severity in the western US vary along proxy gradients representing fuel amount and fuel moisture

Numerous theoretical and empirical studies have shown that wildfire activity (e.g., area burned) at regional to global scales may be limited at the extremes of environmental gradients such as productivity or moisture. Fire activity, however, represents only one component of the fire regime, and no studies to date have characterized fire severity along such gradients. Given the importance of fire severity in dictating ecological response to fire, this is a considerable knowledge gap. For the western US, we quantify relationships between climate and the fire regime by empirically describing both fire activity and severity along two climatic water balance gradients, actual evapotranspiration (AET) and water deficit (WD), that can be considered proxies for fuel amount and fuel moisture, respectively. We also concurrently summarize fire activity and severity among ecoregions, providing an empirically based description of the geographic distribution of fire regimes. Our results show that fire activity in the western US increases with fuel amount (represented by AET) but has a unimodal (i.e., humped) relationship with fuel moisture (represented by WD); fire severity increases with fuel amount and fuel moisture. The explicit links between fire regime components and physical environmental gradients suggest that multivariable statistical models can be generated to produce an empirically based fire regime map for the western US. Such models will potentially enable researchers to anticipate climate-mediated changes in fire recurrence and its impacts based on gridded spatial data representing future climate scenarios.

An analysis of controls on fire activity in boreal Canada: comparing models built with different temporal resolutions

Fire regimes of the Canadian boreal forest are driven by certain environmental factors that are highly variable from year to year (e.g., temperature, precipitation) and others that are relatively stable (e.g., land cover, topography). Studies examining the relative influence of these environmental drivers on fire activity suggest that models making explicit use of interannual variability appear to better capture years of climate extremes, whereas those using a temporal average of all available years highlight the importance of land-cover variables. It has been suggested that fire models built at different temporal resolutions may provide a complementary understanding of controls on fire regimes, but this claim has not been tested explicitly with parallel data and modeling approaches. We addressed this issue by building two models of area burned for the period 1980–2010 using 14 explanatory variables to describe ignitions, vegetation, climate, and topography. We built one model at an annual resolution, with climate and some land-cover variables being updated annually, and the other model using 31-year fire “climatology” based on averaged variables. Despite substantial differences in the variables' contributions to the two models, their predictions were broadly similar, which suggests coherence between the spatial patterns of annually varying climate extremes and long-term climate normals. Where the models' predictions diverged, discrepancies between the annual and averaged models could be attributed to specific explanatory variables. For instance, annually updating land cover allowed us to identify a possible negative feedback between flammable biomass and fire activity. These results show that building models at more than one temporal resolution affords a deeper understanding of controls on fire activity in boreal Canada than can be achieved by examining a single model. However, in terms of spatial predictions, the additional effort required to build annual models of fire activity may not always be warranted in this study area. From a management and policy standpoint, this key finding should boost confidence in models that incorporate climatic normals, thereby providing a stronger foundation on which to make decisions on adaptation and mitigation strategies for future fire activity.

Determining relative contributions of vegetation and topography to burn severity from LANDSAT imagery

Fire is a dominant process in boreal forest landscapes and creates a spatial patch mosaic with different burn severities and age classes. Quantifying effects of vegetation and topography on burn severity provides a scientific basis on which forest fire management plans are developed to reduce catastrophic fires. However, the relative contribution of vegetation and topography to burn severity is highly debated especially under extreme weather conditions. In this study, we hypothesized that relationships of vegetation and topography to burn severity vary with fire size. We examined this hypothesis in a boreal forest landscape of northeastern China by computing the burn severity of 24 fire patches as the difference between the pre- and post-fire Normalized Difference Vegetation Index obtained from two Landsat TM images. The vegetation and topography to burn severity relationships were evaluated at three fire-size levels of small (<100 ha, n = 12), moderate (100-1,000 ha, n = 9), and large (>1,000 ha, n = 3). Our results showed that vegetation and topography to burn severity relationships were fire-size-dependent. The burn severity of small fires was primary controlled by vegetation conditions (e.g., understory cover), and the burn severity of large fires was strongly influenced by topographic conditions (e.g., elevation). For moderate fires, the relationships were complex and indistinguishable. Our results also indicated that the pattern trends of relative importance for both vegetation and topography factors were not dependent on fire size. Our study can help managers to design fire management plans according to vegetation characteristics that are found important in controlling burn severity and prioritize management locations based on the relative importance of vegetation and topography.

Fire regime zonation under current and future climate over eastern Canada

Fire is a major disturbance in Canadian forests. Along with fuel and ignition characteristics, climatic conditions are seen as one of the main drivers of fire regimes. Projected changes in climate are expected to significantly influence fire regimes in Canada. As fire regime greatly shapes large-scale patterns in biodiversity, carbon, and vegetation, as well as forest and fire management strategies, it becomes necessary to define regions where current and future fire regimes are homogeneous. Random Forests (RF) modeling was used to relate fire regime attributes prevailing between 1961 and 1990 in eastern Canada with climatic/fire-weather and environmental variables. Using climatic normals outputs from the Canadian Regional Climate Model (CRCM), we delineated current (1961-1990) and future (2011-2040, 2040-2070, 2071 2100) homogeneous fire regime (HFR) zones. Heterogeneous response of fire regime to climate changes is projected for eastern Canada with some areas (e.g., western Quebec) experiencing very small alterations while others (e.g., southeastern Ontario) are facing great shifts. Overall, models predicted a 2.2- and 2.4-fold increase in the number of fires and the annual area burned respectively mostly as a result of an increase in extreme fire-weather normals and mean drought code. As extreme fire danger would occur later in the fire season on average, the fire season would shift slightly later (5-20 days) in the summer for much of the study area while remaining relatively stable elsewhere. Although fire regime values would change significantly over time, most zone boundaries would remain relatively stable. The information resulting from HFR zonations is clearly of interest for forest and fire management agencies as it reveals zones with peculiar fire regimes that would have been hidden otherwise using predefined administrative or ecological stratifications.

Identifying the threshold of dominant controls on fire spread in a boreal forest landscape of Northeast China

The relative importance of fuel, topography, and weather on fire spread varies at different spatial scales, but how the relative importance of these controls respond to changing spatial scales is poorly understood. We designed a "moving window" resampling technique that allowed us to quantify the relative importance of controls on fire spread at continuous spatial scales using boosted regression trees methods. This quantification allowed us to identify the threshold value for fire size at which the dominant control switches from fuel at small sizes to weather at large sizes. Topography had a fluctuating effect on fire spread across the spatial scales, explaining 20-30% of relative importance. With increasing fire size, the dominant control switched from bottom-up controls (fuel and topography) to top-down controls (weather). Our analysis suggested that there is a threshold for fire size, above which fires are driven primarily by weather and more likely lead to larger fire size. We suggest that this threshold, which may be ecosystem-specific, can be identified using our "moving window" resampling technique. Although the threshold derived from this analytical method may rely heavily on the sampling technique, our study introduced an easily implemented approach to identify scale thresholds in wildfire regimes.