Introduction and spread of nonindigenous species present a formidable threat to forest health. The emerald ash borer (EAB), Agrilus planipennis, is an East Asian-origin insect that has devastated ash (Fraxinus spp.) trees across the United States and parts of Canada since 2002. Proactive surveillance using high-performing predictive models could aid in mitigating pest risk. Predictor variables and modeling methods are important considerations in such analysis. Therefore, we assessed whether relevant single predictors, a combination of predictors grouped under a certain driver category, or multiple key predictors comprising several drivers, alter the goodness-of-fit of logistic regression models to EAB occurrence data (2002 to 2018) from Canada. The predictors used in models included spatial, topographic/positional, transport pathways/human hotspots, host-related factors, and climate-related variables. Using predictors from the best candidate logistic regression model, we tested the performance of 7 different model types including an ensemble model. Our findings showed that predictors from a wide range of drivers better characterized EAB occurrences than single predictors or a combination of predictors from any given driver category. In multi-model comparisons, random forest outperformed all other models, including the ensemble model. Elevation, infestation pressure, accumulated degree-days (>10 °C), and human population density were important predictors of EAB presence. Random forest and ensemble model forecasts for the city of Edmonton, Alberta, Canada, indicated an area of potential concern for EAB. Our research strongly underscores the utility of comparative multi-model approaches in invasive risk assessments that could have important implications for pest surveillance and management.

Forest fires in Canada's boreal forest can cause a great deal of concern for populations, the environment, and infrastructures. One of the tools developed to predict potential fire activity related to these events is the Canadian Fire Weather Index (FWICAN) system. This study aims to analyze the potential of soil moisture and ocean salinity (SMOS) satellite products [soil moisture (SM), vegetation optical depth (VOD), and root zone SM (RZSM)] to provide additional information on prefire soil and vegetation conditions for forest fire danger rating system. Using the random forest (RF) algorithm, we show that adding SMOS data to the FWICAN system indices slightly increases the accuracy of the predictions of potential fire activity. The RZSM from SMOS data was the variable that best improved the model performance, whereas the VOD provided no additional information. In the aim to evaluate the method for regions with limited in situ meteorological data, we used FWI system indices calculated from ERA5 available over the globe (FWIERA5). Taking into account FWIERA5, SMOS data allow to improve substantially the ability to predict forest fire.

Accurate and fine-scale forest data are essential to improve natural resource management, particularly in the face of climate change. Here, we present SCANFI, the Spatialized CAnadian National Forest Inventory, which provides coherent, 30 m resolution 2020 wall-to-wall maps of forest attributes (land cover type, canopy height, crown closure, aboveground tree biomass, and main species composition). These maps were developed using the NFI photo-plot dataset, a systematic regular sample grid of photo-interpreted high-resolution imagery covering all of Canada's non-arctic landmass. SCANFI was produced using temporally harmonized summer and winter Landsat spectral imagery along with hundreds of tile-level regional models based on a multiresponse k-nearest neighbours and random forest imputation method. This tile-level approach revealed the importance of radiometric variables in predicting vegetation attributes, namely winter radiometry, as the large-scale climate gradients were controlled at the tile-level. SCANFI was validated with rigorous cross-validation analyses, which revealed robust model performance for structural attributes (biomass R 2 = 0.76; crown closure R 2 = 0.82; height R 2 = 0.78) and tree species cover (e.g., Douglas fir R 2 = 0.60). SCANFI attributes were also validated with several independent external products, ranging from ground plot-based tree species cover (e.g., black spruce R 2 = 0.53) to satellite LiDAR height data products (e.g., crown closure R 2 = 0.71). SCANFI total aboveground biomass trends also followed those published by other studies. The methodology presented herein can be used to map time series of these attributes, identify the original training points used to make any given prediction, as well as map additional variables associated with the NFI photo-plots that are challenging to map using traditional remote sensing approaches.

AbstractPeatlands in the Canadian boreal forest are being negatively impacted by anthropogenic climate change, the effects of which are expected to worsen. Peatland types and sub‐classes vary in their ecohydrological characteristics and are expected to have different responses to climate change. Large‐scale modelling frameworks such as the Canadian Model for Peatlands, the Canadian Fire Behaviour Prediction System and the Canadian Land Data Assimilation System require peatland maps including information on sub‐types and vegetation as critical inputs. Additionally, peatland class and vegetation height are critical variables for wildlife habitat management and are related to the carbon cycle and wildfire fuel loading. This research aimed to create a map of peatland sub‐classes (bog, poor fen, rich fen permafrost peat complex) for the Canadian boreal forest and create an inventory of peatland vegetation height characteristics using ICESat‐2. A three‐stage hierarchical classification framework was developed to map peatland sub‐classes within the Canadian boreal forest circa 2020. Training and validation data consisted of peatland locations derived from various sources (field data, aerial photo interpretation, measurements documented in literature). A combination of multispectral data, L‐band SAR backscatter and C‐Band interferometric SAR coherence, forest structure and ancillary variables was used as model predictors. Ancillary data were used to mask agricultural areas and urban regions and account for regions that may exhibit permafrost. In the first stage of the classification, wetlands, uplands and water were classified with 86.5% accuracy. In the second stage, within the wetland areas only, peatland and mineral wetlands were differentiated with 93.3% accuracy. In the third stage, constrained to only the peatland areas, bogs, rich fens, poor fens and permafrost peat complexes were classified with 71.5% accuracy. Then, ICESat‐2 ATL08 spaceborne lidar data were used to describe regional variations in peatland vegetation height characteristics and regional and class‐wise variations based on a boreal forest wide sample. This research introduced a comprehensive large‐scale peatland sub‐class mapping framework for the Canadian boreal forest, presenting the first moderate resolution map of its kind.

Three radar polarimetric indices, Radar Vegetation Index (RVI), Canopy Structure Index (CSI) and Radar Forest Degradation Index (RFDI)-normally associated with quad-pol (QP) synthetic aperture radar (SAR) sensors-were derived using compact polarimetry (CP) data from the Radarsat Constellation Mission (RCM). Indices were generated over a 10,000-hectare temperate mixedwood forest containing a range of complex forest structures. For comparative purposes, the same indices were generated using Radarsat-2 QP data. Agreement between CP and QP indices were assessed across broad vegetated land cover types, at the forest stand-level wherein ground plots and airborne LiDAR data were available, and at the pixel level within validation stands representing different forest types. Agreement was consistently strong for the RVI and weak for the CSI, with agreement stronger when generalized to the stand level. Indices were more informative on differences between vegetation types (e.g. forest and open wetland) than between forest types with different structures. With a radar nominal off-nadir incidence angle at 38 degrees, RCM CP indices had narrower dynamic ranges compared to Radarsat-2 QP indices, especially the CSI, contributing to a lower level agreement for the CSI. CP data enables derivation of RVI, CSI and RFDI indices simultaneously, and provides large spatial coverage and capacity for dense time-series collection, features which are unavailable from current QP sensors.

Satellite forest inventories are the only feasible way to map Canada’s vast, remote forest regions, such as those in the Northwest Territories (NWT). A method used to create such inventories is the k-nearest neighbour (k-NN) algorithm, which spatially extends information from forest inventory (FI) plots to the entire forest land base using wall-to-wall features typically derived from Landsat data. However, the benefits of integrating L-band synthetic aperture radar (SAR) data, strongly correlated to forest biomass, have not been assessed for Canadian northern boreal forests. Here we describe an optimized multivariate k-NN implementation of a 151,700 km2 area in southern NWT that included ca. 2007 Landsat and dual-polarized Phased Array type L-band SAR (PALSAR) data on board the Advanced Land Observing Satellite (ALOS). Five forest attributes were mapped at 30 m cells: stand height, crown closure, stand/total volume and aboveground biomass (AGB). We assessed accuracy gains compared to Landsat-based maps. To circumvent the scarcity of FI plots, we used 3600 footprints from the Geoscience Laser Altimeter System (GLAS) as surrogate FI plots, where forest attributes were estimated using Light Detection and Ranging (LiDAR) metrics as predictors. After optimization, k-NN predicted forest attribute values for each pixel as the average of the 4 nearest (k = 4) surrogate FI plots within the Euclidian space of 9 best features (selected among 6 PALSAR, 10 Landsat, and 6 environmental features). Accuracy comparisons were based on 31 National Forest Inventory ground plots and over 1 million airborne LiDAR plots. Maps that included PALSAR HV backscatter resulted in forest attribute predictions with higher goodness of fit (adj. R2), lower percent mean error (ME%), and percent root mean square error (RMSE%), and lower underestimation for larger attribute values. Predictions were most accurate for conifer stand height (RMSE% = 32.1%, adj. R2 = 0.58) and AGB (RMSE% = 47.8%, adj. R2 = 0.74), which is much more abundant in the area than mixedwood or broadleaf. Our study demonstrates that optimizing k-NN parameters and feature space, including PALSAR, Landsat, and environmental variables, is a viable approach for inventory mapping of the northern boreal forest regions of Canada.

Sustainable forest management requires information on the spatial distribution, composition, and structure of forests. However, jurisdictions with large tracts of noncommercial forest, such as the Northwest Territories (NWT) of Canada, often lack detailed forest information across their land base. The goal of the Multisource Vegetation Inventory (MVI) project was to create a large area forest inventory (FI) map that could support strategic forest management in the NWT using optical, radar, and light detection and ranging (LiDAR) satellite remote sensing anchored on limited field plots and airborne LiDAR data. A new landcover map based on Landsat imagery was the first step to stratify forestland into broad forest types. A modelling chain linking FI plots to airborne and spaceborne LiDAR was then developed to circumvent the scarcity of field data in the region. The developed models allowed the estimation of forest attributes in thousands of surrogate FI plots corresponding to spaceborne LiDAR footprints distributed across the project area. The surrogate plots were used as a reference dataset for estimating each forest attribute in each 30 m forest cell within the project area. The estimation was based on the k-nearest neighbour (k-NN) algorithm, where the selection of the four most similar surrogate FI plots to each cell was based on satellite, topographic, and climatic data. Wall-to-wall 30 m raster maps of broad forest type, stand height, crown closure, stand volume, total volume, aboveground biomass, and stand age were created for a ~400,000 km2 area, validated with independent data, and generalized into a polygon GIS layer resembling a traditional FI map. The MVI project showed that a reasonably accurate FI map for large, remote, predominantly non-inventoried boreal regions can be obtained at a low cost by combining limited field data with remote sensing data from multiple sources.

Burn severity is an important component of the fire regime that has not yet been fully characterized for the forests of Canada. The objectives of this study were to (i) create a Canada-wide geospatial database of burn severity for wildland fires across forested regions of Canada from 1985 to 2015, and (ii) use this database to evaluate seasonal and annual trends in burn severity across Canada and regionally using two different regional units (ecozones and Homogeneous Fire Regime zones). We developed the 30 m resolution geospatial Canadian Landsat Burn Severity (CanLaBS) product from Landsat imagery, using values of pre-fire to post-fire differences in normalized burn ratios (dNBRs) for nearly 60 Mha of area burned across Canada’s forests from 1985 to 2015, complemented with estimates of pre-fire forest composition, crown closure, and biomass. Our results suggest that burn severity is generally lower in spring fires than in summer ones nationally and in almost every regional unit. We found no consistent relationship between burn severity and annual area burned across ecozones. Finally, we observed a small but significant decrease in burn severity from 1985 to 2015 across Canada, although this is regionally variable. The CanLaBS database is publicly available at https://doi.org/10.23687/b1f61b7e-4ba6-4244-bc79-c1174f2f92cd .

In this paper, we demonstrate the feasibility of using TanDEM-X (TX) interferometric coherence data for mapping forest height with 25-m pixels across a study area near Fort Simpson, Northwest Territories (NWT), Canada. Our simplified RVOG model locally estimates forest height by combining an optimized estimation of TX interferometric coherence amplitude with the 20-m resolution Canadian Digital Elevation Model (CDEM) accounting for local slope variations. The initial map of TX height estimates provided R2 values of 0.78 and 0.88, mean errors (ME) of 1.66 m and 1.90 m, and root-mean-square errors (RMSE) of 2.7 m and 2.9 m when compared to independent height estimates derived from field plots and airborne LiDAR, respectively. We corrected the bias of TX height estimates using two variants of a LiDAR-based linear model. An application of three cover-specific linear adjustments provided the final TX height map with absolute ME ≤0.05 m and RMSE ≤2.09 m. The approach was tailored to poorly inventoried northern boreal regions through the use of archived TX data, the CDEM, a land cover map and airborne LiDAR transects. Our encouraging results support the perspective of wall-to-wall mapping of forest height across northern boreal forests in the NWT and beyond.

Canada's forests are shaped by disturbances such as fire, insect outbreaks, and droughts that often overlap in time and space. The resulting cumulative disturbance risks and potential impacts on forests are generally not well accounted for by models used to predict future impacts of disturbances on forest. This study aims at projecting future cumulative effects of four main natural disturbances, fire, mountain pine beetle, spruce budworm and drought, on timber volumes across Canada's forests using an approach that accounts for potential overlap among disturbances. Available predictive models for the four natural disturbances were used to project timber volumes at risk under aggressive climate forcing up to 2100. Projections applied to the current vegetation suggest increases of volumes at risk related to fire, mountain pine beetle, and drought over time in many regions of Canada, but a decrease of the volume at risk related to spruce budworm. When disturbance effects are accumulated, important changes in volumes at risk are projected to occur as early as 2011-2041, particularly in central and eastern Canada. In our last simulation period covering 2071-2100, nearly all timber volumes in most of Canada's forest regions could be at risk of being affected by at least one of the four natural disturbances considered in our analysis, a six-fold increase relative to the baseline period (1981-2010). Tree species particularly vulnerable to specific disturbances (e.g., trembling aspen to drought) could suffer disproportionate increases in their volume at risk with potential impacts on forest composition. By 2100, estimated wood volumes not considered to be at risk could be lower than current annual timber harvests in central and eastern Canada. Current level of harvesting could thus be difficult to maintain without the implementation of adaptation measures to cope with these disturbances.

A methods framework is presented that utilizes field plots, airborne light detection and ranging (LiDAR), and spaceborne Geoscience Laser Altimeter System (GLAS) data to estimate forest attributes over a 20 Mha area in Northern Canada. The framework was implemented to scale up forest attribute models from field data to intersecting airborne LiDAR data, and then to GLAS footprints. GLAS data were sequentially filtered and submitted to the k-nearest neighbour (k-NN) imputation algorithm to yield regional estimates of stand height and crown closure at a 30 m resolution. Resulting outputs were assessed against independent airborne LiDAR data to evaluate regional estimates of stand height (mean difference = −1 m, RMSE = 5 m) and crown closure (mean difference = −5%, RMSE = 9%). Additional assessments were performed as a function of dominant vegetation type and ecoregion to further evaluate regional products. These attributes form the primary descriptive structure attributes that are typical of forest inventory mapping programs, and provide insight into how they can be derived in northern boreal regions where field information and physical access is often limited.

Mapping Canada's forests is a significant challenge given their extent and the interprovincial differences in forest inventories. We created new sets of nationally consistent forest attribute maps for the years 2001 and 2011 by building upon previously published work with the objective to determine if sequential maps of forest attributes could be used to quantify changes over time. We first refined our previously published methodology of using the k nearest neighbors (kNN) prediction method and MODIS spectral reflectance data as predictive variables. The maps were generated using an improved reference dataset and a new analytical kNN workflow. We then evaluated 2001 to 2011 changes in two key attributes, aboveground biomass and percent tree cover, on pixels identified from published sources as having undergone fire, harvest, or postdisturbance regrowth during that period. For all three change types, average changes in both aboveground biomass and percent tree cover between 2001 and 2011 matched expectations relative to the dynamics of Canadian forests. Our results support the use of sequential national maps of forest attributes for evaluating regionally aggregated disturbance-related changes in forest properties. The new forest attribute maps are available from Beaudoin et al. (2017; doi:10.23687/ec9e2659-1c29-4ddb-87a2-6aced147a990) at http://ouvert.canada.ca/data/fr/dataset/ec9e2659-1c29-4ddb-87a2-6aced147a990.

Soil data and soil mapping are indispensable tools in sustainable forest management. In northern boreal ecosystems, paludification is defined as the accumulation of partially decomposed organic matter over saturated mineral soils, a process that reduces tree regeneration and forest growth. Given this negative effect on forest productivity, spatial prediction of paludification in black spruce stands is important in forest management. This paper provides a description of the soil database to predict organic layer thickness (OLT) as a proxy of paludification in northeastern Canada. The database contains 13,944 OLT measurements (in cm) and their respective GPS coordinates. We collected OLT measurements from georeferenced ground plots and transects from several previous projects. Despite the variety of sources, the sampling design for each dataset was similar, consisting of manual measurements of OLT with a hand probe. OLT measurements were variable across the study area, with a mean ± standard deviation of 21 ± 24 cm (ranging from a minimum of 0 cm to a maximum of 150 cm), and the distribution tended toward positive skewing, with a large number of low OLT values and fewer high OLT values. The dataset has been used to perform OLT mapping at 30-m resolution and predict the risk of paludification in northeastern Canada (Mansuy et al., 2018) [1]. The spatially explicit and continuous database is also available to support national and international efforts in digital soil mapping.

Abstract A recent global study reported a net difference between areas of forest cover loss and of forest cover gain of about 3.6% of total forest area across the boreal biome, and of 5.6% for Canada, over a 12‐yr period. Net losses of this magnitude should be of concern given the importance of this biome in global biogeochemical cycles linked to climate change. Our analysis for Canada fails to support these results and suggests that post‐harvest recovery of tree cover is generally strong, while post‐fire recovery of tree cover is weaker but nevertheless prevalent. We find that current large area remote sensing methodologies can fail to properly recognize post‐disturbance recovery from non‐forest to forest status in low‐productivity boreal forests when using short time series. With climate change and human impacts intensifying around the world, it is urgently important to be able to reliably distinguish temporary forest cover loss followed by naturally slow recovery from forest decline requiring policy action. The analysis was in large part based on the new Canada Landsat Disturbance product in which fires and harvest since 1984 are mapped at 30‐m resolution (https://doi.org/10.23687/add1346b-f632-4eb9-a83d-a662b38655ad).

Forest fire activity is projected to increase with climate change in Canada, but vegetation feedbacks are usually not considered. Using new information on the selectivity or avoidance of fire as a function of stand age and composition, we ran simple simulation models that consider the changes in the regional age matrices induced by fire and harvesting to project future burn rates. We also projected estimated future regional vulnerability of timber supply to fire by considering these new burn rates. The inclusion of age-related feedbacks would have a large impact on projected increases in burn rates, mostly in a very fire active zone under aggressive climate forcing. Projected burn rates would still increase, but would be 50% less in 2100 than if projected without this biotic feedback in some zones. Negative feedbacks would be virtually nonexistent when potential burning rates are below 1%, whereas realized burning rates would be lowered by more than a 0.5 percentage point when potential burning rates exceed 2.5%. Including fire–vegetation feedbacks had virtually no impact on total volume harvested. As fire burns more old-growth coniferous stands, slightly negative impacts were projected on conifer harvested almost everywhere. These results underline the need to incorporate fire–vegetation feedbacks when projecting future burn rates.

Forest landscapes at the southern boreal forest transition zone are likely to undergo great alterations due to projected changes in regional climate.

Mapping peatland extent in Canada would contribute important information concerning carbon balance and hydrology. While such mapping, based on air photo interpretation and remote sensing data, has recently improved, maps have been limited to 1:1 million scale. We hypothesized that forest structure information from forest inventory plots could be used to predict the presence of forested and treed peat lands in boreal Canada at the ground plot-level, and that a resulting model could be used to predict the distribution of forested and treed peatlands across Canada. Inventory ground plots from the Canadian National Forest Inventory (NFI) with organic soil depth measurements were used to create a model of the presence of treed to forested (canopy cover ranging from sparse to closed) peatlands (greater than 40 cm organic soil depth) in boreal Canada. The presence of black spruce (Picea mariana) or larch (Larix laricina), in combination with low stand height and stand age greater than 75 years, were the strongest predictors of the presence of peatlands. Bioclimatic variables related to high diurnal and annual temperature variation, consistent with a continental climate, also contributed to the increased predicted presence of treed peatlands. Both logistic and boosted regression tree models showed similar results, with similar to 87% accuracy in the discrimination of treed peatlands when validated against an independent set of ground plots. The boosted regression tree model was propagated across Canada using forest attribute raster data layers at 250 m resolution from the NFI along with bioclimatic layers. Estimates of treed peatland extent agreed with data points from peat cores with 85-95% accuracy in the Boreal Shield ecozone, although prediction was less accurate in the more southern boreal and Great Lakes forest areas. The resulting map can be used as an input to forest carbon modelling, and the improved knowledge of treed peatland extent will be useful in modelling wildfire or peatland drainage. Crown Copyright (C) 2016 Published by Elsevier B.V. All rights reserved.

There is considerable variation in the degree of burn severity in boreal fires. One approach that has been used to capture this variation from field and remote sensing perspectives for western Canadian boreal forests is the Composite Burn Index (CBI) and differenced Normalized Burn Ratio (dNBR). Of interest was how well these methods may perform for fires in eastern Canada. This study investigated the CBI-dNBR relationship for selected fires in the eastern boreal forests of Canada, with a view towards contributing to the generalisation of a Canada-wide model. Results for the sampled region showed no difference in the CBI-dNBR relationship between black spruce- and jack pine-dominated stands, whereas this relationship was best described by a Generalised Additive Model (GAM). The dNBR-derived maps would also be useful in support of research and post-fire management in burns outside the studied territory and time frame covered by the existing burn severity mapping system already used in this region. The Saturated growth model proposed for the western boreal region also performed well for our eastern boreal region, thus further supporting the development of a national model.

Fire is a dominant mechanism of forest renewal in most of Canada’s forests and its activity is predicted to increase over the coming decades. Individual fire events have been considered to be non-selective with regards to forest properties, but evidence now suggests otherwise. Our objective was therefore to quantify the effect of forest properties on fire selectivity or avoidance, evaluate the stability of these effects across varying burn rates, and use these results to map local fire risk across the forests of Canada. We used Canada-wide MODIS-based maps of annual fires and of forest properties to identify burned and unburned pixels for the 2002–2011 period and to bin them into classes of forest composition (% conifer and broadleaved deciduous), above-ground tree biomass and stand age. Logistic binomial regressions were then used to quantify fire selectivity by forest properties classes and by zones of homogeneous fire regime (HFR). Results suggest that fire exhibits a strong selectivity for conifer stands, but an even stronger avoidance of broadleaved stands. In terms of age classes, fire also shows a strong avoidance for young (0 to 29 year) stands. The large differences among regional burn rates do not significantly alter the overall preference and avoidance ratings. Finally, we combined these results on relative burn preference with regional burn rates to map local fire risks across Canada.

Fire is a dominant mechanism of forest renewal in most of Canada’s forests and its activity is predicted to increase over the coming decades. Individual fire events have been considered to be non-selective with regards to forest properties, but evidence now suggests otherwise. Our objective was therefore to quantify the effect of forest properties on fire selectivity or avoidance, evaluate the stability of these effects across varying burn rates, and use these results to map local fire risk across the forests of Canada. We used Canada-wide MODIS-based maps of annual fires and of forest properties to identify burned and unburned pixels for the 2002–2011 period and to bin them into classes of forest composition (% conifer and broadleaved deciduous), above-ground tree biomass and stand age. Logistic binomial regressions were then used to quantify fire selectivity by forest properties classes and by zones of homogeneous fire regime (HFR). Results suggest that fire exhibits a strong selectivity for conifer stands, but an even stronger avoidance of broadleaved stands. In terms of age classes, fire also shows a strong avoidance for young (0 to 29 year) stands. The large differences among regional burn rates do not significantly alter the overall preference and avoidance ratings. Finally, we combined these results on relative burn preference with regional burn rates to map local fire risks across Canada.