Publications - Growth and Yield, Inventory and Monitoring

Uncertainty in photo-interpreted forest inventory variables and effects on estimates of error in Canada’s National Forest Inventory

Wed, 24 Apr 2013

Canada’s National Forest Inventory (NFI) relies on photo-interpreted forest resource data provided by provincial and territorial agencies. NFI data are collected at regular intervals in time from a nominal 20 × 20 km network of 2 × 2 km photoplots. Attribute-specific NFI estimates of precision include contributions from sampling errors and uncertainty in the source data. We assessed this uncertainty in NFI photo-interpreted forest attribute data from New Brunswick and Nova Scotia. Attributes examined were: cover type, age, maturity (class), crown closure, height, volume, and area associated with an attribute. Monte-Carlo simulations, with measurement errors superimposed on NFI data assumed to be error-free, showed that estimates of precision were inflated by an average of 7% (range 0%–36%) due to the uncertainty in the source data. Species misclassification and age determination were the largest sources of uncertainty.

Using height growth to model local and regional response of trembling aspen (Populus tremuloides Michx.) to climate within the boreal forest of western Québec

Tue, 24 Jul 2012

Studies relating site index to climatic variables basically assume that the sensitivity of a species to climate remains stable across the geographic range of their study area. Yet, provenance trials speak to the contrary and show that populations are adapted to their local climatic conditions and tend to respond differently to climate. Spatial and temporal complexity of forest productivity and climate-relationships has been globally reported and recent studies have emphasized the necessity for regional studies on forest growth dynamics of current and future populations. The objective of this study was to determine whether the main climatic and non-climatic drivers of trembling aspen (Populus tremuloides Michx.) growth in Québec should be treated as regional (the study area reacts as a unique population) or local factors (the area is composed of different populations) when modeling the spatio-temporal variability of aspen productivity as measured with site index. Stem analysis data was collected from 124 trees (32 stands) that span a north-south (latitude 46–51◦N) transect in western boreal Québec. Most stands were dense with cover density above 60%, even-aged, 50–90 years old, and very often mixed. The northernmost regions (latitude 48–51◦N) are characterized by either organic or clay deposits, while in the south (latitude 46–48◦N) till or clay deposits predominate. Climate variables that met selection criteria as major regional or local factors that influence aspen productivity were selected. A mixed modeling approach was subsequently employed to identify the categorization unit that could be defined as a population. We then predicted variation in the random error with prior information obtained at stand level. Our results show that aspen height growth is mainly driven by annual sums of degree days and stand age. Surface deposit type, which is an indicator of soil nutritive status and moisture potential, was found to have modulated climate influence. Finally, aspen productivity is better explained with a model that assumes that specific populations have a different response function to climate and are adapted to their local climatic conditions. This has implications when predicting the response to climatic change for forest growth models that assume that conspecifics respond to climate similarly.

Vegetation phenology can be captured with digital repeat photography and linked to variability of root nutrition in Hedysarum alpinum

Wed, 10 Oct 2012

Question

Can repeat (time-lapse) photography be used to detect the phenological development of a forest stand, and linked to temporal patterns in root nutrition for Hedysarum alpinum (alpine sweetvetch) an important grizzly bear food species? Location

Eastern foothills and front ranges of the Rocky Mountains in Alberta, Canada. The area contains a diverse mix of mature and young forest, wetlands and alpine habitats. Methods

We deployed six automated cameras at three locations to acquire daily photographs at the plant and forest stand scales. Plot locations were also visited on a bi-weekly basis to record the phenological stage of H. alpinum and other target plant species, as well as to collect a root sample for determination of crude protein content. Results

Repeat photography and image analysis successfully detected all key phenological events (i.e. green-up, flowering, senescence). Given the relation between phenology and root nutrition, we illustrate how camera data can be used to predict the spatial and temporal distribution and quality of a key wildlife resource. Conclusions

Repeat photography provides a cost-effective method for monitoring vegetation development, food availability, and nutritional quality at a forest stand scale. Since wildlife responds to the availability and quality of their food resources, detailed information on changes in resource availability helps with land-use management decisions and furthers our understanding of grizzly bear feeding ecology and habitat selection.

Hardwood Initiative - Part 5: Development of new processes and technologies in the hardwood industry: Project 16 - Towards a Forest Inventory Capable of Predicting the Properties and Economic Value of Hardwoods in Eastern Canada Development of a Product Matrix by Tree Grade and DBH class for Sugar Maple and Yellow Birch.

Fri, 05 Apr 2013

A history of habitat dynamics: Characterizing 35 years of stand replacing disturbance

Thu, 08 Dec 2011

Landscape change, specifically habitat loss and modification, is thought to have an impact on the health, productivity, distribution, and survival of grizzly bears (Ursus arctos L.). Although grizzly bears may preferentially seek out areas of anthropogenic disturbances for foraging opportunities, research has found that grizzly bears experience greater mortality in these areas as a result of increased human access. Additional insights on the location and rates of anthropogenic-driven landscape change are required to better understand related impacts upon grizzly bears. In this study, a time series of 14 Landsat MSS, TM, and ETM+ images were used to retrospectively document and quantify the rate of landscape change over a 35-year period from 1973 to 2008 in a 13507 km² analysis area in western Alberta, Canada. The study area is located within a larger region that contains the highest density of grizzly bears in Alberta and has experienced increasingly intensive forest harvesting and oil and gas exploration activities during this period. To accommodate the differing spectral channels from MSS to TM/ETM+ sensors, the arctangent of the angle of the Tasseled Cap greenness-to-brightness components was computed for each image year, with sequential image pairs differenced and a threshold applied to identify stand-replacing disturbance events.

Results indicated that 11% of the analysis area experienced some form of stand-replacing disturbance (e.g., cutblocks, roads, oil and gas well sites, seismic lines, power lines, pipelines, blowdown) between 1973 and 2008. The greatest proportion of this change (by area) occurred between 2004 and 2006 (24%), while the lowest proportion occurred between 2000 and 2001 (2%). Although the number of change events has fluctuated over time, with a minimum of 2888 change events between 1976 and 1978 (2%) and a maximum of 36623 change events between 2004 and 2006 (29%), the mean size of change events has decreased over time: prior to 1995, mean event size was greater than 1.5ha; after 1995, it was less than 1.5ha. The annual rate of change was greatest between 2004 and 2006 (−1.25%), and lowest between 1981 and 1990 (−0.04%). Consideration of changes within the context of units relevant to grizzly bear management (i.e., grizzly bear watershed units and core or secondary habitat areas) indicate that the amount and rate of change was not spatially or temporally uniform across the study area. While the average change event size has decreased over time, the increasing number of change events has resulted in a larger aggregate area of change in more recent years. Landsat imagery provided a large-area, synoptic, and consistent characterization of 35years of stand-replacing disturbance in our study area, providing information that enables an improved understanding of the complex interactions between grizzly bear distribution, abundance, health, survival, and habitat.

A Horvitz–Thompson-type estimator of species richness

Mon, 05 Dec 2011

A Horvitz–Thompson-type estimator of species richness for plot (cluster) sampling is constructed by considering species sampling as sampling with an unequal probability. Inclusion probabilities are estimated from sample-based estimates of relative species incidence. Bias is addressed by adding, to each observed species, the expected number of unseen species with the same relative incidence. A Hansen–Hurwitz estimator of variance is adopted and augmented by the anticipated variance from sample-based inclusion probabilities and the number of observed species. In Monte Carlo simulation of simple random plot (cluster) sampling from 11 large finite populations of forest trees and three sample sizes, the proposed estimator achieved the best overall ranking in terms of relative root mean square error efficiency when compared to 12 alternative estimators. The proposed estimator ranked third in terms of bias. The augmented Hansen–Hurwitz estimator of variance was liberal (median −13%). No richness estimator was uniformly best across populations and sample sizes. Across all settings, the performance of the best four estimators was similar, both in terms of bias and relative root mean square error efficiency.

A modified bootstrap procedure for cluster sampling variance estimation of species richness

Tue, 05 Apr 2011

Variance estimators for probability sample-based predictions of species richness (S) are typically conditional on the sample (expected variance). In practical applications sample sizes are typically small and the variance of the input parameters to a richness estimator should not be ignored. We propose a modified bootstrap variance estimator which attempts to capture the sampling variance by generating B replications of the richness prediction from stochastically resampled data of species incidence. The variance estimator is demonstrated for the observed richness (SO), five richness estimators, and with simulated cluster sampling (without replacement) in 11 finite populations of forest tree species. A key feature of the bootstrap procedure is a probabilistic augmentation of a species incidence matrix by the number of species expected to be ‘lost’ in a conventional bootstrap resampling scheme. In Monte-Carlo simulations the modified bootstrap procedure performed well in terms of tracking the average Monte-Carlo estimates of richness and standard errors. Bootstrap-based estimates of standard error were as a rule conservative. Extensions to other sampling designs, estimators of species richness and diversity, and estimates of change are possible.

A new design for sampling with adaptive sample plots

Tue, 25 Oct 2011

Adaptive cluster sampling (ACS) is a sampling technique for sampling rare and geographically clustered populations. Aiming to enhance the practicability of ACS while maintaining some of its major characteristics, an adaptive sample plot design is introduced in this study which facilitates field work compared to “standard” ACS. The plot design is based on a conditional plot expansion: a larger plot (by a pre-defined plot size factor) is installed at a sample point instead of the smaller initial plot if a pre-defined condition is fulfilled. This study provides insight to the statistical performance of the proposed adaptive plot design. A design-unbiased estimator is presented and used on six artificial and one real tree position maps to estimate density (number of objects per ha). The performance in terms of coefficient of variation is compared to the non-adaptive alternative without a conditional expansion of plot size. The adaptive plot design was superior in all cases but the improvement depends on (1) the structure of the sampled population, (2) the plot size factor and (3) the critical value (the minimum number of objects triggering an expansion). For some spatial arrangements the improvement is relatively small. The adaptive design may be particularly attractive for sampling in rare and compactly clustered populations with an appropriately chosen plot size factor.

A stand dynamic model for red pine plantations with different initial densities

Mon, 16 Jan 2012

A stand dynamic model was developed to predict the growth response in even-aged forest planations of different initial planting densities. The model is based on the integration of three subcomponents: height growth, self-thinning, and diameter increment. The integrated model uses the height of dominant trees to simulate stand response to site quality and internal growth potential. An extended self-thinning submodel is used to simulate mortality in stands due to crowding and inter-tree competition. A diameter increment submodel is used to link the height growth and self-thinning submodels. The height growth submodel is based on an application of the "Pipe Model" theory. The three-parameter self-thinning submodel is developed from an extended self-thinning law that captures self-thinning in stands before they attain full stocking. The diameter increment model is based on the assumption that diameter increment is related to height growth and available growing space described by stand density. The integrated model is applied to data collected from a 45-year-old red pine (Pinus resinosa Ait.) plantation subsectioned with different initial planting densities. For the data used, only two parameters were required to capture 99% of measured variation in height growth. Additional data from sites with different planting intensities are required to formulate a more generalized height growth model. The slope of the linear self-thinning limit for red pine is approximately -1.5. Model predictions are consistent with field measurements.

An adaptive composite density estimator for k-tree sampling

Tue, 04 Oct 2011

Density estimators for k-tree distance sampling are sensitive to the amount of extra Poisson variance in distances to the kth tree. To lessen this sensitivity, we propose an adaptive composite estimator (COM). In simulated sampling from 16 test populations, a three-component composite density estimator (COM)–with weights determined by a multinomial logistic function of four readily available ancillary variables–was identified as superior in terms of average relative absolute bias. Results from a different set of nine validation populations–with widely different stem densities and spatial patterns of tree locations—confirmed that relative root mean squared errors (RRMSE) of COM were, on average, considerably lower than those obtained with the three-component k-tree density estimators. The RRMSE performance of COM improved with increasing values of k. With k = 6 and sample sizes of 10, 20, and 30, the average relative bias of COM was between −5 and 5% in seven validation populations but in an open low-density savanna-like population bias reached −12% (1979 data) and 7% (1996 data). For k = 6 and n = 10, the RRMSE of COM was, in six of the nine validation populations, within 3.3 percentage points of the RRMSE for sampling with fixed-area plots. Jackknife estimates of the precision of COM estimates of density were negatively biased, leading to under-coverage (7%) of computed 95% confidence intervals.

An architectural model of trees for estimation of forest structural attributes using terrestrial lidar.

Thu, 07 Mar 2013

Terrestrial lidar (TLiDAR) has been used increasingly over recent years to assess tree architecture and to extract metrics of forest canopies. Analysis of TLiDAR data remains a difficult task mainly due to the effects of object occlusion and wind on the quality of the retrieved results. We propose to link TLiDAR and tree structure attributes by means of an architectural model. The proposed methodology uses TLiDAR scans combined with allometric relationships to define the total amount of foliage in the crown and to build the tree branching structure. It uses the range (distance) and intensity information of the TLiDAR scans (i) to extract the stem and main branches of the tree, (ii) to reconstruct the fine branching structure at locations where the presence of foliage is very likely, and (iii) to use the availability of light as a criterion to add foliage in the center of the crown where TLiDAR information is sparse or absent due to occlusion effects. An optimization algorithm guides the model towards a realistic tree structure that fits the information gathered from TLiDAR scans and field inventory. The robustness and validity of the proposed model is assessed on five trees belonging to four different conifer species from natural forest environments. This approach addresses the data limitation of TLiDAR scans and aims to extract forest architectural metrics at different structural levels.

An architectural model of trees to estimate forest structural attributes using terrestrial LiDAR

Thu, 24 Mar 2011

Terrestrial lidar (TLiDAR)has been used increasingly over recent years to assess tree architecture and to extract metrics of forest canopies. Analysis of TLiDAR data remains a difficult task mainly due to the effects of object occlusion and wind on the quality of the retrieved results. We propose to link TLiDAR and tree structure attributes by means of an architectural model. The proposed methodology uses TLiDAR scans combined with allometric relationships to define the total amount of foliage in the crown and to build the tree branching structure. It uses the range (distance) and intensity information of the TLiDAR scans (i) to extract the stem and main branches of the tree, (ii) to reconstruct the fine branching structure at locations where the presence of foliage is very likely, and (iii) to use the availability of light as a criterion to add foliage in the center of the crown where TLiDAR information is sparse or absent due to occlusion effects. An optiization algorithm guides the model towards a realistic tree structure that fits the information gathered from TLiDAR scans and field inventory. The robustness and validity of the proposed model are assessed on five trees belonging to four different conifer species from natural forest environments. This approach addresses the data limitation of TLiDAR scans and aims to extract forest architectural metrics at different structural levels.

Aspects statistiques de la gestion forestière.

Wed, 11 Jan 2012

Assessment of standing wood and fiber quality using ground and airborne laser scanning: A review

Mon, 04 Apr 2011

Accurate information on the wood-quality characteristics of standing timber and logs is needed to optimize the forest production value chain and to assess the potential of forest resources to meet other services. Physical and chemical characteristics of wood vary with both tree and site characteristics. At the tree scale, crown development, stem shape and taper, branch size and branch location, knot size, type and placement, and age all influence wood properties. More broadly, at the stand level, stocking density, moisture, nutrient availability, climate, competition, disturbance, and stand age have also been identified as key determinants of wood quality. Such information is often captured in polygon based forest inventory data. Other terrain-related spatial information, such as elevation, slope and aspect, can improve assessments of site conditions and limitations upon plant growth which impact wood quality. Light Detection And Ranging (LiDAR) is an emerging technology, which directly measures the three-dimensional structure of forest canopies using ground or airborne laser instruments, and can provide highly accurate information on individual-tree and stand-level forest structure. In this paper, we explore the potential of LiDAR and other geospatial information sources to model and predict wood quality based on individual-tree and stand structural metrics. We identify a number of key wood quality attributes (i.e., basic wood density, cell perimeter, cell coarseness, fiber length, and microfibril angle) and demonstrate links between these properties and forest structure and site attributes. Finally, the potential for using LiDAR in combination with other geospatial information sources to predict wood quality in standing timber is discussed.

Beta-diversity gradients of butterflies along productivity axes.

Thu, 28 Jul 2011

Aim: Several lines of evidence suggest that beta diversity, or dissimilarity in species composition, should increase with productivity: (1) the latitudinal species richness gradient is most closely related to productivity and associated latitudinal betadiversity relationships have been described, and (2) the scale dependence of the productivity–diversity relationship implies that there should be a positive productivity–beta-diversity relationship. However, such a pattern has not yet been demonstrated at broad scales. We test if there is a gradient of increasing beta diversity with productivity. Location: Canada. Methods: Canada was clustered into regions of similar productivity regimes along three remotely sensed productivity axes (minimum and integrated annual productivity, seasonality of productivity) and elevation. The overall (β_j), turnover (β_sim) and nestedness (β_nes) components of beta diversity within each productivity regime were estimated with pairwise dissimilarity metrics and related to cluster productivity with partial linear regression and with spatial autoregression. Tests were performed for all species, productivity breadth-based subsets (e.g. species occurring in many and a moderate number of productivity regimes), and pre- and post-1970 butterfly records. Beta diversity between adjacent clusters along the productivity gradients was also evaluated. Results Within-cluster β_j and β_sim increased with productivity and decreased with seasonality. The converse was true for β_nes. All species subsets responded similarly; however, productivity–beta-diversity relationships were weaker for the post-1970 temporal subset and strongest for species of moderate breadth. Between-cluster beta diversity (β_j) and nestedness (β_nes) declined with productivity. Main conclusions: As predicted, beta diversity of communities within productivity regimes was observed to increase with productivity. This pattern was driven largely by a gradient of species turnover. Therefore, beta diversity may make an important contribution to the broad-scale gradient of species richness with productivity. However, this species richness gradient dominates regional beta diversity between productivity regimes, resulting in decreasing between-productivity dissimilarity with productivity driven by a concurrent decline in nestedness.

Biodiversity, ecosystem thresholds, resilience and forest degradation.

Tue, 24 Apr 2012

Biweekly disturbance capture and attribution: case study in western Alberta grizzly bear habitat

Tue, 06 Dec 2011

An increasing number of studies have demonstrated the impact of landscape disturbance on ecosystems. Satellite remote sensing can be used for mapping disturbances, and fusion techniques of sensors with complimentary characteristics can help to improve the spatial and temporal resolution of satellite-based mapping techniques. Classification of different disturbance types from satellite observations is difficult, yet important, especially in an ecological context as different disturbance types might have different impacts on vegetation recovery, wildlife habitats, and food resources. We demonstrate a possible approach for classifying common disturbance types by means of their spatial characteristics. First, landscape level change is characterized on a near biweekly basis through application of a data fusion model (spatial temporal adaptive algorithm for mapping reflectance change) and a number of spatial and temporal characteristics of the predicted disturbance patches are inferred. A regression tree approach is then used to classify disturbance events. Our results show that spatial and temporal disturbance characteristics can be used to classify disturbance events with an overall accuracy of 86% of the disturbed area observed. The date of disturbance was identified as the most powerful predictor of the disturbance type, together with the patch core area, patch size, and contiguity.

Canadian Forest Service Science Highlights. Can satellites help monitor biodiversity more effectively?

Thu, 08 Dec 2011

Canadian Forest Service Science Highlights. How can a national forest inventory monitor forest sustainability?

Wed, 11 May 2011

Canadian Forest Service Science Highlights. How can boreal forest countries learn best practices from each other?

Tue, 10 May 2011

Canadian Forests Exposed to the Winds of Change. Branching out from the Canadian Forest Service, Laurentian Forestry Centre. No. 70.

Tue, 20 Dec 2011

With climate change, Canada’s forests will be exposed to rapid changes in their environment, including variations in temperature and precipitation. Tree species will have to migrate to find the growing conditions that meet their needs. However, the rate of change in climatic conditions is expected to exceed their ability to migrate by a factor of 5 to 10. This raises the question of whether planning strategies for plantations should be adjusted accordingly.

Challenges and needs in fire management: a landscape simulation modeling perspective.

Wed, 15 Jun 2011

Changement climatique et productivité forestière : une forêt de questions – Prise 2. L'Éclaircie du Service canadien des forêts, Centre de foresterie des Laurentides. No. 63.

Wed, 06 Apr 2011

Researchers at the Canadian Forest Service are working on methods to estimate forest productivity at different spatial scales. Variations in forest productivity can best be estimated at the tree and stand levels.

Characterizing stand replacing disturbance in western Alberta grizzly bear habitat, using a satellite-derived high temporal and spatial resolution change sequence

Thu, 27 Jan 2011

Timely and accurate mapping of anthropogenic and natural disturbance patterns can be used to better understand the nature of wildlife habitats, distributions and movements. One common approach to map forest disturbance is by using high spatial resolution satellite imagery, such as Landsat 5 Thematic Mapper (TM) or Landsat 7 Enhanced Thematic Mapper plus (ETM+) imagery acquired at a 30 m spatial resolution. However, the low revisit times of these sensors acts to limit the capability to accurately determine dates for a sequence of disturbance events, especially in regions where cloud contamination is a frequent occurrence. As wildlife habitat use can vary significantly seasonally, annual patterns of disturbance are often insufficient in assessing relationships between disturbance and foraging behaviour or movement patterns.

The Spatial Temporal Adaptive Algorithm for mapping Reflectance Change (STAARCH) allows the generation of high-spatial (30 m) and -temporal (weekly or bi-weekly) resolution disturbance sequences using fusion of Landsat TM or ETM+ and Moderate Resolution Imaging Spectroradiometer (MODIS) imagery. The STAARCH algorithm is applied here to generate a disturbance sequence representing stand-replacing events (disturbances over 1 ha in area) for the period 2001–2008, over almost 6 million ha of grizzly bear habitat along the eastern slopes of the Rocky Mountains in Alberta. The STAARCH algorithm incorporates pairs of Landsat images to detect the spatial extent of disturbances; information from the bi-weekly MODIS composites is used in this study to assign a date of disturbance (DoD) to each detected disturbed area. Dates of estimated disturbances with areas over 5 ha are validated by comparison with a yearly Landsat-based change sequence, with producer's accuracies ranging between 15 and 85% (average overall accuracy 62%, kappa statistic of 0.54) depending on the size of the disturbance event. The spatial and temporal patterns of disturbances within the entire region and in smaller subsets, representative of the size of a grizzly bear annual home range, are then explored. Disturbance levels are shown to increase later in the growing season, with most disturbances occurring in late August and September. Individual events are generally small in area (<10 ha) except in the case of wildfires, with, on average, 0.4% of the total area disturbed each year. The application of STAARCH provides unique high temporal and spatial resolution disturbance information over an extensive area, with significant potential for improving understanding of wildlife habitat use.

Characterizing the state and processes of change in a dynamic forest environment using hierarchical spatio-temporal segmentation

Mon, 18 Apr 2011

Discrete changes in forest abundance, distribution, and productivity are readily detectable using a number of remotely sensed data sources; however, continuous changes such as growth and succession processes are more difficult to monitor. In this research we explore the potential of spectral trajectories generated from a 35-year (1973–2008) time-series of Landsat imagery to characterize change processes in a dynamic forest environment in northwestern Alberta, Canada. We propose a method of hierarchical spatio-temporal segmentation that enables the characterization of change processes that are spatially diffuse and temporally imprecise. Calibrated imagery from Landsat sensors are radiometrically normalized and two metrics derived from the Tasseled Cap Transformation components, greenness and brightness, are used to generate the Tasseled Cap Angle (TCA). The TCA is a measure of the proportion of vegetation to non-vegetation (the occupation state), and its derivative, the Process Indicator (PI), is a measure of change in this proportion through time. These indices condense information from the visible and near-infrared wavelengths, and facilitate lengthy time series analysis of forest landscape change using data from all Landsat sensors.

A combination of the original TCA and its derivative sequence are input to a three level hierarchical segmentation process with the highest and lowest levels defining homogeneous objects at the initial and final date, and the intermediate level identifying trajectories with similar change processes. The development through time of the TCA and PI are described, and the spatial and temporal associations of processes are statistically assessed using the Moran's Index.

A full range of change types were identified on the landscape, from stand replacing disturbances to more subtle growth and succession processes. Results indicate that the study area is in a constant state of change, and maintains a high average proportion of vegetation to non-vegetation. The amount of total landscape modified per decade increased from 18% and 14% in the 1970s and 1980s respectively, to more than 30% and 33% in the 1990s and 2000s. On average, the proportion of vegetation to non-vegetation was increasing prior to 1981, decreasing between 1981 and 1997, and increasing post-1997. There was a high degree of spatial autocorrelation amongst change processes, with a maximum Moran's I of 0.79 in 1973; landscape change became more spatially disperse and widespread after 1981. Temporal correlation of change processes was observed locally, with the period 1990–1995 having the most persistent change.