TY - JOUR
AB - This article contains all presentation abstracts from the “Forest Monitoring Science” track of the 2014 Society of American Foresters National Convention, held October 8-11 in Salt Lake City, Utah. Forest Monitoring Science James Westfall a Corresponding author (jameswestfall@fs.fed.us). bUSDA Forest Service, Newtown Square, PA. Selection of Forest Inventory Cycle Length Based on Growth Rate and Measurement Variability 3 As pressures on forest resources mount, there is an increasing need for up-to-date information upon which sound management and policy decisions can be made. This is evident in the adoption of shorter inventory cycle lengths for national forest inventories in the past few decades. While these intensified efforts are meritorious, substantial efficiency can be lost if cycle lengths are too short in relation to growth in the population. Specifically, if growth is small in relation to measurement variation, the signal-to-noise ratio is small and considerable expense is incurred to obtain data of limited usefulness. To address this issue, the relationships between measurement variation and growth rates were examined across a spatial gradient in the Eastern United States. As tree measurements are often not used directly, these relationships were also analyzed in the context of tree biomass prediction models of varying complexity. The results indicated that for latitude range N 30–38 degrees, where many fast-growing plantations are found, cycle lengths as short as 3 years may be acceptable if the biomass model only relies on tree diameter; however, increased cycle lengths should be chosen for more northerly latitudes. Longer cycle lengths are also suggested when employing more complex biomass models requiring additional inputs such as tree height and crown ratio, which are subject to higher levels of measurement variability. The methods provide a framework for determining cycle lengths that may considerably improve the efficiency and effectiveness of ongoing forest inventory and monitoring efforts. Daniel Unger I-Kuai Hung Ken Farrish a Corresponding author (unger@sfasu.edu). bStephen F. Austin State University, Arthur Temple College of Forestry and Agriculture, Nacogdoches, TX. Forest and Agriculture Landscape Changes due to Petroleum Exploration 35 The Haynesville Shale lies under areas of Louisiana and Texas and is one of the largest gas plays in the United States. Encompassing approximately 2.9 million ha, this area has been subject to intensive exploration for oil and gas, while over 90% of it has traditionally been used for forestry and agriculture. In order to detect the landscape change in the past few decades in particular in forest and agriculture lands, Landsat Thematic Mapper (TM) imagery for six years (1984, 1989, 1994, 2000, 2006, and 2011) was acquired. Unsupervised classifications were performed to classify each image to four cover types: agriculture, forest, well pad, and other. Change detection was then conducted between two classified maps of different years for a time series analysis. Finally, landscape metrics were calculated to assess landscape fragmentation. The overall classification accuracy ranged from 84.7% to 88.3%. The total amount of land cover change from 1984 to 2011 was 24%, with 0.9% of agricultural land and 0.4% of forest land changed to well pads. The results of Patch-Per-Unit index indicated that the well pad class was highly fragmented, while agriculture (4.4–8.6 per sq km) consistently showed a higher magnitude of fragmentation than forest (0.8–1.4 per sq km). Pat Murphy Roger Bohringer a Corresponding author (patriciad.murphy@wisconsin.gov). bWisconsin Department of Natural Resources, Eau Claire, WI. cWisconsin Department of Natural Resources, Boscobel, WI. Improving Plantation Success: Wisconsin's Reforestation Monitoring Program 46 A statewide Reforestation Monitoring Program, developed in 2007 by the Wisconsin Department of Natural Resources' (DNR) State Nursery Program, evaluates conifer, hardwood and shrub planting successes and failures on public and private lands. DNR research scientists now are analyzing the data collected from 911 sites and 55,000+ seedlings. Orders of over 3,000 seedlings are randomly selected for first year surveys; subsequent surveys are conducted at 3- and 7-year intervals. All trees within a 1/100th acre plot are evaluated. Hypotheses being evaluated include: Certain herbicides decrease survivability in some species. Plantations with forester involvement and written planting plans have higher success rates. Planting depth and seedling orientation impact survival and long term growth. Competition level and type at the root zone strongly correlates with survival and growth rates. Deer and rodent browse have serious impacts on plantings. Stock planted early does better than that planted later. The time between lifting/shipping date and shipping/planting date impacts survival. Stock condition (dormancy) at shipping affects survival rate. Shipping method correlates to seedling survival; proper refrigerated transportation increases survival. Storage method after leaving the nursery is critical. Site preparation methods lead to varying degrees of planting success. Roots frequently are over pruned. Planting culls reduces vigor and overall plantation survival. Previous cover type has a bearing on plantation success. Species planted off site shows poor performance. Findings are being used to develop reforestation and management recommendations. Stephen Fairweather a Corresponding author (sfairweather@masonbruce.com). bMason, Bruce and Girard, Inc., Portland, OR. Characterizing Sources of Uncertainty in a Stand-Based Forest Inventory 50 Stand-based forest inventories are typically kept current with a combination of cruising, growth modeling, and adjustments to represent harvest activity. At any point in time the inventory will have some stands with recent cruise data, some stands which have never been cruised but carry estimates for the stratum they belong to, and some stands which were cruised some time ago and have been grown each year using a growth model. Estimating a confidence interval for the inventory for the total collection of stands using traditional approaches is problematic due to the mix of stand inventory methods. In this paper we use Monte Carlo simulation to estimate the error on the total inventory estimate, and show how the approach can be used to evaluate annual inventory update methods with regard to cost, accuracy, and precision. Stella Cousins John Sanders Robert York John Battles a Corresponding author (stella.c@berkeley.edu). bUniversity of California, Berkeley, CA. cUniversity of California–Berkeley Center for Forestry, Georgetown, CA. Improving Estimates of Carbon Characteristics for Standing Dead Trees in the Sierra Nevada 71 As a direct result of rising forest morbidity and mortality, standing dead trees are becoming increasingly important players in forest carbon dynamics. Tree death marks a crucial transition: the shift from carbon sink to carbon source. Warming and drying climate in the West is expected to drive additional mortality and increase the ecological importance of standing dead trees. In this study, we quantify the roles of standing dead trees in forest carbon dynamics. We measured characteristics of six species across a gradient of decay in Sierra Nevada mixed conifer forests. Using dimensional analysis, we developed species-specific density reduction functions that describe how wood and bark change with advancing decay. Our approach is designed for application with the USFS-FIA decay classification system. Results suggest that current estimates of standing dead carbon based on live tree characteristics are typically overestimates. Declines in wood density drive decreases in the total carbon density of standing dead trees: analysis of white fir showed first-stage standing dead trees contained 90% of the carbon found in live trees of the same volume; fifth-stage individuals contained only 42.7%. Across Sierra mixed conifer species, we observed that wood basic specific gravity declines with decay class. Additionally, specific gravity of standing dead trees is more varied than that of live trees, as follows from the many weathering and decomposition processes underway. Our estimates of standing dead tree carbon pools, and how they change during decay, will be used to generate greenhouse gas budgets that better reflect environmental conditions. David Colville James MacKay Randy Milton Sean Basquill Michael Gemmell a Corresponding author (David.Colville@nscc.ca). bApplied Geomatics Research Group, NSCC, Lawrencetown, Nova Scotia, Canada. cNova Scotia Department of Natural Resources, Kentville, Nova Scotia, Canada. Monitoring and Modeling Forest Cover Change to Support Forest Landscape Management in Nova Scotia 72 The Wildlife Division of the Nova Scotia Department of Natural Resources (DNR) is responsible for the management of the province's wildlife, including species at risk such as the mainland moose. In 2009 DNR initiated a mainland Moose Habitat Research Program (MHRP). Objectives of the program included identifying key habitat parameters, defining suitable moose landscapes, and investigating moose movement dynamics among landscape elements. Research findings will be used to define land management options (particularly related to forest harvesting), contribute to population recovery planning, and contribute to the establishment of landscape guidelines for the maintenance of ecosystem integrity. In 2012 DNR approached researchers at the Applied Geomatics Research Group (AGRG) for assistance with the creation of a GIS-based Landscape Modeling Framework (LMF): software tools for defining, characterizing, and integrating wildlife habitat and habitat related functions into the Department's spatial programming systems and decision support structures. AGRG worked closely with DNR throughout 2013 to create this new data analysis environment. The LMF now provides DNR with customized tools to produce landscape composition, structure, and arrangement metrics as an aid in the assessment of forested landscapes and the changes taking place within them. Critical to the task was a land cover data layer which was temporally coincident with wildlife (i.e., moose) occurrence data. A time-aware spatial database was created based on three cycles of provincial forest inventory data and forest change mapping derived from satellite imagery. Within the LMF this powerful temporal database allows for efficient display, analysis, and comparison of landscape conditions over the last 25+ years. Emma Vakili Chad Hoffman Robert Keane Yvette Dickinson Monique Rocca a Corresponding author (vakili.emma@gmail.com). bColorado State University, Fort Collins, CO. cUnited States Forest Service, Missoula, MT. Fuel Treatment Effects on Surface Fuels in Dry Ponderosa Pine Forests of the Southern Rockies 144 Recent research has shown the important influence of fuel variability on wildfire behavior and effects. This has resulted in a need for more accurate quantification of fuels across spatial scales, including a description of variability. However, few studies to date have assessed the spatial variability of wildland fuels or investigated the inherent scales at which they vary. This study investigates the characteristic length scale and spatial variability of surface fuelbed components in dry ponderosa pine forests on the front range of the southern Rocky Mountains pre- and post-fuel reduction treatment. Preliminary results suggest that treatment effects on variability are mixed in both size and direction but that patch sizes are consistently small, measuring between 1 and 14 meters regardless of fuel particle size. This indicates that common sampling techniques do not capture the full variability of surface fuels, and that regardless of the sophistication of the fire model used, fire behavior will not be fully captured. The results from this study will contribute to further developing the next generation of fuel models and creating more accurate and efficient sampling techniques to help land managers assess what effects their treatments have on fuelbeds. Susan Hummel Kevin Halverson Kevin Megown Jeremy Webb a Corresponding author (shummel@fs.fed.us). bUSDA Forest Service, Portland, OR. cUSDA Forest Service, Ogden, UT. dUSDA Forest Service, Salt Lake City, UT. Tree Mortality Estimates from Multi-Year Ground and Lidar Data Compare after Mixed-Severity Wildfire 155 Can remotely sensed imagery be used to count individual trees after a mixed-severity fire and help identify which survive and which die over time? If so, what can be learned about the variation in fire effects on trees in low, moderate, and high burn severity classes? To investigate, burn severity classes were mapped for a ∼3600 ha study site using the relative differenced normalized burn ration (RdNBR). Aerial photos and airborne laser scanning data (ALS or lidar) were acquired in two sets for the site: the first (t1) at post-post fire and the second (t2) after three years. Ground plots (n = 90) were randomly selected from the three burn classes and all trees on the 10m plots were recorded in both time periods. The t1 and t2 photos and lidar data were “clipped” to a 20m plots by using the the ground plot GPS coordinates and the count of live and dead trees was interpreted for photo (PI), lidar (LI), and a merge of the two (M). The count of live and dead trees on each ground (G) plot was tallied directly and estimates were derived for expanded plot sizes of 15m and 20m. The Wilcoxon signed rank test of differences in the median number of live trees was calculated by severity class for each pair (G-PI, G-LI, and G-M) in t1 and t2. Results indicate that live trees were insensitive to interpretation method and plot size; no difference existed in median estimates of live trees in the moderate and high burn severity classes. This means that the median number of live trees counted via interpretation did not differ from what was measured on the ground. In contrast, dead trees were sensitive to both interpretation method and plot size. The ability to count dead trees using the merge of photo and lidar was better than either the PI or LI methods alone, which implies that intensity values available from lidar added key information. This is important, because it can be difficult to sample rare but key forest structural elements like dead trees. Results also offer direct information on variation in tree mortality within burn severity classes and illustrate the importance of studying the direction and magnitude of second-order fire effects, particularly in forests burned with moderate and low severity. Susan Hummel Frank Lake a Corresponding author (shummel@fs.fed.us). bUSDA Forest Service, Portland, OR. cUSDA Forest Service, Arcata, CA. Good Conditions Identified by Tribal Weavers for Harvesting Beargrass Can Inform Forest Management 158 We blended methods from scientific and traditional ecological knowledge to describe forest conditions on sites considered good (G), marginal (M), or poor (P) for harvesting the leaves of a plant (beargrass or X. tenax) used in tribal basket weaving. We relied on voluntary participation of six expert tribal weavers, a stratified, randomized field sample, and discriminant analysis (DI). We accepted each weavers qualitative classification of a forested site (G, M, or P) for beargrass harvest and then sampled site and plant attributes on two plots in each site class (n = 36). Variables included: basal area (BA), fuel loading, canopy cover, beargrass abundance, and leaf color. We used DI resubstitution to identify the combination of variables that maximized the number of correctly classified good sites. We found trends in BA, coarse wood, and leaf color on good sites versus marginal or poor ones across all tribal weaving styles and forest types. On our sites in Washington, Oregon, and California, the average tree BA was between 180–200 ft2/ac. On sites considered by tribal weavers as good for beargrass harvest, however, the BA was distributed on fewer trees than it was on sites they identified as marginal or poor for harvest. Similarly, the average amount of coarse wood (>3”) was lowest on good sites and increased inversely as the suitability of sites for beargrass harvest declined. Taken together, the forest conditions identified by tribal weavers as good for harvesting beargrass had, on average, fewer, larger diameter trees and less coarse wood than did less desirable sites. These results imply that structural elements associated with managing fire behavior and severity in western forests are consistent with the site conditions preferred by tribal weavers for beargrass harvest. Benktesh Sharma William Stewart Jeremy Fried Brandon Collins Jason Moghaddas a Corresponding author (benktesh@yahoo.com). bUniversity of California, Berkeley, CA. cUSDA Forest Service, Pacific Northwest Research Station, Portland, OR. dUSDA Forest Service, Pacific Southwest Research Station, Davis, CA. eSpatial Informatics Group, Pleasanton, CA. Fuels Hazard Reduction Options in California Mixed Conifer: A Combinatorially Exhaustive Evaluation 179 Using the system for Bioregional Inventory Oriented Simulation Under Management (BIOSUM), we compared generic, multi-decade, silvicultural prescriptions with respect to merchantable and energy wood recovered and gross and net costs of treatment implementation in California's mixed conifer forest type. Impacts of policies that render aspects of these treatments infeasible on selected ownerships are also evaluated. We formulated 144 silvicultural prescriptions to reduce fire hazard and maintain forest growth, and then applied them to 680 relevant Forest Inventory and Analysis (FIA) plots representing 4 million forested acres. Prescriptions involved periodic thinning from below, evenly, and from above of trees between 8 and 40” dbh to target residual basal areas of 75, 100, and 125 square feet/acre over a forty year period. These diverse crown fuels approaches were paired with diverse surface fuel treatment strategies directed at smaller trees and shrubs: lop and scatter, mastication, and prescribed fire at 10 or 20 year intervals. Effectiveness was assessed via the Forest Vegetation Simulator, its Fire and Fuel Extension, and its REPUTE module for simulating regeneration. These prescriptions reduced surface flame length from an average (and range) of 4.79 feet (3.49–11.24) to 2.44 feet (0.57–12.06) and probability of torching from 0.72 (0–1) to 0.42 (0–1). Thinning combined with a surface fuel treatment at year 10 and 20 minimized fire hazard. Lop and scatter, and mastication, proved less effective than prescribed fire. Accounting for forest growth, treatment costs, product yields, and fire hazard change over a 30-year time frame provides more complete and realistic guidance than what can be learned from single entry simulations. Max Bennett a Corresponding author (max.bennett@oregonstate.edu). bOregon State University, Central Point, OR. Results of Multi-Party Monitoring on the Middle Applegate Pilot Project 180 The Middle Applegate Watershed Pilot is a collaborative restoration forestry project involving the Bureau of Land Management, Southern Oregon Forest Restoration Collaborative, Applegate Partnership, and others. One of three western Oregon pilot projects designated by the Secretary of the Interior, the Pilot is intended to restore dry Douglas-fir forests in the 80,000 acre Middle Applegate watershed using an ecological forestry approach. The first phase of the project, including a timber sale and service work, has been implemented. Baseline, implementation, and some effectiveness monitoring have been completed by a multi-party monitoring team convened by the Southern Oregon Forest Restoration Collaborative. In this presentation we report on monitoring indicators associated with five project objectives: increase forest ecosystem resistance and resilience, increase spatial heterogeneity, conserve and improve habitat for the northern spotted owl, generate jobs and support regional wood products manufacturing, and gain public support for active management in federal forests. The project provides an excellent case study of the social and technical challenges, as well as the benefits, of a multi-party monitoring program on federal lands. David Shear a Corresponding author (david.shear@eagleimaging.net). bEagle Digital Imaging, Inc., Corvallis, OR. Integrating Image-Based Stand Analysis with Mobile Devices to Enhance Forest Management 235 The availability of super high-resolution images (ground sample distance of less than 1”) enables the extraction of additional information on the status of a stand. These images allow the analyst to see details that are not visible with lower resolution. This image-based stand analysis approach can provide stocking hole and competing vegetation maps, trees per acre, species identification, and many other types of data. Advances in mobile devices and GIS software now allow this captured information to also be used in the field, significantly increasing the effectiveness of the information. The analysis enabled by the super high resolution images enhances forest management by allowing foresters to prioritize stand visits, directing them to the areas of interest for on-site analysis, and facilitating the creation of work orders for replanting, vegetation control, pre-commercial thinning, helicopter flights, and other management actions. Kristen Pelz Frederick Smith Robert Hubbard Charles Rhoades a Corresponding author (kapelz@gmail.com). bColorado State University, Fort Collins, CO. cUSDA Forest Service, Rocky Mountain Research Station, Fort Collins, CO. Species-Specific Effects of Variation in Mountain Pine Beetle-Caused Mortality on Tree Regeneration 242 The recent mountain pine beetle outbreak has killed lodgepole pine on millions of hectares in western North America. Mortality in affected forests varies substantially and is likely to cause variation in future forest development trajectories. At Fraser Experimental Forest in Colorado, USA, we studied the effects of high (75% of basal area) and moderate (40% of basal area) mortality on the regeneration of three conifer species that commonly co-occur in mountain pine beetle-affected forests: shade-intolerant lodgepole pine, and shade-tolerant subalpine fir and Engelmann spruce. Forests of both mortality levels had 90% lodgepole pine overstory basal area with mature spruce and fir. We found high densities of regeneration in both mortality levels, but there was significantly more lodgepole in the high mortality sites and significantly more subalpine fir in moderate mortality sites. Distance to nearest mature tree of each species did not explain differences in regeneration density. Height growth of tree regeneration was greater in the high sites for all species; lodgepole pine growth was 4x greater and Engelmann spruce and subalpine fir growth were 2x greater in the high mortality areas than low mortality areas. In high mortality areas lodgepole pine was the fastest growing species, while in moderate mortality areas Engelmann spruce and subalpine fir grew fastest. The differences among species' responses to high versus moderate overstory mortality suggest variation in mortality rate will lead to substantially different future forest compositions, despite similar pre-outbreak compositions. Ramesh Sivanpillai a Corresponding author (sivan@uwyo.edu). bUniversity of Wyoming, Laramie, WY. Identifying Data Requirements for the Next Landsat Missions: A Panel Discussion 256 This is a panel discussion proposed by the Remote Sensing And Geospatial Applications Working Group (A2-WG) with the goal of soliciting input data requirements for future Landsat missions. Following the successful launch of Landsat 8, discussions are underway for designing the next Landsat satellite (9 and possibly 10). This presents an opportunity for the forestry community to make recommendations for data necessary to monitor, map, and manage forest resources. Input gathered from this panel discussion will be summarized as an A2-WG report and submitted to the US Geological Survey, the agency responsible for collecting data on user requirements. Organization: First, participants will be introduced to the Landsat program and select forestry applications that have taken advantage of Landsat data will be highlighted. This will be followed by discussion and compilation of future data requirements. Scientists from academia and a federal agency will be part of this panel. Demetrios Gatziolis a Corresponding author (dgatziolis@fs.fed.us). bUSDA Forest Service, Portland, OR. Unbiased and Precise Estimation of Canopy Closure Using High-Density LiDAR Data 288 Canopy closure determines the amount of solar illumination present in forest stands, affects tree growth, regeneration potential, wildlife habitat, and various other ecological processes. Distributed estimates of canopy closure obtained from models utilizing costly field measurements and spectral airborne or satellite imagery are often substantially biased and imprecise. This study estimates canopy closure via discrete-return LiDAR data in voxel space. Initially, a stand is partitioned into a continuum of cubical elements (voxels) labeled ‘filled’ if they contain LiDAR return(s), or ‘empty’ otherwise. A set of rays with trajectories initiated at a given location and directed towards the enveloping hemisphere is used to calculate canopy closure as the ratio of rays that intercept at least one filled voxel to the total number of rays cast. Distributed estimates of canopy closure are obtained by repeating the ray-casting and tracing process at desired locations. Optimized software and processing in parallel reduce computational times and enable estimates at fine spatial intervals. The methodology was evaluated against canopy closure estimates derived from hemispherical photographs co-temporal to high-density LiDAR data at 64 temperate rainforest and open, eastern Oregon forest stands comprising a variety of cover type and tree density conditions. It was determined that the LiDAR-based estimates of canopy closure are unbiased and precise (root mean square discrepancy 1.7 percent). Precision improves with the number of rays cast but the improvement is marginal above 30 rays. The proliferation of high-density LiDAR data sets renders this methodology potentially valuable in support of stream shading assessment, forest restoration, and other efforts. David Chojnacky John McGee Laura Johnson Steve Prisley a Corresponding author (dchojnac@vt.edu). bVirginia Tech, Blacksburg, VA. Urban Forest Inventory, Monitoring, and Projection on Private Lots in Falls Church, Virginia, USA 311 In the small City of Falls Church—mostly built over 50 years ago within an eastern Piedmont-type deciduous forest in the Washington DC area—towering 100-cm dbh tulip poplar and oak surround many older homes. However, redevelopment is replacing these homes and changing the urban forest structure. Our city–university collaboration resulted from the need to scientifically inform redevelopment management decisions. Each residential redevelopment in Falls Church is required to retain or plant enough trees for 20% canopy cover within 10 years. The City Arborist has two years during site redevelopment to influence tree cover on these private lands through redevelopment plans. However, implementation guidelines lack information about how long it takes for various tree arrangements to reach 20% cover. We monitored tree cover before and up to 16 years after redevelopment. Two inventory datasets were compared: (1) initial measurements from redevelopment plans (both retained and newly planted trees) and (2) current remeasurements (20 lots out of 300 randomly sampled in strata [years since redevelopment]). Two metrics were calculated and compared: (1) a canopy-cover metric and (2) traditional forest stand metrics based on basal area and trees per ha. Also, growth projection methodology was developed for estimating future tree status, as needed by urban foresters seeking to mitigate redevelopment effects. Results show the canopy-cover metric—although easily understood by the public, developers, and planners—masks tree size changes easily accounted for by traditional metrics. Pros and cons are discussed for use of each metric for monitoring urban forests. Andrew Gray a Corresponding author (agray01@fs.fed.us). bUSDA Forest Service, Pacific Northwest Research Station, Corvallis, OR. Land Use and Carbon Change: Role of Spatial and Temporal Scale in Addressing Uncertainty 312 A variety of field- and remote sensing-based approaches are available for measuring changes in forest ecosystems. The objective of this study was to assess changes in land use and carbon storage in the forests of Oregon and Washington, USA. Methods included repeated measurement of Forest Inventory and Analysis (FIA) and National Forest inventory plots, aerial photography, and satellite imagery. Air-photo analysis indicated 1.16 million acres of nonfederal forest and agriculture land use were converted to residential and urban land uses between 1976–2006 in Washington state (0.13%/yr), twice the rate of change in Oregon (0.064%/yr). Estimates of forest loss compared favorably with plot-based analyses but provided better spatial resolution and the ability to quantify housing density in forest land (WUI). Live tree carbon increased at 2.3 Mg/ha/yr (0.2%/yr) across both states, but most of the increases occurred on public lands west of the Cascade Mountains, and most of the decreases on private lands east of the Cascades. Satellite imagery detected most of the disturbance and management events that were recorded on inventory plots, as well as providing current estimates of change since the last plot measurement. Plot-based assessments can provide great detail on causes of change and be definitive over large areas and time spans. Remote-sensing analyses can provide useful estimates of general types of change for smaller areas and short time spans. In practice, most assessments employ a blend of data sources. Modeling carbon change with satellite imagery and plot measurements is an active area of current research. Stephanie Bohlman a Corresponding author (sbohlman@ufl.edu). bUniversity of Florida, Gainesville, FL. Determining Species Distributions and Growth Rates of Tree Crowns Using Hyperspectral Remote Sensing 319 Technological advances in remote sensing has steadily provided new types and levels of detail of information for forest inventory and monitoring on large spatial scales. The increasing access to high spatial resolution hyperspectral images (1–2 m pixel resolution; > 200 narrow wavelength bands in the visible and infrared spectral regions) provides new proven and promising opportunities to map species distributions and provide information on individual tree characteristics and performance. Here we discuss the use of these images to map crowns of individual species and assess their growth performance in a temperate in Florida and tropical forests in Panama. First, we assessed the accuracies of different statistical techniques to predict species on a crown level across these landscapes. Beyond mapping individual species, there is growing evidence that hyperspectral signatures can measure within and among species variation in canopy chemistry and structure, such as nitrogen content and specific leaf area, which can impact growth rates. Because of this link between canopy chemistry and growth rates, and canopy chemistry and hyperspectral signatures, we investigated if hyperspectral variation within and among species reflects growth rate variation. This study, performed at a species trial experiment in Panama where over 50 species were planted in blocks, indicated that hyperspectral variation among crown within a species varied systematically with growth rates. We discuss future tests of hyperspectral detection of growth rate variation in temperate forest sites via the National Ecological Observatory Network (NEON). Rachel Riemann Doug Griffith James Ellenwood Andrew Lister Barry T. Wilson Frank Krist a Corresponding author (rriemann@fs.fed.us). bUSDA Forest Service, Northern Research Station, Troy, NY. cUSDA Forest Service, Newtown Square, PA. dUSDA Forest Service, Fort Collins, CO. eUSDA Forest Service, St. Paul, MN. Comparative Assessment of Two Nationwide Datasets of Tree Species Distributions 321 Two sets of nationwide tree species distribution datasets have been developed independently by USFS Forest Inventory and Analysis (FIA) and by USFS Forest Health Technology Enterprise Team (FHTET), using different methods and with slightly different original goals. The FHTET datasets were created in order to subsequently develop risk maps to disease and pest infestations. The FIA datasets were developed with more general applications in mind. Both were modeled using multiple input datasets including 30m and 250m satellite imagery and relevant environmental parameters. Using a series of nationwide hexes at different scales (3.5 million ha, 866,000 ha, 216,500 ha, and 78,100 ha), we compare the FHTET and FIA tree species datasets in terms of differences in local average and variance values between the two datasets, which can vary by region and scale. In addition, some of the assessment protocols described in Riemann et al. (2010) were applied to examine differences in data distributions, and systematic vs. unsystematic differences, also at several scales. Pattarawan Watcharaanantapong Donald Hodges a Corresponding author (pwatchar@utk.edu). bUniversity of Tennessee, Knoxville, TN. Factors Influencing Global Carbon Emission from Land-Use Change and Forest Activities 327 Carbon dioxide (CO2) emissions arise not only from energy production and industrial processes, but also from land-use change, deforestation, forest degradation, and land clearing for agriculture. The goal of this study is to evaluate the factors that influence global CO2 emissions due to forest conversion. A tobit model was developed to evaluate the factors influencing CO2 emissions from the conversion of forest areas globally. The level of CO2 emissions per country was the dependent variable; independent variables included land and forest conditions, economic information, demographics, agricultural production, industrial production, employment, school enrollment, and housing and commodity demand. This research will be helpful for researchers, scientists, and policy makers by providing useful information for monitoring the global environmental impacts of land use, land use change, and forest activities. Matthew Tuten Andrew J. Sanchez Meador a Corresponding author (matthew.tuten@gmail.com). bUSDA Forest Service, Montrose, CO. Northern Arizona University, Flagstaff, AZ. Ecological Restoration and Fine-Scale Structural Regulation in Southwestern Ponderosa Pine Forests 367 Fine-scale forest structure regulation objectives are currently a major component of forest plans within Southwestern National Forests. Despite their importance to the management of these forests, questions remain regarding the assessment of silvicultural treatments designed to meet these objectives. We compared stand and sub-stand-scale spatial forest structure attributes associated with the implementation of two silvicultural thinning techniques commonly applied in dense ponderosa pine forests: historical evidence-based ecological restoration and northern goshawk (Accipiter gentilis) foraging area management guidelines. We also assessed vegetation structural stage (VSS; a classification of fine-scale forest structural development) composition and distribution resulting from these approaches. Our results indicate that these silvicultural approaches are largely compatible at stand and sub-stand scales, resulting in minor differences in spatial pattern and stand-scale attributes. Comparison of VSS composition and distribution resulted in minor differences that highlight the importance of spatial-scale, structural complexity and classification methods in the assessment and regulation of fine-scale forest structure. These results have implications for both the development of desired forest spatial structure objectives and the implementation and assessment of silvicultural treatments designed to accomplish these objectives. Bianca Eskelson Vicente Monleon a Corresponding author (bianca.eskelson@ubc.ca). bThe University of British Columbia, Vancouver, British Columbia, Canada. cUSDA Forest Service, Corvallis, OR. Spatiotemporal Models to Assess Whitebark Pine Health Status and Trends in California, Oregon, and Washington 383 Whitebark pine, a keystone species, is declining throughout its range. Its health is affected by a combination of white pine blister rust, mountain pine beetle outbreaks, altered fire regimes, and climate change. Available information about the current whitebark pine health status and trends are based on studies from selected stands and geographic locations rather than a species range-wide inventory. We compiled all available US Forest Inventory data (including off-grid, intensification plots) to provide estimates of the current health status and trends of whitebark pine populations in California, Oregon, and Washington. We analyze the presence of blister rust and bark beetle infested trees on inventory plots as well as the number of trees with blister rust and bark beetle attacks out of the total number of trees. The inventory plots were measured with various sampling intensities at different dates, some measured more than once but with varying measurement periods. Thus, standard design-based approaches cannot be used to estimate health status and trends. In addition, the response variables (e.g., presence of blister rust) are not normally distributed. Spatio-temporal models that incorporate spatial dependency, account for repeated measurements in some plots, and allow a non-normally distributed response are needed. We developed models that account for the complexity of the sampling design with its uneven spatial and temporal intensities. Our results inform forest managers of the current status and health trends of whitebark pine populations in the Pacific coast states. Our models could be applied to other problems from similarly complex forest inventory data. Copyright © 2015 Society of American Foresters
TI - Abstract
JF - Journal of Forestry
DO - 10.1093/jof/113.1.106
DA - 2015-01-01
UR - https://www.deepdyve.com/lp/springer-journals/abstract-ytO0Mz0EmC
SP - 106
EP - 111
VL - 113
IS - 1
DP - DeepDyve
ER -