|Lixia Lambert||FIA Use in Developing the Forest Sustainable and Economic Analysis Model (ForSEAM)||Tuesday||1c||Cherokee||Advances in modeling distributions, structure, and sustainability||The use of Forest Inventory and Analysis (FIA) database in developing the Forest Sustainable Economic Analysis Model (ForSEAM) to conduct spatial optimization analysis on timber resource for renewable energy development.||The Forest Sustainable and Economic Analysis Model (ForSEAM) is a dynamic linear optimization model developed to determine where conventional wood and energy feedstocks could be acquired from timberland. The model is compartmentalized into three sections including supply, demand, and sustainability. The supply component includes timber-sector production activities for 305 production regions in the lower 48 U.S. states. Each region is composed of a set of production activities including sawlog, pulpwood, and energy feedstock (woody biomass). The model currently considers two sources of energy feedstock: 1) logging residue and 2) removal of whole pulpwood and non-merchantable trees as energy feedstock. The growth rates and yields of different size and types of timber were calculated based on information from the Forest Inventory and Analysis (FIA) database. The conventional timber demand component is based on U.S. Forest Service Scenarios determined by the U.S. Forest Products Module. The sustainability component insures that harvest in each region does not exceed annual timber growth, forest tracts are located within reasonable distance of the roads,and current year forest attributes reflect previous period harvests and removals.|
|Kathryn Baer||Species distribution models predict shifting climatic suitability for important subsistence species in interior Alaskan forests||Tuesday||1c||Cherokee||Advances in modeling distributions, structure, and sustainability||Changing climatic conditions in subarctic forests may shift the distribution of understory vegetation, including important berry-producing species. We utilized occurrence records of berry species from Interior Alaska FIA data paired with current climate data and future climate models to estimate berry species distributions in the present and predict future shifts. Models predicted a decrease in highly suitable habitat along with shifts towards higher elevations and away from population centers.||Rapidly changing climatic conditions in subarctic forests have the potential to drive dramatic shifts in the distribution and abundance of their component vegetation. Of particular concern in Alaska is the potential for changes in the distribution and productivity of berry-producing understory species, which represent an important cultural and subsistence resource. FIA inventory data for Interior Alaska offer a unique dataset from which models can be constructed to estimate current berry species distributions and project how these distributions may shift under future climate change. In this study, we utilized records of berry species presence and absence paired with current climate data to estimate the current extent of berry species distributions. We then applied these models to three global circulation models for future climate to predict how these distributions might shift in the study area under future climatic conditions. Models predicted a decrease in the incidence of highly suitable habitat along with shifts towards higher elevations and away from population centers. These results indicate that subsistence berry species may become less abundant in interior Alaska and suggest areas for intensified future monitoring.|
|Margaret Evans||Using forest inventory data to build demography-driven models of the geographic distribution: a case study of Pinus edulis (two-needle piñon)||Tuesday||1c||Cherokee||Advances in modeling distributions, structure, and sustainability||We present a case study using FIA data to parameterize a demographic range model a model of the geographic distribution of Pinus edulis based on core variables of the annualized FIA Program along with PRISM climate data. This demography-driven range model suggests that negative density-dependent regulation, in addition to climatic factors, are important in shaping the distribution, and further that fire plays an important role in excluding this species from more productive forest types.||Climate change is expected to affect species distributions. Species distribution models based on occurrence data, using bioclimatic predictors alone, are widely criticized as phenomenological. Here we view the FIA database as the worlds largest repository of demographic data on trees, and use these data to parameterize a demographic model, an integral projection model, for a species whose entire geographic distribution is contained within the FIA domain, Pinus edulis (two-needle piñon). We evaluate competing hypotheses for piñons distribution: climate alone limits this species vs. climate and competition are both important limiting factors. We find support for the importance of both, with respect to the response of vital rates (growth, mortality, and regeneration) to variation in climatic and competitive conditions, and in terms of the fit between projected population growth rate and observed occurrences of Pinus edulis. Drawing on both disturbance ecology and a deep time perspective on the importance of fire in shaping the evolution of pines, we further suggest that fire may a missing factor that leads to a mismatch between occurrence data and predicted population growth rate above a mean annual precipitation of ~750 mm.|
|Patricia Manley||Data solutions for large-scale modeling and planning: Using FIA-based SilviaTerra Data for resilience planning in the central Sierra Nevada||Tuesday||1c||Cherokee||Advances in modeling distributions, structure, and sustainability||The Tahoe Central Sierra Initiative is a collaborative of 10 organizations in California that is restoring forest resilience across 2.4 M acres of high value watersheds. SilviaTerra data was identified as a singular, high quality data source for vegetation data across the entire landscape. Our detailed assessment addressed forest structure and composition, wildlife habitat, and timber supply over 100 years using the Landis II dynamic growth model to inform restoration options.||Western coniferous forests are exhibiting substantial stress from lack of fire, drought, and changing climates. Large-scale mortality events have been increasing in extent and frequency, and they are raising concerns that the opportunity to restore forest resilience may be slipping away. A common barrier to increasing the pace and scale of restoration is the availability of vegetation data that is consistent across large areas and multiple ownerships, and that describes forest conditions with sufficient detail, accuracy, and precision to support management actions. The Tahoe Central Sierra Initiative, a collaborative of Federal, State and non-profit organizations in California was formed to restore forest resilience across 2.4 M acres of mixed land ownership in some of the most ecologically and socially important watersheds in California. SilviaTerra data was identified as a singular, high quality data source for vegetation data across the entire landscape that provided spatially explicit, high resolution data on forest structure and composition. We conducted a detailed assessment of forest structure and composition, wildlife habitat, and timber supply, and then modeled changes in these parameters over 100 years using the Landis|
|David Bell||An application of generalized joint attribute modeling for modeling forest structure and composition||Tuesday||1c||Cherokee||Advances in modeling distributions, structure, and sustainability||Wall-to-wall maps of forest structure and composition are needed for landscape monitoring and planning. We apply a new hierarchical Bayesian modeling approach for complex, multivariate ecological data to predict multivariate forest inventory structure and composition based on environment and multispectral remote sensing. We compare the results of this model with those of an existing multivariate nearest neighbor imputation approach.||Wall-to-wall maps of forest structure and composition are needed for landscape monitoring and planning. Any such mapping strategy should generate statistically rigorous estimates of multiple response variables (e.g., species-level abundances) that are consistent with emergent properties (e.g., total abundance and species richness). In this study, we will apply the generalized joint attribute modeling (GJAM) approach, a hierarchical Bayesian modeling method for complex, multivariate ecological data, to predict multivariate Forest Inventory and Analysis structure and composition data based on environmental gradients and multispectral remote sensing. We will compare GJAM with an existing multivariate approach (gradient nearest neighbor imputation) in terms of (1) individual species basal area, (2) total tree basal area, and (3) total species richness. Comparisons will include plot-level accuracy assessments as well as comparisons of resulting map products, including both mean predictions and prediction precision (e.g., standard deviations). By comparing GJAM with existing nearest neighbor imputation methods, we will provide new guidance regarding the mapping of complex, multivariate forest attribute datasets.|
|Brian Miranda||Combining FIA data with the LANDIS forest simulation model to improve projections||Tuesday||1c||Cherokee||Advances in modeling distributions, structure, and sustainability||FIA data has been used widely with the LANDIS forest simulation model to provide information about current forest conditions and to calibrate and validate model initialization and growth parameters. The USDA Forest Service Northern Research Station is now pursuing improved integration between the LANDIS modeling community and the FIA research program to draw on the strengths of these two lines of research.||Forest simulation models allow users to project potential forest conditions into the future under different conditions and scenarios. Forest inventory and analysis (FIA) data provides rigorous information about past forest conditions. When used in combination, FIA data and forest simulations models, such as LANDIS, can utilize the information from the recent past to inform the projections of future conditions. FIA data has been used widely in LANDIS simulation studies to provide information about current forest conditions and to calibrate and validate model initialization and growth parameters. The USDA Forest Service Northern Research Station is now pursuing improved integration between the LANDIS modeling community and the FIA research program to draw on the strengths of these two lines of research. FIA data can be used to develop seamless forest condition maps meeting the specific model requirements for the simulation starting conditions, and for statistically rigorous evaluation of model behavior at different spatial scales utilizing the FIA data structure. In turn, the LANDIS model has the potential to inform aspects of the FIA carbon accounting by modeling growth and disturbance processed between plot measurement years.|
|Robert McGaughey||Overview of Digital Aerial Photogrammetry||Wednesday||1c||Cherokee||Use of photogrammetrically-derived 3D point clouds to support large-area forest inventory and monitoring||Digital aerial photogrammetry (DAP) is a technique that derives 3D information from overlapping aerial images. DAP products include digital surface models and 3D point clouds. The combination of broad-scale image acquisition programs such as the National Agriculture Imagery Program (NAIP), DAP-derived data products, and existing terrain data offer the ability to characterize vegetation height and canopy cover over large areas for relatively low cost.||Digital aerial photogrammetry (DAP) is a technique that derives 3D information from overlapping aerial images. The required overlap can be obtained using a series of images captured with frame-based cameras or from the multiple look angles captured by continuous scanners (push-broom sensors). DAP products include digital surface models and 3D point clouds. DAP-derived data is often compared to lidar. However, DAP can only produce measurements for areas in an image that are well illuminated and that contain enough contrast to allow image matching to work. This means that DAP-derived data products can contain gaps and only represent the ground surface in open areas. Combined with USGS 3DEP elevation data, DAP-derived data products offer the ability to characterize vegetation height and canopy cover over large areas for relatively low cost. Such information has many applications in the FIA context including pre-field conditions assessment, improving the precision of county-level estimates, and reducing bias associated with non-sampled plots.
This presentation will present an overview of DAP technology, describe the basic data products, and compare the technology to lidar.
|James Ellenwood||The potential for augmenting statewide forest estimations with canopy height from remotely sensed products the status of national programs||Wednesday||1c||Cherokee||Use of photogrammetrically-derived 3D point clouds to support large-area forest inventory and monitoring||Lidar and digital photogrametric data is widely available and it is possible to produce a nationwide Canopy Height Model (CHM) at a 1-m resolution. Benefits to the Forest Inventory and Analysis program include the potential to improve Image-based Change Estimation (ICE) and pre-field efficiencies; the potential to incorporate better measurements in non-forest areas such as agriculture and urban forest areas; and improved small area estimation and modeling.||Since the initial development of LIDAR and digital aerial photogrammetric technologies, their use has been limited to local project assessments and analysis. In recent years, a number of developments have increased the potential for the use of these datasets for larger-area assessments. The National 3D Elevation Program (3DEP) allows for the cooperation among federal agencies and partners to join forces in funding common and adjacent areas to improve project costs and gather more usable data. The National Agriculture Imagery Program (NAIP) allows for the optional purchase of the digital aerial photogrammetric point clouds, as of 2018. The University of Minnesota Polar Geospatial Center is conducting a large-area digital photogrammetric project based upon all available US Government purchased imagery from Digitalglobe for global coverage. It is possible to produce a nationwide Canopy Height Model (CHM) at a 1-m resolution. Benefits to the Forest Inventory and Analysis program include the potential to improve Image-based Change Estimation (ICE) and pre-field efficiencies; the potential to incorporate better measurements in non-forest areas such as agriculture and urban forest areas; and improved small area estimation and modeling.|
|Andrew Lister||Comparing the relationship between tree canopy height information from LiDAR, phodar and forest inventory data in northeastern forests||Wednesday||1c||Cherokee||Use of photogrammetrically-derived 3D point clouds to support large-area forest inventory and monitoring||The measurement of tree height with phodar over large areas is relatively recent, and little is known about its usefulness compared to LiDAR. The current study assesses phodars potential by comparing LiDAR and phodar-based height products with height information from forest inventory plots from Connecticut and Maryland. Bivariate correlation between the data types and other metrics will be compared with the goal of assessing the utility of phodar for various FIA business processes.||Detailed tree canopy height information is valuable to natural resource monitoring, science and management specialists. Canopy height is commonly measured from the ground on forest inventory plots, but this information is expensive and prone to error. Alternatively, 3-d models of canopy height and structure can be made using aerial LiDAR or stereo pairs of digital aerial photographs, sometimes referred to as phodar. The operational use of phodar over large areas is relatively recent, and there is thus a gap in the understanding of how phodar compares to LiDAR or ground measurements with respect to producing useful canopy height products. The current study seeks to fill this gap by comparing data from wall-to-wall LiDAR and phodar-based height products with height information from US Forest Service Forest Inventory and Analysis (FIA) plots in the northeastern states of Connecticut and Maryland. Bivariate correlation between the data types and other metrics will be compared with the goal of assessing the utility of phodar for various FIA business processes. The goal of this pilot analysis is to assess the potential for this new technology to provide useful 3-d vegetation structure information in northeastern forests.|
|Benjamin Branoff||Evaluating the potential of NAIP point clouds to support operational forest inventory applications in the southeastern U.S.||Wednesday||1c||Cherokee||In this presentation we evaluate the quality of 3D point clouds developed through photogrammetric processing of National Agriculture Imagery Program (NAIP) stereo imagery. NAIP canopy height surfaces for Tennessee and Virginia are compared with measured tree heights collected in the field by FIA, and with canopy heights developed from airborne Lidar data. Issues affecting data quality are evaluated and used to recommend standards needed for broader use of 3D point clouds within the FIA program.||The U.S. Forest Inventory and Analysis (FIA) program regularly collects tree and stand-level data to produce statistical estimates which are used to analyze the current status and condition of Americas forests. Land cover data from the National Land Cover Database (NLCD) are typically used to post-stratify these estimates but the information in these maps is often more related to forest area than tree density or size, thus there is a need for alternative products that can help operationally improve estimates of forest structure-related variables like volume and biomass. In this presentation we evaluate the quality of 3D point clouds developed through photogrammetric processing of high resolution stereo imagery collected by the National Agriculture Imagery Program (NAIP). NAIP canopy height surfaces developed for the states of Tennessee and Virginia are compared with measured tree heights collected in the field by FIA, as well as with canopy height products developed with airborne Lidar data. Issues affecting data quality such as image acquisition date, post-processing techniques and precision of field GPS locations are evaluated and used to recommend standards needed for broader use of 3D point clouds within the FIA program.|
|Jacob Strunk||An Examination of DAP, Landsat, and Environmental Variables for Modeling Volume, Biomass, and Carbon on FIA Plots||Wednesday||1c||Cherokee||Digital Aerial Photogrammetry (DAP) for WA State was evaluated in combination with Landsat and environmental gradients to model forest volume, live biomass, and live carbon. For 570 plots in Washington State we used principal components, linear regression, and variable selection to evaluate relationships among data types. Initial results indicate that use of multiple auxiliary datasets enables superior mapping products to support forest monitoring, management, and planning efforts.||Digital Aerial Photogrammetry (DAP) has recently received increased interest for forest inventory augmentation due to its low cost and availability over large areas including the entire conterminous USA every 2-3 years. There are additional opportunities for large area inventory augmentation, such as the integration of DAP with satellite imagery (e.g., Landsat multispectral data) and environmental gradients (e.g., slope, aspect, elevation, and temperature and precipitation attributes). This study explores the relationships between DAP, Landsat, and environmental gradients and contrasts their abilities to independently and jointly explain variation in measured tree volume, aboveground live biomass, and live carbon on Forest Inventory and Analysis plots. For 570 plots in Washington State, we looked at associations between principal components for the three auxiliary datasets, fit linear models for volume, biomass, and carbon to principal components, and performed a variable selection exercise with all subsets regressions. Initial results indicate that integration of multiple sources of auxiliary information in forest mapping efforts enables superior products to support forest monitoring, management, and planning efforts.|
|Nick Eliopoulos||Estimation of Tree Diameter at Breast Height Using Close Range Stereo Photogrammetry||Wednesday||1c||Cherokee||Video footage taken using a stereo camera was used to report the diameter at breast height. Our algorithm involves performing a frame-by-frame analysis of each image in the video footage to report a diameter at breast height. Depth information from each frame is extracted and interpreted independently without generating a point cloud. Our method reported a diameter at breast height root mean square error of 1.32 cm over 40 trees, with footage taken 3 meters away from each tree.||Forestry inventory analysis is time-consuming and expensive. Contemporary solutions such as terrestrial laser scanning are not convenient for small-scale landowners due to their cost. State of the art solutions involving the use of stereo photogrammetry have the advantage of being mobile, relatively low-cost, and do not require training to use. Our method captures the mobility and low-cost benefits of stereo photogrammetry, while surpassing diameter at breast height accuracy compared to similar groups. Our improvement is an algorithm that is used with stereo footage to report diameter at breast height. Two types of video footage were recorded for use in our algorithm: video captured standing still, and video captured in motion walking through a plot. The best diameter at breast height root mean square error reported for video standing still was 1.32 cm over 40 trees. Our algorithm produced a root mean square error of 1.11 cm on the video captured in motion, which included 18 trees. Footage taken standing still over 20 trees took 5 minutes to record, but only 28.7 seconds for our algorithm to report diameters for each tree.|
|Demetrios Gatziolis||Evaluating the utility of pushbroom photogrammetry-derived point clouds for estimating tree canopy cover||Wednesday||1c||Cherokee||Tree canopy cover is a parameter challenging to measure in the field, yet integral to many forest inventory operations and data analyses. Remotely sensed data conducive to an accurate and precise estimate of cover, namely LiDAR, are typically too costly, especially over large areas. Digital Aerial Photogrammetry (DAP) for NAIP stereo imagery has emerged as a potentially economically feasible alternative. We evaluate the potential of DAP for tree canopy cover estimation across Washington State.||Tree canopy cover is an important biological and ecological parameter often used as a criterion for land classification and other purposes. Definitions of forestland, a critical parameter in assessing rates of forest gain, loss and degradation, are based on a minimum cover threshold (e.g. 10 percent). Because it is time consuming and challenging to measure it with acceptable accuracy and precision during field visits of inventory plots, canopy cover is often estimated via remote sensing. LiDAR data arguably yield the best estimates, but their acquisition cost often leads to sporadic availability. Manual, photointerpretation-based estimates from airborne imagery, such as the one acquired periodically by the NAIP Program for the continental US, require substantial analyst involvement and are susceptible to overestimation owing to the wide field of view and minimal overlap between image swaths. Limitations to photo-interpretation can be potentially overcome by Digital Aerial Photogrammetry (DAP) for NAIP stereo imagery, with tree canopy cover estimates obtained by processing the point clouds generated by DAP. We evaluate this potential with dense LiDAR point clouds co-temporal to the NAIP imagery in the State of Washington.|