|Laurel Haavik||Historical activity of major pests in the Lake States||Tuesday||2a||Salon D/E||Damage, Disease, Disturbance: Forest Health Research with FIA||This project integrates Forest Health Highlights, Forest Health Monitoring, and Forest Inventory and Analyses data to create a series of technical reports and story maps that summarize and synthesize the historical patterns of major pests in the Lake States.||Our goal is to create a series of technical reports and story maps that summarize the historical patterns of major pests in the Lake States. Detailed information on insect activity has been collected by state agencies (Forest Health Highlights) and the USDA Forest Service (Forest Health Monitoring, Forest Inventory and Analysis) since the 1950s. Although these data and observations have been carefully documented in annual and periodic reports, the existing format is not easy to use. Historical patterns in insect activity and concurrent forest conditions can be a useful predictive tool for future forest conditions. We will synthesize this information and present it in a visually intuitive format that forest managers can readily use. Each installment in the series will include a different major pest, with brief sections on basic biology and ecology, historical activity, forest conditions, and management implications. The series will focus most on patterns of insect activity over time in relation to changing forest conditions. The first installment is spruce budworm, Choristoneura fumiferana, a native insect with a well-documented history of outbreaks affecting balsam fir and white spruce in the Lake States.|
|Sarah Jovan||Lichen and Moss Metrics for Monitoring Air Quality in Wilderness and Urban Areas||Tuesday||2a||Salon D/E||Damage, Disease, Disturbance: Forest Health Research with FIA||Land managers commonly use lichen bio-indicator data to supplement instrument-based air monitoring networks. Client priorities focus on environments where our typical community-based metrics dont work, like high elevation Wilderness and urban forests. Recent research uses low cost assays of pollutants in lichen and moss tissue as an alternative. Three ongoing studies are highlighted. In a nutshell, results help decision-makers allocate scarce resources, like air instruments, more effectively.||Its widely acknowledged that monitoring mandates under the Clean Air Act (CAA) arent well met using existing instrument-based networks. Federal land managers (FLMs) in the West often supplement with bio-indicator datasets to gain smaller-scale perspectives on air quality, including the 9,000+ lichen community surveys collected by FIA and NFS since 1989. Stakeholder priorities, however, increasingly focus on air quality in high elevation ecosystems and urban forests. Both environments are hostile to most lichens, making our typical community-based metrics unreliable. Three studies are underway to develop the use of low cost chemical assays of lichen and moss tissue as an alternative: 1) The CAA tasks FLMs with monitoring air quality in Wilderness, many acres of which lie above the elevation limit for epiphytic lichens. We collaborate with R6 NFS to establish an N mapping protocol for their Wilderness Monitoring Plan. 2) We also work with air regulators, state, city, and municipal organizations in Portland, OR, to map air toxics regulated under the CAA using a hardy moss. Results guide placement of limited air instruments. 3) A 1-yr urban study comparing pollutants in moss to PM10 and precipitation is also underway.|
|KaDonna Randolph||The rise and fall of individual and multifactor damage agents in forests of the United States: a retrospective look at FIA damage collection protocols||Tuesday||2a||Salon D/E||Damage, Disease, Disturbance: Forest Health Research with FIA||The FIA Program has included damage collection protocols in its inventory since the initiation of forest surveys in the 1930s. These protocols have naturally undergone revisions over time as information needs and technologies for data collection, storage, and analysis have evolved. An examination of the protocols utilized at various points in history highlights shifts in the agents considered to be most important.||The Forest Inventory and Analysis (FIA) Program of the U.S. Department of Agriculture, Forest Service, has included damage collection protocols in its inventory since the initiation of forest surveys in the 1930s. These protocols have naturally undergone revisions over time as information needs and technologies for data collection, storage, and analysis have evolved. Although changes in the damage collection protocols may impede long-term trend analyses, an examination of the protocols utilized at various points in history highlights shifts in the agents considered to be most important. Many agents such as dwarf mistletoe and fusiform rust have persisted in importance over the years whereas other agents, e.g., turpentining, are no longer of great concern. Agents on the rise include declines, diebacks, and wilts resulting from multiple interacting factors and individual agents introduced into the U.S. or moved outside their native ranges within the U.S.|
|Mark Ambrose||Using EVALIDator to Report on National Scale Mortality Trends for the Forest Health Monitoring Annual Report||Tuesday||2a||Salon D/E||Damage, Disease, Disturbance: Forest Health Research with FIA||In recent years EVALIDator has been used to report on mortality for the Forest Health Monitoring national report. EVALIDator output by ecoregion section has been used to identify regions of highest mortality, forest types affected, and major mortality-causing agents. Examples of analyses from the most recent FHM national report will be presented, as well as other uses of EVALIDator that may provide greater insight into tree mortality trends.||Since 2001, the national Forest Health Monitoring Program (FHM) has produced an annual report on the forest health status and trends across the US using monitoring data from a variety of sources. The method of analyzing mortality has changed several times since the first FHM report. Most recently, EVALIDator has been used to produce growth and mortality estimates by ecoregion section for the most recent cycle of measurements for each State and to identify regions with unusually high mortality. EVALIDator output also has been used to identify the forest types suffering the greatest mortality and major mortality-causing agents. Limitations on the use of EVALIDator for national-scale analyses are failures when the system times-out waiting for a response. In this talk, examples of analyses from the most recent FHM national report will be presented, as well as other uses of EVALIDator that may provide greater insight into tree mortality trends.|
|Zackary Holden||Using climate and biophysical attributes to predict the distribution of root disease across the U.S. Northern Rocky Mountains||Tuesday||2a||Salon D/E||Damage, Disease, Disturbance: Forest Health Research with FIA||We used a database of 15,000 root disease measurements at FIA plots collected across the U.S. Northern Rocky mountains coupled with high resolution (30 m) gridded climate and soil water balance datasets to examine the climatic and biophysical factors associated with presence and severity of root disease. Root disease occurrence was moderately well explained by dewpoint temperature, evapotranspiration and climatic water deficit (AUC = 0.77) with higher root disease prevalence at moist, humid site||We used a database of more than 15,000 root disease measurements at FIA plots collected across the U.S. Northern Rocky mountains coupled with high resolution (30 m) gridded climate and soil water balance datasets to examine the climatic and biophysical factors associated with presence and severity of root disease. Root disease occurrence was moderately well explained by dewpoint temperature, evapotranspiration and climatic water deficit (AUC = 0.77) with higher root disease prevalence at moist, humid sites with low or moderate soil water balance deficits. Root disease severity was not well explained by climatic and biophysical factors alone (R2 = 0.19). Variogram analysis indicated little spatial autocorrelation beyond a 5 km range, suggesting that the distribution of FIA plot samples may not be well suited for capturing the spatial processes governing root disease spread. We used the fitted models to estimate root disease probability of occurrence and severity across the US Forest Service Northern Region and to produce spatially explicit maps that will be used to help identify and quantify potential impacts and help prioritize areas for broad scale planning and analysis.|
|Sunil Nepal||The relationship of 19th century vegetation to 21st century oak decline and mortality in Missouri Ozark Highlands||Tuesday||2a||Salon D/E||Damage, Disease, Disturbance: Forest Health Research with FIA||Recurrent droughts and high stand stocking levels are often blamed for oak decline and mortality in the Ozark Highlands. Historical land use change, especially changes in species composition and structure oftentimes overlooked. We have examined the impacts of historical vegetation conditions in current oak decline and mortality by using early nineteenth-century trees records and current FIA data. The highest red oaks mortality was found in historical oak/pine closed woodland forest condition.||To examine the role of historical vegetation condition change in current oak decline and mortality, we investigated annual FIA data and early nineteenth-century public land survey (PLS) in the Missouri Ozark Highlands. Historical tree data revealed that white oaks were the most frequent species (50%), followed by red oaks (30%), shortleaf pine (8%), hickories (5%), and other species (7%). At present, white oaks, red oaks, and pine decreased to 34, 25, and 6% respectively, but hickories and other species increased to 10 and 25% respectively. Changes in species composition varied among historical vegetation conditions. Hickory and other species increased in all conditions. In contrast, oaks decreased in oak closed woodland, oak forest, oak open woodland and oak savanna, and increased in Oak/pine closed woodland, oak/pine forest, and oak/pine open woodland. Red oaks mortality reached up to 2.6% per year which was significantly higher than other species. The highest red oak mortality was found in the Oak/pine closed woodland. In the future, we plan to develop a modeling system that includes all factors contributing to oak mortality at different spatial scales to aid in the development of applicable mitigation methods of oak decline.|
|Charles Perry||Can I get fries with that? Successes and continued challenges for BIGMAP||Wednesday||2a||Salon D/E||Large-scale computing applications in FIA||FIA entered into a partnership to deliver BIGMAP (Big Data Mapping and Analytics Platform): a cloud-based system for scalable production, analysis, and delivery of integrated geospatial products. Authoritative, science-based outputs support decision-making for private and public land management planning, resource allocation, and landscape sustainability. We will also review persisting challenges associated with partnerships and IT applications in a rapidly evolving political environment.||The Forest Inventory and Analysis (FIA) Program entered into a public:private partnership with the USDA FS CIO Geospatial Services Branch and Esri, Inc. to configure BIGMAP: the Big Data Mapping and Analytics Platform. We are entering the final year of our four-year agreement to deliver a prototype cloud-based system for scalable production, analysis, and on-the-fly delivery of massive, integrated, geospatial products. This solution is notable for its use of a commercial off-the-shelf (COTS) software solutions in a secure cloud environment. The resulting authoritative science-based products are designed to support decision-making for private and public land management planning, resource allocation, and landscape sustainability assessments. BIGMAP is designed to address several agency priorities, and this presentation will highlight several applications available for hands-on exploration during the digital engagement session. We will also review several persisting challenges associated with this type of partnership and IT applications in a rapidly evolving political environment.|
|Barry Wilson||Development of BIGMAP: a cloud-based national scale modeling, mapping, and analysis environment using Esri Raster Analytics||Wednesday||2a||Salon D/E||Large-scale computing applications in FIA||This presentation describes efforts to configure BIGMAP, a cloud-based national scale modeling and analysis environment built upon Esri software in Amazon's cloud. A pilot project was designed for testing BIGMAP with a set of use cases covering the range functionality required. This presentation focuses on use cases to support for national carbon reporting and small area estimation. It provides an overview of the architecture and configuration, example workflows and outputs, and lessons learned.||With the burgeoning remote sensing sector and the growing volume of imagery available at finer resolutions, new approaches are needed to support FIAs techniques research and national scale applications. Cloud-based solutions provide the flexibility to marshal the computing resources needed, and place algorithms and data together in a common environment.
This presentation summarizes efforts to configure the BIGMAP environment, built upon Esri software utilizing Amazon Web Services (AWS). A pilot project was designed to test this environment with a set of use cases intended to cover much of the range of functionality required. This presentation focuses on two of these use cases: a) support for national carbon reporting, and b) small area estimation.
The presenters will provide a brief overview of Esri Raster Analytics. They will discuss the process of architecting the AWS infrastructure and subsequent software configuration and performance testing of environment. They will address the technical challenges of taking use case workflows that were originally developed under different programming paradigms and adapting them to fit within BIGMAP. Finally, they will present some of the outputs produced to date and lessons learned.
|Sean Healey||Cloud computing supporting land cover change mapping||Wednesday||2a||Salon D/E||Large-scale computing applications in FIA||We highlight uses of a cloud platform for: detecting forest disturbances in the US; mapping land cover and use change across East Africa; and, merging a global sample of lidar data with wall-to-wall image data from Landsat to make statistical estimates of biomass over user-requested areas. Were learning lessons about forest cover and land use change that would have been inconceivable without huge expense just five years ago.||Since the beginning of the Landsat program in the 1970s, scientists have been using the platforms imagery to map land cover and land cover change. When the Landsat archive became free in 2008, algorithms using annual or even all acquisitions began to proliferate, with demonstrable benefits in sensitivity and accuracy. However, these algorithms have only begun to reach their potential as cloud computing has become more accessible. Several developments have driven the entire remote sensing field toward cloud platforms: 1) emerging evidence of the effectiveness of new, computationally intensive algorithms; 2) the sheer number of images now available; 3) the need to share ideas and code across increasingly diverse teams, and 4) the need for on-the-fly access to imagery for new user engagement applications.
We highlight uses of a cloud platform for: detecting forest disturbances in the US; mapping land cover and use change across East Africa; and, merging a global sample of lidar data with wall-to-wall image data from Landsat to make statistical estimates of biomass over user-requested areas. Were learning lessons about forest cover and land use change that would have been inconceivable without huge expense just five years ago.
|Kasey Legaard||Use of multi-objective machine learning and high performance computing to reduce prediction bias in forest maps||Wednesday||2a||Salon D/E||Large-scale computing applications in FIA||We have developed machine learning techniques that reduce systematic error when mapping forest attributes from multispectral satellite imagery and FIA plot data. Our approach combines support vector machines with a genetic algorithm to simultaneously minimize total and systematic error. Large-scale applications are supported through the use of modular and highly parallelized software executed on University of Maine cloud computing resources and computing clusters.||Methods used to produce maps from satellite imagery and forest inventory plot data generally result in patterns of error that are detrimental to many applications. We have developed machine learning techniques that reduce systematic error when mapping forest attributes from multispectral satellite imagery and geospatial data. Our approach is based on the optimization of support vector machines using a multi-objective genetic algorithm designed to simultaneously minimize total and systematic error. Using FIA plot data we have mapped tree species occurrence and relative abundance across Maine and obtained outcomes that compare well against other approaches. Our methods are highly generalizable and automatable, but also computationally demanding. Large-scale applications require use of high performance computing resources. We have developed efficient software for use on University of Maine cyberinfrastructure, including modular image processing and data handling workflows implemented on the cloud, and highly parallelized model training and map production code executed on a computing cluster. Our intention is to support a broad set of research and management applications through the production of high-quality spatial data at low cost.|
|Tony Chang||Utilizing cloud-computing for implementing a deep convolutional neural network based forest structure and classification model||Wednesday||2a||Salon D/E||Large-scale computing applications in FIA||Development and implementation of new biomass/carbon models require increased computational hardware that include large amounts of memory and access to GPUs. Contemporary cloud-compute platforms allow for such computational loads for parameter fitting state-of-the-art deep learning models. This research demonstrates a cloud-based approach to perform both classification and regression concurrently based on a large FIA training data set and these next generation models.||In recent years, the increased availability of high-resolution (<30 m) imagery and advancements in machine learning algorithms have opened up a new opportunity to fuse multiple datasets of varying spatial, spectral, and temporal resolutions. Here, we present a new model, based on a deep learning architecture, that performs both classification and regression concurrently, thereby consolidating what was previously several independent tasks and models into one stream. The model, a multi-task recurrent convolutional neural network that we call the Chimera, integrates varying resolution, freely available aerial and satellite imagery, as well as relevant environmental factors (e.g., climate, terrain) to simultaneously classify five forest cover types (conifer, deciduous, mixed, dead, none (non-forest)) and to estimate four continuous forest structure metrics (above ground biomass, quadratic mean diameter, basal area, canopy cover). We demonstrate the models ability and performance on the Azure cloud-computing platform for model training on 9967 georeferenced FIA field plots within California and Nevada, and then implement a large scale prediction for the Sierra Nevada region to understand the viability of this new approach.|
|David Bell||An integrated disturbance and vegetation mapping fra,ework highlights landscape level changes in forest dynamics during a multiyear drought||Wednesday||2a||Salon D/E||Large-scale computing applications in FIA||NEED||NEED|
|Rebecca Tavani||The importance of NFIs in the international context||Thursday||2a||Salon D/E||Global view of national forest inventories (NFIs): how they have progressed, ways they have utilized partnerships, and possibilities for the future.||Rebecca Tavani is a Forestry Officer working for FAO in support of national forest inventories. She has 12 years of experience in providing technical and programmatic NFI support to countries in the African region (e.g. The Gambia, Uganda, Liberia, Ethiopia, Zambia) under the National Forest Monitoring and Assessment (NFMA) & REDD+ programmes at FAO. She has a B.A. in Biology from Brown University and a Masters of Forestry from the Yale School of Forestry and Environmental Studies (FES).||For more than 50 years, FAO has been supporting countries to collect forest information through national forest inventories (NFI) that meet national and international information needs operating on the premise that better information leads to improved decisions, which leads to more effective action in the forest sector and beyond. Today, FAO is supporting multipurpose NFIs in over 20 countries. The data collected has supported countries in both national and international reporting, particularly in meeting the reporting requirements of the Paris Agreement. NFI data has been critical to the construction of national & sub-national forest reference levels as well as in providing improved biomass estimates for updating IPCC default emission factors. In this presentation, current FAO support to global NFI efforts will be presented focusing on the application of the data to REDD+ processes and related challenges.|
|Andrew Lister||Progress on National Forest Inventories||Thursday||2a||Salon D/E||Global view of national forest inventories (NFIs): how they have progressed, ways they have utilized partnerships, and possibilities for the future.||The panel will discuss elements of the progress of the participants national forest inventories. These include technical challenges and solutions; efforts to achieve institutionalization; management of data processing, reporting and information technology; and use of technologies like data recorders and remote sensing to help improve the inventories. Other topics will include how to manage change and adjust inventory design as lessons are learned.||See summary -- this is for the panel 2 of the international session|
|Andrew Lister||Thursday||2a||Salon D/E||Global view of national forest inventories (NFIs): how they have progressed, ways they have utilized partnerships, and possibilities for the future.|
|Kaffo Eric||State of the National Forest Inventories in Cameroon: challenges and perspectives||Thursday||2a||Salon D/E||Cameroon carried out its last national inventory of its forest resources with the technical support of the FAO between 2002 to 2004, FAO. The process covered only 45% of the country's area. As part of regular small inventories for forest concessions and other small areas, Cameroon is planning to realize a new national forest inventory which will help the country updating most of the forest data but also provide information for the REDD+ process.||Cameroon carried out a national inventory of its forest resources with the technical support of the Canadian International Development Agency (CIDA) between 1982 and 1990 on about 14 million hectares, or 29.47% of the territory. From 2002 to 2004, FAO decided to support a second national forest inventory, which covered 45% of the country's area this time around.
The methodology used in these inventories was based on stratified sampling. As part of the work, partnerships were established with academic and research institutions, as well as the national herbarium. As the main results, the FAO NFI helped for development of volume rates for the estimation of standing tree volumes as well as updating the phytogeographic map of Cameroon.
In addition to these inventories, others are carried out on a regular basis on smaller areas, including management inventories, as well as exploitation inventories, in each parcel of forest to be logged. Unfortunatly, most of the data are out of date and qualified personnel are retired.
As prospects, it is envisaged to the realization of a new forest inventory to update the volume rates, and estimate the the biomass, as well as carbon stocks in connection with the REDD + mechanism.
|Sara Goeking||Partnerships in national forest inventories: Benefits, challenges, and characteristics||Thursday||2a||Salon D/E||This panel will discuss partnerships in national forest inventories (NFIs), which often involve cooperation among multiple in-country institutions, neighboring countries, and global organizations. Panelists will discuss benefits and challenges derived from partnerships, including their effects on information sharing and data transparency. Other topics include decision-making for international reporting and strategies for engaging stakeholders who can offer critical feedback to the NFI.||See summary. This Summary applies to a single Panel discussion, labeled as Panel 1.a and 1.b in the session submission.|
|Sara Goeking||Thursday||2a||Salon D/E|
|Sonja Oswalt||Forest Resources of the United States, 2017: Informing the Nation, the Continent, and the Globe||Thursday||2a||Salon D/E||This presentation shares the outcome of the 2017 U.S. Forest Resources Report, progress on the North American Forest Database as well as lessons learned, and the process by which the United States participates in the global Forest Resources Report.||The FIA program contributes statistics to the federally mandated Resources Planning Act every 10 years, with 5 year updates. These statistics contribute not only to domestic policy action, but have international implications, as well. Currently, through the work of the North American Forest Commission, forestry agencies in the U.S., Canada, and Mexico have worked to align data definitions from their respective forest inventories in a manner sufficient for combination in a continent-wide database. Additionally, FIA statistics continue to feed into the United Nations Food and Agriculture Organization (UN FAO) report on global forest resources. This presentation shares the outcome of the 2017 U.S. Forest Resources Report, progress on the North American Forest Database as well as lessons learned, and the process by which the United States participates in the global Forest Resources Report.|
|Dooley Kerry||Possibilities for national forest inventories||Thursday||2a||Salon D/E||This panel discusses possibilities for national forest inventories (NFIs) growing into the future. Discussions will include how to manage competing demands of producing consistent data over time versus incorporating data of emerging importance, including how to anticipate changing demands and building in resilience for inventories. Panelists will share specific examples of challenges they have overcome; how NFIs improve research, management or policies; and their hopes for NFIs in the future.||See summary|
|Dooley Kerry||Thursday||2a||Salon D/E|
|Aurea Erica Aponte||"The challenge of the Peruvian State in leading the conservation of relict forests of Polylepis through the National Forestry and Wildlife Inventory: Challenges and opportunities to make decisions"||Thursday||2a||Salon D/E||In recent years, communities around the world have had a central role in local forest management and ecosystem conservation. In the Andes of Peru, peasant communities have a long tradition of forest management based on a particular world view. The information generated by the National Forestry and Wildlife Inventory (Inventario Nacional Forestal y de Fauna Silvestre -INFFS in Spanish) provides information on the current situation of the relict forests of Polylepis "Queuña."||In recent years, communities around the world have had a central role in local forest management and ecosystem conservation. In the Andes of Peru, peasant communities have a long tradition of forest management based on a particular world view. The information generated by the National Forestry and Wildlife Inventory (Inventario Nacional Forestal y de Fauna Silvestre -INFFS in Spanish) provides information on the current situation of the relict forests of Polylepis "Queuña." In this context, the Government of Peru, through SERFOR, implements the INFFS in all forests. Through 1855 sample units, of which 489 are in the Andes ecozone, mountain forests containing "queñuales" (local name for Polylepis trees) are examined. The challenge is to characterize the composition, structure, state of conservation and benefits that the Polylepis forests generate for rural communities. The methodology consists in organizing permanent sample units in "L" shaped conglomerates of ten circular sampling subunits of 500 m2 each, where trees larger than 10cm dap are measured. Currently, 67 Sample Units (SU) have been evaluated, of which Polylepis trees are found in 13 SU. Additionally, it should be noted that in Peru there are approximately 21 Polylepis|
|Dooley Kerry||International session Wrap-up||Thursday||2a||Salon D/E|