User Conference 2011

From Syncrosim
Jump to: navigation, search

Presentations from the 2011 Landscape State-and-Transition Simulation Modeling Conference


The first ever Landscape State-and-Transition Simulation Modeling Conference was held from June 14-16, 2011 in Portland Oregon. The conference was hosted by Apex Resource Management Solutions and ESSA Technologies, in collaboration with the U.S. Forest Service, The Nature Conservancy, the Oregon University System's Institute for Natural Resources, and the LANDFIRE project. The conference brought together users of state-and-transition simulation models in an effort to provide opportunities for sharing experiences with different applications of the available tools.

Proceedings of the conference were published in 2012 as a U.S. Forest Service General Technical Report.


Oral Presentations

Forecasting the Benefits of Conservation Strategies for Land Managers

Greg Low

Applied Conservation Inc.

Greg Low View presentation video View presentation slides (PDF)

“If you had a dollar to spend on conserving the ecological systems in a landscape, where would you invest it first? To help land managers answer this question, staff from Applied Conservation and The Nature Conservancy’s Nevada office have deployed Landscape Conservation Forecasting ™ (aka Enhanced Conservation Action Planning) at national forests, national parks and BLM lands in four states from California to Tennessee. The methods include: (1) using satellite imagery to assess the health of ecological systems; (2) employing ecological models using VDDT to demonstrate how those systems will change over time; (3) utilizing VDDT simulations to assess how alternative management actions can influence those changes; (4) measuring success by calculating an ecosystem’s departure from its natural range of variability, on a scale of 1 – 100, with and without various management actions; and (5) considering the cost of each strategy in order to help land managers prioritize their on-the-ground actions to get the highest conservation return on investment. Conservation forecasting – when used by any land management agency – identifies strategies that can be easily embedded in the agency’s plans, provides a science-based foundation for its NEPA documents, and positions the agency to be eligible for additional funding sources.

Evaluating the Costs and Benefits of Alternative Weed Management Strategies for Three Montana Landscapes

Brad Bauer2, Leonardo Frid1 (presenter) , David Hanna2, Nathan Korb2, Katy Bryan1, Brian Martin2, Brett Holzer3

1ESSA Technologies Ltd.; 2The Nature Conservancy, Montana; 3Private

Leonardo Frid View presentation video View presentation slides (PDF)

Invasive plant species management at the landscape scale in the Western U.S. is generally based on fine-scale experience and arbitrary decisions (“rules of thumb”) with limited understanding of long-term outcomes across broad areas or over long periods. In order to develop the best strategies to maintain landscape values and prevent the spread of invaders, quantitative tools are needed to compare the effectiveness of various management strategies at different spatial scales and over several decades. We used TELSA to model the spread of invasive species at the landscape scale for three Montana landscapes. Using TELSA allowed the comparison of several management strategies under a variety of budget constraints to evaluate the long-term advantages of different approaches to native plant communities, identify appropriate resource allocation levels to maximize benefit, and assess costs and benefits of strategies within an economic analysis framework. The model results provide a tool to guide real-world decision making across large landscapes.

Modeling on the Grand Scale: LANDFIRE Lessons Learned

Kori Blankenship1, Jim Smith1, Randy Swaty1, Ayn Shlisky2, Jeannie Patton1, Sarah Hagen1

1The Nature Conservancy; 2USDA Forest Service Pacific Northwest Region, Portland, OR

Kori Blankenship View presentation video View presentation slides (PDF)

Between 2004 and 2009, the LANDFIRE project developed state-and-transition models for all major Ecological Systems in the United States using the Vegetation Dynamics Development Tool (VDDT). To complete this effort, LANDFIRE staff: hosted modeling workshops, web conferences and individual meetings across the country, engaged more than 700 experts in developing and reviewing models, taught experts and managers how to model with VDDT, and facilitated the creation of over 2,000 models quantifying pre-settlement reference conditions. An undertaking of this type invariably has highs and lows, and provides an opportunity to learn about the model building and model review process. In this presentation, we will discuss what we learned from this modeling odyssey that may help others efficiently build more useful models in the future.

Using TELSA to Model Regional Vegetation Dynamics in the Southeastern US

Jen Costanza1, Todd Earnhardt1 (presenter) , Adam Terando1, Alexa McKerrow1,2

1Biodiversity and Spatial Information Center, NC State University, Raleigh, NC; 2US Geological Survey Core Science Systems, Raleigh, NC

Todd Earnhardt View presentation video View presentation slides (PDF)

Predicting future landscape change is essential for developing robust long-term conservation strategies. In the southeastern US, significant change due to rapid urbanization and intensive forest management has impacted ecosystems. These impacts are expected to continue, and are likely to be exacerbated by projected changes in climate. As part of the USGS’s Southeast Regional Assessment Project (SERAP), we are incorporating urban growth, forest management trends, and climate change effects into VDDT and TELSA models of vegetation dynamics across nine-states in the southeastern US through the year 2060. We are generating possible future landscapes and examining how the region may look in the future with respect to land cover. These projections will be used as an input for predicted species distribution modeling and to inform regional conservation decision-making. Here, we discuss our use of existing data from LANDFIRE and the US Forest Service Forest Inventory and Analysis program in our regional vegetation dynamics modeling. We illustrate our derivation of initial conditions, use of fire disturbance area and frequency adjustments, incorporation of projected climate change impacts via changes to fire frequency, incorporation of urbanization projections via polygon forcing, and the inclusion of forest management activities as landscape disturbances. Our most recent results from the Apalachicola-Chattahoochee-Flint River watershed show major changes to the distribution and characteristics of vegetation on the landscape if recent trends continue. Our model outputs highlight the interaction among drivers of change that produce qualitatively different landscapes in the future. These results will be useful for informing species distribution modeling and conservation planning in the Southeast.

Predicting Cheatgrass and Juniper Invasion in Eastern Oregon Shrub Steppe

Megan Kanaga Creutzburg1, Joshua Halofsky2, Miles Hemstrom3

1Institute for Natural Resources; 2Washington State Department of Natural Resources; 3USDA Forest Service Pacific Northwest Research Station

Megan Creutzburg View presentation video View presentation slides (PDF)

Many threats are jeopardizing the shrub steppe of the Columbia Basin, including the spread of invasive species such as cheatgrass (Bromus tectorum) and expansion of Western juniper (Juniperus occidentalis) into shrub steppe beyond its historical range. Native shrub steppe provides important habitat for many wildlife species and is valued for its biodiversity, and land managers are in need of tools to aid in assessing risk of shrub steppe conversion and management options to maintain native shrub steppe. For the Integrated Landscape Assessment Project (ILAP), we use a state-and-transition modeling approach to project changes in shrub steppe vegetation over the next 50 years. We construct models using both empirical data, including empirically-derived fire probabilities, and expert opinion for transitions that are still poorly understood, such as livestock grazing effects. Without aggressive management, cheatgrass is expected to increase in many warm, dry Wyoming big sagebrush communities, and juniper invasion is likely to expand throughout much of the cool, moist mountain big sagebrush habitat where seed sources are readily available. We use our model projections to develop landscape-scale maps of cheatgrass and juniper invasion risk to aid land managers in prioritizing watersheds for restoration treatments. We conclude that predictive state-and-transition models provide a useful framework for conceptualizing vegetation dynamics of shrub steppe systems, for making projections of future vegetation conditions, and for prioritizing areas for management treatments.

Quantifying Conversation: Engaging Experts and Incorporating Uncertainty in Restoration Modelling

Christina A. Czembor, William K. Morris, Brendan A. Wintle, Peter A. Vesk

School of Botany, University of Melbourne, Australia

Christina Czembor View presentation video View presentation slides (PDF)

Simulation models assist in forest restoration by predicting the outcomes of alternative management options. We tackle a real-world case study of forest restoration in Victoria, Australia, by constructing state-and-transition models in VDDT to predict how three alternative management options alter forest structure over time. The lack of empirical ecological data to parameterize models necessitated the use of expert opinion. We constructed models for five participants, from a pool of 19 applicable experts, and found substantial differences among the expert models’ predictions. Uncertainty can affect model predictions, but it is rarely acknowledged in expert models. We quantified prediction variance to determine the relative importance of three sources of uncertainty: disagreement between experts, self-assessed imprecision, and modelled system stochasticity. We found that disagreement between experts contributed more to prediction variance than both self-assessed imprecision and modelled system stochasticity, and that the between-expert parameter was the most uncertain. Using insights from this case study, we provide advice on engaging experts and creating simulation models using expert opinion. We also suggest that modelling strategies relying on a single expert opinion, forced consensus between experts, or that only incorporate uncertainty due to system stochasticity could produce biased models and overconfident predictions. Management decisions based on biased or overconfident predictions can result in inefficient conservation investments and poor outcomes.

Estimating Landscape Potential in Ontario’s Great Lakes-St. Lawrence Forests using VDDT and TELSA

Phil Elkie1, Michael Gluck1, Colin Daniel2, Brian Naylor1, Fred Pinto1, Glenn Watt1, Murray Woods1

1Ontario Ministry of Natural Resources; 2Apex Resource Management Solutions Ltd.

Phil Elkie View presentation video View presentation slides (PDF)

Ontario’s Crown Forest Sustainability Act requires forest managers to plan for the long term health and vigour of Crown forests by practicing forest management that, within the limits of silvicultural requirements, emulates natural disturbances and landscape patterns. We ran stochastic simulations using VDDT and TELSA to estimate the natural potential of managed landscapes within the Great Lakes-St. Lawrence Forest Region of Ontario, Canada (11,864,872 ha, 29,318,737 ac). VDDT combined with TELSA provided us with a unique modeling environment that we were able to customize using 26 forest classes while incorporating crown fire, surface fire, cyclic spruce budworm outbreaks, and wind events. The results of multiple model runs (i.e., the simulated ranges of natural variation of forest composition, pattern and structure) are being used as targets when developing 10 year forest management plans. The simulation results also provide an estimate of the potential of each landscape to provide habitat for several evaluative indicators (e.g., barred owl, black bear, beaver, fisher, Kirtland’s warbler, lynx, marten, moose, northern and southern flying squirrel, northern goshawk, pileated woodpecker, pine warbler, red shouldered hawk, red squirrel, ruffed grouse, snowshoe hare, spruce grouse and wood duck). The simulation results are included as milestones in Ontario’s science based ‘Forest Management Guide for Great Lakes–St. Lawrence Landscapes’ (

The Integrated Landscape Assessment Project

Miles Hemstrom

USDA Forest Service, Pacific Northwest Research Station, Portland, OR

Miles Hemstrom View presentation video View presentation slides (PDF)

The Integrated Landscape Assessment Project (ILAP) was funded in 2009 by the American Reinvestment and Recovery Act. ILAP is building landscape-wide data and models for all major upland ecological systems (except for croplands and urban lands) in Arizona, New Mexico, Oregon, and Washington. ILAP products include: 1) existing vegetation and potential natural vegetation data; 2) state and transition models that allow forecasting potential future conditions under differing management scenarios; 3) interpretations of wildlife habitat, fire and fuel conditions, treatment finances, and community economic impacts that might result from alternative management scenarios; and 4) decision support frameworks. Decision support methods will integrate model projections and other information to allow managers, policy makers, and other interested people to view the potential results of alternative management scenarios in flexible ways, depending on users’ interests. The project also integrates watershed conditions and other resource, social, and economic information into the decision support framework. A climate change module connects mid-scale state and transition models to broader-scale climate change and dynamic vegetation models to allow examination of potential climate change effects on local vegetation, natural disturbances, and resource values in conjunction with the influences of management actions. The models, data, and interpretations will be useful to public land management planners, policy analysts, local and regional restoration collaborators, and many others. Most ILAP products will be delivered by the end of calendar year 2011.

Using TELSA to Model Buffelgrass Spread and Control in the Sonoran Desert

Tracy Holcombe1, Jeff Morisette1, Aaryn Olsson2, Leonardo Frid3

1US Geological Survey; 2Northern Arizona University; 3ESSA Technologies Ltd.

Tracy Holcombe View presentation video View presentation slides (PDF)

Buffelgrass (Pennisetum ciliare) is invading the deserts of the southwestern United States, growing in dense stands and introducing a wildfire risk to an ecosystem not adapted to fire. We used TELSA to develop a simulation model to address this urgent issue and evaluate the potential efficacy of alternative mitigation strategies. To develop a state and transition model for buffelgrass spread and control dynamics we held a workshop with ecologists and land managers in Tucson, Arizona in May, 2010. The model incorporated key aspects of buffelgrass spread and management including: inventory, treatment, and maintenance activities. We tested our model on the Catalina Mountain study area just north of the city of Tucson, AZ. The model was calibrated using a spatial time series of buffelgrass spread. Other inputs to the model include a habitat suitability model and current buffelgrass distribution. We asked about the role of management effectiveness on the outcome of buffelgrass spread over time, developing five simulation scenarios that included combinations of no management, management with unlimited resources, management with limited resources, and highly effective outcomes or ineffective outcomes. Our simulation results show that in the short term, management with limited resources and effective outcomes performed better than the unlimited management with ineffective outcomes. This suggests that if there is a choice to be made between allocating resources to treat more area and increasing the efficiency of treatment for a smaller area, the latter may be better. If buffelgrass patches can be effectively discovered early in the invasion process, the total area to be treated and maintained could be substantially reduced.

Climate Change Considerations in VDDT: Using Output From a Dynamic Global Vegetation Model to Parameterize a State and Transition Model

Becky Kerns1, Miles Hemstrom1, Dominique Bachelet2, David Conklin2, Josh Halofsky3, Michelle Buonopane1, Jessica Halofsky4

1USDA Forest Service Pacific Northwest Research Station; 2Conservation Biology Institute, Oregon State University; 3Washington State Department of Natural Resources; 4 University of Washington

Becky Kerns View presentation video View presentation slides (PDF)

Evaluating how climate change may potentially impact ecosystem sustainability and ecosystem services is a key challenge for land managers. State and transition models can be used to develop management strategies that anticipate potential changes, but not all landscape models explicitly incorporate climate. Our objective was to link output from the Dynamic Global Vegetation Model MC1 to a suite of VDDT models. We used two example landscapes to develop our process: (1) eastern Oregon Cascades; and (2) Arizona’s Mogollon Rim. First we calibrated 30-arcsec MC1 output using historical climate data plus maps and information about existing vegetation. We then developed a crosswalk between the broad MC1 vegetation types (VTs) and VDDT potential vegetation types (PVTs). We created a local calibration for MC1 for each landscape, and used this calibration to run MC1 using future climate data from two A2 GCMs. We aggregated the VDDT PVTs into a single “mega-model” that incorporated representative current and future major PVTs for each landscape. Transition multiplier files were then developed using output from MC1 that: (1) changed fire probabilities annually; and (2) directed annual shifts through time from one VDDT PVT type to another. Output from these VDDT models showed far less landscape change than comparable MC1 output, partially due to inherent inertia in vegetation as modeled by VDDT. While there is considerable uncertainty in our projections, the models allow managers and others to explore some potential outcomes of climate change on landscape-level vegetation dynamics using the VDDT platform.

Landscape Conservation Forecasting for Great Basin National Park

Louis Provencher1, Tanya Anderson1, Greg Low2

1The Nature Conservancy; 2Applied Conservation Inc.

Louis Provencher View presentation video View presentation slides (PDF)

National parks have the dual responsibilities of preserving natural communities and restoring communities degraded by humans. At Great Basin National Park (GBNP), habitat degradation can be traced to fire exclusion and the introduction of non-native species. Natural communities with highly departed ecological conditions can have profound impacts on fire, wildlife and vegetation management. In 2009, GBNP and The Nature Conservancy (TNC) collaborated to: (1) create highly detailed maps of potential natural plant communities, the vegetation classes within each community, and ecological departure from reference condition; and (2) simulate management actions and budgets required to reach reference condition using VDDT and Path. TNC mapped 21 potential natural communities. Nine were slightly departed from the reference condition, 10 were moderately departed, and only 2 smaller systems were highly departed. Ecological departure was caused primarily by sagebrush systems lacking the earliest succession classes and aspen-conifer systems that were over-represented by late succession classes. Eleven communities were not targeted for active management because they were projected by the computer simulations to benefit from periodic wildfires. Ten communities were chosen for simulated management that required multiple strategies over 50 years. Computer simulations of cost-effective management actions achieved lower ecological departure for all 10 focal natural communities. The total cost of implementation was $3,595,268 over 50 years. Many actions were implemented fully in the first years of simulation. The next step is to update simulated management actions given that climate change may affect the natural communities’ ability to reach target conditions.

A Methodology for Using FVS to Calibrate VDDT

Donald C.E. Robinson1, Sarah J. Beukema1, Miles Hemstrom2, Don Vandendriesche3

1ESSA Technologies Ltd., Vancouver, BC; 2USDA Forest Service, Pacific Northwest Research Station, Portland, OR; 3USDA Forest Service, Forest Management Service Center, Ft. Collins, CO

Donald Robinson View presentation video View presentation slides (PDF)

Land managers commonly use FVS and VDDT as planning aids. Although complementary, the models differ in their approach to projection, spatial and temporal resolution, simulation units and required input. When both are used there is a need to harmonize the parameters and dynamics of the two models for purposes of comparison and consistency. We used the transition probability matrix as the basis for comparing the two models, using side-by-side simulations with FVS and VDDT to project 250 mixed conifer stands. We designed and carried out a simulation experiment with managed and unmanaged scenarios, to: (1) explore the consequences of different approaches to smoothing out side-effects such as jitter, which results from creating categories from continuous data; and (2) create more robust FVS outputs by removing censored and rare observations. Our analysis includes verification of the Preside system, comparison of matrix behavior, as well as the behavior of VDDT with FVS runs imported into VDDT. Three useful conclusions are: (1) including rare transitions from FVS is important to getting reasonable temporal dynamics; (2) initialization and censoring issues can be ignored; and (3) smoothing and jitter can also be ignored. One surprising conclusion is that very different assumptions about regeneration cause VDDT and FVS results to be profoundly different for species, size and canopy structure. One nagging question is “how can we tell which model is right?” Field observations would help, and iterative model revision of both FVS and VDDT models is also helpful to a point. Our best advice is to use each model to challenge and improve the assumptions of the other, using each model to illuminate the “blind spots” of the other.

Use of VDDT/Path Modeling in National Forest Planning in the Pacific Northwest

Ayn Shlisky1 and Don Vandendriesche2

1USDA Forest Service Pacific Northwest Region, Portland, OR; 2USDA Forest Service, Forest Management Service Center, Ft. Collins, CO

Ayn Shlisky View presentation video View presentation slides (PDF)

Effective National Forest planning depends on scientifically sound analyses of land management alternatives relative to desired future conditions and environmental effects. The USDA Forest Service Pacific Northwest Region is currently using Path and/or VDDT to simulate changes in forest composition and structure for the revision of five Land and Resource Management Plans in Oregon and Washington. We illustrate the use of Path to examine current and desired forest conditions, develop Forest Plan Environmental Impact Statement (EIS) alternatives, and evaluate environmental effects for the Colville and Okanogan-Wenatchee National Forests (WA). Model parameters include natural and human-caused disturbance. Vegetation state transitions were derived from empirical studies, published literature, and expert opinion. Forest growth rates and biomass yield coefficients were extrapolated from Forest Vegetation Simulator (FVS) modeling of National Forest inventory plot data. Some model parameters were also derived from FVS post-processors, such as the Preside program. Preside was used to classify forest inventory plots into VDDT states, estimate mean residence times (within a state) and transition probabilities (between states), and summarize the alternative pathways between states. Path is being used to simulate the effects of alternative combinations of forest treatments such as forest thinning, regeneration harvest, and prescribed burning. The model is used to estimate the effects of treatments on a suite of decision criteria including forest structural states, departure from reference conditions, woody biomass yield, wildlife habitat, and fire severity and frequency. Path models used in Forest Plan revisions have proved useful in testing assumptions, developing alternative restoration scenarios, and documenting knowledge.

State-and-Transition Modeling in LANDFIRE

James L. Smith

The Nature Conservancy, LANDFIRE Project Lead

James Smith View presentation video View presentation slides (PDF)

State-and-transition models played a critical role in the LANDFIRE Project. The reverse is also true, LANDFIRE played (and continues to play) a key role in the development of state-and-transition models, modeling software, and modeling opportunities. Both cases will be described to provide additional context about the current status of vegetation modeling, and to help identify future opportunities.

Landscape Composition in Aspen Woodlands under Various Modeled Management Scenarios

Eva K. Strand, Lee A. Vierling, Stephen C. Bunting

Rangeland Ecology and Management, University of Idaho, Moscow, ID

Eva Strand View presentation video View presentation slides (PDF)

Quaking aspen is declining across the western United States. Aspen habitats are diverse plant communities in this region and loss of these habitats can cause shifts in biodiversity, productivity, and hydrology across spatial scales. Western aspen occurs on the majority of sites seral to conifer species, and long-term maintenance of these aspen woodlands requires periodic fire. We use field data, remotely sensed data, and fire atlas information to develop a spatially explicit landscape simulation model to assess the effects of current and historic wildfire regimes and prescribed burning programs on landscape vegetation composition in the Owyhee Mountains, Idaho. The model is run in the TELSA environment. Model outputs depict the future structural makeup and species composition of the landscape at selected time steps under simulated management scenarios. Under current fire regimes and in the absence of management activities, loss of seral aspen stands will continue to occur. However, a return to historic fire regimes, i.e., burning 12-14% of the modeled landscape per decade, maintains the majority of aspen stands in early and mid seral woodland stages and minimizes the loss of aspen. A fire rotation of 70-80 years was estimated for the historic fire regime while the current fire regime resulted in a fire rotation of 340-450 years. Implementation of prescribed burning programs that treat aspen and young conifer woodlands according to historic fire occurrence probabilities, are predicted to prevent conifer dominance and loss of aspen stands.

Calibrating State and Transition Models with FVS: A Case Study

Reuben Weisz1, Don Vandendriesche2, Monica Boehning3, Melinda Moeur4, Linda Wadleigh1, Jack Triepke1, Mitchel White3, Chris Nelson3, Judith Palmer3, Jim Youtz1, Bruce J. Higgins7, Tessa Nicolet1, Pam Bostwick1, Dan Mindar5, Mike Pitts1, Mike Manthei6, Wayne Robbie1v

1USDA Forest Service, Southwestern Region, Albuquerque, NM; 2USDA Forest Service, Forest Management Service Center, Ft. Collins, CO; 3Apache-Sitgreaves NF, Springerville, AZ; 4USDA Forest Service, Pacific Northwest Region, Portland, OR; 5USDOI NPS, Springerville, AZ; 6Coconino NF, Flagstaff, AZ; 7Kaibab, NF AZ

Reuben Weisz View presentation video View presentation slides (PDF)

This paper describes how the Forest Vegetation Simulator (FVS) was used to calibrate State and Transition Models (STMs) in Arizona and New Mexico. A standard set of silvicultural and fire transitions were evaluated using FVS simulations on FIA plots that have been binned up by each model state within each Potential Natural Vegetation Type (PNVT). A range of outputs from FVS (such as natural growth in the absence of disturbance, the probabilities of transitions to destination states resulting from natural and anthropogenic events, harvest volumes, and vegetation characteristics such as carbon values) were captured and linked to transitions through the modeling framework.

Squeezing Riparian Vegetation Dynamics into a State-and-Transition Modeling Paradigm

Steven M. Wondzell1, Agnieszka Przeszlowska1, Peter A. Bisson1 and Miles A. Hemstrom2

1USDA Forest Service, Pacific Northwest Research Station, Olympia, WA; 2USDA Forest Service, Pacific Northwest Research Station, Portland, OR

Steven Wondzell View presentation video View presentation slides (PDF)

State and transition models (STMs) are commonly used to project landscape changes resulting from the combined influence of plant succession, natural disturbance, and land management decisions. The models used by the US Forest Service typically focus on upland plant communities, even though many critical management issues revolve around the management of riparian zones and the links between land management decisions and the state of aquatic ecosystems. Our objective was to develop Aquatic-Riparian STMs that address critical management issues, could be easily linked with upland STMs, and thus allow a more comprehensive approach to landscape planning. Building these models proved problematic. First, a large number of models was necessary to simulate all possible combinations of potential riparian vegetation types and the potential morphologic types of mountain stream channels. Secondly, a number of attributes not commonly tracked in upland models were necessary to simulate the interactions between riparian vegetation and stream channels. Tracking attributes such as large wood and riparian shrubs required a large number of state classes within each model. Lastly, while linkages between upland and riparian disturbances and stream channel morphology are widely recognized, quantitative data describing these linkages were scarce. Thus, the Aquatic-Riparian STMs were heavily reliant on expert opinion to qualitatively rank channel conditions and habitat suitability for anadromous salmonids. This presentation will explore some of these problems and present results of model analyses from pilot watersheds in the interior Columbia River basin and northern Oregon Coast Range. VDDT Project web site:

Timber, Biomass and Carbon Pool Modeling with Path/VDDT

Xiaoping Zhou and Miles Hemstrom

USDA Forest Service, Pacific Northwest Research Station, Portland, OR

Xiaoping Zhou View presentation video View presentation slides (PDF)

Using the VDDT model, the integrated landscape assessment project (ILAP) explores the dynamics of broad-scale, multi-ownership landscapes and prioritizes management activities over time by evaluating and integrating broad resource information across all lands in the study region. Products from ILAP will help land managers, planners, and policy-makers evaluate management strategies that reduce fire risk, improve habitat, and benefit rural communities. The long-term trends of these important resources are vital for land managers, policy-makers, and other interested people. Our case study illustrates a methodology for modeling timber production, biomass supply potential and forest carbon sequestration over time in central Washington with the output of VDDT simulations. It also presents how the Forest Inventory and Analysis (FIA) data and other data were applied to assist with interpretation of the VDDT simulation output.

Poster Presentations

Using TELSA to Model Southern Pine Beetle Dynamics in the Southeastern US

Jennifer Costanza1, Jiri Hulcr1, Todd Earnhardt1, and Alexa McKerrow1,2

1Biodiversity and Spatial Information Center, North Carolina State University, Raleigh, NC; 2USGS-Core Science Systems, Raleigh, NC

View poster (PDF)

In the southeastern US, the southern pine beetle (Dendroctonus frontalis, SPB) has a major effect on forest dynamics. Many SPB infestations occur in loblolly pine (Pinus taeda) stands that are managed for timber production, and occasionally result in elimination of entire stands. Periodic regional outbreaks of SPB can result in widespread tree mortality and economic loss. We used VDDT and TELSA to develop a spatially-explicit model of SPB outbreaks in managed loblolly pine stands and project forest dynamics through 2060. In our model, susceptibility to SPB outbreak depends on a stand’s age and proximity to recent outbreaks, as well as stand density, which is determined by a stand’s management history. We used empirical data from past beetle activity to model two outbreak scenarios: a scenario of latent beetle density with low frequency of infestation, and a scenario in which outbreaks have high probabilities and are regionally autocorrelated. We also incorporated two management levels: current management, which included frequent thinning, as well as a reduced management scenario, which included less thinning. Our results show that under the background outbreak scenario, current high levels of management resulted in relatively low SPB activity over time, and decreased management caused only slightly increased beetle activity. However, under the high outbreak probability scenario, both intense management and non-management resulted in major tree mortality. This is the first landscape-level model of the interactions between pine management, beetle effect, and stand age, and can be easily integrated into models of regional vegetation dynamics via VDDT and TELSA.

Land Development and Natural Resources in Central Oregon

Jim Duncan1 and Theresa Burcsu2

1World Bank Institute, World Bank; 2Institute for Natural Resources - Portland, OR

View poster (PDF)

Conversion of land from uses such as forestry to rural and exurban development has implications for wildlife, local economies, and broad scale land development patterns, among other things. In central Oregon, land managers, policy makers, and residents are concerned with such impacts. Understanding the patterns and interactions between development and habitat informs land managers and policy makers. In this poster, we explore changes in vegetation and land use patterns under two potential development scenarios – one where development is constrained geographically, and one where it is not. We found that under the constrained scenario development was geographically concentrated relative to the other scenario, allowed for sustainable generation of open old forest conditions in some areas, and provided suitable habitat for wildlife across a large area. Under the unconstrained scenario, the speed and magnitude of potential development was highly localized, a wide swath of wildlife habitat may be negatively impacted, and opportunities for forest restoration may be reduced on parcelized, non-industrial, private forestland. Additionally, the potential for increased amounts of wildfire fuels may put rural and exurban residences at greater risk of destructive wildfires.

Estimating Sage-Grouse Habitat Abundance Using a State-and-Transition Modeling Framework

Louisa Evers1, Miles Hemstrom2, and Rick Miller3

1BLM Oregon State Office, Portland, OR; 2USDA Forest Service, Pacific Northwest Research Station, Portland, OR; 3Oregon State University, Corvallis, OR

How much sage-grouse (Centrocercus urophasianus) habitat might have existed prior to 1850 and how might that amount have differed from the current recommendations of sage-grouse biologists? We attempted to answer these questions using VDDT to develop reference conditions for vegetation, assign sage-grouse habitat types to the community phases included, and estimate the abundance of breeding, brood-rearing, and wintering habitat. We selected the Malheur High Plateau major land resource area in eastern Oregon as our study area. We used ecological site descriptions and climate and fire occurrence data for this area, and the sagebrush and sage-grouse literature to develop a set of four sagebrush models, determine which disturbances to include, estimate the probabilities of those disturbances and their impact, and make calls on sage-grouse habitat type and quality. Our models indicated the median amount of sage-grouse breeding habitat may have been slightly more abundant than levels currently recommended by sage-grouse biologists, brood-rearing habitat may have been as or more abundant, but wintering habitat may have been less abundant. These results suggest that either sage-grouse do not need as much wintering habitat as sage-grouse biologists recommend or that the amount of wintering habitat may have constrained sage-grouse populations.

Decision Support for Integrated Resource Assessment

Sean Gordon

College of Forestry, Oregon State University

View poster (PDF)

Integrated resource assessments face the challenge of how to summarize and present large, diverse data sets. One approach being used by the Integrated Landscape Assessment Project is the Ecosystem Management Decision Support system (EMDS). EMDS brings fuzzy logic and multicriteria decision analysis tools into the GIS environment for assessing and combining diverse indicators and presenting results at multiple levels of aggregation. A prototype assessment is underway for a 2.3 million acre mixed ownership area in central Washington. Major challenges and decision points include: choosing indicators relevant to management concerns; assessing indicators to a common scale for aggregation; choosing levels for conceptual, temporal and spatial aggregation; and handling uncertainty in the modeling framework.

Simulating Fire Hazard Across Landscapes Over Time Through Integration of the Vegetation Dynamics Development Tool (VDDT) and the Fuel Characteristic Classification System (FCCS)

Jessica Halofsky1, Stephanie Hart1 , Miles Hemstrom2, Morris Johnson3, Joshua Halofsky4, Roger Ottmar3

1University of Washington, School of Forest Resources, Seattle, WA; 2USDA Forest Service, PNW Research Station, Portland Forestry Sciences Lab, Portland, OR; 3USDA Forest Service, PNW Research Station, Pacific Wildland Fire Sciences Laboratory, Seattle, WA; 4Washington Department of Natural Resources, Olympia, WA

View poster (PDF)

The Vegetation Disturbance Dynamics Tool (VDDT) and the Fuel Characteristic Classification System (FCCS) are two valuable products used by many land managers throughout the United States. VDDT is a state and transition model that simulates changes in vegetative composition and structure across a landscape under different disturbance regimes and management scenarios. The FCCS is a software application that allows users to record fuel characteristics as fuelbeds and analyze fire potential of wildland and managed fuels. Although the utilities of VDDT are many, VDDT does not directly assess fire hazard for different vegetation states. We are integrating VDDT and FCCS to enhance the utility of VDDT and enable simulation of vegetation composition, structure and related fire potential across a landscape over time. Multiple FCCS fuelbeds will be created, based on plot data, for every VDDT state class (vegetation structure and cover combination) in mid-scale (5th field watershed) models covering the states of Oregon, Washington, Arizona and New Mexico. Fire potential, including fire behavior potential, crown fire potential, and available fuel potential, will be calculated for each fuelbed in FCCS, and mean fire potential will be calculated for each VDDT state class. The resulting link between VDDT state classes and fuelbed fire potential will allow users to assess the effects of disturbance regimes and management activities, such as fuel treatments, on vegetation communities and related fire hazard across a landscape over time.

Informing Wildfire Transitions in VDDT with a 25 Year Wildfire Record

Joshua S. Halofsky1, Miles A. Hemstrom2, Michael Polly3, Leonardo Frid4

1Washington State Department of Natural Resources, Olympia, WA; 2USDA Forest Service, PNW Research Station, Portland Forestry Sciences Lab, Portland, OR; 3Oregon State University, Institute for Natural Resources, Portland, OR; 4ESSA Technologies Ltd., Vancouver, BC

View poster (PDF)

VDDT model construction often relies on expert opinion when informing transition probabilities. While decades of field experience is invaluable in creating VDDT models, we have noticed a greater interest from land managers to move away from expert opinion towards more quantitative estimates of transition rates. In this vein, we recently developed an approach using Monitoring Trends in Burn Severity (MTBS) data to quantitatively derive wildfire probabilities used in VDDT. This poster illustrates an overview of the approach, which can be applied anywhere in the continental United States for the same 25 year record, or in other countries where spatial wildfire information exists.

Integrated Landscape Assessment Project Overview

Miles Hemstrom1, Janine Salwasser2, Lisa Ekman3

1USDA Forest Service, PNW Research Station, Portland Forestry Sciences Lab, Portland, OR; 2Institute for Natural Resources; 3formerly with USDA Forest Service

View poster (PDF)

The Integrated Landscape Assessment Project (ILAP) supports watershed-level prioritization of restoration in Arizona, New Mexico, Oregon, and Washington. The project explores the dynamics of broad-scale, multi-ownership landscapes over time by evaluating and integrating information on current and future vegetation and fuel conditions, wildlife habitat, watershed conditions, and the potential costs and benefits of management treatments. Products from the project will be used by land managers, planners, and policymakers to evaluate management strategies that reduce fire risk, improve habitat, and benefit rural communities. Science Delivery, one of three project modules, consists of two parts: 1) the compilation and production of map data used to characterize current condition; and 2) the generation of state-and-transition models to project ecosystem trends into the future. Consistent mid-scale vegetation data, potential vegetation data, watershed boundaries, and other necessary mapping is being produced for all major ecosystem types in the four states. Vegetation Dynamics Development Tool (VDDT) software is used to model ecosystem dynamics. The initial prioritization of broad-scale, multi-ownership landscapes over time is based on integrating Knowledge Discovery module outputs about: 1) existing fuel conditions and how they might change over time; 2) selected wildlife habitats and whether fuel treatments might affect them; and 3) the potential costs and benefits of different management treatments, including the economic potential of the material removed by the treatments and opportunities for new products. The Decision Support module integrates and evaluates the Knowledge Discovery module outputs using both a static, largely aspatial multi-attribute approach, and through spatial techniques for guiding decision making.

Using Spatially Explicit Modeling to Evaluate Effects of Fuel and Timber Management on Forest Landscapes in southwestern Oregon

David Hockman-Wert

U.S. Geological Survey

Spatial modeling with a GIS provides a visual, intuitive interface that makes it possible for scientists to: (1) compare different management options; and (2) demonstrate how these scenarios play out in a dynamic landscape. The landscape modeling approach allows scientists and managers to provide feedback both in the development of alternative scenarios and in the refinement of the models. The Landscape Scenario Analysis Project, initiated and funded by the Bureau of Land Management (BLM), facilitated interactions between scientists from the USGS Forest and Rangeland Ecosystem Science Center and BLM managers regarding broad-scale implications of vegetation management for forest ecosystems in western Oregon. In this project, scenarios were developed for the Applegate Watershed in southwest Oregon, an area with many rural residents living in a fire-prone landscape. Spatially explicit modeling was performed with ESSA Technologies’ TELSA and VDDT software packages developed specifically for modeling interactions between natural disturbances and forest management. The spatial module in TELSA integrates directly with a geographic information system and provides the functionality to build models using existing spatial data layers such as forest stand characteristics, land ownership, riparian reserves, and fire regime zones. Maps and graphs of temporal trends generated by the models have helped managers visualize the effects of natural and management disturbances on forest landscape and stand structure. In particular, results from the model were used to help assess cumulative effects from timber harvest since data were compiled across ownerships and land-use allocations.

Updating Non-forested Vegetation Geospatial Layers for LANDFIRE: Incorporating Disturbance

Gretchen Meier1, Don Long2, Brian Tolk3, Jeff Natharius3

1Arctic Slope Regional Corporation Research and Technology Solutions (ARTS); 2USDA Forest Service, Missoula Fire Science Lab, MO; 3Stinger Ghaffarian Technologies (SGT)

LANDFIRE is a multi-partner mapping project that generates consistent, comprehensive spatial data for the entire United States. The maps and data layers describe existing vegetation type (EVT), existing vegetation cover (EVC) and height as well as a suite of fire-related fuel layers such as fire regime condition class, simulated historical fire regimes, and current departure from those regimes. LANDFIRE produced the first complete suite of products in the fall of 2009 and project plans include conducting regular updating to ensure that data layers and other products remain current. In areas with significant disturbance, products will be updated biennially. In 2010, LANDFIRE produced updated vegetation layers that incorporated post-2001 managed and natural disturbance information. Specific examples of incorporated disturbance data are: fire; vegetation management; and insect damage and disease. Disturbance data included attributes associated with disturbance year, type, and severity. The disturbance layers were derived through use of: (1) historic and current fire burn data from the Monitoring Trends in Burn Severity Program (MTBS); (2) Landsat time-series data analysis; and (3) local agency disturbance polygons. Here, we present the results of the LANDFIRE updating process using examples of disturbed areas in the Columbia Plateau. Data sets for this area have been updated to include data on EVT, EVH, and EVC. Herbaceous EVTs replace forested EVTs in locations impacted by high-severity disturbances such as high-severity fire and logging. Herbaceous EVTs also replaced non-forested EVTs in areas affected by moderate to high-severity fire such as in instances of shrubland fire and management treatments.

Balancing Model Output and Precision of Wildlife Analysis in Planning for Natural Resources

Anita T. Morzillo1 and Miles Hemstrom2

1Department of Forest Ecosystems and Society, Oregon State University; 2USDA Forest Service, Pacific Northwest Research Station, Portland, OR

View poster (PDF)

Wildlife conservation is often a central focus in planning for natural resource management. For planning efforts at regional or national levels, evaluation of wildlife habitat involves balancing the desire for information about habitat characteristics and the feasibility of completing analysis across large areas. Our objective is to describe tradeoffs made in assessment of wildlife habitat within a multiple-objective modeling framework. The Integrated Landscape Assessment Project (ILAP) is a broad-scale effort to evaluate impacts of land management strategies on fuels condition, timber production, wildlife habitat, and community economics across Oregon, Washington, Arizona, and New Mexico. The Vegetation Dynamics Development Tool (VDDT) is the central model within the multi-objective framework, but the forest characteristics of the model output may be misinterpreted in wildlife assessment. First, the observational unit for wildlife analysis is habitat, which does not provide information about actual species occurrence or distribution. Second, subjectivity exists in researcher interpretation of qualitative species-habitat relationships. For quantified species-habitat relationships that exist, only those which match output criteria may be used for analysis. Third, visual interpretation of results may vary based on the scale of analysis used in the modeling effort. When preparing knowledge for application to natural resource planning, there is a need to focus on consistent and defensible information and emphasize the limitations of knowledge derived from data analysis.

Learning to Predict Vegetation Change: a Step-wise Progression for Acquiring State and Transition Modeling Skills

Colleen Ryan1 and Eva Strand2

1National Interagency Fuels, Fire, and Vegetation Technology Transfer; 2University of Idaho Moscow

View poster (PDF)

Understanding the relationships between vegetation dynamics and disturbance processes can challenge today’s land manager. State-and-transition modeling using the Vegetation Dynamics Development Tool (VDDT) and Path software has proven to be useful for natural resource professionals in addressing a wide range of questions related to forest and rangeland management, fuels planning, wildlife habitat management, and ecosystem carbon modeling. The LANDFIRE project has developed a set of vegetation dynamics models for U.S. ecosystems using the VDDT software. These models distill ecological knowledge provided by thousands of local experts, offering valuable information for resource managers, who can adapt these models to address a range of resource management questions. To facilitate this process, the National Interagency Fuels, Fire, and Vegetation Technology Transfer (NIFTT) is developing a set of online courses to assist users in learning to create, modify, and use vegetation dynamics models with the new Path modeling software. These courses will be organized into a “learning pathway” that will begin with basic modeling concepts for novice users and continue through the introduction of advanced applications. Students will be able to enter the pathway at any point, depending on their individual experience and needs. The learning pathway will include coursework describing the development of the LANDFIRE vegetation models, as well as techniques for adapting these models to address specific land management applications.

Simulating Treatment Effects in Pine-Oak Forests of the Ouachita Mountains

Ayn Shlisky1 and Kori Blankenship2

1USDA Forest Service Pacific Northwest Region, Portland, OR; 2The Nature Conservancy, Bend, OR

View poster (PDF)

Effective land management project decision-making depends on scientifically sound analyses of management alternatives relative to desired future conditions and environmental effects. In this poster presentation we illustrate the use of a state-and-transition model to evaluate likely future landscape conditions in a pine-oak forest on the Ouachita National Forest, Arkansas based on current and potential future alternative management actions. We used the model, developed with the Vegetation Dynamics Development Tool, to simulate the effects of current management compared to alternative woodland thinning, regeneration harvest, and prescribed burning treatments. At the time of the study, a National Forest interdisciplinary team was completing a project-level environmental assessment of alternative management scenarios across a 16,700 acre watershed. Potential interactions between climate change and management scenarios were also simulated. Effects included not only relative timber outputs, but also smoke, fire frequency and insect and disease outbreak frequency over the long-term. Our results indicate that a woodland management emphasis generally yielded landscape structure, insect outbreak frequency, and fire regime conditions closer to the desired future condition specified in the Ouachita National Forest Plan compared to a seed tree management emphasis. When potential climate effects were considered, the woodland management emphasis also yielded greater woody biomass harvest output, and less smoke output than the seed tree management emphasis. While the model outputs have proven to be useful, the process forced the team to test assumptions and document knowledge – two intangible but valuable outcomes.

Creating an Effective and Efficient Conservation Strategy: Landscape Conservation Forecasting

James Smith1, Sarah Hagen2, Jeannie Patton3, Randy Swaty4, Kori Blankenship5

1The Nature Conservancy Jacksonville, FL; 2The Nature Conservancy Jacksonville, Minneapolis, MN; 3The Nature Conservancy Jacksonville, Boulder, CO; 4The Nature Conservancy Jacksonville, Marquette, MI 5The Nature Conservancy Jacksonville, Bend, OR

In a complex landscape, choosing the restoration strategy that provides the most benefit for the available budget is challenging for resource managers. One method for meeting that challenge is Landscape Conservation Forecastingtm (LCF). LCF was developed by staff in the TNC Nevada Field Office to improve the effectiveness and efficiency of their conservation planning process. LCF uses public domain software tools, combined with local and/or nationally available LANDFIRE data to explore a range of management activities, costs and potential conservation benefits. An example from the Bodie Hills region in California/Nevada is briefly presented.