- 1 Oral Presentations
- 1.1 From land cover mapping to land change science: Dynamic monitoring of a changing landscape
- 1.2 State-and-transition simulation models: theory and practice
- 1.3 Assessing landscape impacts of potential bioenergy production scenarios using spatially explicit state-and-transition simulation models
- 1.4 Development of a state-and-transition simulation model to simulate the impacts of land-use and vegetation change on ecosystem carbon dynamics: A methodological approach for national resource assessment
- 1.5 Application of state-and-transition simulation modeling to reconstruction of historical trends in carbon storage in the conterminous U.S.
- 1.6 Simulating the historical range of variability in fire-adapted ecosystems
- 1.7 Representing temporal variability in wildfire disturbances in state-and-transition models used for land management planning and assessment
- 1.8 State-and-transition simulation modeling of land use threats to protected areas in Mediterranean California
- 1.9 Modeling management and climate disturbances to evaluate landscape scale conservation efficacy
- 1.10 State-and-transition simulation modeling of oil sands mine reclamation in the boreal forest of northern Alberta, Canada
- 1.11 Downscaling and spatial reapportionment of coarse scale global land-use/land-cover projections for use in region-level state-and-transition simulation modeling
- 1.12 Combining species distribution models and state-and-transition simulations to support resource management under climate change
- 1.13 Climate change, management and future sage-grouse habitat in southeastern Oregon
- 1.14 Landscape Conservation Forecasting™ with federally listed species in southwest Utah
- 1.15 Feed the engine! Utilizing spatial imagery to feed the ST-Sim engine
- 1.16 Using state-and-transition models in an integrated landscape assessment to inform Forest Plan Revision in the Arizonan Sky Island Ecosystem
- 1.17 Methods used to parameterize the spatially-explicit components of a state-and-transition simulation model
- 1.18 Developing state-and-transition simulation models for the Rangeland Vegetation Simulator using NRCS ecological site descriptions
- 1.19 Optimizing buffelgrass management with differing levels of aridity and treatment options
- 2 Poster Presentations
- 2.1 Holding the Line: Modeling Invasive Buffelgrass in the Cactus Forests of Arizona
- 2.2 Changes in community exposure to tsunami hazards in the U.S. Pacific Northwest based on future land change scenarios
- 2.3 To burn or decay: States and transitions in mesic and dry mixed-conifer models for the Arizonan Sky Island ecosystem
- 2.4 Fusing State-Transition Models from Ecological Sites with the Rangeland Vegetation Simulator
- 2.5 Predicting Coastal Wetland Restoration Outcomes Using State-and-Transition Simulation Models
- 2.6 Fire Regime Syntheses in the Fire Effects Information System (FEIS)
- 2.7 State-and-Transition Simulation Modelling: Status and Future Use of Parallel Computing
- 2.8 Landscape Conservation Forecasting: Creating Effective and Efficient Conservation Plans
- 2.9 Supporting State-and-Transition Modeling: Online Resources from LANDFIRE
From land cover mapping to land change science: Dynamic monitoring of a changing landscape
Dr. Thomas R. Loveland
Senior Scientist, Co-Director of the Geographic Information Science Center of Excellence, South Dakota State University and USGS
Land cover mapping has evolved from early 1960’s and 1970’s efforts designed to provide static characterization of the major land uses and cover to today’s focus on a more integrated, holistic, and dynamic mapping of a wide range of land change variables. Historically, land cover maps were produced by a small number of remote sensing specialists with the tools and expertise to craft one-off maps for their study areas. Now, advances in Earth observation systems, computing, and image-processing tools have expanded the frequency and geographic coverage of land cover maps. The maturing field of land change science is now pushing for land cover products better tailored and more flexible for a wider range of applications, including model parameterization, resource management, and decision support. Driven by land change science objectives to explain how the patterns, processes, and consequences of changes in land use, cover, and condition at multiple spatial and temporal scales affect people and nature, the next grand challenge in land cover from space is continuous monitoring. Today’s land cover monitoring projects must cover larger geographic areas, span broader periods of time, be produced with higher temporal frequency, and quantify both the state and condition of the landscape. Real time mapping of change as it is occurring requires tapping into the rich archives of Landsat and other Earth observations missions. This challenge will lead to richer, more flexible land change information that will improve our geographic knowledge of land change, and will also improve model parameterizations and scenarios definition needed for more informed projections.
State-and-transition simulation models: theory and practice
Colin Daniel1,2 and Leonardo Frid2
1University of Toronto
2Apex Resource Management Solutions Ltd.
There is a wide array of modelling environments and approaches available for predicting landscape dynamics. One such approach involves the use of state-and-transition simulation models (STSMs), whereby transitions between discrete states of a landscape are simulated probabilistically through time across a spatial grid. The following paper provides an overview the STSM approach, from its origins in the early 1990s through to latest in STSM software. We begin by outlining some of the theoretical underpinnings for STSMs, including their relationship to Markov chain models. STSMs differ from stationary Markov chains in several important ways, including their ability to capture the age and time-since-transition of states, to allow transition probabilities to be conditional upon age and time-since-transition, to set targets for transitions between states, and to modify transition probabilities over space and time. We also show how it is now possible to generate a wide range of spatially-explicit dynamics in STSM simulations, using software tools such as ST-Sim to invoke external models to calculate transition probabilities dynamically for any grid cell and timestep as a function of the current state of landscape. Finally we demonstrate some of the key STSM concepts through a case study example exploring the effects of changes in wildfire on the distribution and abundance of tree species in the boreal forest of Ontario, Canada.
Assessing landscape impacts of potential bioenergy production scenarios using spatially explicit state-and-transition simulation models
Jennifer K. Costanza1, Robert C. Abt2, Alexa J. McKerrow3, Jaime A. Collazo4
1North Carolina Cooperative Fish and Wildlife Research Unit, Department of Applied Ecology, North Carolina State University
2Department of Forestry and Environmental Resources, North Carolina State University'
3U.S. Geological Survey, Core Science Analytics and Synthesis
4U.S. Geological Survey, North Carolina Cooperative Fish and Wildlife Research Unit, Department of Applied Ecology, North Carolina State University
Much of the debate on renewable fuel alternatives has focused on the potential for broad-scale bioenergy production from annual and perennial crops, but the landscape impacts have not been fully assessed. In the southeastern U.S., where both forests and agricultural crops could be important sources of biomass for biofuels, addressing these impacts is essential. We expect four major types of landscape change: conversion of natural forests to intensively managed forests; increased management intensity in already-managed forests; conversion of non-forest to forest; and conversion of forest to purpose grown crops. Our team simulated these changes in North Carolina through 2050 under six biofuel production scenarios that varied crops used and quantities of biofuels produced. We linked forest timber supply and supply chain economics models with spatially explicit state-and-transition simulation models (STSMs). Results indicate that by 2050, large-scale land conversion to biofuel crops and intensification of land management could result from liquid biofuel demand. The impacts of those changes will depend on which types of land those changes effect. When biomass feedstock supply comes from forests, the result is more forest area because of higher demand for forest products. However, conventional forestry can only supply a small amount of biomass for biofuels, and the rest would likely be met by purpose grown crops. Impacts to forest ecosystems are reduced if those crops are grown on lands already in agriculture. The STSM framework is critical for helping identify tradeoffs between landscape impacts and society’s need for renewable fuels.
Development of a state-and-transition simulation model to simulate the impacts of land-use and vegetation change on ecosystem carbon dynamics: A methodological approach for national resource assessment
Benjamin M. Sleeter1, Jinxun Liu1, Colin Daniel2,3, Leonardo Frid2, Zhiliang Zhu4
1U.S. Geological Survey, Western Geographic Center
2Apex Resource Management Solutions Ltd.
3University of Toronto
4U.S. Geological Survey, National Carbon Sequestration and Greenhouse Gas Flux Assessment
Scenario-based simulations are an important tool for global change assessments, such as those conducted periodically by the Intergovernmental Panel on Climate Changes (IPCC). Scenarios for future changes in land cover and land use can be applied to process-based model projections of ecosystem carbon dynamics. However, these projections are complicated by 1) the complexity of process-based ecosystem models, 2) the relatively high computational demands required to run simulations, and 3) the lack of integrated modeling systems capable of simulating the feedbacks between carbon and land-use change in an open and transparent way. We have extended the use of state-and-transition simulation models (STSMs) to simulate changes in vegetation and land-use under a range of future global change scenarios from the IPCC 4th and 5th Assessments. Using these scenario simulations, we have integrated a fast and transparent stock-and-flow carbon accounting model to simulate the transfer and fate of carbon among a set of defined pools in forest, shrub, grass and agricultural ecosystems. To parameterize the carbon model we used the process-based Integrated Biosphere Simulator (IBIS) to derive baseline carbon flux values that reflect the effects of vegetation type, forest age, climate trend and variability, and human and natural disturbances. This approach allows us to perform a wide range of scenario experiments and in-depth uncertainty analysis of carbon trends. We present this integrated modeling framework as a novel approach in terms of its degree of integration, its representation of uncertainty, and its spatially explicit nature. ________________________________________________________________________________________________________________________
Application of state-and-transition simulation modeling to reconstruction of historical trends in carbon storage in the conterminous U.S.
Eric T. Sundquist, William Condon, Kate Ackerman
U.S. Geological Survey, Woods Hole, MA
State-and-transition simulation modeling (STSM) provides a probabilistic framework for evaluating how ecosystems are affected by human activities and natural processes. By applying statistical and deterministic demographic relationships, STSM can be used to approximate spatial and temporal trends in the distribution of basic subgroups (states) and age cohorts within ecosystem populations that are subject to a variety of transitions due to growth, disturbance, and management. To apply STSM to ecosystem carbon accounting, the demographic representations of STSM must be related quantitatively to rates of carbon storage, transfer, and loss. These relationships are commonly defined in the form of empirical functions or look-up tables that link carbon storage to ecosystem states and transitions. For applications to large geographic areas with diverse ecosystems, a central challenge is to develop empirical carbon functions that are consistent, transparent, and data-constrained. We have developed and tested an empirical methodology for application of STSM results to estimates of historical forest biomass carbon in the conterminous U.S. We have combined forest inventory and yield data from the U.S. Forest Service (USFS) with STSM models from the LANDFIRE program to produce a set of empirical forest carbon equations for ~240 tree-dominated ecosystems. These equations have a consistent sigmoid form that includes growth terms and loss probabilities that can be adjusted through time. Ecosystem distributions and transition probabilities are constrained by available historical records for changes due to land use, management, and fire. The methodology is tested by its capacity to reconstruct reasonable time-dependent forest demographic and carbon trends for the period 1700-2000 CE.
Simulating the historical range of variability in fire-adapted ecosystems
Kori Blankenship1, Leonardo Frid2, Jim Smith1 and Randy Swaty1
1 The Nature Conservancy LANDFIRE Team
2 Apex Resource Management Solutions Ltd.
Reference ecological conditions provide important context for land managers as they assess the condition of their landscape and set desired future conditions. State-and-transition models (STMs) are commonly used to estimate reference conditions which can be used to evaluate current ecosystem conditions and to guide land management decisions and activities. The LANDFIRE program created more than 1,000 STMs and used them to assess departure from a mean reference value for ecosystems in the United States. While the mean provides a useful benchmark, land managers and researchers are often interested in the range of variability around the mean. This range, frequently referred to as the historical range of variability (HRV), offers model users improved understanding of ecosystem function, more information with which to evaluate ecosystem change and potentially greater flexibility in management options. We developed a method for using LANDFIRE STMs to estimate the range of variability around the mean reference condition for each model state in ecosystems where fire is the dominant disturbance. The approach is flexible and can be adapted for use in a variety of ecosystems. The addition of a range in HRV estimates acknowledges spatial and temporal variation which may be more relevant in a changing climate. In this presentation we will demonstrate our method for calculating the HRV and discuss its strengths and limitations.
Representing temporal variability in wildfire disturbances in state-and-transition models used for land management planning and assessment
Priya C. Shahani
U.S. Forest Service, Southwest Region
State-and-transition models are often used by land management agencies to predict the effects of candidate management regimes on vegetation communities. Wildfire is a key factor in shaping these communities, and stochasticity in wildfire dynamics should result in variation in vegetation structure (and species dominance) in the plant communities being modeled. However, dramatic variation in model outcomes adds a challenge to assessment and planning of land management activities, and this variation in outcomes may differ in importance to planning efforts at different landscape scales. Through this work, I compare the predictions generated by models for three southeastern Arizona vegetation communities with different temporal patterns in wildfire dynamics under three different analysis scenarios: a) wildfire probabilities are modified by Monte Carlo multipliers to represent observed patterns in strings of high and low fire years; b) paired “high wildfire frequency” and “low wildfire frequency” models are run to portray the extremes of possible outcomes; and c) average wildfire probabilities are used without other modifications. In addition to comparing the predictions made via these three approaches, I examine the value that each type of prediction has in land management planning. Finally, I explore approaches to output data summarization and error characterization to maximize utility for land management planning.
State-and-transition simulation modeling of land use threats to protected areas in Mediterranean California
Tamara S. Wilson, Benjamin M. Sleeter
U.S. Geological Survey, Western Geographic Science Center
Land use in Mediterranean California will likely present a greater challenge to biodiversity than climate change this century. While protected areas in this region are intended to safeguard biodiversity, they are often surrounded by intensive human land use which can pose a major dispersal obstacle for species. Non-local human land use also influences regional ecological flows and processes, having potentially broad reaching impacts on protected areas far removed. Developing future scenarios of regional land use can help identify protected areas at greatest risk from both local and regional land use intensification. Using a state-and-transition simulation model, we modeled spatially-explicit (1 km2) land use from 2000-2100 under two alternative land-use and emission scenarios for three ecoregions in Mediterranean California, USA. Land use change projections were based on the Intergovernmental Panel on Climate Change’s (IPCC) Special Report on Emission Scenarios (SRES). We analyzed scenario-based land use conversion threats from agriculture and development and calculated a conversion threat index (CTI) to identify protected areas with highest projected land use conversion probability within closest proximity to existing protected areas. State-and-transition simulation modeling of future land use scenarios can help identify ecoregions and protected areas at greater risk from regional land use threats, information useful in prioritizing future conservation resources, strategies, and targets.
Modeling management and climate disturbances to evaluate landscape scale conservation efficacy
Amanda Swearingen1, Janet Silbernagel1, Jessica Price1, Randy Swaty2, Nicholas Miller3, Kristina Nixon4
1University of Wisconsin at Madison- The Nelson Institute
2The Nature Conservancy Global Fire Team
3The Nature Conservancy
Conservation of biodiversity and human values at a meaningful scale often requires strategies that are expensive. Deciding on the best strategies among many is often facilitated by collaborative modeling. Using LANDFIRE spatial data and workshops designed to incorporate local expert knowledge, we generated and modeled landscape scenarios that reflect possible futures for the Wild Rivers Legacy Forest in Northern Wisconsin using the VDDT/TELSA modeling suite. In all, four scenarios were designed that represent different management regimes: current management, ecological forestry or restoration, increased easement purchases and no conservation action. All scenarios were modeled out 100 years under current and increased windthrow and fire disturbances. The results of this analysis show the change of the landscape over time, indicating the spatial implication of such management and climatic disturbances. The goal of this analysis is to inform ongoing management and conservation efforts. Here we present the results of these 100 year projections, discuss potential implications of the results, and reflect on the modeling process.
State-and-transition simulation modeling of oil sands mine reclamation in the boreal forest of northern Alberta, Canada
Gillian M. Donald1, Leonardo Frid2 and Colin J. Daniel2
1Donald Functional & Applied Ecology Inc. and Fort McMurray Métis Local 1935
2Apex Resource Management Solutions Ltd.
Oil sands mine operators in northeastern Alberta, Canada, are required to reclaim disturbed land to target the establishment of a self-sustaining, locally common boreal forest integrated with the surrounding area. The Cumulative Environmental Management Association (CEMA) is a non-profit, multi-stakeholder organization based in Fort McMurray, Alberta, whose role is to produce recommendations to government regulators pertaining to the cumulative impact of oil sands development in northeastern Alberta. These recommendations include guidance documents for reclamation to forest ecosystems and a criteria and indicator framework for reclamation certification. The development of a state-and-transition simulation model (STSM) as a decision-support tool for reclamation and closure planning was initiated by CEMA to assist in evaluating the assumptions identified in the reclamation guidance documents to predict which reclaimed areas are likely to achieve reclamation certification criteria. The model was created with input from regulators, industry practitioners, Aboriginal groups and non-profit organizations. Land units were defined for a reclamation classification system based on soil moisture regime, slope position, aspect and reclamation cover material placement prescriptions. Data from a long-term monitoring program and other projects conducted by CEMA were used to parameterize the model. Target and non-target states were defined using reclamation certification indicator thresholds including site index, characteristic species and crown closure. Management interventions such as infill tree planting, understory species planting and fertilization were incorporated into the model. The modelling scenarios demonstrated how the STSM could be used as a decision-support tool for reclamation and closure planning in the mineable oil sands region.
Downscaling and spatial reapportionment of coarse scale global land-use/land-cover projections for use in region-level state-and-transition simulation modeling
Jason T. Sherba, Benjamin M. Sleeter, Adam Davis, Owen Parker, Rachel R. Sleeter
U.S. Geological Survey, Western Geographic Science Center
Global land-use/land-cover (LULC) projections and historical datasets are typically available at coarse grid resolutions that are often incompatible with scenario modeling applications at regional and national scales. This spatial incongruity has been a barrier for effectively utilizing global LULC datasets within a state-and-transition simulation modeling (STSM) framework. In order to transform global gridded LULC data into spatial scales and thematic LULC classes appropriate for further analysis, three downscaling techniques were tested. For each downscaling approach, Intergovernmental Panel on Climate Change (IPCC) Representative Concentration Pathway (RCP) land-use projections at the 0.5 degree cell level were first downscaled to four Level III ecoregion boundaries in the Northwestern United States. RCP transition values at each cell were downscaled based on the proportional distribution between ecoregions of either (1) cell area, (2) land-cover composition derived as a zonal histogram from the 2006 NLCD dataset, or (3) historic land-cover transition values from the USGS Land Cover Trends dataset. The resulting disaggregated cell zones, from each downscaling approach, were aggregated according to their bounding ecoregion and “cross-walked” to relevant LULC classes. To demonstrate the effectiveness of our disaggregation techniques, reapportioned LULC transition values were applied as input parameters in a STSM for modeling ecoregion-level LULC change from 2005 to 2100. Although all three downscaling methods may be used effectively, downscaling using the Trends dataset produced projected LULC change estimates most consistent with historical observations. Regardless of the downscaling method, some LULC projections remain improbable and require further investigation.
Combining species distribution models and state-and-transition simulations to support resource management under climate change
Brian W. Miller1,2, Jeffrey T. Morisette1,3, Andrew Hansen4, Tony Chang4, Leonardo Frid5
1Department of the Interior North Central Climate Science Center
2Colorado State University
3U.S. Geological Survey
4Montana State University
5Apex Resource Management Solutions
State-and-transition simulation models (STSMs) are known for their ability to explore the combined effects of multiple disturbances, ecological dynamics, and management actions on vegetation. However, integrating the additional impacts of climate change into STSMs remains a challenge. We began exploring a means of addressing this challenge by combining STSM with species distribution modeling. Species distribution models estimate the potential habitat for a given species using field data that consist of presence (and possible absence) locations of a given species as well as the values of environmental or climatic covariates thought to define the species’ habitat suitability at these locations. Thus, in order to account for changes in habitat suitability due to climate change we used a species distribution model to generate continuous surfaces of species occurrence probabilities. These data were imported into ST-Sim, a state-and-transition simulation-modeling platform, where they dictated the probability of each cell transitioning between alternate potential vegetation types at each timestep. The simulation model was parameterized to capture additional processes of vegetation growth, disturbance, and management. We compared historical model runs against species observations, and then projected the simulation into the future in order to compare outcomes under several management scenarios. Preliminary results indicate that species distribution models and STSMs are complementary tools, and combining them is an effective way to explore management options while accounting for the anticipated impacts of climate change.
Climate change, management and future sage-grouse habitat in southeastern Oregon
Megan Creutzburg1, Emilie Henderson1, David Conklin2
1Institute for Natural Resources, Portland State University
Contemporary pressures on sagebrush steppe from climate change, exotic species, wildfire, and land use change threaten sagebrush-obligate species such as the Greater sage-grouse (Centrocercus urophasianus). To effectively manage sagebrush steppe landscapes for long-term goals, managers need information about the impacts of climate change, disturbances, and management activities on rangeland vegetation. We integrated information from three different models to project future vegetation dynamics and sage-grouse habitat constraints under multiple scenarios of climate change and management across 23.5 million acres in southeastern Oregon. Specifically, we linked a sage-grouse habitat climate envelope model, a dynamic global vegetation model of climate-related vegetation shifts, and state-and-transition simulation models of vegetation community dynamics and management. Our results illustrate two climate-related constraints to future sage-grouse habitat: expansion of cool-moist sagebrush steppe and concurrent juniper encroachment; and expansion of xeric sagebrush, where hot, dry summer conditions are unfavorable for sage-grouse. All climate scenarios showed increases in cool-moist sagebrush and juniper woodland expansion, but climate scenarios varied widely in their projections of xeric sagebrush extent. While vegetation management is unlikely to affect drought-related sage-grouse habitat loss, active management through juniper cutting and prescribed fire can slow the threat of woodland expansion. However, current levels of management appear unlikely to prevent further degradation of sage-grouse habitat over the next century. Our results provide information about likely joint effects of climate change and management in shaping future vegetation condition and sage-grouse habitat throughout southeastern Oregon.
Landscape Conservation Forecasting™ with federally listed species in southwest Utah
Louis Provencher1, Gen Green2, Joel Tuhy3, Dan Fletcher4, Elaine York2
1The Nature Conservancy, Reno, NV 2The Nature Conservancy, Salt Lake City, UT 3The Nature Conservancy, Moab, UT 4Bureau of Land Management, Cedar City Field Office, UT
In 2012, the Bureau of Land Management Cedar City Field Office (CCFO) and The Nature Conservancy (TNC) entered into a cooperative agreement to implement TNC’s Landscape Conservation Forecasting™ for the CCFO’s Hamlin Valley and Black Mountains supporting populations of greater sage-grouse (GRSG) and federally listed Utah prairie dog (UPD in Black Mountains only). Primary goals were to (1) map potential vegetation types and their current vegetation classes using remote sensing, (2) determine current condition of all ecological systems, (3) use STSM software to forecast anticipated future condition of ecological systems under alternative management scenarios, and (4) calculate ecological return-on-investment of each scenario. Eighteen and twenty three ecological systems were mapped, respectively, in Hamlin Valley and Black Mountains. Using a new metric of Unified Ecological Departure (UED), five systems were not departed from reference and management conditions, seven were moderately departed, and six were highly departed in Hamlin Valley. In the Black Mountains, two systems were not departed from reference and management conditions, six were moderately departed, and fifteen were highly departed. Minimum, maximum, and preferred management scenarios were simulated with spatial S&T models in the STSM software for only systems that were departed and required sensitive species management in each landscape. Simulations were primarily constrained by budget limits ($12.5 mil vs. $1.25 mil per year per landscape) and designed to reduce problematic vegetation classes and increased GRSG and UPD habitat suitability calculated using resource selection functions built with experts. The preferred management scenario improved both GRSG and UPD habitat suitability.
Feed the engine! Utilizing spatial imagery to feed the ST-Sim engine
Spatial Solutions, Inc.
With the continued advances in sophistication and functionality of state-and-transition modeling, one is reminded of the old adage of “garbage in, garbage out”, or the converse “quality in, quality out.” One basic truth persists for modeling: simulation results will only be as accurate and reliable as the base data feeding the modeling engine. Utilizing spatial imagery from satellite and airborne platforms can provide outstanding opportunities for developing new, custom land cover characterizations or updating existing data sets for an ever-increasing scope of budgets, resolutions, and timelines. This presentation will focus on the sources, opportunities, challenges, and solutions for employing remotely-sensed spatial data for powerful state-and-transition modeling. More specifically, we will discuss issues of (a) allocation of time and cost between mapping potential vegetation types and their current vegetation classes; (b) client choice of imagery resolution given that normal applications of ST-Sim will required more coarsely resampled geo-data, (c) the necessity of field surveys, (d) our evolution from sparse traditional training plots to high-volume rapid observations supported by new technology, and (e) the role of object-based imagery analysis in very heterogeneous landscape.
Using state-and-transition models in an integrated landscape assessment to inform Forest Plan Revision in the Arizonan Sky Island Ecosystem
A.M. Lynch1, P.C. Shahani2, J.M. Ruyle3, C.P. Wilcox3, J.R. Malusa4, F.J., Triepke2, J.M. Salwasser5, D.A.Falk4
1U.S. Forest Service, Rocky Mountain Research Station
2USDA Forest Service, Southwest Region
3USDA Forest Service, Coronado National Forest
4University of Arizona
5Oregon State University, Institute for Natural Resources, Portland Oregon
VDDT-based state-and-transition models (STMs) were used to project and compare landscape-scale effects of proposed management alternatives in the Coronado N.F. Forest Plan Revision process.Models were parameterized using approaches developed by the US Forest Service and the Integrated Landscape Assessment Project.In order to provide scientifically sound analyses of land management alternatives relative to desired conditions, we formed a team of analysts and experts from the National Forest Systems (NFS), Universities, and Forest Service Research.The team adapted existing STMs to local conditions.Landscape-level vegetation conditions were projected for Forest Service lands for each prevalent ecosystem type.Each model was run under different management regimes for 10 Monte Carlo simulations over 1000 years, and examined at years 10, 1000, and 1000, with the 1000-year time horizon used to characterize steady-state conditions.This approach allowed us to test Forest Plan Alternatives using large data sets on current vegetation conditions and ecosystem dynamics.Substantive changes were made to the Draft Plan based on analyses results.Modeling results also helped focus Plan documentation on important aspects of vegetation dynamics.Our most significant challenges related to vegetation mapping errors, ecosystem changes during the analysis process due to large wildfires, expected vs. realistic analysis time frames, complications related to the agency-academic partnership, and turnover in key team positions.Significant accomplishments were related to the confidence and quality of the completed analysis and planning documents, formation of lasting partnerships, significant gains in understanding of local ecosystem processes, and the effectiveness of proposed management actions.
Methods used to parameterize the spatially-explicit components of a state-and-transition simulation model
Rachel R. Sleeter, William Acevedo, Christopher E. Soulard, and Benjamin M. Sleeter
U.S. Geological Survey, Western Geographic Science Center
Spatially explicit state-and-transition simulation models (STSMs) of land-use and land-cover (LULC) change increase our ability to assess regional landscape characteristics and associated carbon dynamics across multiple scenarios. However, without characterizing appropriate spatial attributes such as forest-age, patch metrics and land-use distribution, a STSM would randomly place LULC change on the landscape to meet projected demand. This paper describes the methods, source data and rationale used to parameterize the initial LULC conditions used in a spatially explicit application of a STSM. Our approach for setting initial conditions was to adapt data derived from multi-temporal harmonization of independent datasets that monitor land-use in addition to land-cover. Harmonization combines many national-level, LULC datasets through a pixel-based data fusion process to generate LULC data coherent with the classification structure of USGS Land Cover Trends (LCT) data. Existing data from the North American Carbon Program was used to parameterize forest-age. To constrain or magnify potential LULC transitions, spatial multipliers were developed that reflect external drivers influencing the probability of change at each pixel. Distance-decay theory was used to parameterize variables related to forest harvest, agriculture, grass-shrub, and urban development. Preliminary results include a 2000 LULC dataset at 1-km resolution with 8 Level-1 classes consistent with the LCT statistical estimates. STSM runs are presented to compare each major land-use transition given the presence or absence of spatial multipliers. By incorporating highly parameterized model inputs into a spatially explicit application of a STSM, the pattern and spread of LULC transitions are grounded in spatially representative landscape characteristics.
Developing state-and-transition simulation models for the Rangeland Vegetation Simulator using NRCS ecological site descriptions
Leonardo Frid1, Colin Daniel1,2 and Matt Reeves3
1Apex Resource Management Solutions Ltd.
2University of Toronto
3USDA Forest Service, Rocky Mountain Research Station
The Rangeland Vegetation Simulator (RVS) is a module for the Forest Vegetation Simulator (FVS) currently being developed by the US Forest Service Rocky Mountain Research Station. While FVS can simulate tree lists and stand level vegetation dynamics for forests; the RVS is intended to extend the capability of FVS such that it can estimate succession, fuels and biomass for non-forested vegetation communities under alternative management and natural disturbance scenarios. In order to extend FVS to non-forested vegetation communities, the work underway for the RVS will build on a large existing body of work by the USDA Natural Resource Conservation Service (NRCS) on classifying and describing rangeland vegetation dynamics according to ecological sites. The NRCS has built up a nationwide database of Ecological Site Descriptions (ESDs). With the NRCS Ecological Site Information System, every acre of rangeland is assigned to an ecological site, based on its long-term ecological potential due to climate, soils and topography. When completed, the database of ESDs will contain a detailed description of the biophysical characteristics of each ecological site including a conceptual state-and-transition model of the vegetation dynamics for that ecological site. In this presentation we use a sample ecological site to demonstrate how these conceptual models can be digitized into state-and-transition models as a tool for quantifying rangeland vegetation dynamics for the RVS.
Optimizing buffelgrass management with differing levels of aridity and treatment options
Catherine S. Jarnevich, Tracy R. Holcombe, and Catherine Cullinane Thomas
U.S. Geological Service, Invasive Species Science
In southern Arizona and elsewhere in the U.S., African buffelgrass (Pennisetum ciliare) is spreading exponentially and across multiple habitats and jurisdictions, creating a novel fire risk and transforming natural ecosystems. Resource managers have limited resources to expend, and need to know the optimal division between treatment, monitoring, and control. Our objective was to examine the potential effect of different precipitation/ mortality effects and management results on the distribution and abundance of buffelgrass in Ironwood National Monument. We analyzed treatment effectiveness with different levels of aridity and 12 management scenarios using TELSA. Scenarios suggest a direct relationship between resources spent and the magnitude of invasion on the landscape, and the rate of invasion over time decreases with increased management effort. To achieve an actual reduction and stabilization of buffelgrass populations, management unconstrained by fiscal restrictions and across all jurisdictions and private lands is required; without broad and aggressive management, buffelgrass populations are expected to increase over time. Large upfront investments with minimal spending in the future results in the most efficient use of resources to achieve lowest invaded acreage in the future. _________________________________________________________________________________________________________________________
Holding the Line: Modeling Invasive Buffelgrass in the Cactus Forests of Arizona
Jim Malusa1 and Priya Shahani2
1School of Natural Resources and the Environment, University of Arizona, Tucson, Arizona 85721
2US Forest Service, Southwestern Region Office, 333 Broadway Blvd SE, Albuquerque, NM 87102
We provide an example of how State and Transition Modeling can help land managers visualize the consequences of an invasive grass on a native ecosystem with high habitat and aesthetic value. The invasive is the perennial buffelgrass (Pennisetum ciliare), and the native ecosystem is Sonoran Paloverde Mixed Cacti Desert Scrub (SPM) system in southern Arizona. The hallmark species of the SPM ecosystem are the giant saguaro cactus and the diminutive palo-verde tree. The native species are not adapted to fire; in contrast, buffelgrass not only facilitates fire, but increases in density afterwards. Buffelgrass also out-competes native shrub species such as brittlebush. Across the border in Mexico, buffelgrass has already transformed large swaths of the SPM ecosystem into a grass savanna.
We built a 10-state model describing this ecosystem, as part of the Coronado National Forest Plan. Three of the 10 states are characterized by buffelgrass: initial colonization, where buffelgrass comprises < 5% cover; acute infestation with 5-50% cover; and >50% cover and the loss of native shrubs/trees due to competition and fire. We estimated, with difficulty, the relative proportions of these three states, and the rate of spread, and found that the Coronado NF needs to kill at least 1250 acres of buffelgrass per year to hold the line on the spread of buffelgrass at about 15% of the 37,000 acres of SPM within the Coronado NF. In contrast, treatment of 500 acres per year will result in the loss of about 90% of the SPM ecosystem in 50 years.
Changes in community exposure to tsunami hazards in the U.S. Pacific Northwest based on future land change scenarios
Christopher E. Soulard1, Nathan J. Wood2, Tamara S. Wilson1
1US Geological Survey, Western Geographic Science Center, 345 Middlefield Road, Menlo Park, CA 94025
2US Geological Survey, Western Geographic Science Center, 2130 SW 5th Avenue, Portland, OR 97201
Tsunamis have the potential to cause considerable damage to coastal communities throughout the Pacific Northwest. The relative impact of tsunamis can change as communities expand within and outside of the tsunami-inundation zone. Unmitigated land use change may leave communities more susceptible to hazard-related losses due to increased overall exposure of developed land, yet hazard vulnerabilities may be resolved by adaptive management of growth. To understand potential future community vulnerability to tsunamis, four spatially explicit future scenarios of developed land use were modeled spatially out to 2050 using state-and-transition models and examined spatially against Cascadia Subduction Zone tsunami-inundation zone maps. For each scenario we assessed how tsunami vulnerability would change for communities along the coasts of California, Oregon, and Washington through 2050. Our results can assist local governments in visualizing alternative future development patterns, improving land-use and emergency planning, and minimizing hazard-related losses.
To burn or decay: States and transitions in mesic and dry mixed-conifer models for the Arizonan Sky Island ecosystem
C.P. Wilcox, A.M. Lynch, P.C. Shahani, R.L. Biggs, C.H. Stetson, R.,Weisz, J.M. Ruyle, Y. Begay
U.S. Forest Service, Rocky Mountain Research Station
We modified and calibrated Southwestern mesic mixed-conifer (MMC) and dry mixed-conifer (DMC) state-and-transition models (STMs) for forests in southern Arizona for use in the Coronado N.F. Forest Plan Revision process.The general Southwestern STMs represent a wider range of conditions and disturbances than what is found locally, particularly with respect to tree species composition and insect outbreaks, making direct application or calibration to local conditions difficult.The modified models are somewhat simpler, replace some insect-driven transitions with decay-driven transitions (there are fewer tree species and insect disturbance agents in the Sky Island mixed-conifer ecosystems), and incorporate transition rates calibrated to local disturbance histories, particularly post-fire transitions observed after multiple events of low-, moderate-, mixed-, and high-severity fire.The DMC model is significantly more complicated than the MMC model, though ecologically it appears to be a simpler system.To some extent this represents greater local knowledge about the more abundant DMC (88% of the MC), with more states represented and known.But it also reflects the disruption of DMC processes by fire-exclusion.The DMC is the most productive of the fire-driven ERUs, and it quickly and robustly responded to fire exclusion, resulting in increased tree densities and fuels, and uncharacteristic fire.
Fusing State-Transition Models from Ecological Sites with the Rangeland Vegetation Simulator
Matt Reeves1, Leonardo Frid2 and Colin Daniel3
1USDA Forest Service, Rocky Mountain Research Station
2Apex Resource Management Solutions Ltd.
3University of Toronto
The Rangeland Vegetation Simulator (RVS) is designed to estimate growth, succession, fuels and effects of disturbance of non-forest systems. Presently, the RVS uses the Biophysical Settings (BPS) data product from the LANDFIRE Project to describe vegetation succession. Biophysical Settings are designed to depict succession (e.g. changes in species composition and structure) from a Clementsian, prior to Euro-American settlement. Contemporary ecological theory, brought on by the presence of invasive species and new understanding of vegetation dynamics, suggests that the linear succession inherent in the BPS data product is flawed for describing ecological change. Therefore, the RVS is being reprogrammed to utilize State-Transition models from Ecological Sites that have been digitized and processed with St-Sim, providing Monte Carlo analysis of vegetation composition and structure corresponding to user-designed management scenarios. This poster presents a conceptual flow and data processing of the RVS from user input to resulting information and output.
Predicting Coastal Wetland Restoration Outcomes Using State-and-Transition Simulation Models
Chantal Vis1, Colin Daniel2 and Josh Keitel2
1Parks Canada, Protected Areas Establishment and Conservation Directorate, 25 Eddy St., 4th floor, Gatineau, QC K1A 0M5
2University of Toronto, Dept. of Ecology and Evolutionary Biology, 25 Willcocks St., Toronto, ON M5S 3B2
3Parks Canada, Lake Louise, Yoho and Kootenay Field Unit, P.O. Box 220, Radium Hot Springs, BC V0A 1M0
Fire Regime Syntheses in the Fire Effects Information System (FEIS)
Robin J. Innes, Ilana L. Abrahamson, Kris Zouhar, Janet L. Fryer, and Jane Kapler Smith
Fire Modeling Institute, Missoula Fire Sciences Laboratory, Rocky Mountain Research Station, Missoula, MT
Managers need up-to-date, science-based information on fire regimes to make sound decisions for wildland management. The Fire Effects Information System (FEIS, http://www.feis-crs.org/beta/) is now providing a product that addresses this need: a collection of Fire Regime Syntheses, which synthesize the growing body of knowledge on historical fire regimes and potential fire regime changes. Fire Regime Syntheses bring together information from the scientific literature and state-and-transition models for LANDFIRE Biophysical Settings (BpS) and their associated geospatial data (http://www.landfire.gov/fireregime.php). Each Synthesis describes a group of BpSs, selected on the basis of available literature and similarity in ecology or fire regime. Syntheses contain information on historical fire frequency, spatial pattern, extent, seasonality, ignition sources, and typical patterns of fire intensity and severity, plus an analysis of information on contemporary changes in fire regimes, especially regarding changes in fuels. Each Fire Regime Synthesis links to related FEIS Species Reviews. In December 2014, fire regime information will be available in FEIS for all 1,100 Species Reviews. Fire Regime Syntheses will enable LANDFIRE to incorporate the latest science on historical fire regimes into BpS model revisions.
State-and-Transition Simulation Modelling: Status and Future Use of Parallel Computing
Michael S. O’Donnell
U.S. Geological Survey, Fort Collins Science Center, 2150 Centre Avenue, Bldg. C, Fort Collins, CO 80526, USA
State transition modelling requires ecological data and knowledge of vegetation states (communities) and transitions between communities, but it also requires a knowledge of computing resources that ecologists and rangeland managers often lack. Executing complex models and/or using regional data with serial processing could prohibit research because of unrealistic computation time, which often results in curtailing hypotheses. To address this challenge we explore decomposing Monte Carlo simulations of state transition models into parallel tasks to reduce overall computing time. ST-STM1, a plugin for SyncroSim1 software, supports decomposition of models on a single multicore machine; though vertically scaling single multicore machines has limitations. We demonstrate that complex models will scale better in distributed computing environments, also known as horizontal scaling. High-throughput and high-performance computing (distributed computing) enable researchers to distribute the workload on multiple machines, thereby supporting larger landscapes (spatial extents), higher spatial resolution vegetation products, more complex models, and sufficiently abundant Monte Carlo simulations in a timely manner. The objectives of this study include identifying challenges of executing state-and-transition models, identifying common bottlenecks of computing resources, developing software that enables parallel processing of Monte Carlo simulations, and evaluating the advantages and disadvantages of different computing resources. Our results indicate significant advantages to using decomposition and parallel computing of larger and more complex state-and-transition models with distributed computing. Smaller data sets and moderately complex models do not necessarily warrant using high-throughput or high-performance computing, but decomposition and execution of simulations on a multicore machine will reduce overall computation time.
1Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Landscape Conservation Forecasting: Creating Effective and Efficient Conservation Plans
Jeff Campbell, Louis Provencher, Leonardo Frid, Greg Low and Jim Smith
1Spatial Solutions Inc.
1The Nature Conservancy of Nevada
1Apex Resource Management Solutions Ltd.
1The Nature Conservancy LANDFIRE Team
An overview of Landscape Conservation Forecasting (LCF) is presented. We will show the basic steps involved in applying the LCF concept, and the information requirements. The benefits of the process will be demonstrated using a real example from an LCF partner.
Supporting State-and-Transition Modeling: Online Resources from LANDFIRE
Kori Blankenship, Randy Swaty, Sarah Hagen, Jeannie Patton and Jim Smith
The Nature Conservancy LANDFIRE Team
The LANDFIRE Program has been utilizing state-and-transition modeling since its inception in 2004, and provides a vast, national library of reference condition models to any interested party free of charge. To assist those accessing LANDFIRE models and all state-and-transition modelers, LANDFIRE and partners have created a number of support resources that are available online. We will describe those resources and how to access them.