LINK-3: Interregional fire-fuels-climate linkages

Title:  Predicting future and understanding past fire dynamics in the western U.S., Tasmania, and New Zealand

Investigators:  Keane, Yospin, Perry, all PIRE investigators

Project Description

Background:  While field studies of wildland fire dynamics provide the foundation for fire sciences, ecological modeling has a critical role by allowing field data to be integrated with other ecological research to explore fire interactions in space and time.  Landscape simulation modeling provides an effective, standardized, and objective context to evaluate fire regimes and ecological change and can be used to explore fire, climate, and vegetation interactions and to quantify fire regimes in space and time. Most importantly, modeling can help predict potential fire dynamics under future climates to provide fire scientists with critical information to mitigate any adverse effects. Last, and most importantly, models can be used to extrapolate and expand field study results across large spatial scales and test hypotheses about past and future changes across multiple temporal scales.   

This synthesis project uses modeling and iterative data-model comparisons to explore the interactions of vegetation, climate, and disturbance at different spatial scales and temporal scales in WildFIRE PIRE study areas.  The modeling experiments will be driven by insights from the paleo- and historical information, as well as future climate projections.  They will serve as a tool to identify the drivers of major thresholds in landscape dynamics, including the consequences of temporal and spatial changes in vegetation (e.g. composition, distribution, successional stage), climate (e.g., fuel moisture, fire weather), fire (e.g. frequency, extent, ignition distribution), and fuel (e.g. fuel loadings, fuel classes). 

Different grid-based models and modeling approaches are available for the western U.S., Tasmania, and New Zealand that relate to WildFIRE PIRE.  Dynamic ecosystem process models (Fire-BGCv2, Firescape and other GCTE models) have been used to simulate landscape dynamics in several vegetation types in the northwestern U.S. and southwestern Tasmania, including areas where WildFIRE PIRE studies are underway (Cary et al., 2006; Keane et al., 2010; King et al., 2008a,b, 2006).  In New Zealand, landscape fire-succession models have been developed to integrate modules and functions that explicitly represent human activity (Perry et al., in review).  In this approach, plant-functional types (including flammability traits) are used to represent spatial and temporal competition for resources (water and light) in a rule-based modeling framework, and wildfire behavior is represented using a cellular-automata model of fire spread.   

Both modeling approaches require accurate parameterization of key landscape-level processes. Data are needed to describe the abiotic (soils, topography, and climate) and biotic (overstory and understory vegetation and fuels) landscape conditions currently present across a study area.  Such parameterization of FireBGCv2 has been done in the western U.S., but not Tasmania or New Zealand.   Some datasets and output for southwestern Tasmania are available from the Firescape modeling experiments of Geoff Cary and Karen King (King et al., 2008a, b; 2006).  In New Zealand, parameterization of a fire simulation is underway by Perry.

Objectives:  We propose a flexible modeling approach in Syn-3.   At the least, FireBGCv2 will be used to simulate fire-climate-vegetation dynamics for Rocky Mountain landscapes.  However, there are few data sets for parameterization, initialization, and validation of Tasmania and New Zealand landscapes for the FireBGCv2 model.  As a result, FireBGCv2 will NOT be used in New Zealand and a less data-intensive landscape fire-succession modeling approach will employed to match the resolution of the available New Zealand data.  For Tasmania, the quality, quantity, resolution, and availability of FireBGCv2 data requirements are unknown but it is possible that there are enough data for an acceptable FireBGCv2 simulation.  We will spend the first six months in 2012 evaluating the data for FireBGCv2 simulation, and if sufficient, we will apply FireBGCv2 to a Tasmanian landscape.  This Tasmania modeling is a new effort that will be useful for interpreting new PIRE datasets and comparing with Cary and King’s modeling studies in southwestern Tasmania and modeling experiments underway in the northwestern U.S. 

Modeling efforts will focus on testing specific PIRE-related research hypotheses, such as:

  • What would happen to New Zealand’s vegetation if fire frequency increased or decreased during Holocene (NZ-1), the IBP and subsequent Maori period (NZ-2), and at present and in the future (NZ-3)? 
  • Are Tasmania fire regimes and vegetation patterns explained by targeted versus random ignitions during the Holocene (Tas-1, Tas-3)?  What role has fire played in the expansion and maintenance of Athrotaxis forest (Tas-1, Tas-2)?  How does fire contribute to fine-scale vegetation mosaics over the last 3000 years (Tas-4)?
  • How have changes in fire regime and climate altered lower forest dynamics in the northwestern U.S. during the Holocene (US-1) and in the last few centuries (US-2, US-3)? 
  • How does fire influence the probability of plant invasion in native forest and steppe communities (Syn-1, NZ-2, NZ-3)?  How are disturbance synergies amplified under different climate and land-use scenarios (Syn-1)?
  • What shared linkages exist between fire, climate, and land-use change in PIRE study areas that help understand response to projected climate change in the future?

Work Plan:

The Fire BGCv2 effort will focus on two 50-100K ha landscapes, one in Yellowstone National Park (YNP) and the other in interior Tasmania within the current range of Athrotaxis, providing sufficient data exist for running FireBGCv2.  Landscapes that include heterogeneous vegetation types will be targeted for landscape fire-succession models. The resolution, extent, and detail of the landscapes will be decided in future, and all input map layers, parameterizations, and initial conditions will be documented and posted to the web site.  The YNP study landscape has been sampled and all digital maps have been collected and can be posted by September 2011.  Information on the Tasmania study landscape will be collected if not already available and a decision to simulate the Tasmania landscape will be made by June 2012.

We will explore landscape-climate-fire-vegetation dynamics using FireBGCv2 and all other models using a simulation experiment that employs scenarios to describe model behavior under various conditions.  Under this format, we will specify a set of scenarios that are designed to emphasize differences in model behavior over diverse climate, fire, and vegetation conditions.  Presented here is the draft experimental design; the actual simulation experiment design will be formalized over Year 2 through interaction with modelers and Wildfire-PIRE collaborators.  The following primary scenarios and levels will be simulated:

  • Climate.  Four different climate scenarios will be simulated
    • Historical climates.  Paleoclimatic scenarios will be developed by the Wildfire-PIRE scientists and used as input to the model.
    • Contemporary scenario.  Taken from weather data collected for the target landscapes.
    • A Warm Moist future scenario will be compiled from the seven GCMs for each landscape.
    • A Warm Dry future scenario will be compiled from the seven GCMs for each landscape.
  • Fire Ignitions. Several fire scenarios will be simulated to mimic past, current, and future fire ignition patterns.
    • Lightning ignitions. Random ignition patterns will be simulated.
    • Targeted ignitions.  Ignition patterns will be simulated based on assumptions of prehistoric land-use activities and travel routes
    • Future ignitions. A digital layer of future ignition probabilities will be developed based on predicted land use and fire management plans.
  • Exotic Invasion.  Two scenarios will be used to explore the effect of exotic invasions on fire regimes and vegetation dynamics.
    • No exotics.  Only native species will be modeled.
    • Exotics.  Exotic species will be included in the simulation.

Many response variables will be used to detect changes between scenarios and models.  These response variables are scale dependent and reflect the variables needed to successfully complete the simulation objectives.  They are output in tables and digital maps, and they can be augmented with others later.  FireBGCv2, for example, will focus on

  • Landscape. Fire rotation (yr), fire return interval (yr), carbon (kg/m2, NPP, NEP, GPP, NEE, fireC, abovegroundC, belowgroundC)
  • Stand. Fuels (coarse woody, fine woody, litter, duff, shrub, herb), carbon (same as above), structure (dominant species, canopy cover, basal area, density), exotics (abundance)
  • Fire. Fire intensity, severity, tree mortality, fuel consumption, carbon emissions.
  • Tree. Age structure, species distribution and size classes.

Models will be initialized using local vegetation, fire, topography, soils, and climate data collected for each landscape.  Parameterization of the models will be done for each landscape using data collected during past studies, literature reviews, and Wildfire-PIRE data collection.   We will run the model for 500 years, and output response and explanatory variables are available as decadal timesteps.  We will also output maps of fire intensities, severities, and vegetation composition.  Results will be analyzed using MANOVA, regression tree, and multivariate techniques to determine the subtle differences and similarities between and across the three landscapes by scenario.

WildFIRE PIRE will hire a post-doc to lead Syn-3.  This person will serve as an interface between modeling and empirical activities, and lead synthesis publication efforts.  We seek a scientist with ecological training and strong numerical skills who can work closely with all WildFIRE PIRE PIs to understand the research questions and methodological approaches. 

Specific tasks of the post-doc:

  • Evaluate data availability, quality, and resolution in Tasmania, and if sufficient, parameterize FireBGCv2 to run on an interior Tasmania landscape (a 40 km2 area near Cradle Mountain was discussed given the intensity of PIRE-related work there).  This task will require a stay of at least 6 months at the USDA Forest Service Fire Science Lab, working with Keane’s group, as well as time in Australia working with Cary, King, Bowman, Haberle, Fletcher, and King. 
  • Work with the PIRE team to design, run, and analyze model experiments that test paleo, historical, and modern hypotheses concerning the drivers and consequences of landscape change.
  • Coordinate modeling efforts underway in the western U.S., Tasmania, and New Zealand to ensure that approaches help achieve WildFIRE goals for interregional and interhemispheric comparison.
  • Take the lead in developing, writing, and publishing synthesis products and publication related to WildFIRE PIRE projects.

Related Activities 

These synthetic activities tie into ongoing modeling work of Keane and Rachel Loehman (FSL) in the western U.S., Perry and others in New Zealand, and Cary and colleagues in Tasmania.  It also provides a platform for broader understanding and comparison of all WildFIRE PIRE projects. 


Manuscripts will present the results of the simulation experiments; manuscripts on data-model comparisons using historical and paleodata of postfire dynamics, disturbance interactions, and fire-climate-human interactions.

Year 2 Update:


This project represents the culmination of several PIRE activities through modeling and data-model comparisons, and its activities are just beginning.  At the moment, Syn-3 focuses on the use of FireBGCv2 and other fire models to develop scenarios of climate change and human activity.  Once the FireBGCv2 model is parameterized for Tasmania, for example, we plan on simulating a wide variety of fire management, climate, and fuel scenarios and comparing the results against similar scenarios in comparable U.S. settings.  This will help us understand the short- and long-term controls on fire regimes and fire occurrence in different settings.

Discussion of Syn-3 activities is still underway and will continue at the 2012 US PIRE meeting (Bozeman, June) and the 2013 All-Hands PIRE meeting (Dunedin, January 19-20).  We have written and submitted a conceptual paper outline a framework for examining interregional fire-fuel-climate linkages and WildFIRE PIRE goals (McWethy et al., in review).