• 15-16 Nov. 1999 Landscape Fire Modeling Workshop

• 18-19 Nov. 1999 Fire Model Comparison Workshop

• Global change impacts on landscape fires

15-16 Nov. 1999 Landscape Fire Modeling Workshop, Pacific Forestry Centre, Victoria, British Columbia, Canada. Contacts: Mike Flannigan or Brad Hawkes

About 50 people from around the world attended a landscape fire modelling workshop in Victoria, BC at the Pacific Forestry Centre on Nov 15 &16 1999. This workshop was sponsored by the Canadian Forest Service, the Canadian Climate Action Fund and was supported by GCTE Activity 2.2 - Landscape Processes. The purpose of this workshop was to explore opportunities and identify limitations of landscape fire models for boreal and temperate ecosystems. Keynotes presentations included an overview of landscape fire models, dynamic global vegetation models, fire effects/succession modelling, level of protection modelling and application in forest/vegetation management. A number of fire models and modelling shells were presented. Breakout sessions on dynamic global vegetation models, modelling gaps and needs, application of fire models for forest management and future directions were held on day 2 of the workshop. This workshop was an excellent preamble for the GCTE Fire Model Comparison Workshop held in Victoria on Nov. 18-19. A publication that will summarize the results of the Landscape Fire Modelling Workshop will be available by April 2000 from Mike Flannigan, Canadian Forest Service, 5320 - 122nd St, Edmonton, AB T6H 3S5, Canada

18-19 Nov. 1999 Fire Model Comparison Workshop, Pacific Forestry Centre, Victoria, British Columbia, Canada.
Contacts: Sandra Lavorel, Mike Flannigan

Executive summary

Eighteen participants from North America, Europe, Australia and South Africa attended a landscape fire model comparison workshop at the Pacific Forestry Center, Victoria, British Columbia on the 18-19 Nov. 1999. The objectives of the workshop were to initiate a landscape fire comparison task by: identifying main models and interested authors; designing a plan for model comparison experiments; devising an implementation strategy for the comparison exercise.

Three questions were identified, which should be addressed by the comparison exercise:

1. How well do different landscape fire models reproduce fire statistics under current climate, both with respect to each other and with respect to fire history records at selected sites?
2. At what spatial and temporal scales does landscape pattern matter to fire regime?
3. Under which weather conditions is fire propagation (in different models) sensitive to fuel landscape pattern, and how often under present / future climate is the threshold of insensitivity crossed?

General principles for model comparisons guided the discussions. Comparison starts by spelling out hypotheses and domain of applicability for each model. Each model also needs to be submitted to an individual sensitivity analysis as a preliminary step to comparison. Then only can models be run against one or a few selected data sets. Comparisons are made on outputs from multiple simulations of each model.

A model ID card was developed to serve as a framework for sampling existing models. Fields to be filled out for each model included:

• Information about model context in terms of ecosystem type, climate, scales, initial purpose and modelling / computing constraints;
• Description of vegetation representation and succession methods;
• Description of fire module: separating fire modelling into initiation, ignition, propagation and extinguishment. For each of these the type of approach and parameters used need to be specified;
• Model outputs for vegetation structure and fire regime; priority outputs to be required for comparisons were identified.

Landscape fire model comparisons will now follow two parallel streams of activity:

1. Full model comparisons: will parameterize existing models for a selection of sites and corresponding data sets. A preliminary list of target sites was outline based on representativity of major world ecosystems (except tropical rainforests) and known data availability and modelling expertise. This activity will be based on expert workshops held at the sites of interest. The objectives will be to compare vegetation and fire regime outputs from alternative models against spatial and temporal records.
2. Model component, or 'toolbox', comparisons: will implement methods from different models for each fire component (initiation, ignition, propagation, extinguishment) under a common modelling environment, for example LAMOS (Davies et al. 1999, LAMOS - A LAndscape MOdelling Shell, Technical reference document). Systematic comparison for each component will be carried out for a small number of selected data sets. The general principle is that of standardization of all other model components while studying one component at a time. Additionally, new models may be created by hybridizing methods across existing models. A proposal to NCEAS for a series of working group and a research fellowship is being prepared to cover this activity over the mid 2000 - end 2001 period.

Global change impacts on landscape fires. Contacts: Mike Flannigan and Sandra Lavorel.

NCEAS first working group – Santa Barbara 30 Oct. – 3 Nov.

The objectives of this first workshop were to design and organise landscape fire succession model comparisons to be carried out over one year of work.

We first identified general goals for these comparisons with respect to coarse scale modelling (regional, global), landscape modelling, and land management. Common goals included the need to identify a minimum set of processes and corresponding methods to be included in models so as to reach maximum simplicity, ability to move across scales, and capture the main sensitivities of fire regimes to climate and landscape patterns.

The group will produce a classification scheme for landscape fire succession models that will be used as a key to the large number of available models by specifying their main assumptions and domains of application.

We developed a strategic plan for comparing existing models and methods, of which the first step will be carried out by the working group over the next year. This first factorial comparison experiment will examine process sensitivities within existing models by using alternative parameterisation for climate scenarios, topography and vegetation pattern. The group will analyse effects of modelling approaches for a list of output variables tat capture fire regimes and landscape vegetation pattern.

Later steps will involve mixing and matching of modules  from components in different models using landscape modelling shells (e.g. LAMOS, SELES), as well of re-parameterisation of original models across biomes.

View the summary for NCEAS landscape fire modelling working group30 OCT. - 3 Nov. 2000 powerpoint presentation.