Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Australie → Forest Flammability : Modelling and Managing a Complex System

University of New South Wales (2011)

Forest Flammability : Modelling and Managing a Complex System

Zylstra, Philip John

Titre : Forest Flammability : Modelling and Managing a Complex System

Auteur : Zylstra, Philip John

Université de soutenance : University of New South Wales

Grade : Doctor of Philosophy (PhD) 2011

Résumé
Wildland fire has considerable influence on both natural and anthropogenic environments and consequently, the ability to understand, predict and manage it has become a growing priority as human populations have increased their influence upon and awareness of the natural, fire prone environment. Despite this, in Australia a disconnection and failure to transfer traditional understandings of the subject into the modern context has coupled with an inadequate and frequently simplistic grasp of fuels and their implications for management. This has been manifested in ineffective fuel management, a low level of confidence in fire behaviour predictions and a paucity of peer-reviewed science on the subject. Concepts of forest flammability still do little to address concerns and observations that were raised in some priority fuel arrays over a century ago. This thesis attempts to address the dilemma by identifying where the gaps in Australian fire knowledge lie and by offering a unique approach to the situation using complex systems modelling. A structure is proposed that by taking a literal approach to modelling the propagation of flame through and between plants and fuel strata creates a transparent framework for understanding and analysing fire behaviour. Fire spread is conceptualised as the interaction of the three aspects of flammability – ignitability, combustibility and sustainability with the geometry of the fuels. Fire spread occurs when a critical state is satisfied ; once the new fuels are alight the resulting flame dimensions are calculated and the process is repeated. Fuels are defined as any dead or live plant material that undergoes combustion and thereby contributes to the fire behaviour. This definition excludes a priori measurements as the three dimensional structure of a forest imposes a circular feedback between flame dimensions and fuel availability. This level of complexity is critical as the same vegetation that acts as fuel may also act to suppress fire spread by affecting drying processes or reducing wind speed at different heights in the forest. As a result, the concept of simple fuel reduction or the reduction of carbon storage or biomass as a proxy for managing flammability is shown to be fundamentally flawed. It is proposed that effective management of the flammability of forests or any other fuel array is only possible by understanding the complex relationships between potential fuels and forest structure. To achieve this, the necessary sub-models required to create a fully operational fire behaviour model are identified, sourced from the literature or developed where necessary. New models were constructed to describe the ignitability, combustibility and sustainability of flame in individual leaves, as well as describing external influences on fire behaviour such as the effect of slope on flame angle and the leaf area index of the forest on wind speeds at different heights. Plant moisture models were developed for six species to enable validation of the model, and the Keetch-Byram Drought Index was tested as a model of soil moisture. The models used are provided as “first generation” models – capable of fulfilling the role while identifying where future work is required to improve on them. The complete fire behaviour model was constructed in an Excel spreadsheet and validated with an extreme condition test, predictive validation and a credibility analysis. The extreme condition test demonstrated that the model was not subject to the same domain considerations of empirical models but that modelled fire behaviour was inherently limited by the physical processes which underlie the model. Model accuracy was compared with the performance of three widely used Australian models and provided a improvement of 4 to 12 times greater accuracy to all three on rate of spread, which was statistically significant for two of the models It also demonstrated up to 12 times greater accuracy when estimating flame heights, although this was only significant against one model. The credibility analysis tested all four models against decision thresholds for prescribed burn planning, determining initial attack success against unplanned fires, determining the attack method to be employed against unplanned fires and as tools for forward planning in wildfire management. The model was slightly out performed on one test but performed more reliably than two of the other three on all other counts except for forward planning in incidents, where it significantly out-performed all other models. Suitable applications of the model were examined and examples given, demonstrating that new areas of research into fuel management, fuel-weather interactions and feedbacks between fire and climate change were now possible due to the model. The post-fire succession of an area of sub-alpine forest was used to model the changes in flammability with time since fire as an example application. The results demonstrated that in these conditions, the fuel-age paradigm of fire management was an inappropriate simplification that if followed would produce counter-productive results in some environments. Further examples of fire-weather interactions were given and the implications for fire management and climate change examined.

Mots Clés : Fire ecology — Fire behaviour — Fuel management

Présentation

Version intégrale (6,32 Mb)

Page publiée le 29 novembre 2014, mise à jour le 15 juillet 2017