Generating a framework for fuel inputs to future fire behaviour models: reviews, recommendations and remote sensing

Abstract

Land managers use models to understand potential fire behaviour, which provide insight into the social, economic and environmental implications of fire. Although combustion is a fundamental process, numerous fire behaviour models exist across the world, each with model-specific inputs and outputs. While these are useful for local-level fire management, vegetation- or country-specific fire behaviour models limits knowledge-sharing between jurisdictions – largely because model inputs (particularly fuel arguments) and outputs differ, meaning they cannot be readily compared. At the same time, advancements in remote sensing techniques have resulting in accurate methods to estimate fuel for current fire behaviour models that have not been fully integrated. This project has two key aims: a) to utilise remote sensing and field-based measurement approaches to develop a comprehensive set of model parameters that can be used to universally characterise fuel attributes fundamental to fire behaviour, and b) based on knowledge gaps identified, undertake expert elicitation to identify fuel attributes for fire behaviour not currently utilised in models, and propose methods to estimate these values using remote sensing. To achieve this, we reviewed 25 fire behaviour models to identify similarities and differences in model inputs and outputs. We then subset model inputs to the fuel parameters, and linked each to current remote sensing methods. Following the review, and in conjunction with an expert elicitation process, we developed a list of fundamental fuel attributes missing from current fire behaviour models. From this list, we developed novel fuel parameters that fully utilise information contained within remote sending datasets. The final step in this process is to validate the importance of each fuel argument using further expert elicitation and field burning experiments*. We found many common parameters, including a fuel value, short- and long-term fuel moisture, temperature and wind variables, however, the physical fuel parameters are typically where models diverge. Models across the world use fuel inputs that describe the amount, horizontal and vertical arrangement of fuel, bark hazard and the importance of crown fuels. Preliminary expert elicitation identified shared and divergent opinions related to fuel characteristics that drive fire behaviour. While common parameters related to horizontal and vertical arrangement of fuel were important, gaps exist in our capacity to describe ladder fuels. From these results, we have developed novel remote sensing metrics to describe vertical and horizontal connectivity. Throughout 2022, we aim to validate these preliminary results through further expert elicitation and field observations of fire behaviour. Through this work, we have developed a complete list of fuel parameters for fire behaviour and related these to remote sensing methods. This provides a framework for current models to shift to remotely sensed inputs, and a basis for future fire behaviour models. Overall, we argue that this work provides the foundation for a universal fuel assessment methodology to be developed, that is scalable and captures change over time, that can link physical fuel structure and remote sending techniques to fire behaviour models of the future. * Note that this work is planned for late 2022.