Developing a Risk Assessment Approach for Forest Fire at the Rural-Urban Interface: Potential of the Wildfire Threat Analysis Framework. Final report

EXECUTIVE SUMMARY
The report covers a scoping study on risk assessment for forest fire at the rural-urban interface undertaken by the University of Manchester, working alongside Forestry Commission England (FCE) and Forest Research (FR). The research was funded by the Natural Environment Research Council (NERC) under the Probability Uncertainty and Risk in the Environment (PURE) Associates Programme.
The scoping study tests the applicability of the Wildfire Threat Analysis (WTA) technique from New Zealand to a UK context and at a local scale. WTA treats wildfire threat as the cumulative combination of three aspects: ignition potential (risk of ignition), potential fire behaviour (hazard of fire spread), and values put at risk as a result (assets, including life and well-being). These are represented by three GIS modules, each comprising a set of layers of geographical information, organised into components and sub-modules.
The threefold WTA framework mapped reasonably well onto two existing risk assessment frameworks used in the UK. In the National Risk Register, risk is a function of likelihood and impact. Risk here approximates to threat in WTA, and likelihood to risk of ignition together with hazard. NRR impact is captured partially by the values at risk inventory, and partially by magnitude of the hazard. Crucially, it also requires a better understanding of the relationship between intensity of the fire hazard and degree of damage produced.
In the Crichton risk triangle used for flooding, and by the IPCC for climate change related risks, risk is seen as a function of hazard, exposure of assets to the hazard and also their vulnerability. Risk here again approximates WTA threat, and hazard to WTA’s risk of ignition together with hazard. Exposure equates to the chance of WTA values coming into contact with fire, thus in a way reflecting the extent of the area affected by fire under the hazard module. Vulnerability is an attribute of values at risk, describing the potential harm to people or assets if they are exposed to fire. It has no direct counterpart in the WTA framework, so we modified values at risk to include social vulnerability, but did not include vulnerability of ecosystems services or infrastructure.
Appropriate spatial information was collected for an 11 km by 12 km case study area on the Berkshire, Surrey and Hampshire borders, centred on Crowthorne Wood and Swinley Forest. The three local Fire and Rescue Services (FRS) have attended nearly 1000 vegetation fires in four years in this area, including a major, politically significant fire in Crowthorne Wood in Spring 2011 (the ‘Swinley Forest’ fire).
Two of the three GIS modules, that is, risk of ignition and values at risk, were developed using a 25m cell size, which is appropriate to the local scale at which management is undertaken. The hazard module could not be constructed due to lack of fire weather data at the time, but simulations of fire spread for the Crowthorne 2011 fire were done by King’s College London. Values at risk from wildfire spread were grouped into three sub-modules: human health and wellbeing; property and infrastructure; and ecosystem services.
The UK’s Incident Recording System (IRS) records vegetation fires attended by the approximately 50 regional FRS and was an extra source of information used in the risk of ignition module. A panel of experts were consulted through two stakeholder meetings on the appropriateness of the WTA framework and the choice of layers. The experts were asked to evaluate individual GIS layers within each of the three modules and to assess issues such as the variation in risk of ignition with land cover type. Opinions were also gathered on how to threshold layers, for instance, into classes representing distance from access points.
The expert panels were then asked to weight the relative importance of the resulting components within each module. The weights were applied and the resulting maps were presented for further evaluation and feedback. An overall risk of ignition map was established for the study area. In combining the three values at risk sub-modules, maps showing values at risk for human life were weighted more heavily than infrastructure, which, in turn, was given greater weight than ecosystem services.
The outcome of the wildfire threat analysis exercise was a set of agreed maps showing risk of ignition and values at risk across the study area. These had been refined through intensive discussion with groups of end-users during follow-up meetings. Potential applications of the maps include: for Local Authorities, guiding local development plans; and for Fire Services, informing Integrated Risk Management Plans (IRMPs), including the deployment of emergency services in the event of a fire. For the Forestry Commission (FC), Ministry of Defence (MoD) and other land managers, it targets resources for public fire awareness and other fire prevention measures and fuel management measures such as thinning or replanting.
Work remains to be done on development of a hazard module showing potential fire spread at this local scale. Kings College London (KCL) recently developed 2km Fire Severity sub-indices for the Met Office under a parallel PURE Associates grant. This is still too coarse for a 25m local WTA but would be appropriate at the national scale. The Fine Fuel Moisture Code 2km data could be incorporated into a regional or national risk of ignition module to give the more meaningful probability of sustained ignition. The other fire severity sub-indices could be used with other layers to create a national or regional scale worse-case wildfire hazard map. A nested approach may be the most appropriate, developing a coarser scale WTA to identify national wildfire hotspots, where a more intensive local finer scale WTA (as here) is required.
Further ecosystem services should also be added to the values at risk module. Finally, work is also required to test whether WTA works at the coarser regional and national scales appropriate to strategic risk assessment by national agencies. For risk of ignition, this is likely to require replacing stakeholder weighting with mathematical modelling. More accurate risk of ignition models will be possible, however, if the geo-reference recorded in IRS is standardised to the estimated point of ignition.