This study looks at wildfires from the ground up rather than from the satellite down, as the previous post. Many studies and news articles seem to make the connection “climate change (via heat and drought) will lead to worse fires (I think they mean or imply “worse in terms of damage to things important to humans”).” What stands between weather and damage are a) fuel treatments, suppression folks and their technology, tactics and strategies, and b) peoples’ behavior in terms of developing fire resilient communities, including evacuation technologies and so on. This study looks specifically at many of the variables related to a), and how they interact to contain wildfires. The study is from New South Wales, Australia. I wonder whether readers with experience in the US find that these observations match up with their own experiences?
In order to understand the landscape of the different scientific approaches used in the studies of wildfire, I think it’s helpful to specify the questions asked and the methods used.
Specifically, we asked what is the relative importance of environmental and human factors in containing grass and forest fires at various time periods from when the first ground crews arrived at the fire. From the findings of previous studies, we hypothesise that:
1. Factors which influence fire behaviour – fuel, weather and topography – will be important in determining the probability of containment.
2. The number of resources and the response time will be important in determining the probability of containment.
Research Methods Data from 2219 forest fires and 4618 grass fires. “Random forest” (from machine learning?) approach.
Who Did the Study? Who Funded It? Scientists from Australian universities. It does not appear to have been funded by anyone.
Here are some paragraphs from the discussion.
Unsurprisingly, the more resources available to control the fire, the more likely the fire will be contained. Grass fires are generally easily accessible to tankers and containment is achieved by directly applying water to the fire edge. If the fire spread is too fast or the flame height too high, then direct attack is made on the flanks of the fire, working from the rear to the head (Cheney and Sullivan, 2008; Luke and McArthur, 1978). The more resources available, the faster the fire will be contained. Forest fires may be directly attacked at the fire edge if it is safe and accessible to firefighters or contained by indirect attack which involves burning back from control lines to provide an effective barrier against the main fire (Fried and Fried, 1996; Luke and McArthur, 1978).
Indirect attack cannot be achieved unless a suitable control line is established, hence the more crews available to prepare the control line and ensure the back burn is contained within the control line, the faster the fire will be contained. Fires are successfully contained when the fire spread has been stopped, therefore factors which influence fire spread, fuel load, weather conditions and topography (Cruz et al., 2015) are also important factors influencing fire containment. Our results align with Tolhurst and McCarthy (2016) who characterised fires burning when the fire danger index<50 as mostly fuel- and topography-dominated fires and fires burning when the fire danger ndex > 50 as mostly weather-dominated fires. In our study, fuel and topography were the dominant environmental variables in forest models with weather less important. However in our study, most (97%) forest fires occurred when the fire danger index < 50. In New South Wales, a fire danger index>50 occurs on average only 1.9% days each year (calculated using 3pm weather data from the Bureau of Meteorology weather stations in NSW over a 30 year period from 1982 to 2013). Studies that focus on fires above FDI 50 find weather conditions are the strongest predictor of fire spread (e.g. Jin et al., 2014; Moritz et al., 2010; Price et al., 2015) and therefore we may expect suppression effectiveness to be more strongly linked to fire weather in these conditions.
Collins2018_suppression is a link to the entire study.