4.1.5 Herding Biochemical Cats: Quantifying the Qualitative Properties of Fire

“If flame were hiding in forests ready-made, Not for one moment could the fires be hid, But everywhere they’d burn the woods, turn trees to ashes” Lucretius, circa. 56 BC.

Wildfire is inherently heterogeneous in character (Lertzman et al, 1998), and with the possible exception of volcanic eruptions, perhaps the foremost so within the category of geological and meteorological hazards. Whereas some categories of hazardous natural phenomenon are essentially mechanical in nature (i.e. geological structural failures such as earthquakes and landslides), therein confined to problems of an essentially Newtonian quality, as biochemical events, wildfires embody an Einsteinian dynamic. In strength, wildfires are no less variable than in any other respect. Several are the definitions for fire severity, however, at the time of writing, the foremost useful definition for the purposes of this thesis is that of USDA Forest Service biologist Colin Hardy and colleagues, whom in recognition of the coevolution of fire and biota describe fire severity as “the intensity of the fire as it affects the bio- geochemical environment” (Hardy, 2005, p.5 in reference to Hardy et al, 1998, 2001). This description is concordant with that of several other parties whose research examines the relationship between biota and fire (Ryan and Noste, 1985; Morgan et al, 2001).

A wide variety of variables are used to gage fire severity, including flame length, fire size, resistance to control, rate of spread, and fuel consumed, amongst others (Hardy, 2005). Generally, mortality in overstory biomass is considered the primary indicator of fire severity (Agee, 1993; Morgan et al, 1996). However, whereupon assessing fire severity within wildlands, fire ecologists examine the percentage loss in organic biomass (Keeley, 2009; Lenihan et al, 1998), including soil matter and changes thereto [i.e. depth of heat penetration and impacts thereof] (Wells et al, 1979; Ryan and Noste, 1985), floral and faunal mortality rates (Chappell and Agee, 1996; Larson and Franklin, 2005), spatiotemporal shifts in the patterning of flora [i.e. species richness and invasive plant populations] [Keeley et al, 2005; Wang and Kemball, 2003; Ryan, 2002; Turner et al, 1999; Whelan, 1995; Lea and Morgan, 1993; Morgan and Neuenschwander, 1988; Ryan and Noste, 1985), together with the fire’s general behaviour (Turner et al, 1994). Fire severity varies across the several qualitatively different fire regimes, as does fire intensity, which a quantitative measure describes “the physical combustion process of energy release from organic matter” (Keeley, 2009, p.116). Fire intensity, like fire severity, is subject to several interpretations. However, its use in this thesis is aligned to that of the seminal mathematical firespread model (Rothermel, 1972), which today remains the cornerstone of fire behaviour metrics, for which “the energy per unit volume is multiplied by the velocity at which the energy is moving” (Keeley, 2009, p.117). A variation on fire intensity is fireline intensity, which quantifies the rate of heat transfer at the flame front (Byram, 1959). Studies evidence a coupling of fireline intensity and flame length in forest and shrubland fires (Keeley, 2009; Fernandes et al, 2000; Johnson, 1992; Andrews and Rothermel, 1982). However, the metrics affiliated to an ecosystem’s historical natural fire regime [52] (Hardy et al, 2001) are indicative, but not prescriptive in their affiliation, acting as a guide, not a hard and fast rule. Within the context of fire ecology, as with frequency and severity, changes in fire intensity promotes or excludes particular species, thus tends alter the composition of biotic assemblages (Archibald, 2013).

Across the conterminous United States, Remote Automatic Weather Stations [RAWS] continuously monitor meteorological variables including air temperature, dew point, pressure, wind speed and direction, relative humidity, precipitation, solar radiation, fuel temperature and fuel moisture at 1,731 sites (USGS, 2016). First introduced by the USDA Forest Service in 1976, and today monitored by the National Interagency Fire Centre [NIFC], together with data live-streaming from satellites and other terrestrial and space information and communication technology [ICT] devices, RAWS enable the provision of daily fire weather [53] updates to an array of local, and national government agencies. The Forest Service utilise RAWS data to construct wildfire indices that act as ballpark guides on event probabilities. Developed in 1972, revised in 1978, and subject to modifications thereafter (Deeming et al; 1977; Bradshaw et al, 1983), the National Fire Danger Rating System [NFDRS] works in much the same way as many other hazard classification systems, such as the Volcanic Explosivity Index [VEI] and the Saffir-Simpson Hurricane Wind Scale (Schott et al. 2012), ranking event probabilities by number.

Physics-based nonlinear dynamic equations generate numerical values for NFDRS indices including Spread Component [SI] (Schoenberg et al, 2007), which, highly variable, and constructed from data including fuel moisture, relative humidity, wind speed/direction, and topography (slope class), (sensu Rothermel, 1972) is calculated “with the fuel elements weighted by surface area” (Bradshaw et al, 1983, p.23); Energy Release Component [ERC] (Schoenberg et al, 2007), which utilises readings of fuel moisture to estimate the available energy [i.e. heat release per unit at the fire front], therein potential fire intensity (Bradshaw et al, 1983); and Ignition Component, which spanning 0-100, and calculated from readings of fuel moisture together with the temperature of the receptive fine fuels, ranks the probability of a firebrand igniting a fire in a fine fuel complex [54] (NIFC, n.d). Further NFDRS indices include the Haines Index, which ranked 1-6, translates atmospheric stability and moisture content readings into probabilities for plume-dominated fires [55]; Lightening Activity Level [LAL], also ranked 1-6, which interprets storm data to give estimates of cloud-to- ground lightning strikes; and the Keetch-Byram Drought Index [KBDI], which ranked 0-800, and customised to individual geographic regions, indicates the moisture content within the topsoil [upper 8 inches], based on recent precipitation measurements as they relate to annual rainfall patterns (Ibidem).

The Burning Index [BI] is the foremost commonly used of the several NFDRS indices (Schoenberg et al, 2007). Ranked 0-110, the BI number is a function of SC and ERC values, and is indicative of fire intensity; its value divided by 10 equal to flame length [FL] in feet at the flame front (Pyne et al, 1996; Andrews and Bradshaw, 1997; Mesonet, 2016). For example, a BI number of 0-28 equates to a flame length of up to 2.8 feet, therein a low intensity fire. Whereas a BI number of 110 equates to a flame length of 10.10 feet, therein a high intensity fire [Fig. 20]. However, invaluable though composite indices are in providing a bird’s eye view of fire-prone landscapes, and the condition of the biomass and weather systems thereof, they alone cannot fully quantify how and why a wildland fire will become manifest therein (Cocke et al, 2005). For example, LANDSAT’s Normalized Burn Ratio [dNBR] (USGS, 2004) is used to monitor fire severity and ecosystem response in regions including California’s chaparral shrublands. Analysis of the region’s ecosystemic response to the wildfires of late October 2003, which, one of the case studies used in this thesis, burned through 200,000 ha, revealed that the dNBR that was attributed to the event by USGS’ EROS data center did not in fact correlate with the real-world results (Keeley 2009; Keeley et al, 2008). Therein, in and of itself, remote-sensing data has “limited predictive ability” in projecting the ecosystem impacts of wildland fire, hence requires coupling with other methods including field studies (Keeley, 2009).