Wildfire fuels and fuel moisture are important factors controlling fire spread and fire hazard. Traditional methods, based on field sampling and meteorology have been expanded to include remote sensing and GIS. Remote sensing is playing an increasingly important role for mapping wild fire fuels and monitoring how they change seasonally.
Fig 1) A species level of the Santa Monica Mountains. This map was generated using Multiple Endmember Spectral Mixture Analysis (MESMA), where the number and types of endmembers (pure spectra) are allowed to vary on a per pixel basis. The endmember selected corresponds to a specific species or type of material, providing a map of plant species.
Remote sensing contributes to fuels mapping in a number of ways. Probably the most traditional approach is through classification. In this way, remotely sensed data are classified into some form of a land-cover map. Two common fuel classification systems include the Anderson fuel classification system and National Fire Danger Rating System (NFDRS). In the example from the Santa Monica Mountains, a species-level map generated using AVIRIS (Fig 1) is "cross-walked" to an Anderson fuels map for use in fire spread modeling (Fig 2).
Fig 2) A species-level map translated to Anderson Fuel models.
An alternative approach is to estimate fuels directly from remotely sensed data. For example, the ratio of green live foliage to dead materials is an important determinant of fire hazard. One method for estimating live and dead canopy components is Spectral Mixture Analysis (Fig 3). Fuel moisture is potentially the most important factor impacting fire hazard. In chaparral fires, much of the fire is carried by green, live foliage as a crown fire. As fuels dry seasonally, the amount of energy required to burn off the water decreases and fire hazard increases. Direct measures of canopy moisture can be derived from AVIRIS (Fig 4). Woody biomass can be estimated using radar (Fig 5).
Fig 3). Fraction images for Green vegetation, Non-photosynthetic vegetation (NPV) and Soil mapped for a spring/fall pair of AVIRIS images mapped using Spectral Mixture Analysis. A pixel is typically comprised of several materials within the field of view. Spectral Mixture Analysis decomposes this mixed spectrum into several pure spectra, called endmembers, weighted by the proportion of the endmember within the field of view. The Green vegetation fraction provides an areal estimate of green leaf cover and the NPV fraction provides an estimate of the dead fraction.
Fig. 4) Liquid water and water vapor images generated from AVIRIS. Water vapor images respond primarily to topography, creating an inverted DEM. Seasonal changes in water vapor respond to specific humidity. Liquid water is a measure of water present in canopies. It varies both as a product of the number of leaves (ie, green live biomass) and leaf water content (canopy moisture). The total canopy moisture provides an estimate of the amount of water a fire would need to evaporate to spread.
Fig. 5) Figure showing radar backscatter measured over several fire scars and a plot of the relationship between stand age and backscatter. Radar backscatter responds to the size of elements within a canopy and their moisture content. Long wavelength radar, such as L and P band, respond primarily to woody components. As canopies age and accumulate woody biomass, radar backscatter increases. Above a threshold of biomass, backscatter saturates. Biomass levels in chaparral suggest that saturation should not occur.
Remote sensing is also considered an increasingly important tool for monitoring post-fire recovery following an event. Fire scars can be readily mapped using a diversity of techniques and post-fire recovery assessed by revisiting the site (Fig 6).
Fig. 6) AVIRIS liquid water images acquired in 1996, 1997 and 1998 over the Calabasas fire scar. The fire scar becomes less evident as vegetation recovers.
