Rapid Fire Mapping

Project History

Although the funding for this project was officially received in 2003, actual research began in the fall of 2001. Over this time, the goal of the project to develop a fire fuel load / behavior and fire risk map has remained constant. The methodology used to arrive at this end product has been shaped as a result of experience and knowledge gained over the duration of the project.

When this project was first mentioned to me in 2001, my experience was focused mainly on GIS, remote sensing and geosciences. My experience with fire was limited to lighting the pilot light on my stove. A major factor influencing the changes in the project methodology can be related to the three wildland fire academies I have attended since 2001 when I knew I would be charged with this project. The experience gained at these academies has provided invaluable insight to the properties of fire, and environmental variables affecting it. These academies provided me with not just practical knowledge, but gave me an insight to the type of questions and research I should be looking at to improve my methodology.

Original Methodology

Originally, Rhode Island Forest Community Types would be delineated using Landsat ETM+ satellite imagery. The end result of this analysis would be used as a fuel load index for the state. Forest community types are comprised of a variety of different tree species. As such, it is important to see if fuel loading and fuel model classification can be linked to the forest community type, based on the Rhode Island forest community classification.

Problems with Original Methodology

A quick review of the original methodology by instructors at the New York Wildfire and Incident Managemtn Academy (NYWIMA) revealed the following flaws.

Individual tree species produce distinct spectral signatures. Forest community types are made up of different species of
trees making signature extraction difficult to impossible.
Oak trees are not mutually exclusive to 1 forest community
type
Forest community types cannot reliably be linked to
fuel loadings necessary to produce a fire behavior map.
Because.
Federal fire fuel models are based on the SIZE (diameter) of DEAD and downed woody debris
Total biomass is reported vs available biomass. This is a problem, as more biomass does not necessarily mean more fire behavior, as live fuels can actually inhibit fire spread.
Species type can influence, but does does not determine the amount of available fuel when determining fire behavior. A young sapling stand will burn much differently than a mature pine stand with needle understory.

Landsat ETM+ imagery has a spatial resolution of 30m. This means that each pixel will represent several tree canopies. If these trees are of different species, a "mixed" value will be reported for the pixel.

Fire behavior is influenced mainly by the dead, fine surface fuels below the canopy. Landsat ETM+ imagery is not able to penetrate the canopy of a timber stand to detect these fine fuels.

Fortunately, the flaws exposed at the NYWIMA were met by equally helpful suggestions on how to improve the methodology.

Menu