Lab 4: Fire Hazard Analysis

A beginning step, and possibly one of the most important steps, is to appropriately prepare the collected data for analysis. The analysis of fire hazards relies on multiple factors, two of them being the slope of land and type of vegetation coverage. Because slope is a critical factor and is based on the DEM data, it is critical that the DEM data is in appropriate units for the slope calculation. To ensure this, the first step is to change the projection of the DEM, which is originally in degrees, into a projection that uses unit distance, or in this case meters. Additionally to be safe, the FRAPS landscape coverage should be converted as well.

Following this initial preparation a hillshade and slope layer can be created. The analysis in this scenario is a product of how slope and vegetation are reclassified and then added together using map algebra. In this project I used a 10 point NFPA Hazard Points ranking system to reclassify the slope percent rise values into fire danger. This is fairly robust system as, though fire hazards are highly subject to local conditions, slope by itself is reasonably well classified. For the vegetation reclassification I created my own numerical classification also based on a 0-10 scale that while loosely based on the NFPA Hazard Point system demonstrates some challenges in accuracy. Without intimate knowledge of the local area, seasonal conditions, longitudinal trends it is difficult to create a vegetation classification system that provides relatively accurate predictions.

There are also some additional challenges in this fire hazard analysis. Even if the individual fire hazard classification for each factor has an acceptable degree of accuracy there is the issue of weighting each factor. This issue is first present in the reclassification system, because when the map algebra analysis consists of summing the values of each factor, it is possible to weight a factor by giving it a larger scale. In my analysis, I equally weighted slope and vegetation in reclassification by providing a 10 point scale for each. Additionally, it is also possible to weight each factor during the process of raster calculation by providing a multiplication coefficient, however, without a greater knowledge of which factor has greater importance, this model again only uses equal weighting. Thus, my results final results, while demonstrating a clear trend of increased fire hazard in areas of dry vegetation and greater slope, can only provide general predictability, especially as other important factors are missing and with issues of factor weighting.