Analysis using the MSN imputation produced a fuel loading spatial dataset of forested areas throughout the study area. Fuel model 8 was assigned to the majority of polygons, totaling 65.0% or 1803 out of 2775 polygons. Fuel model 9 was next totaling 24.4% or 678 out of 2775 polygons, followed by model 10 with 10.6% or 294 out of 2775 polygons. Model 8 had the greatest overall accuracy with 1421 of 1803 total polygons classified correctly, with 285 being classified as model 9 and 97 being classified as model 10. Model 9 saw 291 of 678 classified correctly, while 271 were classified as model 8 and 116 as model 10. In model 10, 64 of 294 were classified correctly as model 10, while 120 were classified as model 9 and 110 as model 8. Accuracies for the analysis were based on a modified so called jackknife assessment of accuracy, whereby, each of the 2775 polygons were used as an accuracy assessment point. Additionally, Chi squared statistics were computed for the true validation run, as well as a randomization run where Y variables were randomized with respect to the X variables (the relationship between X variables is maintained, however). The true validation statistics, as found in table 5, reflect high Chi squared statistic values, with a sum of 443.4 while the Kappa statistic was reported as 0.2886. The randomized validation run had a total Chi Squared value of 12.6 and a Kappa statistic of -0.02479. Of note is the fact that the analysis was least able to discriminate between fuel model 9 and fuel model 10. This may be due to the fact that fuel model 10 was sampled least frequently, and is also likely due to the fact that fuel models 9 and 10 share similar spectral characteristics. However, functionally, it is more important to be able to discriminate fuel model 8 from fuel models 9 and 10 as the fire behavior output from model 8 is much less than that of models 9 or 10.

The Fire behavior map produced from the FlamMap analysis does not depict much variation in fire behavior between the fuel classes 8, 9 and 10 throughout the study area. This may be due, in part to the fact that fuel models 9 and 10 have very similar fire behavior outputs in comparison to fuel model 8. In addition, the steepest slope found during analysis of the DEM was 28 degrees, however, the average slope throughout the state was 4 degrees. Even when modeled as 28 degrees, fuel models 8 and 9 and 10 still respond more to an increase in wind than to an increase in slope.

Note the decrease in fireline output as 1-hour fuels are increased to a critical value of approximately 17 tons per acre.

Heat output vs Fuel

Another graph illustrating a decrease in activity when fuel loading becomes too high.

Fuel Model Properties

Similarities between fuel models 9 and 10 in terms of fireline behavior - flame length. It should be noted that although model 9 and 10 have similar flame lengths, model 9 can be expected to have a higher rate of spread.