Imputation is a method of assigning unknown or unsampled areas values of fuel model. We take advantage of the fact that Landsat ETM+ imagery exists for the entire study area and use this to impute the reference areas which we have sampled to those which are unsampled. Neighbors which are closer are weighted higher than those which are more distant.

Initial attempts to correlate fuel loads with forest community types were unsuccessful. As such, a most similar neighbor (MSN) imputation method (Moeur and Stage 1995) was used to assign values of fuel loading for un-sampled areas based on 3024 VFRDB sampled point locations and data available globally for the entire study area. The MSN imputation used data from the VFRDB locations along with Landsat ETM+ data, specifically, tasseled cap components one, two and three, along with NDVI to assign values to unknown and un-sampled locations.

click for a flash animation of the MSN procedure (new window)