Multi-Criteria Evaluation (MCE)

An MCE allows more than 2 criteria to be evaluated at once (Idrisi guide to GIS). In this This MCE had 9 factors, size, density, land cover, temperature, precipitation, temperature, slope, aspect, wildfires and prescribed burns. None of these datasets had distinct points at which they suddenly became unsuitable so a fuzzy set was created from each one in the decision wizard that assigned a number on a scale from 0 to 255 according to how suitable the pixel was. Each dataset was treated differently here according to its characteristics.

Slope: The steeper a slope the faster a fire will move up it (Granger & Schelling 2003) therefore fuzzyslope used a sigmoidal monotonically increasing shape with control points of 0 and 90.

Aspect: In the northern hemisphere south facing slopes are know to receive more sunlight than other aspects (Shoenberg et al 2003). This extra heat causes fires to burn more fiercly on south facing slopes. Fuzzyaspect used a symmetrical j-shape with the control points 90, 180, 180 and 270.

Wildfire and Prescribed burns: Research has shown that an increase in time since last fire will increase the burn area until about 30 years old. After 30 years there is no longer an increase (Shoenberg et al 2003). Fuzzyyear used a symmetrical sigmoidal shape with the control points 0, 1900, 1974 and 2004. This was necessary to prevent anything with a value of zero from being counted as very highly flammable.

Temperature: Burn area increases with temperature until about 21 centigrade when it begins to level out (Shoenberg et al 2003). For fuzzytemp a monotonically increasing sigmoidal shape was used with control points 0 and 21.

Precipitation: Precipitations above 20mm begin to cause a decline in burn area (Shoenberg et al 2003) therefore a symmetrical sigmoidal shape was used with the control points 0, 1, 20 and 2500

Size: a symmetrical sigmoidal shape was used for size because while multilayered vegetation is most conducive to burning, all of the smaller sizes were not much different. The control points 0, 2, 4 and 6 were used with 6 being medium to large trees as these are less conducive to burning than the other sizes (Shoenberg et al 2003).

Land Cover: A monotonically increasing sigmoidal shape was used on the ranked land cover data set to obtain fuzzyveg. The control points were 0 and 9 as the intensity of fire and likely hood of burning increased with the number.

Density: A symmetrical sigmoidal shape was used on the density dataset to create fuzzydensity with conrol points at 0, 1, 3 and 4 because very high and very low density are not conducive to intense fires (Omi 2005).

The result of the MCE assigned a number to the pixels according to how ‘suitable’ they were for a high intensity wildfire. These values were then reclassified into 10 classes with 0 being the lowest ‘suitability’ and 9 the highest.