Priorities for Fuels Treatment

Dendrogram showing how the overall fire

A decision model for determining priorities of subwatersheds for fuels treatment was graphically designed with Criterium DecisionPlus (InfoHarvest, Inc., Seattle, WA), which uses both the analytic hierarchy process (AHP, Saaty 1992) and the Simple Multi-Attribute Rating Technique (SMART, Kamenetsky 1982) to support planning activities such as priority setting, alternative selection, and resource allocation. We used a decision model structure that was nearly identical to that of the logic model (figure at right). In the context of decision models based on the AHP, the concept of topics is replaced by criteria. Thus, in the decision model for fuels treatment, the first level of the model contained the three criteria, fire hazard, wildfire behavior, and ignition risk. However, for purposes of setting treatment priorities for subwatersheds, we also added a fourth criterion, percentage of subwatershed area classified as wildland-urban interface (WUI), to illustrate expanding the scope of analysis to include additional logistical factors that can influence decisions about priorities. Note that numerous other criteria and subcriteria could be included to account for other logistical considerations that might influence decisions about treatment priorities.

Weights for each criterion at the first level of the decision model were derived from the standard pair-wise comparison procedure of the AHP (Saaty 1992) in which a decision maker is asked to judge the relative importance of one criterion versus each of the others. We provided the judgments on relative importance for our example application. Weights for sets of subcriteria under each criterion (the second level of the decision model) were derived in the same manner. For purposes of subsequent discussion, criteria at the lowest level of an AHP model are commonly referred to as attributes of a decision alternative, and these attributes correspond to the elementary topics of the logic model (Table: Logic outline).

A SMART utility function was specified for each attribute of a subwatershed, and this function represented the mirror image of the fuzzy membership function of its corresponding elementary topic; i.e., the fuzzy parameters defining no support and full support (Table: Definition of data inputs) were now used to define utility values of 1 (full utility) and 0 (no utility), respectively, on the SMART utility scale of [0, 1]. Note, however, that the WUI criterion is both a primary (first level) criterion of the decision model and an attribute of a subwatershed for which there is no corresponding elementary topic in the logic model. In this case, the critical values corresponding to full and no utility were separately specified as 67 and 0 percent, respectively, and represent the maximum and minimum of observed WUI percentages.