REP17182 - Laserfiche WebLink

Home Browse Search

it '1 classification equation given the small sample size. Finally, slope and distance to water are easily measured from a topographic ~ , map while accurate classification of land form and vegetation usually requires field observation. A brief Monte Carlo type study was conducted of the LDFS for ' the combinations of predictor variables discussed above. Results indicated that regardless of whether water and slope, or slope, ' vegetation and land form were used as predictors, slope was the "fastest moving" variable. This is, large changes in predicted ' group membership occurred in response to small corresponding changes in slope, with site probability increasing as slope ' decreased (this was also indicated by the relative weight of the discriminant coefficients associated with slope in the above equations). It could be said that slope was an energizing variable ' that "drove" probabilities of site presence or absence, while the other categories were mediating variables that could either weaken or strengthen probabilities in either direction. A predictive model based on either of the two LDFS discussed above would predict a high frequency of site locations within Dowe Flats proper. The overriding characteristic of this area is an ' absence of marked slopes. A predictive model may be enhanced by inclusion of geomorphic variables. The identification of these variables must await a geomorphological investigation of the study area. Since this pilot study was based primarily on data gathered from Indian Mountain and ' Rabbit Mountain, the inclusion of geomorphologic variables relevant to Dowe Flats as well as the enclosing mountains may significantly ' alter and refine the prediction equation with regard to Dowe Flats as a discrete areas. Applicable RP3 Contexts As mentioned in the previous section, prehistoric ' investigations in the Dowe Flats study area relate primarily to the ' 90