The scale of maps useful for predictive modelling and potential mapping depends on the degree of accuracy required by the end user. Regional planners may be happy with a 1 :250,000 scale map that provides very general information regarding whether areas have archaeological potential or not. A research officer at the Archaeology Branch, on the other hand, requires a map at 1 :50,000 or larger scale that provides more information and accuracy regarding potential than what their experience alone can provide. The potential maps at 1 :250,000 created by Bussey and Alexander (1992) appear, in the absence of field testing, to be quite powerful in that nearly 100% of the known sites are predicted in about 25% of the total area. However, these maps could not have been produced without the extensive and intensive probabilistic field inventory work which has taken place in the area, and the correlation of archaeological and ethnographic information with environmental parameters at finer scales. The 1:250,000 maps are useful for regional overviews, but their use is limited to alerting non-archaeological planners of potential archaeological concerns. For archaeological resource managers, there may by a problem with large areas of low potential having many unmapped pockets of very high potential (e.g. lesser known quarries, pockets of grasslands or subalpine parklands and other concentrated resource locales), and of many areas with low or nil potential mapped as having potential. Comparing the 1:250,000 map portion mapped at 1:50,000 by Eldridge and Eldridge (1979), virtually all the smaller Dean River study area was mapped by Bussey and Alexander as having "potential", whereas 94% of this area was classed as low to very low potential by Eldridge and Eldridge. Many of the predictive modelling projects in the United States have found that using a 1:250,000 scale of data produces information that is of little utility. The archaeological model will tend to have poor accuracy, especially if the quality of environmental data available at 1:250,000 is not at maximum resolution for the scale.
To increase the accuracy and sophistication of the models it is recommended that, whenever possible, predictive modelling and potential mapping be carried out at 1 :20,000 scale and the results exported to the smaller scale when required. The operations associated with changing scale in a GIS - rectification of feature placement through rubbersheeting and manual shifting - would be a major problem even if just the archaeological potential cover was exported. Since the coarser grained maps would probably lump finer resolution potential classifications in any case, and relatively few polygons would be mapped, it is probably more cost effective to manually transfer the combined polygons onto a clean hardcopy 1 :250,000 map and then re-digitize. In the interim, while electronic 1 :20,000 maps are unavailable to the Archaeology Branch, it will be useful to continue low resolution modelling at 1:250,000 scale following the example of Bussey and Alexander (1992). The 1:250,000 scale has the advantage of relatively low cost for the map base and also integrates with the CORE initiative standard.