The output of the LAWIPROG-model is a raster data set of continuous hazard level values. The values are transformed into a colour ramp to coincide with current standard avalanche danger descriptors.  The utility of the model lies within the ability to detect individual cell values, and to determine areas of equal value, as well as to determine changes in hazard as the levels vary morphologically. Moreover, new input parameters can result in the change of calculated output immediately.  This is a great advantage to forecasters who want to assess the accuracy of their predictions just prior to release, or for updating current forecasts rapidly as conditions change.

The weight factors were defined by avalanche experts, but not empirically tested under field conditions. The first requirement of the model is to compare the predicted hazard grid with observed events and known historical avalanche paths. Released avalanches can be mapped, digitized and finally superimposed on the map of the predicted hazard. Moreover, it is difficult to localize the release of potential avalanches. Avalanches manifest themselves only by external forces such as ski tourists, animals, or explosives. Further, since avalanches are rare and infrequent it is difficult to gather large enough samples for statistical testing.

Methodological and Operational Errors

In most forecast systems a great many sources of uncertainty are inherent. All weight factors for the LAWIPROG-model are derived directly from the values within the 25 meter resolution DEM of the North Vancouver region. The derivation of slope, aspect and development of elevation values data processes involve uncertainty and error. The multiplicative nature of the model and reclassification schema further increases the uncertainty of results as error propagation occurs during each of the computational stages. Further, the simple nature of the model provides a general spatial distribution of hazard regions but the results would still require serious field validation. 

The resolution of display on the World Wide Web lacks sufficient detail to exhibit the subtle changes of hazard level produced by the Idrisis 32 software.  Higher resolution of display is required prior to distribution of hazard prediction results.

Conclusion

The results of the application of the of the LAWIPROG-model demonstrate areas of high avalanche hazard. As applied by experts, the model may serve as a useful tool for hazard prediction. 

A critical question arises: Should these hazard classification maps be distributed to the public at a general level? The correct interpretation of their content, and an accurate assessment of cartographic display requires trained professionals, who possess additional in-depth landscape and snow pack condition knowledge. As well, experts require knowledge of the limitations and the generalized results produced by GIS raster based analysis.

It follows that the development of simplified prediction models must undergo rigorous empirical testing and model development refinements before any such map may be distributed to the public, and before distribution over the internet could be encouraged.