Item: Improving statistical models by modifying the avalanche variable
Title: Improving statistical models by modifying the avalanche variable
Proceedings: Proceedings of the 1980 International Snow Science Workshop, Vancouver, BC, Canada
Authors: Richard T. Marriott, USDA Forest Service Avalanche Warning Centre~ Seattle, WA.
Abstract: Statistical models which attempt to minimize the statistical noise caused by errors in the observational data and by the use of independent variables which cannot completely describe the variance of the dependent variable-are being developed at the Northwest Avalanche Forecast Office in Seattle. This has involved both the modification of the input data used in the model development and rethinking of the forecast variable. The quality of the models being produced must be gauged by the assistance they provide to a forecaster in a central office. The philosophy is that models provide information which can act as a parallel, redundant system to the human thought process. The statistical models should not be expected to replace or short-cut the judgmental decisions necessary in forecasting, but should be designed to facilitate them.
Language of Article:
Keywords: artificially triggered, forecasting, snowpack stability, statistical models
Digital Abstract Not Available