Minimizing the effects of inaccurate sediment description in borehole data using rough sets and transition probability

Lithologies for water wells in Canada constitute a major data source for describing the subsurface geology. Generally the datasets contain inconsistent descriptions and consequently have limited utility for deriving relevant geological information. The Geological Survey of Canada (GSC) has developed a standardization scheme to address the problem but the scheme does not thoroughly resolve the data inaccuracy problem. This study applies rough sets and transition probability to evaluate the GSC scheme using a protocol based on high quality borehole data. Our results demonstrate that misclassifications exist in the description of fine geologic materials, particularly, clay and silt. The results based on transition probability show that the data inaccuracy problem persists even when borehole data are standardized using the GSC scheme. The study provides information for the borehole data that is equivalent to metadata for users to quantify the level of uncertainty associated with inconsistent sediment description in the borehole data.