Daniel Jakubek
Predicting the Contamination between Sites of Sediment Core Measurement in Lake Ontario © 2002
Research activities involving the collection of sediment core samples are extremely time consuming and expensive to fund. This research project utilized data from the Environment Canada Great Lakes Sediment Assessment Program. A total of 32 contaminants were measured at 70 sediment core sampling locations in Lake Ontario. Methods for the estimation of sediment contamination levels were investigated. The Sediment Quality Index (SQI) was calculated and assessed as being a satisfactory measure for the areas in Lake Ontario where the sediment quality is frequently threatened or impaired.
Ordinary kriging was identified as the optimal spatial interpolation model. Individual prediction maps were successfully produced for 20 of the contaminants and cross-validation was used to further assess 'ordinary kriging' as an appropriate method for predicting the spatial distribution of sediment contamination. The results rendered from cross-validation provided an assessment of the relative success of each of the interpolation procedures. Limitations including the limited number of sampling sites, and minimal data at Lake Ontario Areas of Concern, were the main factors for inaccurate prediction surfaces.