Geographic information analysis of uranium in groundwater near the Lake Ontario Ordnance Works (LOOW) and Niagara Falls storage site in Niagara County, New York

In this thesis standard geostatistical methods are applied to the study of groundwater contamination by uranium in the Lake Ontario Ordnance Works (LOOW) site in northwest Niagara County, New York. Geospatial statistical methods are described and compared to traditional methods of statistical analysis of groundwater data. Recent studies comparing both geostatistics and traditional methods for groundwater analysis are reviewed to evaluate the strengths and weaknesses of each approach. Conclusions from this comparison are: (1) Geostatistics not only explores the statistical trends of environmental data in the whole estimation area but can provide spatial interpolation for unknown sample locations, and prediction error estimation. (2) Geostatistics requires the combination of analytical results and geographic locations for every monitoring program. The choices and number of sample locations plays a more important role in the geostatistical model than that of traditional statistics methods. Geostatistics models are functions related to distance. (3) Geostatistical software provides the direct visualization of every step of data exploration and prediction. It makes it easily to understand the procedure and reduces the complexity of mathematical content. In the second part of the thesis, geostatistics are applied to the study of uranium concentrations in groundwater on the LOOW site with special attention near the Niagara Falls Storage Site (NFSS), where uranium and radium from the US Manhattan Project are stored. Maps of interpolation of uranium concentrations are developed from optimized geostatistics using Ordinary Kriging and Inverse Distance Weighted (IDW) methods. The results are compared to the geographical distribution of data and sampling sites. (1) The concentrations of uranium predicted near the NFSS site have higher estimates than those of other sites.