Environmental Geostatistics and Statistics

Geostatistics is a specialized form of statistics that focuses on understanding the spatial distribution of a variable. Geostatistical analysis can provide quantitative estimates of the values of environmental, geological, and hydrological variables at unsampled locations, as well as an assessment of the uncertainty attached to the estimates. Thus, geostatistics can be used to map the probability that the concentration of a contaminant in groundwater exceeds regulatory standards (Figure 1). Geostatistics can be used to design sampling programs that will reduce the uncertainty with which parameters are estimated.

Geostatistics can also be used to provide conditional simulations of hydrogeological parameters such as hydraulic conductivity, which can then be used as input to fate and transport models to estimate relevant probabilities (e.g., the probability that a plume will reach a specified location in a given period of time). One powerful aspect of geostatistics is that it provides data fusion tools to incorporate secondary data, such as seismic velocities or geophysical well log data, into the estimation and simulation of hydrogeological variables, thus decreasing the level of uncertainty and increasing the efficient use of all relevant data.

In addition to geostatistics, Battelle also has strong capabilities in environmental statistics, especially multivariate analysis of environmental data. We use both classical and machine learning methods to understand the structural patterns within the data. This can be useful for identification of the associations within a set of environmental variables and for the identification of important clusters of observations in an area.

Figure 1 shows the estimated tritium concentration in groundwater for a portion of the Hanford Site in Washington State. Figure 2 shows the probability that the tritium concentration exceeds the drinking water standard.

Relevant Projects:

• Mapped thickness and concentration of contaminated sediments off the coast of the Palos Verdes Peninsula, Southern California. Provided estimate of the total mass of contaminated sediment at the site.
• Used geostatistics to map the distribution of critical ecological habitat at the Yakima Training Center for the US Army
• Used multivariate methods to determine structural patterns within contaminated groundwater at the El Campo site, Texas
• Mapped concentration of contaminated groundwater and provided uncertainty estimates at the DOE Hanford Site.
• Used geostatistical methods to optimize the well network for monitoring the distribution of contaminated groundwater at the Hanford Site.
• Provided numerical models of subsurface geology of the Hanford Site based on geostatistical methods for use in groundwater flow and transport modeling.
• Discriminated between multiple sources of TCE groundwater contamination at metal fabrication facility.