On-Demand Webinars

STATISTICAL ISSUES IN GROUND WATER MONITORING APPLICATIONS

This webinar provides an academic overview of advanced-level statistical methods that are used in analysis of ground-water monitoring data at waste disposal facilities. The logic behind and application of the statistical methodologies is the focus of the webinar, rather than the mathematical derivation of the statistical methods. The use of quality statistical methods in environmental monitoring can more rapidly detect real impacts to the environment and are less prone to producing costly false positive results. Use of these methods in other areas of determining environmental compliance (e.g., soil sampling, assessment and corrective action monitoring) are highlighted as well.

Protection of our natural resources from the disposal of hazardous materials continues to be a major environmental and health concern in the United States and throughout the world.

A major concern is the impact on groundwater from waste disposal facilities. Since both municipal solid waste and hazardous industrial waste contain constituents that are potentially harmful to human health and the environment (although typically at very different concentrations and likelihood of detection), this concern has given rise to regulations in the United States requiring extensive chemical testing of groundwater quality at waste disposal facilities (e.g., RCRA solid and hazardous waste landfills, and TSCA, PCB disposal landfills).

The result of these new regulations is that thousands of previously unregulated facilities in the United States are now required to provide extensive geochemical assessments of groundwater on a regular basis (i.e., quarterly or semiannually) and, on each occasion, to determine if the facility has contaminated groundwater on the basis of these analytical results.

The logic behind and application of the statistical methodologies is the focus of the webinar, rather than the mathematical derivation of the statistical methods.

What is the problem?

Groundwater monitoring decisions historically have been rooted in statistical theory. Although the methods were often inappropriate, statistics has nevertheless played a critical role in environmental impact decisions, beginning with formal regulatory discussion of these issues [see RCRA regulation USEPA, 1982].

The problem is technically interesting:

Given a new monitoring measurement for a groundwater monitoring well, drilled in a particular aquifer, and analyzed for a particular substance, what is the probability that measurement represents an effect of some unnatural source (e.g., a hazardous waste disposal facility)?

Presented this way, this problem is seen by the statistician as a problem of statistical prediction. Given a collection of historical or background measurements for the substance from an area geographically removed from the potential source of contamination (e.g., groundwater monitoring wells located upgradient of a waste disposal facility), what limit or interval will contain the new measurement with a desired level of confidence (e.g., 95% confidence)?

Although this seems to be a natural view of the problem, it took many years for USEPA and industry to acknowledge its utility for ground-water monitoring applications.

If this were the full extent of the problem, the solution would be elementary and routine application of statistical prediction and tolerance intervals would suffice. The problem, however, is far more complicated. Key problems involve the issue of multiple comparisons produced by repeat testing of many monitoring wells, each analyzed for multiple constituents, censoring of often large amounts of data below detection limits, skewed distributions, and balancing false positive and false negative rates at acceptably low levels.

What you will learn:

• Develop a working knowledge of ground-water monitoring statistical methods.
• Understand the differences between statistical issues related to detection, assessment, and corrective action monitoring.
• Understand why certain statistical methods should be avoided in environmental monitoring applications.
• Develop a better understanding of regulatory issues.
• Obtain an intuitive understanding of the strengths and weaknesses of statistical approaches to environmental monitoring.
• Understand the history of statistical methods in ground-water monitoring programs.

Learn how to apply what you learn:

• Obtain a working knowledge of normal, lognormal, gamma, and nonparametric prediction limits.
• Obtain a working knowledge of confidence intervals for comparison of concentrations to regulatory standards in assessment and corrective action monitoring.
• Understand the differences between inter-well and intra-well monitoring programs.
• Understand how to incorporate "non-detects" into the statistical analysis of your data.
• Learn how to detect outliers, test for normality, and test for trend in environmental data.