Request Expert

Expert is a statistician with many years of experience in environmental statistics, especially related to site characterization and remediation. He has developed numerous environmental sampling designs for field projects and experimental designs for studies and has analyzed and reported the resulting data. He is an expert in multivariate, spatial and environmental statistics, in the statistics of measurement (analytical chemistry and microbiology) and in statistical quality. He developed software (R, MATLAB, Visual Basic) and database applications (MS SQL Server, MySQL, MS Access) and performed mathematical modeling and scientific and engineering computation for numerous projects.

He developed and implemented innovative adaptive sampling and statistical methodology and software for multivariate spatial Bayesian prediction for a large nationwide soil remediation program. This was used at many large properties in several small towns, saving the client several million dollars.

He developed and wrote up the approach for data analysis in Institutional Review Board (IRB) protocol for a pilot clinical trial of using autologous stem cells to treat concussion.

He developed experimental designs, performed data analysis and prepared reports for FDA validation of two new analytical methods for radiopharmaceuticals. He used Reproducible Reporting using knitr and RStudio. He also developed an IQ/OQ report for the R installation. FDA approved the IQ/OQ and both analytical methods.

He provided statistical, software development and documentation support to a variety of EPA research programs. He developed an improved Lowest Concentration Minimum Reporting Limit (LCMRL) for analytical chemistry reporting limits (RL) to support the Unregulated Contaminant Monitoring Rule. He developed statistical methods and software for real-time analysis of freshwater clam behavior as an early warning indicator of toxin introduction into drinking water supplies. He developed failure models of drinking water system pipe breakage data using Cox models with shared frailty to support cost models for condition assessment and management of water supply infrastructure. He developed models for improved estimation of total phosphorus in streams. This effort used a new statistical framework (SSN, Spatial Stream Network) for spatial statistics on a flow directed network. Paper published in Journal of the America Water Research Association. He performed various statistical analyses related to contaminants in first order streams (published paper) and to land application of treated sewage sludge. He provided database and statistical support on a large study of the impact of a major urban lead remediation program, including impact on blood lead levels in children. He analyzed a large spatio-temporal data set of moisture sensor readings from a sensor network in an experimental porous pavement installation in Cincinnati. The purpose was to develop criteria for design of monitoring networks to assess the performance of porous pavement in alleviating urban flash flooding.

Expert developed statistical methodology and performed statistical analysis for several anthrax response projects, including the Brentwood and Trenton USPS Facilities and Anacostia Naval Annex. At the USPS Brentwood facility, he developed risk-based statistical sampling design approaches for anthrax sampling for verification of decontamination. He developed estimators for spore surface concentration density based on the number of samples, the frequency of non-detects and various characteristics of the sampling and testing procedures. He also developed methodology based on local-likelihood to construct aggregate or 2-D/3-D estimates of spore kill probability based on analysis of biological indicator results following fumigation.

He developed a database architecture using MS SQL Server that tied together disparate data sources (such as meteorological data, ambient air network data, control system data, HVAC data, building sensors and laboratory data) on a real-time to near real-time basis for control, analysis and archiving. This architecture, along with the various display and analysis tools he created for it, played a central role in analysis and control of the tracer gas tests and ClO2 fumigation.

Expert helped to resolve a dispute between USAF and Versar over the quality of work in a runway resurfacing project. He developed a spatial sampling plan for the green space around the runway, analyzed the resulting data and prepared a report.
For Hunters Point Shipyard, he developed an adaptive sampling/monitoring design for excavation and screening of mixed (chemical, rad, UXO) waste that rescued a troubled project and saved the client (USN) $6 million.

For PG&E, he used environmental forensics, compositional analysis and hydrogeology to determine that seasonal surface water quality permit exceedances on McDonald Island in the Sacramento-San Joaquin River Delta were caused by seasonal upwelling of deeper water due to drop in precipitation and river levels rather than to leaching from residual sediments in a remediated former drilling pond.

For a confidential utility client, he developed statistical methodology using charts based on an SPRT statistic for monitoring quality of inspections in a project in which over 750,000 homes were screened and/or inspected for mercury contamination. He developed the software for data entry and management, statistical analysis and automated reporting and managed the database server.

Expert developed the T+2 statistic for simultaneous multi-analyte comparison to background and successfully negotiated with Ohio EPA to use it for a RCRA site closure. The T+2 statistic is a “one-sided” version of Hotelling’s T2 statistic. OEPA had initially insisted on a statistical protocol which would have led to sampling and remediation over much of the 50-acre site, and would have made closure financially impossible, even though the suspect activity was confined to a small area. The site was successfully closed at a reasonable cost.

He modeled the performance of complex, multi-tiered sampling, compositing and testing designs for the housing areas at Treasure Island, CA, based on existing data. He determined the most cost-effective design in terms of expected sampling, analytical and remediation costs. This sampling design was also highly protective of public health.

Expert programmed (S-Plus language) a model of the fuel droplet dispersion resulting from the collision of an F-16 with a private plane and its subsequent settling. He used models of spray nozzle dynamics to estimate droplet size distribution based on aerodynamic shear. The subsequent dispersion was modeled using a Gaussian plume model, the droplet settling terminal velocity distribution and meteorological data. This model indicated the locations of most likely contamination from jet fuel on the ground leading to reduced environmental sampling costs.

Expert developed innovative sampling strategies for characterization and for verification of dioxin contaminated areas on Johnston Atoll. This involved the use of areal composite sampling with a composite cutoff based on a hot spot threshold and the Maximum Entropy composite model, which he developed. Computer simulations based on existing data and several hypothesized hot spot distributions showed good ability to detect hot spots with acceptable false positive rates. A senior EPA statistician at the ORD laboratory in Las Vegas reviewed this work and helped to sell it to the EPA project team. The strategy simultaneously improved both estimation of average residual dioxin concentrations (as 2,3,7,8-TCDD TEQ) and hot spot detection capability. It also gave the project team a quick and easy decision rule to guide excavation. The design also included detailed procedures for sample preparation and compositing and for laboratory subsampling. These procedures gave data with much greater precision than the previous data generated by another contractor. This saved the client about $50,000 in analytical costs.

In a subsequent phase, he used the existing data from the previous contractor to estimate the total quantity of soil within the HO storage area with average TEQ greater than 1 ug/kg. Under the previous sampling design, the sampling below surface was highly biased and caused the previous contractor to greatly overestimate the quantity of contaminated soil. He used spatial statistical methods with depth and surface staining as additional predictors to produce much more accurate quantity estimates from the original data. He quantified the error of the estimate using the nonparametric bootstrap. This saved the client several million dollars by avoiding unnecessary excavation and incineration of a very large quantity of soil.

Expert developed experimental designs, sampling techniques and performed data analysis for laboratory and field pilot studies to support the use of bioremediation and phytoremediation on several field projects. He developed a carbonate buffer system to mimic seawater so that electrolytic respirometer studies could be carried out on coral sands. This enabled the use of bioremediation at several Navy facilities in the Pacific.

Expert designed a new RCRA groundwater detection-monitoring approach based on control charting downgradient minus upgradient differences in each monitoring parameter. He modified the existing Part B facility permit for a private industrial waste landfill, including use of the new technique and other improvements. He successfully negotiated the change with Ohio EPA after only one meeting with the agency. The landfill had been reporting detection (false alarms) in one or more parameters each quarter for several years. There have been no false positives since the new design was adopted.

Expert was the Quality Assurance Officer for several incineration Trial Burns and Performance Demonstration Tests of thermal treatment systems. He analyzed data from several pilot scale and demonstration scale treatment systems, including thermal desorption, chemical treatment and soil washing to support optimization and measurement/demonstration of effectiveness. He conducted engineering pilot studies to determine the most effective cleaners to remove PCBs from concrete flooring and electrical wiring.

He developed computational fluid dynamics computer models for pipe sizing in soil vapor extraction (SVE) systems, for natural gas flare design, for steam heating of the subsurface and for design of porous pipe for air curtain groundwater contamination barriers. He performed extensive flow and pipe sizing calculations to support design of the largest Soil Vapor Extraction system in the world, which was installed on Midway Island.

Experience Highlights:
 Developed and implemented large scale spatial multivariate Bayesian prediction program for remediation and final status of hundreds of large properties in SW US contaminated with smelter waste
 Developed and managed DBMS, sampling design and all statistical analysis for anthrax fumigation of USPS P&DC facilities, Brentwood and Trenton
 Developed and applied innovative approaches to composite sampling
 Performed DQO development, sampling design, data analysis and DQA for many site investigation and remediation projects
 Developed limit of quantification method (LCMRL) and software for USEPA Office of Groundwater and Drinking Water
 Designed, analyzed and reported results of analytical chemistry method validation studies to FDA
 Helped develop and teach courses on control charting, statistics of laboratory quality and statistics of process analyzers