ARCHIVED REPORTS XR0011755 - Laserfiche WebLink

Home Browse Search

Evaluation of Natural Attenuation: 7500 West Eleventh Street, Tracy, CA. Page 45 stability based on changes of the distribution of the mass of analytes of concern within y the plume with time. Natural attenuation of analytes of concern within the plume can be expected to approximate a first order exponential decay if the plume is stable and analyte .:� concentrations are decreasing. However, this simple relationship can be obscured by data !` scatter related to such phenomena as variations in analyte concentrations with rise and fall of groundwater levels. It is, therefore, necessary to assess the "goodness of fit" of a regression curve drawn through the data from any given well. If the fit is poor, as characterized by a low correlation coefficient, then a data trend cannot be established with confidence. The MAROS software computes the correlation coefficient to determine whether a data trend can be established with an adequate level of confidence. The procedure used by the software is essentially similar to the trend lines used by SJC to evaluate the history of analyte concentrations in such wells as MW-3 that are illustrated on Figures 20-30. The Mann-Kendall test is a statistical procedure that is well-suited for analyzing trends in data over time (Gilbert 1987). It is a non-parametric test for the zero slope of the first- >' order regression of time-ordered concentration data vs. time. It does not require any assumptions as to the statistical distribution of the data (e.g., normal, lognormal, etc.) and 11 can be used with data sets that include irregular sampling intervals and missing data. The MAROS software applies a very conservative decision-maker for characterizing the results of the Mann-Kendall test. For example, for the «ond of an analyte in a given well to be classified as "increasing" or "decreasing," the slope of the trend must be greater than zero or less than zero with a confidence of greater than 95%. Trends are classified as "probably increasing" or "probably decreasing" if the slope of a trend line can be determined with a confidence of 90-95%. In cases where the trend line cannot be computed with a confidence of 90% or more, the data is classified as having "no trend." The results of the plume stability analysis are presented in the MAROS Plume Analysis Summary presented in Appendix A. Examination of that table shows that, for the BTEX compounds and MTBE, linear regression analysis shows that the time/concentration data are, with minor exceptions, either decreasing, stable or have no trend. When the more sophisticated Mann-Kendall results are considered, no analyte in any of the wells has a trend vs. time that is classified as "increasing," nor even "probably increasing." These results demonstrate that the primary plume emanating from the 7500 West Eleventh Street Site is stable, as SJC has previously concluded, based on conventional plume stability observations, such as those described in Section 7.1.2 of this report. To further evaluate the condition of a plume and assess whether it is stable or shrinking, the MAROS software computes the moments for each sampling event. The primary moment calculation is the calculation of the zeroth moment, an estimate of the dissolved mass of a given analyte of concern relative to the source of the released fuel (i.e., about the point of origin of the well coordinate grid). Variations of the zeroth moment of a —i given analyte over time can show high variability, largely due to fluctuating concentrations at the most contaminated wells, as well as changes in the number and sic j.