FIELD DOCUMENTS - Laserfiche WebLink

Home Browse Search

"Test Methods for Evaluating Solid Waste, Volume II: Field Manual,Physical/Chemical Methods" (EPA, 1986). The statistical evaluation was conducted to further evaluate the concentration of lead within soil at the site. The statistical evaluation addressed the following items: • Calculation of mean; • Determination of the distribution of the sample data; and • Calculation of the 80% Confidence Interval (CI) which provides a corresponding 90%Upper Confidence Level (UCL), interpreted as a 0.90 probability that the true mean for a given population is no higher than the calculated UCL. The data from all intervals were combined into one data set for analysis as Caltrans construction plans typically call for excavation of soil to 0.6 meters (2 feet) for road base preparation. Evaluation of the soil data resulted in a mean (average) concentration of total lead of 36.0 mg/kg (Appendix E). A histogram of the total lead results was constructed to evaluate the distribution of the total lead concentrations within the data set. The data were found to be heavily skewed to lower concentrations (Appendix E). The data are not normally distributed. Therefore, the UCL of the 80% CI (90%UCL) can be approximated if the number of samples is sufficiently large. Gilbert (1987) states that for data with a highly skewed distribution a sample population of 50 or more samples may be needed for analysis of the UCL. The number of samples in the total sample population is 165. For samples with lead concentrations reported as non-detect above the analytical method reporting limit, a value of one-half of the reporting limit was used in the statistical evaluation. Calculations were performed within Statistix 7. The 90%UCL was calculated using the following equation from Gilbert (1987) (modified from SW-846 to provide the upper one-sided confidence limit): 90%UCL=x + (t0.90)(s/n12) where x =mean of the data toxo=the quantile for the "t"distribution for the 90%UCL S = standard deviation of the data n =number of samples SncDP-N:IrnncjpllProjectsl8?I8 1?641ReportU?0 Austin report.doc 4_3 5/21/07