Background Trace Element Concentrations in the Franciscan Complex, San Francisco, CA: Background Trace Element Concentrations in the Franciscan Complex, San Francisco, CA MS Thesis Defense Megan Simpson March 16, 2004


The Problem: The Problem Superfund clean-up sites are overwhelming problem in US -Hunter’s Point Shipyard -Presidio To assess a site for contamination, background trace element levels need to be determined


Introduction: Introduction This study examines background trace element levels in the Franciscan Complex (chert, sandstone, greenstone, serpentinite) Samples analyzed for trace element levels (chromium, cobalt, nickel, lead, strontium) Performed statistical analyses and created trace element distribution maps


Purpose: Purpose Will aid in understanding of the distribution of background trace elements data reflects naturally occurring levels found in native bedrock Assist environmental clean-up projects by providing source of baseline data for measuring soil quality Expand current trace element level database


Previous Research: Previous Research Work by Kearney Foundation in 1996 looked at trace elements in soil often referenced in remediation projects, land use planning only two samples close to San Francisco Area Work by Schlocker in 1974 looked at potassium feldspar in sandstone fewer samples (5-7) for each rock type My work enhances the range of sample locales and quantities


What elements are we looking for?: What elements are we looking for? Previous work shows Cr, Co, Ni, Pb and Sr to be prominent in soil and rock within SF Presidio and Hunter’s Point have high levels much time and energy dedicated to clean-up of these sites


Slide7: Geologic Background San Francisco is part of the Mesozoic Franciscan Complex, which formed in accretionary wedge Subduction zone ~200-80 mya Source: U.S. Geological Survey Franciscan Complex


Franciscan Complex: Franciscan Complex Serpentinite Greenstone Sandstone Melange up to 10,000 ft thick Also chert, shale, limestone, conglomerate


Depositional Environment: Depositional Environment Most likely a low latitude marine environment: marine fossils in some clastic rocks, Radiolaria in chert highly fractured, interbedded greenstones (rapid cooling of hot lava) Conglomerate and massive graywacke beds created by turbidity currents further evidence: shale layers between graded beds, small scale current bedding, ripple marks


Methods and Materials: Methods and Materials Prior to sampling aerial photographs Sampling Samples were collected from 45 accessible outcrops around SF Clean hands/dirty hands technique (minimizes cross contamination) At least 5 fresh rock chips collected from outcrop and pooled Latitude/longitude of location noted with GPS unit


Aerial Photography: Aerial Photography USGS collection from 1940s-1970s Provided insight into previous industrial activity Showed sampling areas free from major contaminating factors No evidence of alteration of background levels Presidio Area


Slide12: 45 sample locations Presidio Twin Peaks/ Glen Park Potrero Hill McLaren Park


Chert Sample Locations: Chert Sample Locations 15 samples Glen Canyon Park, Twin Peaks light tan to red distinct bedded layers with folding


Greenstone Sample Locations: Greenstone Sample Locations 8 samples McLaren Park, Corona Heights, Twin Peaks highly fractured extensively weathered dark gray to dark reddish-brown


Sandstone Sample Locations: Sandstone Sample Locations 10 samples McLaren park, Castro large, thick bedded outcrops randomly fractured highly weathered brown to gray


Serpentinite Sample Locations: Serpentinite Sample Locations 12 samples Potrero Hill, Presidio, Baker Beach greenish-gray to blue highly weathered, sheared soft, friable


Geochemical Analysis: Geochemical Analysis 48 samples submitted to SGS Mineral Services in Toronto (45 plus 2 blind duplicates and 1 reference sample) Tested for 40 trace elements using ICP-AES (Inductively Coupled Plasma-Atomic Emission Spectrometry) Samples decomposed using mixture of HCl, Nitric, Perchloric, HF Digestion is aspirated and elemental emission signal is measured for elements


Slide18: Identification of outcrop samples Geochemical results Statistical analysis Mann-Whitney and ANOVA Developed trace element distribution maps Results and Discussion


Identification of Outcrop Samples: Identification of Outcrop Samples Field Identification Thin sections from each rock type (Univ of Utah) X-ray Diffraction (XRD) evaluation


Slide20: Thin Sections (USGS Petrographic Microscope)


Slide21: Thin Sections (USGS Petrographic Microscope)


X-ray Diffraction: X-ray Diffraction Scattering of x-rays from a crystal where resulting interference pattern determines structure of crystal Performed by Mineral Services Lab at USGS Showed minerals consistent with each rock type Serpentinite (lizardite, chrysotile) Chert (quartz) Sandstone (quartz) Greenstone (albite, diopside)


Slide23: XRD of Chert Quartz 0 7000 40 2 theta Count/s


Geochemical Results: Geochemical Results Results for Cr, Co, Pb, Ni, Sr compared to data from Schlocker show: Chromium Lead Cobalt Nickel Strontium


Statistical analysis comparing my data to Schlocker’s data: Statistical analysis comparing my data to Schlocker’s data Assists in description and analysis of data generated descriptive statistics (mean, std dev, ranges, etc.) histograms (show distribution and observations of sample) tests for normality (data fits bell shaped curve) Komolgorov-Smirnov, Shapiro-Wilk Q plots (compare data to linear ideal)


Statistical Testing: Statistical Testing If data is ‘normally distributed’, randomly collected, Student ‘t’ test can be used If data is not ‘normally distributed’, Mann-Whitney U test is appropriate Tests on Schlocker data showed data not normally distributed, therefore Mann-Whitney U analyses most appropriate to demonstrate significant differences between datasets


Mann-Whitney U Test: Mann-Whitney U Test Data obtained from two random samples (n1 and n2) Samples are combined, each value assigned a rank (smallest is rank 1, largest is rank n1+ n2) U statistic based on totals of ranks (Ta, Tb) The smaller value of either Ta or Tb, the greater the evidence that samples from different populations Tests were performed using 95% confidence level (95% of all samples give interval that includes the mean, 5% would give interval that does not)


Hypothesis: Hypothesis Null hypothesis states: ‘the difference between the mean ranks of the datasets is not statistically significant’ OR average level of specific trace element found in given rock type in Schlocker’s data is not statistically different from average level of same element, same rock type found in my data


Comparison of Trace Element Levels between Schlocker and Simpson Datasets using Mann Whitney Test: Comparison of Trace Element Levels between Schlocker and Simpson Datasets using Mann Whitney Test Chromium significant differences seen in serpentinite (means of 564 vs. 1384 mg/kg) and chert (means of 3 vs. 22 mg/kg) data no significant differences seen in greenstone and sandstone Cobalt significant differences seen in sandstone (means of 25 vs. 7 mg/kg) and serpentinite (means of 59 vs. 81 mg/kg) no significant differences seen in chert and greenstone


Slide30: Nickel significant differences seen in all rock types: chert (means of 43 vs. 22 mg/kg), greenstone (means of 180 vs. 46 mg/kg), sandstone (means of 57 vs. 27 mg/kg), serpentinite (means of 3000 vs. 1773 mg/kg) Lead significant differences seen in chert (means of 0.25 vs. 40 mg/kg) and sandstone (means of 2.6 vs. 14 mg/kg) no test on greenstone and serpentinite Strontium significant difference seen only in chert (means of 5 vs. 38 mg/kg)


ANOVA: ANOVA Determines significant differences in the means of two or more datasets Performed on my data comparing means of same trace element between four rock types Data is normally distributed, equal variances, from randomly collected samples Null hypothesis: means of sample populations are statistically equal


Slide32: Mean (mg/kg) and Standard Deviation for each rock type in conjunction with ANOVA results a Identical letters indicate no significant difference at the 95% confidence level. b,c Non-identical letters indicate a significant difference at the 95% confidence level Chromium and nickel concentrations are ~15-35 times higher in serpentinite


Trace Element Distribution Maps: Trace Element Distribution Maps Created to illustrate levels of trace elements throughout San Francisco Within sampled areas, data is accurate Outside of sampled areas (west, upper northeast, southeast corners) data is interpolated ArcInfo/ArcMap 8.3 Inverse Distance Weighted (IDW) interpolation method assumes each sample point has a local influence that decreases with distance


Slide34: N Chromium Distribution Map


Slide35: Cobalt Distribution Map


Slide36: Strontium Distribution Map


Slide37: Lead Distribution Map


Slide38: Nickel Distribution Map


Conclusion: Conclusion This study: enhances understanding of background trace element distribution in Franciscan Complex will assist in development of future environmental clean-up studies expands trace element database Improvements to this dataset: increasing sample size/location examining other trace elements sampling soil