Page 74 - Geologic Investigations in the Lake Valley Area, Sierra County, New Mexico

Acknowledgments
All sample preparation was performed in the Denver labo ratories of the U.S. Geological Survey, assisted by Dave Firewick. We acknowledge XRAL Laboratories, Inc., for ICP AES analysis and R. Knight of the Denver laboratories of the U.S. Geological Survey for INAA analyses. We acknowledge the Bureau of Land Management for help and assistance in accessing sampling sites and providing background information about past mine development and current reclamation interests. Finally, we acknowledge comments on the manuscript by J.M. O’Neill, J.C. Ratté, W.R. Miller, and G.A. Desborough.
Methods
Field Methods
Study Site Identification and Sampling Strategy
Most of the mines in the various districts are shaft or adit openings with the mine spoil piled around or near the openings. In the initial phase of this study, random mine dump piles were selected and two samples consisting of a surficial sample and a sample about 30 cm depth were collected from each mine dump. This sampling allows comparison of surface material, presum ably more subject to weathering and leaching, with slightly deeper material, presumably subject to less exposure and weath ering. In addition, each sample was sieved into three size frac tions, with size intervals at 2 mm and 0.15 mm (8 and 100 mesh, respectively), in order to examine partitioning of trace elements into the various size fractions (Herring and others, 1998). The second phase of the project included resampling of the Lake Val- ley district, where again randomly selected mine dump piles— including some of the same piles that were sampled initially— were sampled but only for surficial material. This time, the sam pling design was altered to ensure representative sampling of the entire surface of the pile in a single sample, which requires obtaining at least thirty, 30-g incremental sub-samples taken ran domly over the surface of the pile. These are combined into a single sample that is statistically representative of the material at the surface of the dump pile. Obviously, it provides no charac terization of material in the inside of the pile. Prior to analysis, the sample was sieved to separate the < 2 mm fraction for analysis. The coarser material was discarded. The strategy behind this type of analysis is to concentrate on this finer grained material, assuming that this material with its large total particle surface area has a greater potential for geoenvironmental interaction compared to the coarser material.
Analysis
Samples were analyzed for 40 elements using inductively coupled plasma–atomic emission spectrometry (ICP-AES). Other analytical techniques were used for As, Cr, Hg, Sb, total S,
and CO2. Herring and others (1998) presented analytical methodology and complete analytical results.
Water Leach Studies
Aqueous extracts of splits of the analyzed samples were obtained using a 20:1 deionized water:rock leach ratio. Leach ing time was 24 hours, and the leach experiments were passive except for a single inversion with return to original position after 1 hour. The leachate was filtered at 0.2 μm, and the pH and conductivity were measured immediately on the filtered sample. The filtrate was acidified to pH 1 and subsequently analyzed by ICP Mass Spectrometry (ICP-MS).
Statistical Techniques and Central and Expected Ranges of Occurrence
Statistical analyses require numeric data sets. When some, but less than or equal to one-half, of the reported values are non- numeric (censored) and below their lower limit of detection (LLD), these censored values have been replaced with arbitrary values equal to 70 percent of their LLD. For the elements with censored distributions, the geometric means and deviations were estimated by the technique of Cohen (1959) for singly truncated distributions. Mineralized samples are unique among crustal rocks in that they represent highly enriched concentra tions of some otherwise minor or trace elements. These concen tration distributions include some enormously elevated concentrations, which greatly skew the arithmetic mean toward higher values. For that reason, the geometric means of sample concentrations are also reported. These log-transformed con centrations remove the weighting bias caused by a few largely elevated concentrations against a field of otherwise much lower values. Further discussion of data treatment is included in the report by Herring and others (1998).
Mine dumps are frustratingly heterogeneous. They are composed of a variety of rock types, including soil, gangue, wall rock, and bedrock. The material can range from weathered to nonweathered and can vary in size. Consequently, the task of modeling the distribution of elements within mine dumps tran scends the enormous past effort expended in attempts to under- stand and model more well behaved distributions of elements, such as lognormal, in rocks and soils that occur in an undis turbed (unmined) environment (Miesch, 1976). Nonetheless, the initial approach used here is to compare concentration data that are based on assumed log normal distributions of various elements. Herring and others (1998) noted that several of the trace elements from the Lake Valley area have concentration distributions that exhibit apparent single- or multi-modal geo metric mean normal distributions. Geometric means are reported along with their arithmetic counterparts in table 1. In the case of a few extremely high concentrations, the geometric mean will always be lower than the arithmetic mean and moder ate the effect of those few extreme concentrations.
Estimates of the distributions about the geometric mean (GM) are provided by the geometric deviation (GD). The
68 Geologic Investigations in the Lake Valley Area, Sierra County, New Mexico