USGS Home
Contact USGS
Search USGS

U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2008–5167

Back to Table of Contents

Data Acquisition and Methods

Stratigraphic data were collected from multiple sources and used to test hypotheses of statistical similarity of the frequency distributions of sediment interbed thickness. Data were collected from several sources and by various methods, and were analyzed by a selected suite of statistical tests.

Data Collection

Subsurface stratigraphic information for boreholes in this study was derived from lithologic data and natural-gamma geophysical logs. Independent stratigraphic interpretations based on only natural-gamma logs and interpretations based on a combination of natural-gamma and lithologic log information were catalogued in a spreadsheet database (appendix A). Unless otherwise specified, “interbed thickness” data in this report refers to what are considered the most accurate inferences of thicknesses derived from a combination of lithologic and natural-gamma log information.

Lithologic logs from “USGS series” holes 127 through 134 were obtained from the USGS INL Project Office in electronic format (table 2). Lithologic information from holes designated Middle 2050a and Middle 2051 was obtained from photographic logs of recovered core. Visual distinction between sediment and basalt in the core-box photographs was fairly straightforward. Lithologic information for hole Middle 1823 was available only from a depth of 500 ft to 1,654 ft (Mazurek, 2005); the first 500 ft of the hole were not cored (table 2). Natural-gamma logs were provided by the USGS INL Project Office for all coreholes in this study and were the only type of geophysical data used in this analysis.

Each lithologic unit encountered in a corehole was assigned to a composite stratigraphic unit by correlating the corehole’s stratigraphy with that of nearby holes in which a composite-unit stratigraphy had been previously defined by Anderson and Liszewski (1997). Boreholes that were used for this purpose are listed in table 3. Where direct correlation of individual units was difficult or ambiguous, a comparison of depth of interbed and depth of composite units in surrounding holes aided in composite unit assignment.

Five coreholes (USGS 129, USGS 130, USGS 133, USGS 134, and Middle 1823; fig. 2) terminated within a sedimentary interbed, thus providing only a minimum thickness estimate. Such interbeds were excluded from subsequent analysis. Such omission ensures that all thickness data used in this analysis are tied to best estimates of actual thicknesses. Considering the rarity of this occurrence (5 interbeds of 122 observed), omission of these beds does not significantly affect the statistical tests.

Analysis Approach

Welhan and others (2006) concluded that the proportion of sediment relative to basalt in composite unit 1 was larger than that in other composite units. Composite unit 1 also tends to lack thin interbeds (less than about 10 feet) (figs. 3 and 4), which may be the result of relative volcanic quiescence over the past 200 Ka (Anderson and others, 1997; Champion and others, 2002; Welhan and others, 2006).

Welhan and others (2006) tested the frequency distribution of sediment abundance within individual composite units in 333 boreholes across the INL and concluded that these frequency distributions did not vary in a statistically significant manner among composite units 2 through 7 (and less confidently, between composite unit groups 2–7 and 8–14) (Welhan and others, 2006, p. 12 and 16). This study tests the stationarity of the frequency distribution of individual interbed thicknesses within a composite unit rather than the distribution of total sediment thickness in a composite unit. The term distribution will be used hereafter in the statistical sense (that is, synonymous with frequency distribution).

The analyses conducted here differ from the approach taken by Welhan and others (2006) because of the small sample size of the new corehole data set. The analyses conducted here allow more refined tests of temporal stationarity because they consider more information about the statistics of individual rather than aggregate interbed thicknesses, whereas the approach taken by Welhan and others (2006) is better at testing spatial stationarity across time. The two approaches are equivalent if the relative rates of sedimentation and volcanism that affect sediment thickness distributions have been invariant in time and space.

Nonparametric statistical tests of similarity were performed on distribution shapes, medians, and variances of interbed thicknesses to determine if the different sample populations displayed statistically significant differences or if they were statistically stationary. The Kolmogorov-Smirnov (K-S) two sample test was used to compare distribution shape (form, skewness, peakedness, and degree of tailing) among pairs of sample populations; the Mann-Whitney (M-W) test was used to determine similarity of two medians; the Kruskal-Wallis (K-W) test was used to determine similarity of multiple medians; and Levene’s (L) test was used to determine the similarity of two variances.

Back to Table of Contents