Many countries have completed, or are in the process of acquiring, national geochemical datasets, including the UK (McGrath and Loveland, 1992), Finland (Elo and others, 1992), the Baltic states (Riemann et al, 2000), Slovakia (Curlík and efcík, 1999) China (Xie and Ren, 1993), Mexico (unpublished), Canada (Friske and Hornbrook, 1991; Hornbrook, 1989; see also the Canadian National Geochemical Reconnaissance website), and many others. In the United States, there are two datasets in existence that can be considered national geochemical coverages. Shacklette and Boerngen (1984) collected a suite of ~1300 soils across the 48 conterminous states, achieving a sample density of 1/6000 km2, equivalent to the collection of samples on a 75-km grid across the country. These were analyzed for ~40 elements; modern maps of ~20 elements are presented in Gustavsson and others (2001). The low density of these data limits their usefulness for mineral and environmental studies except at very small scales, but they do establish broad trends in elemental concentrations in soils across the country, and are widely cited. A much higher resolution national geochemical coverage of the US is provided by the airborne gamma-ray spectrometry data collected in the late 1970's and early 1980's by the National Uranium Resource Evaluation (NURE) program, reported by Duval and Riggle (1999; additional references cited therein) for the conterminous states, and Duval (2001) for Alaska. This method provides data for maps of K and "equivalent" Th and U at an effective resolution of a few km, but is limited to these three radioactive elements.
The NURE program also conducted an extensive "Hydrogeochemical and Stream Sediment Reconnaissance" (HSSR) program (see Smith, 1997), resulting in the collection of several hundred thousand samples of stream sediment, soils, lake sediments, and a few other solid sample media, and a comparable number of stream and well water samples across ~65% of the land area of the US. The average density of NURE HSSR sampling is ~1 per 20 km2, equivalent to sampling on a 4-5 km grid. NURE may then be considered a 2/3 complete national geochemical coverage. Unfortunately, the NURE program did not adopt a single set of analytical protocols. Four laboratories conducted analyses, each using its own set of methods, each measuring a different set of elements. This lack of consistency, combined with poor quality control for many methods, renders the NURE data difficult to use for geochemical studies, especially at a national scale (see Grossman, 1998; there are many other studies in the literature that use NURE data). Although the chemical data generated by the NURE program are problematic, the vast sample archive that resulted from the program is a valuable resource, the existence of which ultimately enabled the creation of the National Geochemical Survey.
Geochemical sampling programs within the USGS
Since the 1960's, the USGS (like many other public and private organizations) has been engaged in a wide variety of
geochemical sampling projects at scales ranging from a few km (e.g., around individual mines) to 10's of km (e.g., in small wilderness areas) to 100's
of km (e.g., in mineral resource assessments of 1 x 2 degree quadrangles) to
state-scale (e.g., a soil-sampling study of Missouri; Miesch, 1976). These
have resulted in a patchwork of fairly dense sample coverage across a small fraction of the US: ~10% of the conterminous
states and ~50% of Alaska (Bailey and others, 1999), for a total of ~15% of the country's
land area. However, a multitude of sampling methods and analytical techniques were used in doing these studies, and the
majority of the chemical analyses were done by semiquantitative (imprecise and
sometimes inaccurate) methods that, although adequate for the purposes of
exploration studies being done at the time, are not suitable for a multi-purpose
In essence, these USGS data suffer from the same problem as the NURE data, being uneven in quality and elemental
coverage. Moreover, the USGS samples are, as a group, more poorly documented than those from the NURE program, both in terms of
having readily accessible field notes and precise geographic coordinates. (Note:
we believe that most of the USGS samples probably are well-documented
somewhere, either in field notes or published reports. There is, however,
no simple way to retrieve most of this information, which has not been archived
in any central database.)
In addition to conducting their own geochemical sampling programs, many projects at the USGS since the mid 1980's and
continuing to the present time, have made use of the NURE sample archive to conduct research and mineral assessments.
In many cases, the original NURE analyses were not considered adequate to
achieve project goals, and samples were reanalyzed by more suitable methods. Several
tens-of-thousands of NURE samples have received such treatment prior to the current work by the NGS, covering a small
fraction (<10%) of the nation.
Regional-scale geochemistry at the USGS
In the early 1990's, the Mineral Resources Program (MRP) at the USGS began an effort to provide assessments of the nation's mineral-resource potential on a regional basis. One of the initial priorities of this effort was focused on the southeastern coastal states, reaching from Virginia to Mississippi. A critical component of this assessment was to be a compilation of existing geochemical data on stream sediments and other surficial materials. Initially, the only available data were those measured in the original NURE program, which offered a fairly well-controlled dataset: nearly all samples were stream sediments processed in a single way and analyzed by a single laboratory (Savannah River National Laboratory), using only two main analytical methods (neutron activation and atomic absorption). Unfortunately, coverage was lacking for nearly all of Florida and Mississippi and large parts of Georgia and Alabama. Also, many important elements analyzed by atomic absorption, such as Pb, were only determined in about half of the covered area.
In order to extend geochemical coverage in the southeastern US beyond areas reached by the NURE program, MRP began in 1997 to
collaborate with state-government agencies to collect and analyze new stream sediment and soil samples. Sample collection was
based on a 10x10 km grid of collection cells, providing a density ~5x lower than was done by NURE, but still sufficiently dense
to delineate large-scale regional geochemical trends while still being affordable. At the same time, MRP also began an effort to
systematically reanalyze NURE samples in the southeastern states, especially in the Gulf Coastal Plain Province (due to
its importance in the areas of agriculture and mineral deposits). The new data would be compatible in terms of
analytical methodology and spatial density with the new samples being collected by state governments.
National-scale geochemistry at the USGS: the NGS
By 1998-1999, it had become clear within MRP that there would be great benefits to combining these early efforts in the southeastern states with other regional geochemical assessments being conducted by the Program in the western US (e.g., Folger, 2000), and expanding to a national scale. A new national effort comprising four interlinked regional "Surveys and Analysis" projects was begun, including a task to undertake a National Geochemical Survey (NGS). A complete national geochemical coverage was sought, using an internally consistent set of analytical methods and principally based on stream sediments. The minimum sample density was planned to be 1 sample per 289 km2 in all land areas of the country (i. e., based on a 17x17 km sampling grid). Analytical methods were chosen to include a 40-element ICP package plus single-element determinations of As, Se, and Hg by atomic absorption for every sample.
To accomplish the goal of having a national coverage, a multifaceted approach for the NGS was devised: (1) NURE samples that were already reanalyzed by the USGS using the appropriate methods would be incorporated into a database. (2) A subset of the existing NURE stream sediments and other solid samples would be reanalyzed by the selected methods in all other areas covered by the NURE program. An archive of all NURE samples was available at the USGS in Denver, Colo. Software was custom-designed to aid in the random selection of NURE samples within a grid framework, and to select samples for quality control purposes. (3) Existing data from previous USGS geochemical sampling programs for stream sediments would be incorporated if collection and analytical methods were compatible with those of the NGS. Data would be extracted from the National Geochemical Database, also administered by MRP at the USGS. (4) Samples taken from the archives of stream sediments collected by earlier USGS sampling programs would be reanalyzed to fill in any areas not already covered. (5) Collaborations would be established between the NGS, state governments, and private industry to collect and analyze new samples in areas where none were currently available. Stream sediments would be collected wherever possible. Data from all of these sources would be combined into a single national database.
Stream sediments were chosen as the principal sample medium in the NGS for several reasons. Stream sediments have traditionally been used in geochemical mapping programs because they integrate all sources of sediment in the stream's drainage basin. Soils, the other commonly used solid medium, are likely to represent a much smaller volume of source material. This potentially allows detection of geochemical anomalies due to mineralization or anthropogenic point-sources of contamination using a much lower-density sampling array of stream sediments than soils. However, stream sediments are not available in all locations. In the NURE program, stream sediments were collected at ~80% of all sites where solid samples were taken, but other media including soils (~12%) and pond sediments (~8%, mostly in Alaska) were also collected out of necessity. Because the NGS relies on NURE as the principal source of samples in much of the US, all of these media were accepted for inclusion into the database and for reanalysis. In those parts of the nation where the NGS and its collaborators would collect new samples, soils would be substituted for stream sediments where necessary. This includes areas of very low relief and poor drainage, such as in southern Florida and the lowlands of southwestern Alaska, where standing water or tundra dominates the landscape, and farm lands where local streams have been largely channelized and diverted for agricultural purposes (e.g., northwestern Mississippi).