Some Considerations in Applying Background Concentrations to Ground Water Studies


Illinois State Water Survey, 2204 Griffith Drive, Champaign, IL 61820; (217) 333-3729; fax: (217) 244-0777;


Ground water scientists, especially those working in the environmental field, frequently use the concept of background values. Originally conceived for use in exploration geology, the background concept has been embraced and modified by the environmental geoscience community, primarily by estimating concentrations of contaminants in pristine areas. The USGS (2004) defines background concentration in its Water Basics glossary as “a concentration of a substance in a particular environment that is indicative of minimal influence by human (anthropogenic) sources.” Behind this deceptively simple concept, however, are a host of complexities that we discussed in some detail in a previous article (Panno et al. 2006). Runnells (1998) pointed out that the misuse or misunderstanding of background can lead to poor economic decisions in the exploration geochemistry field. We believe that the background concept is sometimes used in ground water studies with little thought for its implications, and its misuse can lead to poor remediation/policy decisions. In this commentary, we summarize the problems often encountered in determining background concentrations originally discussed in our earlier article and offer suggestions for a more rigorous approach.

Determining background: Methods, assumptions, and misapplications

Background values can be determined temporally by reviewing data on contaminant concentrations in ground water prior to human activity or spatially by measuring current concentrations in areas unaffected by human activities (Panno et al. 2006). Calculating a temporal background requires the use of historical data. It is assumed that historical data represent values prior to considerable human influence or at least certain types of influence. Whereas these data probably represent the best estimate of background values available, their primary limitation is that such historical data sets are scarce. There is also the possibility that samples in historical data sets were collected using methods considered unacceptable today (e.g., inadequate purging, unfiltered samples) or analyzed with methods that had biases not recognized at the time (Gillham et al. 1983).

For these reasons, spatial background values are more commonly used, primarily by collecting ground water samples from “pristine” areas or upgradient of the contamination. Another method, which may incorporate both spatial and temporal aspects, is evaluating large data sets using descriptive statistical measures to determine background values.

Although evaluating large data sets is the most useful technique for determining background and threshold values on a regional or larger scale, researchers generally must rely upon the sampling and analytical techniques of others over which they had no control (Panno et al. 2006). Data sets are often neither random nor unbiased, and contaminated areas are often overrepresented (Helsel and Hirsch 1997). In addition, rigorous statistical techniques for evaluating such data can be difficult to apply. Further, data obtained from many locations in a large area are unlikely to have a uniform distribution and may have a large percentage of censored values. When analyzing such data, researchers may fail to account for important natural and anthropogenic differences that can exist among ground water basins or watersheds (Panno et al. 2006).

A common misuse of background is to apply values determined on a regional or even nationwide scale to a relatively small study area. Site data are then interpreted based on this large-scale background value that averages the natural spatial variability of parameters which include topography, vegetation domain, recharge rates, aquifer type, well depths, and proximity to natural sources of the ion/chemical of interest. For example, 2 or 3 mg/L of nitrate as nitrogen (NO3-N) is commonly used as a background value for ground water, citing nationwide studies by the USGS (Madison and Brunett 1985; Mueller and Helsel 1996). However, these values are likely to be too large for forested or wetland areas and too small for areas with geologic or biologic sources of NO3.

Determining background concentrations from ground water samples collected in pristine areas is a common practice, but using such values may be inappropriate. Pristine areas may have substantially different physical, chemical, and biological characteristics than the potentially impacted area(s). For example, applying a background NO3 value determined from forested areas to areas dominated by heavily fertilized row crops may be comparing apples to oranges. There are often important topographic, climatic, soil fertility, geological, and other dissimilarities that are inherent in such land-use differences. Additionally, in many parts of the world, locating an appropriate pristine area can be problematic because of the long history of human impacts to the N budget.

In some cases, pristine regions may have elevated concentrations of the chemical of interest. For example, elevated perchlorate levels in ground water have usually been attributed to industrial and military sources such as manufacture of munitions, but recent research has shown that naturally high perchlorate concentrations are found in some semiarid environments (Rajagopalan et al. 2006; Rao et al. 2007). Using background values from such pristine environments would obviously make it difficult, if not impossible, to identify anthropogenic contamination at the site of interest.

Physicobiochemical processes may complicate background determinations, and stating that samples with concentrations below a threshold value are uncontaminated can be misleading. For example, it is common for NO3-N to decrease dramatically with depth due to denitrification when sufficiently anoxic conditions are encountered along a flowpath (Postma et al. 1991; Kelly 1997). Thus, contaminated ground water may lose its contamination indicator and appear pristine.

Determination of background values should take into account the history of land use in a particular study site. In many parts of the world, human activities have been so pervasive for such a long time that it may be pointless to attempt to determine presettlement values, because concentrations from presumed pristine areas in these locations are likely to have been affected. When studying NO3, the anthropogenic contamination sources of greatest interest are usually synthetic fertilizers and livestock and human waste. Even before widespread application of synthetic fertilizers, clearing and plowing of land occurred, exposing soil organic matter to the atmosphere. That initiated the oxidation of a large reservoir of organic N that ultimately leached NO3 to underlying ground water (Stevenson 1986). Although tilling soil is clearly a human activity, it is not typically accounted for when assessing anthropogenic sources. Using chloride (Cl) concentrations as an indicator of contamination from landfill leachate or large-animal facilities may be similarly complicated in metropolitan areas in snowy climes. Runoff from road salting has raised Cl concentrations in many shallow aquifers in the northern United States and Canada, potentially making it difficult to identify other sources of contamination (Howard and Haynes 1993; Kelly and Wilson in press).

In our previous article (Panno et al. 2006), we proposed the concept of “present-day background” to include concentrations of human-related contaminants such as NO3 and Cl that are elevated above presettlement concentrations. This concept is consistent with the fact that anthropogenic activities now dominate global environmental changes, with some scientists suggesting we acknowledge that we have left the Holocene and entered the Anthropocene Epoch (Crutzen 2002; Zalasiewicz et al. 2008). With respect to NO3, present-day background includes sources such as soil organic matter from the residue of fertilized and unfertilized crops, products of combustion, and evaporation of ammonia (NH3) from synthetic and organic N fertilizer and livestock waste, in addition to naturally derived NO3.

This idea of geographically variable background values has long been recognized in exploration geochemistry; concentrations of metals in natural water can be orders of magnitude higher in mineralized areas than in nonmineralized areas (Runnells 1998). The common use of upgradient well(s) to determine background is also a tacit acknowledgment of this idea; the threshold level of concern, above which there is contamination at a site, is whatever the concentration is entering the site. Typically, this background does not represent an environment “indicative of minimal influence by human (anthropogenic) sources” and may be part of a heterogeneous plume entering the site.

Considerations when determining background

When considering background, researchers should ask themselves questions such as (1) How will the background values be used? (2) Do I need a temporal or a spatial background? (3) How is the background value to be determined (e.g., upgradient well(s), historical data, statistical evaluation of a data set)? (4) What does the background value actually represent (e.g., pristine vs. present day vs. contaminated upgradient)? (5) Is this background value appropriate for the study area? (6) Are there potential sources of the contaminant in addition to the source of concern? and (7) Are there physicobiochemical processes that may affect contaminant concentrations within the hydrologic environment?

In general, background values should be determined on the scale of the particular study because these values are not universal entities in three-dimensional space and may be site and/or depth specific. For a small site, the best method for determining background is usually installing several upgradient wells. Ideally, these wells should be screened in the same flow zone(s) as the suspected contaminants. For larger scales such as a watershed, a statistical evaluation should be conducted using a large number of samples (100 or more), preferably collected from ground water with similar geochemical conditions (if the contaminant of interest is affected by geochemical conditions) within the watershed. Rigorous methods for evaluating such data include cumulative probability plots, the iterative 2-σtechnique, and calculated distribution functions (Sinclair 1974, 1991; Nakić et al. 2007). A bonus from probability plots is that additional threshold values are sometimes determined, which may give the researcher greater insight into the sources and fate of the contaminant (Panno et al. 2006). Finally, historical records of land use should be considered, and any historical data should be used in concert with the methods described previously if they were collected in the same (or similar) location as the study area.


We thank Peter Kitanidis and an anonymous reviewer for their helpful suggestions to improve this article.