THE IOWA STATE-WIDE RURAL WELL-WATER SURVEY: SITE AND WELL CHARACTERISTICS AND WATER QUALITY
G.R.Hallberg, K.Woida, R.D.Libra, K.D.Rex, K.D.Sesker, B.C.Kross, L.S.Seigley, B.K.Nations, D.J.Quade, D.R.Bruner, H.F.Nicholson, J.K.Johnson, and K.L.Cherryholmes
Iowa Department of Natural Resources, Geological Survey Bureau,
Technical Information Series 23, 1992, 43 p.
The Iowa Department of Natural Resources (DNR), in conjunction with the University of Iowa, Center for Health Effects of Environmental Contaminants (CHEEC), conducted the State-Wide Rural Well-Water Survey (SWRL) between April 1988 and June 1989. The SWRL survey provided a statistically valid assessment of the proportion of private rural wells and rural Iowa residents affected by various environmental contaminants. Previous reports reviewed the SWRL design, water-quality results, and relationships among individual site characteristics and water-quality. Few site factors occur in isolation, hence, this report focuses on analysis of combinations and interactions of site characteristics, well-siting and well-construction features, and their relationship with water-quality, to refine possible causal associations. This assessment also uses "gradient" measures to evaluate the associations of various factors at varying distances from a well. If a point source is a major causal factor, its association with contamination should be stronger as proximity to the well increases. Many trends and associations are not statistically significant (at 95% confidence) but are presented if logically consistent.
While several site/well system factors are associated with water quality, most are subordinate to, or interrelated with the overriding factors of well (and casing) depth and well-construction type. Shallow (<50 feet deep) wells of any construction type show the greatest proportions of all contaminants. Large-diameter seepage wells, which are typically open to the water table, account for over half of the wells that are 50 feet deep. Thus, they exhibit greater proportions of contamination, particularly for total coliform positives. Drilled, cased wells are generally much deeper and exhibit lower contamination rates. Drilled wells with less than <50 feet of casing are functionally shallow wells, regardless of depth, and have contamination rates for chemical constituents similar to state-wide averages. Drilled wells that are cement grouted are generally deeper wells and show statistically better water quality. Older wells are often presumed to produce water with greater contamination than younger wells, because of aging effects on the well materials and/or less rigorous construction and siting considerations. But this trend is inconsistent, not significant, and is confounded because older wells are also more shallow and because a large percentage are seepage wells. There is no indication that wells constructed after 1982 (initiation of new well-construction standards) produce any better water quality than older cohorts, though the sample size of newer wells is small and little analysis possible.
Sites with sealed wellheads show somewhat less contamination than average, while unsealed wellheads are characterized by somewhat higher contamination rates. Large-diameter seepage wells, even with a sealed well head or with pitless adapter (buried-slab) construction, still show greater than average proportions of most contaminants. For drilled wells the difference between sealed and unsealed conditions was more significant, but this accounts for only a small portion of contaminated wells (about 1%). Water distribution and storage systems did not significantly affect water-quality results, with the exception of the high proportion of coliform positives that are associated with cisterns and other outside water-storage structures (even concrete tanks). Sites where more than one operable well is present showed greater than average occurrences for nitrate-N >10 mg/L and coliform positives. However, this is largely a function of the use of multiple seepage wells at the majority of these sites. Sites with only one operable well were dominated (70%) by deeper drilled wells.
Previous analysis of the SWRL data showed minor differences in the rates of some contaminants at sites with abandoned wells, no septic systems, or chemical storage on the property. When these factors are stratified by distance to the water-supply well, and by number of abandoned wells, consistent proximity effects are not apparent. For example, wells located within 15 feet of chemical storage and handling areas are uncommon, occurring at <0.6% of rural sites, and none of these wells contained pesticides or nitrate-N >10 mg/L. One of the most striking associations is that water quality is significantly better in non-farm, suburban housing tracts than agricultural areas. Even though the suburban wells are located closest to septic systems they had significantly fewer sites with >10 mg/L nitrate-N and coliform positives. A prime example of the confounding of variables is that wells with the greatest distance (>200 feet) from septic tanks and chemical storage and handling areas often exhibited the greatest degree of contamination. This relationship results from the high proportion of seepage wells in the greatest distance categories. These wells tend to be located along waterways, or other areas with a high water table (to promote seepage), often quite distant from houses, cattle-yards, storage buildings and chemical-mixing sites. Site studies show that these wells are located an average of one-quarter of a mile from such likely point-sources. While such wells are highly susceptible to impacts from accidents and point sources, the majority are located so far from such sources that there is limited potential for direct impact. Hence, these wells are "under-represented" in point-source cases.
The effects of sinkholes or agricultural drainage wells (ADWs) are not significant in a state-wide context. Sinkholes were identified in the vicinity of only 2.1% of sites and only 0.6% of sites were near ADWs. No sites with ADWs had pesticide detections or nitrate-N >10 mg/L.
Onsite disposal of home and farm refuse shows no effect on water-quality in the SWRL survey. For farmed sites neither the crop nor livestock enterprises, rates of N-fertilization, nor the herbicides applied were reflected in the water quality at the on-farm well. However, these patterns are relatively ubiquitous in Iowa. Neither methods nor locations of disposing of excess pesticide formulation, rinseate, or pesticide containers exhibited any significant association with water-quality results. Wells located in feedlots showed significantly higher concentrations of nitrate, but not bacteria problems. Such sites comprise only about 3% of wells state-wide, and account for only about 1% of the wells with >10 mg/L nitrate-N.
Sites that reported the formulation of pesticides at a hydrant at or near the wellhead and/or spills of pesticides were related to greater than state-wide averages for pesticide detections. However, when the results are stratified by distance from the well, and by well depth and well type, consistent proximity trends are absent. Sites where residents reported that all the herbicides were custom applied showed the same proportion of pesticide detections as those where they mixed all the herbicides on-site.
About 5.4% of farms reported spills, back-siphoning, or other accidents with pesticides and/or fertilizers near their wells. As expected, these sites show a greater proportion of pesticide detections. However, of the detections at these sites, only 36% contained the pesticide involved in the mishandling incident. The relationship between the proximity (i.e., the actual potential for an effect on the well) of mixing and handling of pesticides to the well-head and pesticide detections, while inconsistent, does indicate a significant association within a distance of about 50 feet. The high proportion of pesticide detections at sites where pesticides are mixed (and spilled) within 50 feet of the well head stands up throughout the analysis of the many confounding variables, e.g., the sites are not dominated by seepage wells or inordinately shallow wells. This relationship is a strong indication that such handling, over time, has likely impacted these wells adversely. Spills and mixing near the wellhead still only explains a small portion of the total occurrence of pesticides, however. When all detections from wells within 50 feet of mixing sites are considered, they account for about 1.5% of pesticide detections state-wide; combined with other known spill and accident sites (whether the compound found was related to the spill or not), they account for about 3-3.5% of detections state-wide (of the 13.6% measured). Such detailed analysis is requisite for understanding the magnitude of such impacts. While there is little question that point-source problems (e.g., spills, back-siphoning, repeated mishandling near a well) occur and contaminate wells, it is easy to overstate their occurrence from anecdotes. The estimates of the extent of such incidents derived from the detailed data collected in the SWRL study are similar to the values derived from other detailed field investigations in Iowa. The review of sites where pesticide concentrations exceeded HALs provided similar insights: 25%, were clearly "point- source" cases, one a spill and one back-siphoning accident; the majority, 62.5%, were probable nonpoint sources related to pesticide occurrences in shallow groundwater; 1 case, 12.5%, was equivocal.
Neither simple proximity nor susceptibility are cause and effect. Neither the proximity of handling and mixing to a well, nor the use of a susceptible well-type dictate that point sources are the cause of a pesticide detection. The continuing analysis of the SWRL survey data reinforce prior findings, that: 1. point-source problems clearly contribute to the occurrence of pesticides in water-supply wells; but, 2. they account for a small portion of the problems; and, 3. nonpoint sources also account for many pesticide detections in water-supply wells, which is supported by other detailed research.