Hydrogeologic Controls on the Occurrence and Distribution of Radon-222 in Glacial Aquifers of Southern Rhode Island

A field study was conducted to evaluate the distribution of Radon (222Rn) in the glacial aquifers of the Pawcatuck River Basin, Rhode Island. A total of 95 ground-water samples were collected from private wells in stratified-sediment and bedrock aquifers of the Upper Wood, Queen-Usquepaugh, and Chipuxet river basins. Gamma-ray (uranium, thorium, and potassium) emissions from the regolith material throughout the study area were measured. The ground-water samples were analyzed for basic chemical constituents as well as uranium and 222Rn to help evaluate the factors controlling 222Rn distribution. The granite of the Scituate Igneous Suite underlies the Upper Wood River and Queen-Usquepaugh aquifers. The granite gneiss of the Esmond Plutonic Suite underlies the Chipuxet and the Queen-Usquepaugh aquifers. The granite gneiss of the Sterling Plutonic Suite is found underlying the Upper Wood River aquifer. The uranium-bearing minerals (source of radon) found in the bedrock are zircon, allanite, sphene, and monazite. The average uranium content of the Esmond Gneiss is 1.9 ppm, Sterling Gneiss is 3.3 ppm, and Scituate Granite is 4.1 ppm (Nevins, 199 I). All wells sampled in this study yielded radon levels above the proposed EPA limit of300 pCi/L, with many being an order of magnitude or more greater. Wells in areas underlain by the Esmond Suite had the lowest radon content (range 500 to 30,400 pCi/L, median 1,400 pCi/L), areas underlain by the Sterling Suite were not significantly different but showed slightly higher concentrations (range 700 to 27,300 pCi/L, median 1,600 pCi/L), however, the areas underlain by the Scituate Suite had significantly higher levels (range 1,600 to 83,500 pCi/L, median 5,900 pCi/L). Water chemistry factors play little if any role in influencing radon concentrations. High fluoride concentrations in ground water, however, indicate that the mineral fluorite is present in the underlying bedrock. Fluorite is commonly found with uranium-bearing minerals in A-type granites. The physical processes such as well yield and the siting of uranium are the controlling factors in the distribution of radon between surficial and bedrock wells. Bedrock aquifers exhibited higher radon concentrations than surficial-materials aquifers because surficial-materials aquifers have greater water-transmitting capacity, thus a greater volume of water to dilute the radon. Radon concentrations showed no correlation with the uranium content in the surficial material. However, higher radon levels in ground water correlate with


List of
Introduction  is a radioactive isotope with a half-life of3.8 days. It is produced during the decay ofuranium-238 to lead-206 ( Figure 1). In an aquifer, radon-222 (hereafter radon) is ejected from the solid and into the adjacent pore space, mineral grain, or ground water . Because of its short half-life and the relatively slow flow of ground water in most settings, radon is unlikely to be transported great distances in aquifers. Therefore, when high levels of radon are found in ground water, its source (uranium) is likely to be present in the surrounding rock or sediment. Previous studies, conducted in the United States, comparing radon in ground water to bedrock type and uranium content found higher ground-water radon concentrations in the bedrock containing the most uranium . Because radon is undergoing nuclear disintegration when it decays, its concentration is reported as disintegrations per unit time. A becquerel (Bq) is one disintegration per second. Radioactivity in ground-water is usually reported as picocuries per liter (pCi/L); 1 pCi/L = 37 Bq/m 3 .
The United States Environmental Protection Agency (USEPA) has determined that elevated levels of radon gas in indoor air is the second leading cause of lung cancer, cigarette smoking being the number one cause. Several studies on indoor-air radon in homes with high radon concentrations in their domestic water supplies have shown that higher indoor-air radon concentrations correlate with indoor well-water use ( Figure 2) Wanty et al, I 992, and Folger et al, 1994).
Approximately I to 7 percent of radon-related deaths are the result of radon that is released from well water during normal household activities .
The health risk of radon in well water is not from direct ingestion, rather the degassing of radon into the indoor air poses the threat ). When radon is inhaled, it and its a,-emitting daughter products (see Figure I) deposit on the lung tissue. Therefore, prolonged inhalation of 222 Rn-rich air can cause carcinogenisis in the lung tissue (National Research Council, 1988   At the 1983 National Workshop on Radioactivity in Drinking Water, a population-weighted average radon concentration in public water supplies serving less than 1,000 people was estimated at 780 pCi/L . Because of the potential health threat of radon, the USEPA originally proposed a maximum contaminant level (MCL) of300 pCi/L for radon in drinking water. This MCL was not accepted by Congress and is currently under revision.
The small public-water suppliers and private wells throughout the United States are in greater danger of high radon levels because many of these small, private supplies are wells located in the less productive crystalline rock aquifers. Large public-water supplies consist of surface water bodies.
Furthermore, the time between extraction from the aquifer and actual water use is less than one half life of 222 Rn in private wells. The USEPA predicted higher radon concentrations in much of New England as compared to the rest of the United States . In 1986, the Rhode Island Department of Environmental Management (RlDEM) initiated a statewide, private well water quality survey and radon was detected in each of the 310 private wells tested (RlDEM, 1990). If the USEPA were to adopt an MCL of 300 pCi/L for radon, over 90% of these Rhode Island wells would exceed the standard. Studies were also conducted on indoor-air radon concentrations throughout Rhode Island. Using composition of underlying bedrock, glacial deposit distribution, and indoor radon data collected by Rhode Islanders Saving Energy (RISE) program,  made a geologic radon potential map of Rhode Island.
This radon potential map of the state showed that Rhode Island has moderate to high radon potential according to EPA' s criteria.
In addition to the uranium content of the underlying bedrock, radon concentration in well water is also a function of the physical properties of the rock and sediment comprising the aquifer. Previous investigations of ground-water radon potential in other states have found that several factors including uranium mineralization, fracture aperture in the underlying bedrock, and degree of metamorphism also impact radon concentrations of ground water Wathen, 1987, Krishnaswami andand Peter Folger, Colorado School of Mines, Dept. of Geology and Geologic Engineering, written communication, 1994).
Ground water is the potable water source for approximately 24% of Rhode Island as a whole, and for I 00% of southern Rhode Island. Previous studies of crystalline rocks underlying the state have found significant amounts of uranium . Furthermore, much of the surficial material originated from this uranium-bearing granitic bedrock. Therefore, the potential for high levels of radon exists in wells in surficial material as well as in the underlying bedrock.
The purpose of this study was to try to understand the influence of hydrogeologic factors on the occurrence of radon. The field assessment included the following objectives: • measurement of 222 Rn levels and chemical composition of ground-water in wells screened in glacial material as well as bedrock; • determination of the variability of 222 Rn within the three-dimensional framework of the aquifer; • evaluation of the factors controlling 222Rn levels, for example: uranium content in the underlying bedrock, depth to bedrock, and chemical evolution of the groundwater; and • development of a conceptual model for the distribution of 222 Rn in the aquifers of the Pawcatuck River Basin, Rhode Island.

Study Area/Geology
Rhode Island is located within the A val on lithotectonic zone, a zone containing approximately l 0 structurally and stratigraphically distinct bedrock units (Plate 1-modified after . The study area is the Pawcatuck River Basin, located in southern Rhode Island (Figure 3) Group, dating to the late Proterozoic, approximately 600 million years before present (MYP) . This alaskite gneiss is a quartz-rich granitic gneiss containing sodic plagioclase and microcline, with accessory hornblende, magnetite, biotite, muscovite, sphene, and zircon. The augen granite gneiss of the Esmond Igneous Suite is a late Precambrian (approximately 620 MYP), calcalkaline rock containing quartz, plagioclase, biotite, potassium feldspar, and accessory epidote, chlorite, muscovite, sphene, monazite, apatite, and zircon . The final bedrock type in this study area, a member of the Esmond-Dedham terrane, is the granite of the Scituate Igneous Suite, located in the northern portion of the study area. This granite is Devonian in age (approximately 370 MYP) and is composed of quartz, plagioclase, potassium feldspar, and accessory biotite, allanite, sphene, fluorite, calcic hornblende, calcite, and zircon .
The alaskite gneiss of the Sterling Plutonic Group and the augen granite gneiss of the Esmond Igneous Suite are I-type granites formed from hydrous melts on continental plate edges in island arc environments . The granite of the Scituate Igneous Suite is and A-type granite, an anhydrous granite formed from high temperature melts, containing fluorine, and formed in stable fold belts and tensional regimes in continental crust . A-type granites  Figure 3. Location ofQueen-Usquepaugh, Chipuxet, and Upper Wood River aquifers and the other major aquifers in the Pawcatuck River basin and generalized surficial geology (modified from Johnston and Dickerman, 1985).
differ from I-type granites in that they are enriched in large, high field-strength elements (like uranium),!type granites have lower quantities of these elements .
Crystalline rocks differ from unconsolidated deposits in that they have little primary permeability, and therefore water flow through these crystalline rocks is along fractures. Wells completed in these crystalline bedrock aquifers have lower yields than stratified-sediment aquifers, ranging from 0.5 to 80 (median I 0) gallons per minute Dickerman, 1985, andDickerman and  . Gamma-ray emission data from bedrock outcrops show average uranium concentrations of 5.5 ppm (range 2 to 14 ppm) for the Esmond Gneiss, 5.0 ppm (range 3 to 9 ppm) for the Sterling Gneiss, and 7.2 ppm (range 3 to 20 ppm) for the Scituate Granite (Veeger and Hermes, University of Rhode Island, Geology Department, 1994, unpublished data).
The three aquifers that comprise this study are unconsolidated aquifers of glacial origin deposited in southward-trending bedrock valleys during the retreat of the late Wisconsinan ice sheet .
The sides of these valleys are bedrock highs covered with a thin deposit of till._ Large braided meltwater streams from the glaciers at the head of the valleys flowed south, depositing sediment along the valley floors in a deltaic sequence. This glacio-fluvial/glacio-lacustrine environment created a strongly heterogeneous materials distribution. The glacial deposits in the valleys consist of fine-to coarse-grained sand, with some gravel and silt, derived from granitic igneous and metamorphic rocks to the north. A generalized geologic cross-section of the Chipuxet aquifer is provided in Figure 4. Deposition in the Upper Wood and Queen-Usquepaugh produced a similar accumulation of sediment. The sand and gravel deposits are parts of the stream, delta slope, and delta-plain sequences, whereas the silt, and fine-grained sand deposits are part of a lacustrine environment. The gravel at the base of the stratified deposits in the Chipuxet is a buried esker, or ice tunnel deposit (Jeff Campbell, University of Rhode Island, Dept. of Geology, written communication). The aquifers vary in thickness, ranging from less than 50 feet of saturated thickness to greater than 150 feet of saturated thickness in the deepest portions of the bedrock valleys. Ground-water flow in these unconfined aquifers is in a southerly direction through the porous granular material. Well yields from the stratified sediment deposits generally range from 100 to 900 gallons per minute (Johnston and Dickerman, 1985). These aquifers are complex, and in some areas vertical mixing between the bedrock and surficial material aquifers may occur.

Approach and Methodology
Ground-water samples were collected from the stratified-sediment and bedrock aquifers within the three river basins. Wells completed in shallow surficial material and those close to the bedrock surface were sampled to obtain information concerning the distribution of radon within the surficial material. The samples from the stratified-sediment deposits close to bedrock will either represent the more chemically evolved stratified-sediment aquifer ground water, or a mixture between the underlying bedrock ground water and that of the surficial material just above it. Ground-water samples within the underlying bedrock were taken to evaluate the radon potential of these uranium-bearing bedrock units. ~ u e: u ~ A gamma-ray spectrometer was used to measure the amount of radioactivity in the regolith material that can be attributed to uranium, thorium, and potassium, three naturally occurring radioactive elements that produce gamma rays. These gamma-ray emission data may provide insight into the radon potential of the surficial material throughout the study area.
Well-log data, where available, were obtained from the United States Geological Survey (USGS)-Water Resources Division, Providence, Rhode Island. Information of interest included stratified-sediment aquifer material (sand or gravel, for example), depth to bedrock, and well yield.
Surveys were sent to homeowners throughout the Upper Wood River, Queen-Usquepaugh, and Chipuxet basin aquifers asking for well and aquifer information (Appendix A). Wells were chosen on the basis of this survey and spatial distribution relative to the underlying bedrock. Each well was screened for possible contamination using electrical conductance. According to Johnston and Dickerman ( 1985), background electrical conductivity values in the study area should be less than 100 µSiem. Wells with less than 200 µSiem specific conductance were preferred because they had the least amount of input from anthropogenic sources. Using the homeowner's well pump, ground-water samples were collected after 3well volumes had been evacuated from each well, and pH, temperature, and electrical conductivity had stabilized. The standard sampling procedures and a field sheet are included in Appendices B and C. Field analyses included temperature, pH, electrical conductance, and dissolved oxygen.
In the Upper Wood River aquifer, 29 out of 37 wells were sampled within or above the gneiss of the Sterling Suite, the remaining 8 wells were sampled within or above the granite of the Scituate Suite (Plate 1). Most of the wells (16 out of21) in the Queen-Usquepaugh were sampled within and above granite of the Scituate Suite, 5 wells were sampled from within and above gneiss of the Esmond Suite. A total of 3 7 wells were sampled in the Chipuxet aquifer from within and above gneiss of the Esmond Suite.
Ground-water samples were collected for the following laboratory analyses: radon, uranium, alkalinity, calcium, magnesium, sodium, potassium, iron, manganese, silica, fluoride, chloride, nitrate, phosphate, and sulfate. All samples (except for radon) were filtered through 0.45 µm filters and stored at 4 °c in highdensity polypropylene bottles. Samples collected for cation analysis were acidified with concentrated hydrochloric acid and those collected for uranium analysis were acidified with concentrated nitric acid, both to a pH of 2.
Because of the volatility of radon gas, a sampling procedure developed by the USGS was followed when collecting the 222 Rn samples (Rich Wanty, United States Geological Survey, Denver Federal Center, written communication, 1993). First, the flow rate was reduced so no agitation existed in the discharging well water. A 10 ml sample of ground water was collected from inside the hose with a pipette prior to the water coming into contact with the atmosphere. The sample was then dispensed into a vial containing 10 ml of mineral oil-based liquid scintillation cocktail. The sample was immediately capped and shaken so as to partition the 222 Rn into the scintillator phase. A total of 4 7 gamma-ray emission readings were taken from stratified-sediment and till in the three river valleys. These data were used to calculate ¾K (potassium), eU (uranium), and eTh (thorium) content of the materials.

Water Chemistry
Chemical composition of the ground-water samples are included in Tables 1, 2, and 3. In order to define the background chemistry of each of the aquifers, wells with conductivities above 200 µSiem, chloride above 30 mg/L, or nitrate above 20 mg/L were not included in the water chemistry interpretation Table 1. Well data for Queen-Usquepaugh ground-water sampling sites (in mg/L, except as noted).  Table 1 continued. Well data for the Queen-Usquepaugh ground-water sampling sites (in mg/L, except as noted).     Br (these wells are identified in Tables 1, 2, and 3). Concentrations above these levels show excessive contamination from road-salt runoff, fertilizers, and/or septic leachate. Charge balance errors were calculated for the analyses, and samples with errors of 10% or more were excluded.
There exist·distinct chemical differences between the surficial-materials ground water and the bedrock ground water within the three aquifers; bedrock well water has higher pH, fluoride, and silica values than the surficial-materials well water. Median concentrations of selected constituents in the Upper Pawcatuck aquifers are included in Table 4. When the samples are plotted on trilinear diagrams, the chemical differences between bedrock and surficial wells are readily apparent (Figures 5, 6, and 7). The bedrock wells are dominated by HCO 3 (most greater than 40% of anions) and Ca (most greater than 40% of cations) and fall in the Ca+ HCO 3 field. The surficial wells are dominated by higher chloride (most greater than 40% of anions) and Na and K (most greater than 40% of cations) values. However, there is a good deal of overlap between the surficial and bedrock wells in the cation field. The deep surficial wells that plot in the bedrock field show evidence of greater chemical evolution and possible mixing with bedrock ground water.
The water chemistry of the Queen-Usquepaugh aquifer shows considerable water-rock interaction and geochemical evolution. Alkalinity, pH, calcium, silica, magnesium, and conductivity increase with depth into the surficial material, and then further into bedrock.
In the Upper Wood River aquifer, the alkalinity, silica, and conductivity values increase with depth from shallow surficial wells to deep surficial wells, and into the bedrock aquifer. The shallow and deep surficial materials ground water shows very similar pH, calcium, fluoride, magnesium, and sodium values, however they are less than those seen in the bedrock ground water (except magnesium). The geochemical evolution with depth in the Upper Wood River aquifer can only be considered when comparing all surficial material ground water to bedrock ground water.
The Chipuxet aquifer does not exhibit this relationship. Alkalinity, calcium, and conductivity decrease with depth in surficial wells. The bedrock wells are higher in pH, alkalinity, silica, fluoride, and sodium than the surficial-materials wells.  Wood, on the other hand, has higher alkalinity in its surficial materials aquifer, and higher silica and fluoride values in its bedrock aquifer than the Chipuxet and Queen-Usquepaugh aquifers. The Upper Wood surficial materials wells are more chemically evolved than the other two aquifers.
Dissolved uranium concentrations were greater in ground water from the bedrock aquifers than the surficial-materials aquifers. Median uranium concentrations in ground water from the surficial and bedrock wells were 0.33 ppb (n=l0) and 1.7 ppb (n=7) in the Queen-Usquepaugh, 0.32 ppb (n=l5) and 2.32 ppb (n=9) in the Upper Wood, and 0.27 ppb (n=l8) and 2.45 ppb (n=l l) in the Chipuxet. The Queen-Usquepaugh had slightly lower uranium concentrations in its bedrock aquifer than the other two bedrock aquifers.

Radon Distribution
Radon concentrations in the study area ground water were highly variable, ranging from 500 to 83,000 pCi/L. Median radon levels by aquifer are included in Table 5 and a bar diagram of radon distribution by aquifer is included in Figure 8. The single-factor Anova statistical analysis method (at the 95% confidence level) was used to determine whether radon concentrations were drawn from populations with the same mean . Statistical analysis is included in Appendix F. In the Queen-Usquepaugh aquifer, the Anova test showed that despite very different medians the radon concentrations in the shallow and deep surficial wells do not represent statistically different waters. Therefore, the surficial wells were pooled together and compared to the bedrock ground-water radon readings. The Queen-Usquepaugh surficial ground-water radon concentrations are significantly lower than the radon concentrations in the bedrock wells.  The Anova statistical analysis test was also used to evaluate radon data on the basis of the underlying bedrock. Median radon levels are included in Table 6 and a bar diagram of radon distribution by underlying bedrock is included in Figure 9. Surficial wells in areas underlain by granite of the Scituate Igneous Suite showed no significant difference between shallow and deep surficial radon concentrations.
Therefore, these wells were pooled together and compared to the bedrock wells in the granite of the Scituate Suite. Bedrock ground water has significantly higher radon concentrations than the surficial materials ground water. Wells underlain by gneisses of both the Esmond and Sterling Suites also exhibit the same patterns.

Gamma-ray Data
The data for the gamma-ray emissions in the stratified sediment deposits are included in Table 7.
Only slight differences were found between the measured eU and eTh concentrations of the regolith of the three aquifers.

Discussion
Radon levels within the three aquifers are highly variable. This variability is caused by either chemical or physical processes. In order to evaluate which processes have the most effect on the level of radon in a well, radon concentrations must be compared to both the chemical and physical properties of each aquifer.  The chemistry of ground water in the Pawcatuck river basin is affected by several chemical processes: mineral dissolution, hydrolysis, redox reactions, and ion exchange. Furthermore, anthropogenic input can also influence the chemistry of ground water (fertilizer for instance). As recharge water percolates downward through the aquifer materials, organic acids (from plant decay) react with the material in the aquifer (water-rock reactions), and chemical weathering takes place. Figure 14 shows a correlation between alkalinity represented as% meq of the major anions (bicarbonate, chloride and sulfate) and radon. Because the relationship between alkalinity and radon is not predictive, a rank comparison 29 Gamma readings in regolith underestimate actual regolith concentrations because of bulk density variations.
(Speannan's rank test) was used to compare the two parameters. The Spearman's rank test (Appendix F) indicated that there is a statistically significant correlation between radon and alkalinity, with radon levels greater than l 0,000 pCi/L usually found in bedrock ground water with alkalinity accounting for more than 40% of the meq. The radon concentration is attributable to local contact with the water because radon has such a short half-life, so long residence time does not produce higher radon concentrations. This is not a causative relationship between elevated radon and alkalinity. Surficial materials aquifers are more or less being continuously replenished by precipitation percolating downward through the surficial material.
Bedrock aquifers do not have this direct infiltration, therefore, as a result of bedrock weathering and longer residence time, the ground water is more chemically evolved and has a higher alkalinity than surficial materials ground water.
*not including well #20 There is also a general relationship between radon and fluoride concentrations (Figure 15). The Spearman's rank test was also perfonned on these data (Appendix F), and indicated a statistically significant correlation between radon and fluoride in ground water. As ground-water fluoride values increase, ground-water radon concentrations increase. Ground water with fluoride concentrations greater than I mg/Lis likely to have radon concentrations above 10,000 pCi/L. The single-factor Anova statistical analysis method (at the 95% confidence level) was used to detennine whether fluoride concentrations were drawn from populations with the same mean (Appendix F)  The rock in this study that contains fluorite (a fluoride-bearing mineral) is the granite of the Scituate Suite. The granite of the Scituate Igneous Suite is an A-type granite (Hennes and Zartman, 1992), an anhydrous granite fonned from high temperature melts in stable fold belts and tensional regimes in continental crust . Both gneisses of the Sterling and Esmond Suites are considered Itype granites. However, the primary mineralogy of the gneiss of the Sterling Suite is not preserved, thus making it difficult to do a diagnostic characterization of the parental material. The gneiss of the Sterling Suite exhibits some of the chemical features of an A-type granite as well as its I-type characteristics (0. Don Hennes, 1995, University of Rhode Island, Geology Department, written communication). These Itype granites do not contain fluorite, and are fonned from more hydrous melts on continental plate edges in Andean kinds of plate boundaries . The fluorine is associated with the precipitation of U-bearing minerals and thus higher radon levels are found in A-type granites (fluoride complexes will be discussed later). The A-type granite of the Scituate Igneous Suite underlies the Upper Wood River and Queen-Usquepaugh aquifers. The Chipuxet (the aquifer with the lowest radon concentrations) is underlain solely by the gneiss (I-type) of the Esmond Suite, which does not contain fluorite.
The single-factor Anova statistical analysis method (at the 95% confidence level) was also used to detennine whether uranium concentrations were drawn from populations with the same mean .
Bedrock ground water samples have statistically greater dissolved uranium concentrations (median 2.4 ppb) than surficial wells (median .3 ppb), but there was no difference between aquifers (Appendix F). No statistical relationship was found between radon levels and uranium concentrations in the ground-water samples. The solubility of uranium is a function of oxidation-reduction conditions in the aquifer. The effect of redox conditions on uranium solubility is illustrated through a comparison of dissolved uranium and dissolved iron concentrations ( Figure 16). Because uranium is mobile in oxidizing conditions and iron is mobile only under reducing conditions, high levels of uranium (greater than 2 ppb) are found only in waters with less than 2 mg/L iron. In addition, wells in the area underlain by the Scituate Granite, the bedrock with the highest uranium content, have, on average, the highest radon levels. The source of radon in ground water therefore, is the uranium in the solid phase, not dissolved uranium. Previous studies have also demonstrated this relationship (Gundersen, 1989, Wanty and. Iron m /L

Figure 14. Uranium versus iron for different water-bearing zones in the Chipuxet, Queen-Usquepaugh, and Upper Wood River Basins (not including wells 19 and 49).
Previous investigations of radon in ground water have found that radon levels are affected by physical processes and not chemical processes . In this study of Rhode Island ground water, different radon levels are seen in waters with very similar chemical signatures, the Upper Wood and Queen-Usquepaugh deep surficial materials wells for example. The Queen-Usquepaugh deep surficial-materials wells have much higher radon levels (4,700 pCi/L as compared to 1,100 pCi/L in the Upper Wood), even though the alkalinity, calcium, silica, fluoride, and sodium median values are almost equal. This suggests textural advantages in the Queen-Usquepaugh aquifer material such as the favorable siting of uranium or differences in flow rate (volume of ground water) influencing radon levels.
On the other hand, the deep surficial-materials wells in the Upper Wood and Chipuxet are producing similar radon levels (1,100 pCi/L and 1,400 pCi/L, respectively) but have very different alkalinity, calcium, magnesium, and sodium values. This implies that residence time and water evolution are factors in water chemistry but not on radon levels of ground water. Instead, physical properties such as ground-water flow rate or uranium-siting are more likely the controlling factors.
In an attempt to understand the radon variability within the three-dimensional framework of the aquifer, radon versus depth within the aquifers and radon along the axis of the aquifers was analyzed.
When comparing radon with depth to bedrock in the surficial materials wells, the greatest radon levels (greater than 2,000 pCi/L) are found in wells underlain by the Scituate Igneous Suite, all of which are within 20 feet of the bedrock surface ( Figure 17). However, when radon values are compared to depth, there is no correlation between radon and depth from the well bottom to bedrock in the three aquifers. The Upper Wood River and Chipuxet surficial aquifers showed no variation in radon concentration over a wide range of proximity to bedrock surface values.
The trilinear diagrams suggest vertical mixing between the bedrock and surficial materials aquifers in the Queen-Usquepaugh and Upper Wood River aquifers (Figures 5 and 6). Several of the deep surficial wells just above the bedrock surface, plot in the bedrock ground-water field. Although it appears that there is mixing between surficial and bedrock aquifers in the trilinear diagrams, chemical data do not support this. Therefore, some of the higher radon values seen in surficial materials wells are a result of insitu conditions (uranium siting for example) in the lower portions of these aquifers.
There is no spatial variability of radon in surficial wells along the axes of the aquifers (Figures 12   and 13 ). However, spatial variation is seen in bedrock wells in the study area (Figures l 0 and 11 ). There is an increase in radon concentrations in bedrock ground waters from wells south of the Scituate/Sterling contact (in the Sterling Suite) to wells north of the contact (in the Scituate Suite).
The factors controlling radon concentration seem to be the uranium in the underlying bedrock and the physical properties of the aquifer. Uranium is found in accessory minerals in the crystalline bedrock of the study area. This uranium is either disseminated within the mineral grain, or has been mobilized to the edges of the mineral grains, resulting in concentrated uranium around the mineral grains. As radium (radon's parent) decays, alpha recoil is responsible for moving radon that is close enough to the surface of the grains into the adjacent ground water . Therefore, the siting of the uranium within the mineral is very important in radon mobility. A limitation of this study is that the 30 Well #22 was no included because well depth is unknown.
physical characteristics of the bedrock ( degree of fracturing and possibility of secondary uranium mineralization along the walls of the fracture) where these wells are located are unknown.
Based on the limited sample size used here and unpublished data from Veeger and , there is no relationship between radon levels in the surficial-materials aquifers and the eTh and eU concentrations in the regolith. The median uranium content of the three surficial materials aquifers was less than 2.5 ppm.
Because well yields were available for only one third of the wells, and most of them were qualitative estimates by the well drillers, well-yield data were insufficient to create a quantitative comparison between the amount of flow through an aquifer and radon levels. In previous studies of radon and hydraulic aperture, differences in radon concentration within well pairs were attributed to differences in hydraulic aperture alone (Peter Folger, 1994, Colorado School of Mines, written communication) and inverse correlations were seen between dissolved radon and well yield . A larger fracture aperture allows a greater volume of water to pass through, thus the 222 Rn produced in the rock is in essence diluted by the greater volume of water transmitted through the fracture. In this study, evidence of this relationship is also seen. Two of the bedrock wells sampled in this study, numbers 17 and 20, were located in the same bedrock material but yielded ground water with very different radon concentrations. Well 20 (well depth 85 feet) has a reported yield of 0.25 gallons per minute and radon concentration of 82,900 pCi/L, whereas well 17 (well depth 360 feet) has a reported yield of 20 gallons per minute and a radon concentration of 11,300 pCi/L. The radon in well 17 was probably diluted by a greater volume of water in the fracture resulting in a dramatically lower radon concentration.  found that the siting of uranium in the Scituate and Esmond rocks is within the accessory minerals and not along grain boundaries. She did not analyze a sample from the Sterling alaskite gneiss. Nevins' samples were from a fresh outcrop and not from a well boring. Water-bearing fractures are weathered as a result of water-rock interactions. This weathering process could redistribute uranium from within the lattices to the grain surfaces.  found that although uranium was held within the allanite grain in unaltered samples, weathered samples indicated that the uranium remobilized and adsorbed onto nearby biotite and microfractures.
This study has found that there is a statistically significant relationship between ground-water fluoride and radon concentrations. Previous studies have discovered that the presence of fluoride, phosphate, hydroxide, sulfate, and carbonate in natural waters increase the solubility of uranium minerals and thus mobility of uranium in ground waters . The most important uranyl complexes in the acidic waters of this study are formed with fluorides (because of water pH and species concentrations).
When redox conditions are conducive to uranium complexing in the solid phase, the fluoride species remains in the liquid phase as a marker. Uranyl complexing with fluoride is another reason why there is a statistically significant relationship between ground-water fluoride and radon concentrations ( Figure 15).
Further evidence of uranium remobilization is found in studies conducted on 226 Ra, the parent of radon. Radon concentrations are generally several orders of magnitude greater than the radium concentrations . Therefore, the radium must be concentrated near enough to the water-rock interface (adsorbed onto mineral surfaces) to produce these high dissolved radon concentrations.
Metamictization (the destruction of mineral structure as a result of radiation) may also influence the mobility of uranium. Studies have found that with increased metamictization, uranium-bearing minerals like zircon become amorphous and more easily leached because of the internal radiation damage to the structure of the mineral . In some cases, allanite and zircon are metamict minerals.  found evidence of metamictization in some allanite grains in her Rhode Island bedrock samples.
A further consideration involving uranium siting is the possibility of shear zones in the underlying bedrock surrounding the well. During rock shear, strain is concentrated in narrow fault zones, causing mylonites to develop because of the ductile shear (plastic .deformation). A mylonite is a rock that has undergone a change in microstructure, porosity, permeability, and chemical composition from the parent rock. This reorganizing of the parent rock creates a foliation into which the uranium is redistributed by hot oxidizing fluids . Several studies on shear zones throughout the Appalachian region have found the uranium-bearing minerals concentrated in zones of local melting .
Uranium is a much more effective producer of mobile radon when it is in the mylonite foliation because this foliation may weather preferentially, exposing the uraniferous surfaces. As a result, the radon in soil gas and ground water, and the uranium concentration in the bedrock may increase with increasing shear.
Although data from this study were not comprehensive enough to thoroughly investigate this phenomenon, anomalously high ground-water radon concentrations are seen in wells near the Hope Valley Shear Zone (39,700 pCi/L in a bedrock well) and in the Queen Usquepaugh aquifer near the Scituate/Esmond bedrock contact where some shearing might exist (10,700 pCi/L and 15,400 pCi/L in two surficial wells and 82,900 pCi/L in a bedrock well) (Figures 10, 11, 12, and 13

Summary and Conclusions
All wells sampled yielded radon levels above the proposed EPA limit of 300 pCi/L and many were more than an order of magnitude greater. Although the EPA will probably adopt a standard that is less rigorous than 300 pCi/L, this study reveals that much of the ground water in southern Rhode Island has elevated radon levels that may be cause for concern.
Water chemistry does not play an important role in controlling radon concentrations. However, some chemical parameters can provide clues about the uranium content of the bedrock and the radon content of the ground water. High radon values correlated with high fluoride in the ground water because the bedrock suites that contain the mineral fluorite (A-type granites) are likely to be uranium-rich and fluoride also increases uranium mobility.
Bedrock aquifers yield higher concentrations of radon than surficial materials (derived from local bedrock) aquifers. This relationship is both related to the lower ground-water flow rate in bedrock aquifers as compared to permeable surficial materials aquifers, and the availability of uranium-bearing minerals in both settings. High radon values were discovered in areas where the underlying bedrock contained the most uranium (Scituate Suite) because the source uranium is the solid phase and not the dissolved species (there was no correlation between radon concentration and dissolved uranium).
The greatest radon levels among surficial materials wells (greater than 2,000 pCi/L) are found in surficial wells that are underlain by bedrock with high uranium contents ( The variability of ground-water radon concentrations between bedrock and surficial materials wells cannot be isolated to one specific reason. The variable with a great deal of influence is ground-water flow rate. Higher ground-water radon levels are seen in bedrock wells because of the lower flow rate of ground water through the bedrock aquifer as opposed to the flow through the surficial material.
Furthermore, the more favorable siting of uranium and metamictization in the bedrock aquifer, are also factors in producing higher radon concentrations in bedrock wells as opposed to surficial wells within an aquifer. This study has found that there is a statistically significant relationship between ground-water fluoride and radon concentrations. The presence of fluoride increases the solubility of uranium minerals, the uranium remobilizes to a more favorable siting along the grain boundaries, and the fluoride remains in the ground water as a marker. Differences in radon concentrations between aquifers are caused by uranium-siting and ground-water flow rate as well, but the uranium content of the underlying bedrock and the possibility of influence from a shear zone in the bedrock play the important roles in radon concentrations of the ground water.
Factors controlling radon distribution are so complex that it is impossible to develop a quantitative model to predict radon levels in the ground water with the available data. Information needed to distinguish between these factors would be well yields, fracture size for bedrock wells, and samples of both the bedrock and surficial material to evaluate textural differences (for surficial wells) and the siting of uranium.
This investigation has shown that radon is a concern for ground-water users within the Pawcatuck River Basin, Rhode Island. Although it is not possible to make precise predictions of radon concentrations in ground water, this study shows that bedrock ground-water users in the Queen-Usquepaugh and Upper Wood aquifers, particularly those underlain by the Scituate Igneous Suite, can expect relatively high levels of radon in their ground water (10,000 pCi/L to 85,000 pCi/L), while bedrock ground water users in the Chipuxet can expect relatively low levels (less than 10,000 pCi/L). Those users who withdraw ground water from surficial-materials aquifers are at a lower risk for elevated radon levels in their ground water than bedrock ground water users. We are conducting a study of the water in the Pawcatuck Rjver Basin. This study ts supported by the Rhode Island Water Resources Center. We are investigating the relationship between ground-water quality and geology. As part of this study, a number of private wells in the area "ill be sampled. We are asking for help from homeowners around the Wood Rjver, to develop a data base for our study. If you have answers to any of the following questions, please return them in the self-addressed. stamped envelope we have provided. If your well is included in the study, you will receive a copy of the laboratory results with the complete chemical analysis  Standard method 4500-Si D, molybdosilicate colorimetric method ) using Milton Roy model 1201 spectrophotometer.

Mn
Atomic absorption air-acetylene flame method. Uranium Fluorometric analytical method, laser phosphorimetry with standard addition of sodium hexametaphosphate.

Appendix E. Ion Chromatography Accuracy.
The ground-water samples were analyzed for anions and cations using the Dionex series 4500i ion chromatograph. The ion chromatograph is calibrated using four levels of standards. Samples with values greater than the standards were diluted. The standards used are included in Tables El and E2. These standards were run as samples to detennine the accuracy of the ion chromatograph. The ion chromatograph results for the standards were then used to calculate the percent error for each run. Average percent errors for anion and cation analyses are given in Tables E3 and E4.

Appendix F. Statistical Analysis.
Analysis of radon data.
Statistical analysis of the radon data was done to test whether two populations were significantly different. First, the Anova (single factor) test was done to statistically analyze the data. The populations were assumed to be distributed normally. In order to make the distribution roughly symmetric (normal) without heavy outliers, anomalously high and low values were discarded on a few of the tests. Then, the Ftest was used to test whether the populations were significantly different.
F-test. The hypothesis that the populations are not significantly different is rejected if F > F cnt· Both the F and Fent values are found in the A-nova output.
The data are in log form and are first analyzed by aquifer.

Oueen-Usguepaugh
shallow surficial compared to deep surficial radon levels in wells:

Analysis of fluoride compared to radon data.
Speannan's rank correlation coefficient was used to test whether there is a significant correlation between fluoride and radon. Both the radon data and the fluoride data were ranked and the Spearman' s rank correlation coefficient was calculated. This correlation coefficient was then analyzed using the Student's T-test to detennine its significance. Since tis greater than tcrio the correlation coefficient of the ranks is significant, and therefore there is a significant correlation between radon and fluoride.

Analysis of alkalinity compared to radon data.
Speannan's rank correlation coefficient was used to test whether there is a significant correlation between alkalinity and radon. Both the radon data and the alkalinity data were ranked and the Speannan's rank correlation coefficient was calculated. This correlation coefficient was then analyzed using the Student's T-test to detennine its significance. Since t is greater than tcrio the correlation coefficient of the ranks is significant, and therefore there is a significant correlation between radon and alkalinity.

Anova analysis of fluoride data.
Statistical analysis of the fluoride data was done to test whether two populations were significantly different. First, the Anova (single factor) test was done to statistically analyze the data. The populations were assumed to be distributed normally. In order to make the distribution roughly symmetric (normal) without heavy outliers, anomalously high and low values were discarded on a few of the tests. Then, the Ftest was used to test whether the populations were significantly different.

~-
The hypothesis that the populations are not significantly different is rejected ifF > Fcrit· Both the F and Fent values are found in the A-nova output.
The data are first analyzed by aquifer.

Anova analysis of uranium data.
Statistical analysis of the dissolved uranium data was done to test whether two populations were significantly different. First, the Anova (single factor) test was done to statistically analyze the data. The populations were assumed to be distributed nonnally. In order to make the distribution roughly symmetric (normal) without heavy outliers, anomalously high and low values were discarded on a few of the tests.
Then, the F-test was used to test whether the populations were significantly different.
F-test. The hypothesis that the populations are not significantly different is rejected ifF > F crir· Both the F and F crir values are found in the A-nova output.
The data are first analyzed by aquifer.
Analysis of dissolved uranium data by aquifer.

Oueen-Usguepaugh
shallow surficial compared to deep surficial uranium levels in weIIs: