Ground Water – Surface Water Exchange – Methodology

Ground Water – Surface Water Exchange


Ground Water – Surface Water Exchange
A View from Inside the Unconfined Karstified Floridan Aquifer with Implications for Ground Water Protection


Radon-222 (222Rn) provided an ideal natural tracer for this investigation because concentrations in ground water are typically an order of magnitude or more greater than in surface water. 222Rn is the radioactive decay product of Radium-226 which is ubiquitous in the phosphatic sediments composing all or part of the Alachua Formation, present in outlying hills, and the Hawthorn Formation, present in the Northern Highlands (Kaufmann and Bliss, 1978). 222Rn is a volatile gas with a half-life of 3.8 days and is highly soluble in water. 222Rn is released to ground water from Radium-226 by:

  • (1) alpha recoil in micro-pore walls (Rama and Moore, 1984) and
  • (2) dissolution of aquifer material that supplies soluble Radium-226 which subsequently decays to 222Rn (Ellins et al., 1990).

Because of its high volatility, 222Rn gas quickly dissipates when exposed to the atmosphere creating a significant disequilibria between concentrations in ground water and surface water (Rogers, 1958). 222Rn concentrations are measured in alpha-scintillation counters and reported in Becquerels per liter (Bq/L) which equate to the number of alpha particle disintegrations per second per liter.

Rogers (1958) demonstrated that elevated 222Rn concentrations measured in streams and rivers are indicative of ground water inputs. Furthermore, Ellins and others (1990, 1991) showed that by accounting for gas exchange across the air-water boundary, elevated measures of 222Rn can be used to quantify ground water influx to a stream or river. The same principles were applied in this investigation but in reverse. When sampling the Floridan aquifer from inside the Devil’s Ear cave system, low 222Rn concentrations indicated river water intrusion. The difference between the 222Rn concentration in a sample and that of a pure aquifer standard was used as the basis for quantifying river water intrusion.

Mixing Model
The quantity of intruded river water in the samples taken from the Devil’s Ear cave system was determined using the following equation:


  • Rs = the 222Rn concentration in the sample,
  • Rriv = the background 222Rn concentration in the river,
  • Riv = the decimal fraction of river water in the sample, and
  • Raq = the background 222Rn concentration in the aquifer.

Solving the equation for the percentage of river water in a given sample produces


Background 222Rn concentrations in the aquifer were measured by sampling five wells near the field area but more than 1 km from the river. Raq was determined to be 13.0 Bq/L in February 1992 and 14.2 Bq/L in June 1993 by averaging the values obtained from the five wells. Measuring Rriv in the field area revealed values ranging between 4.2 and 9.0 Bq/L. These values were not considered accurate estimations of the background 222Rn concentration in the river because of large ground water inputs from several springs. Instead, Rriv was determined by averaging several measurements collected by Ellins and others (1991) and Hisert (1994) that were taken just upstream of the field area where there is less ground water input. Averaging these reported 222Rn concentrations produced a value for Rriv of 1.0 Bq/L. The model only assumes fixed end-member concentrations for the mixing waters. Note that a greater value for Rriv would result in an increased value for %Riv.

Delta Oxygen-18 (d18O) Confirmation
Variations in d18O provided a qualitative check on the 222Rn mixing results. Oxygen isotope ratios are expressed in parts per mil (o/oo). Variations in d18O in natural waters result from isotopic fractionation driven by evaporation and condensation. Preferential evaporation of 16O causes a relative 18O enrichment in surface waters producing more positive values of d18O. Ground water does not evaporate so ground water samples are characterized by more negative d18O (Ellins, 1992). Thus, fractionation permits the discrimination between ground water and surface water based on the d18O signatures. Water samples, collected in the 1993 sampling period, were analyzed for d18O to qualitatively check the results obtained from the 222Rn mixing model. Greater d18O values were expected to correspond to sampling locations with small 222Rn concentrations and thus confirm regions of river water intrusion to the cave system. d18O was not directly measured in the river, however, lakes and ponds near the study area yielded a consistent d18O of -1.5 o/oo, whereas ground water values approached -4.0 o/oo (Hisert, 1994).

Sampling the Devil’s Ear Cave System
The Devil’s Ear cave system was sampled twice, in February 1992 and in June 1993. A team of two cave-divers made 21 dives into the system and collected 50 water samples.

222Rn sampling was conducted following the methodology described in Key (1981) and Ellins and others (1990). Water samples for 222Rn measurements were collected in evacuated 250 ml plastic bottles. The bottles were filled with approximately 150 ml of aquifer water leaving 100 ml of head space to collect the 222Rn gas that would be volatilized from the water sample. After surfacing, the samples were transported to a laboratory where the gasses in the head space were extracted from the sample bottles into Lucas-type counting cells. After sufficient time to allow the 222Rn to equilibrate with it’s daughter products, the cells were placed into alpha-scintillation counters where light photons emitted by alpha particle disintegrations were counted and recorded as Becquerels per second per liter of sample. Key (1981) reports that the error associated with this method does not exceed 14%.

d18O samples were collected in 50 ml glass vials, precleaned with nitric acid and filled with distilled water prior to the dive. The vials were flushed with air from the scuba cylinders at the underwater sampling locations then rinsed and refilled with aquifer water.

A more detailed accounting of the methodology used to collect samples for 222Rn and d18O as well as a description of the error involved with the data collection is provided in Kincaid (1994).