Vol. 48, 01, 05ok.qxd 101 ANNALS OF GEOPHYSICS, VOL. 48, N. 1, February 2005 Key words radon – hydrology – tracer – hydro- graph separation 1. Introduction Naturally occurring isotopic and chemical tracers are among the most useful tools in the study of river flow generation. Such tracers are often applied in the context of simple mixing models that attempt to separate the relative con- tributions to river flow from different geological reservoirs. A well-known example is the use of 18O or 2H to separate storm river flow into ‘new’ water (storm precipitation) and ‘old’ water (pre- storm subsurface water) (e.g., Sklash et al., 1976; Hooper and Shoemaker, 1986). The understanding of hydrological flow and river flow generation has gained an increasing importance in a wide range of studies for water resource, contaminant transport and geochemi- cal and even biochemical issues. In the last 40 years, many publications have tried to elucidate the mechanisms by which water is collected in catchments and how it moves down hillslopes into small rivers (Hoehn et al., 1989). Among the many methodologies, those involving natu- rally occurring tracers have become important. Tracer studies generally lead to the separa- tion of river flow into two or more components and eventually, through the radioactive tracers like anthropogenic tritium, natural radon and ra- dium isotopes, give an estimation of the ground- water residence times. Numerous studies demon- Mailing address: Dr. Antoine Kies, Laboratoire Physi- que des Radiations, Centre Universitaire de Luxembourg, 162a Avenue de la Faïencerie, L-1511 Luxembour; e-mail: kies@cu.lu Using 222Rn for hydrograph separation in a micro basin (Luxembourg) Antoine Kies (1), Harald Hofmann (1), Zornitza Tosheva (1), Lucien Hoffmann (2) and Laurent Pfister (2) (1) Laboratoire Physique des Radiations, Centre Universitaire de Luxembourg, Luxembourg (2) Centre de Recherche Public – Gabriel Lippmann (CREBS), Luxembourg Abstract In order to obtain information on the hydrological signature of rivers during and after heavy rain events, small catchment areas are selected as experimental sites. Hydrograph separations based on environmental tracers are performed. Natural isotopic tracers such as 18O, 2H and particularly 222Rn may help to distinguish the components dominating the outflow, particularly of ‘pre-event waters’, ‘event waters’ and ‘post-event waters’. Even with moderate concentrations in groundwater, radon can be a very sensitive indicator of groundwater input into rivers. The selected microbasin under investigation is situated in the western part of Luxembourg and belongs to the At- tert River catchment. At chosen points at the basin’s outflow radon detectors continuously measure radon activ- ity in water. The radon monitors are installed together with high precision thermometers, conductivity meters, flow meters and automatic water samplers for chemical analysis. Besides the continuous measurements, grab water samples are taken at different locations along the stream, most of them during periods of heavy rain events. Presented are the results of a one year measurement campaign. During the dry season i.e. during more or less continuous discharge conditions, the observed mean values do not show substantial variations and can be used as reference values. Fluctuations of the measured data during rain events are discussed and the interplay between the different parameters analysed. 102 Antoine Kies, Harald Hofmann, Zornitza Tosheva, Lucien Hoffmann and Laurent Pfister strate the utility of the stable hydrogen isotope deuterium 2H and the stable oxygen isotope 18O. Studies involve other tracers like calcium (Gene- reux et al., 1993), silicium and various other ma- jor ions, depending on the ionic strength of the wa- ters and their possibility to acquire substantial amounts of those elements from the soil and rocks. Many of common tracer studies focus on the effect of storm flow, the most common type of analyses being a separation of hydrographs into ‘old’ water present in the watershed before the start of the rain event and the ‘new’ water mixing in during the event of interest. For the study of river flow generation it is un- realistic to expect that whatever tracer methods are applied, all the answers to the important ques- tions concerning hydrological flow paths on wa- tersheds and information on the water movement in a catchment during a rainfall-runoff event are given. It is in applying multiple field techniques that an optimal approach can be achieved. In the CYCLEAU project the aim is to use a multifunctional approach. In the present paper only the involvement of the tracer element radon together with temperature and electrical conductivity are presented and discussed. Differences in radon content of subsurface waters arise from differences in radon emanation by porous media (bedrock or soil) and from the differences in the degree of subsurface degassing. The factors influencing the behavior of 222Rn in the subsurface have been the subject of numerous investigations (e.g., Clements and Wilkening, 1974; Schery et al., 1984 for the elder references) and are not part of the present paper. Here we ex- pect only markedly different radon concentra- tions in the natural waters, especially correspon- ding to zones of unsaturated and saturated soils. Simple three-end-member mixing models may provide a useful framework for riverflow generation, the three different waters being su- perficial water (vadose zone water), soil ground water and bedrock water (Genereux et al., 1993). Merot et al. (1995) suppose a four-component hydrograph separation in studying the complex variations of groundwater, riparian zone seep- age, hillslope subsurface flow and event water. 222Rn can be useful in distinguishing be- tween saturated zone water and unsaturated su- perficial water; the former usually has a much higher radon content than the latter. Water in unsaturated zones may lose much of its radon to the atmosphere by degassing to soil air, whereas water in saturated zones generally re- tains most of its radon. Superficial water that enters a saturated zone begins to accumulate 222Rn, acquires the 222Rn signature of the rock and soil underground at a rate controlled by the retention time and the radioactive decay. Superficial water having been in a saturated zone for several days can be considered groundwater. Radon can reach relatively high concentrations in groundwater but, because of its low solubility, it degasses rapidly in surface waters. Therefore it can be a very sensitive in- dicator of groundwater input to streams. The 222Rn content of river water is strongly affected by volatilization to the atmosphere, and this must be accounted for in using radon data to estimate a possible groundwater influx from subsurface water sources, important in river flow generation. In the following we con- sider a given length ∆x of a river and two sam- pling sites at the upstream and downstream ends. A simple one-dimensional model equa- tion can be proposed as Q Q q x kQRn Rn Rn Rn q avg avg2 2 1 1 = + - x∆5 5 5 5? ? ? ? where Q (m3/s) is the river discharge, the sub- scripts 1 and 2 designate the upstream and downstream ends, q (m3s–1m–1) the inflow rate per unit of river length, τ the travel time, k a radon-degassing rate constant; Qavg and [Rn]avg are the mean river discharge and the mean radon volumetric concentration in the stream. We assume that the length is sufficiently small allowing the same radon concentration for lat- eral inflowing water. As the travel time is very short, a loss of radon due to decay (T = 3.8 d) is negligible. Due to the turbulent runoff of the small rivers, the primary mechanism of radon re- moval is gas exchange. 2. Material and methods To obtain information on the hydrological signature of small rivers during and after heavy 103 Using 222Rn for hydrograph separation in a micro basin (Luxembourg) rain events, four small catchment areas of the same river were selected as experimental sites. Ephemeral or first-order streams drain those wa- tersheds of 10-100 ha. The selected sub-catch- ments under investigation are situated in the western part of Luxembourg and belong to the Attert River Basin, the latter being integrated in the European Network of Experimental Research Basins (ERB). The catchment area of the river basin is 318 km2. The altitudes vary from 540 m to 210 m, with slopes presenting a maximum of 16°. The geological underground is formed of De- vonian schist in the highest northern part, Trias- sic marls und mudstones for the middle part and Liassic sandstones for the southern part. The present paper concerns the micro basin H (fig. 1), located in the southern part of the Attert Basin. It is covered to 93% by forest and 7% by pasture; the underlying rocks and soils are over 81% sandstone, 13% marls, 3% marls/mudstones and 3% alluvium deposits. The topography strat- ifies the watershed into two subunits, hilltops and valleys, with altitudes between 380 and 320 m. 2.1. Radon in water measurements Liquid Scintillation Counting (LSC) is used for the determination of α-emitters in environ- mental samples. Presently, this technique is used for radon and radium in water measurements. For radon measurements, normally 14 ml of wa- ter are mixed to 7 ml of scintillation cocktail Be- taplate, out of which 5 ml are counted in the Triathler portable liquid scintillation device. This device provides efficient alpha-beta discrimina- tion. For the very low concentrations of river wa- ter, the sample volume is increased to 250 ml, thus a limit of detection down to 50 mBq/l can be obtained. Samples are prepared by carefully in- troducing the water under the cocktail prepared in a glass vial. Measurements are performed af- ter a waiting time of at least 3 h, permitting equi- librium between radon and the decay products. Concerning continuous radon-in-water measurements, for radioprotection purposes, a detection limit of 10 Bq/l is needed, but for hy- drogeological studies detection limits lower than 1 Bq/l are required. When deciding which Fig. 1. Map of the catchment areas of the Attert River, shown are the four micro basins under study. The pres- ent work concerns micro basin H. 104 Antoine Kies, Harald Hofmann, Zornitza Tosheva, Lucien Hoffmann and Laurent Pfister method to use, an important point is how much measuring time is necessary to obtain an accept- able precision and give a signal that significant- ly differs from the background. Radon concentrations in a spring can vary considerably with time. The sampling frequency has to be adapted to the dynamics of the corre- sponding aquifer. However, one rarely knows how fast an aquifer will react to changing envi- ronmental conditions, e.g., to heavy precipita- tions. Extreme cases are karst springs where dis- charges are reported to increase by two orders of magnitude within hours after a storm (Eisenlohr and Surbeck, 1995). Even springs emerging from an aquifer containing old water may react quickly to precipitation because of a rapid change in hydraulic pressures. In order not to miss important features, our continuous radon- in-water monitors have a temporal resolution of half an hour. There are several possibilities to monitor radon continuously in water. We decid- ed to use the principle of radon gas measure- ments in a closed circuit coupled to the water. The coupling consists either in bubbling air through the water, or replacing the bubbling fa- cility with a diffusion tube (fig. 2). In combining the degassing unit or the diffusion tube with a radon detector based on 222Rn and 218Po, it is pos- sible to have a resolution below 1 Bqm–3. This is necessary if one wants to monitor continuously radon concentrations in river water far away from the sources where nearly all of the radon has degassed. If one puts the diffusion tube into the water it is essential to have a high water flow as a depleted or enriched zone exists around the diffusion tube, not having the same radon con- centrations as the water flowing in. In order to avoid this drawback we let water circulate in the tubing, measuring the radon concentration in the vessel, either by putting a small radon monitor (DoseMan, Sarad company, Dresden) into the exchange vessel, or pumping the air from the ex- change vessel to the measuring chamber of the radon monitor. As the semiconductor devices normally have to work under dry conditions, air passes a drierite column prior to counting. An- other drawback with the diffusion tube is the change in the Oswald coefficient with tempera- ture. Water temperature has to be measured and a correction performed. Preferably we use the diffusion-tube method to monitor the water at springs where temperature variations are small. Temperature and conductivity measure- ments are performed each time a grab sample is taken for radon measurements. Continuous radon measurements are always coupled with continuously working thermometers and con- ductivity meters. A locally installed weather station informs on atmospheric parameters, namely rainfall. Fig. 2. Principle of the use of an Accurel diffusion tube for radon-in-water measurements (Surbeck, 1996). Fig. 3. Variations with time of the grab-samples taken at different sources (with concentrations higher than 10 Bq/l) and at 2 points along the river (H1 and H6). Table I. Summary of results of radon measure- ments in 8 sources of the catchment area under study. Location No. of meas mean min max (Bq/l) (Bq/l) (Bq/l) H4 14 13 9.6 15 H5 9 23 22 25 H6 14 4.8 2.8 8.4 H8 14 10 6.8 12 H10 12 16 9.5 19 H11 14 12 11.0 14 H17 7 13 10.6 14 H18 6 14 12.0 17 water. Nevertheless, if the water output is on a hillside, often prior to the physical output, radon degassing may occur due to underground turbulent flow and small water cascades. Figure 3 shows the temporal evolution of radon concentrations measured at 8 sources. Dur- ing this period no major rain event was observed and the radon concentrations do not experience wide variations. Nevertheless a rainy period at the end of August and at the end of November in- duced a slight increase in radon concentrations. Figure 4 shows radon concentrations meas- ured during a December rainy period at the out- let of the microbasin. Among the data, most in- teresting are those documenting continuous measurements of radon, electric conductivity and temperature; barometric pressure did not in- fluence the measured data. The data of fig. 4 were collected at the outflow point H1 of the mi- crobasin under study. For the period under inves- tigation, every major rain event induced a rapid decrease of the electric conductivity and an in- crease in radon concentrations. Often a time lag between conductivity and radon is observed. The drop in conductivity is due to the input of superficial water to river water, whereas increased radon concentrations are due to groundwater. Infiltrating water to the aquifer has a piston effect on residential groundwater in the frac- ture systems close to the main underground path- ways and canals. Radon-rich water is pressed out of the fracture system and enters the river, some- times with a time delay to the conductivity, the lat- 105 Using 222Rn for hydrograph separation in a micro basin (Luxembourg) 3. Results and discussions To quantify the influx of groundwater to sur- face discharge, it is necessary to define a typical 222Rn value for the groundwaters of the basin. This value was established by measuring the 222Rn concentration of a number of springs in the investigated microbasin (table I). Radon concen- trations in the springs entering the river range be- tween 10 and 25 Bq/l, which is about a factor 10 higher than river radon concentrations. Spring water is generally collected less than 1 m from the point where it first emerged from the ground. Degassing over a short distance can be admitted as insignificant, and no volatiliza- tion correction has to be applied to 222Rn spring 106 Antoine Kies, Harald Hofmann, Zornitza Tosheva, Lucien Hoffmann and Laurent Pfister ter tracing the direct superficial flow. Radon peaks, consecutive to a rain event, last much longer as do conductivity lows thereby documenting a differ- ence in their contribution of hillslope subsurface and groundwater or superficial runoff water. There exists an apparent contradiction be- tween the hydrometric evidence of rapid flow along surface and near-surface pathways and the radon evidence of a high portion of old water con- tributing to storm flow. Storm flow generation re- sults in a rapid increase in groundwater level in- creasing the hydrostatic pressure and thus the un- derground flow in near channel areas (Mulhol- land, 1990). One can expect higher radon concen- trations at the output as water normally retained in underground fracture zones close to the main flow channels is pushed into the flow channels. 4. Conclusions Radon occurs naturally in all groundwater with varying concentrations depending on lithol- ogy and geological structure. Here we describe a methodology that uses 222Rn to provide informa- tion on river inflows by admitting that water from different pools contributing to riverflow differs in 222Rn concentration. Superficial water has a markedly different radon content from ground- water, the latter differing in radon concentrations if originating from saturated soils or from frac- tures in the bedrock. Levels of 222Rn found in rivers are at least one order of magnitude lower than the associated groundwater concentrations, thus radon is a sensitive way of detecting ground- water inflow. 222Rn data can be used in a simple mass-balance equation in conjunction with river discharge data to quantify groundwater inputs to surface flow. In addition, in order to allow the lo- cation and quantification of groundwater and sur- face water deliveries, independent estimates of groundwater discharge and recharge or aquifer storage capabilities are possible. The response of a forested watershed to ad- vective rain events is the main purpose of the present study. The results, even if preliminary, are very promising and partly illustrate the giv- en objectives. Acknowledgements The research was undertaken in the frame- work of the «CYCLEAU» national project based on the study of the interactions between differ- ent parameters in the water cycle, at several scales. CYCLEAU (FNR/01/03/02) is funded by the Luxembourg National Research Fund. Fig. 4. River water radon and conductivity versus time measured at H1, the outflow of the microbasin area. Very high water levels, consecutive to a very strong rain event on 2nd January 2003, upset the radon measuring device. Radon concentrations measured after are those of outside air. 107 Using 222Rn for hydrograph separation in a micro basin (Luxembourg) REFERENCES CLEMENTS, W.E. and M.H. WILKENING (1974): Atmospher- ic pressure effects on 222Rn transport across the earth- air interface, J. Geophys. Res., 79 (33), 5025-5029. EISENLOHR, L. and H. SURBECK (1995): Radon as a natural tracer to study transport processes in a karst system. An example in the Swiss Jura, C.R. Acad. Sci. Paris 321, Ser. IIA, 761-767. FREYER, K., H.-CH. TREUTLER, J. DEHNERT and W. NESTLER (1997): Sampling and measurement of radon-222 in water, J. Environ. Radioactivity, 37, 327-337. GENEREUX, D.P. and H.F. HEMOND (1990): Naturally occur- ring radon 222 as a tracer for streamflow generation: steady state methodology and field example, Water Re- sour. Res., 26 (12), 3065-3075. GENEREUX, D.P., H.F. HEMOND and J.P. MULHOPLLAND (1990): Use of radon-222 and calcium as tracers in a three-end-member mixing model for streamflow gen- eration on the West Fork of Walter Branch watershed, J. Hydrol., 142, 167-211. HOEHN, E. and H.R. VON GRUNTEN (1989): Radon in ground- water: a tool to assess infiltration from surface ground- water to aquifers, Water Resour. Res., 25, 1795-1803. HOOPER, R.P. and C.A. SHOEMAKER (1986): A comparison of chemical and isotopic streamflow separation, Water Resour. Res., 22 (10), 1444-1454. MEROT, P., P. DURAND and C. MORISSON (1995): Four-com- ponent hydrograph separation using isotopic and chemical determinations in an agricultural catchment in Western France, Phys. Chem. Earth, 20 (3-4), 415- 425. MULHOLLAND, P.J., G.V. WILSON and P.J. JARDINE (1990): Hydrogeochemical response of a watershed to storms: effects of preferential flow along shallow and deep pathways, Water Resour. Res., 26 (12), 3021-3036. SCHERY, S.D., D.H. GAEDDERT and M.H. WILKENING (1984): Factors affecting exhalation of radon from a gravelly sand loam, J. Geophys. Res., 89, 7299-7309. SKLASH, M.G., R.N. FARWORDEN and P. FRITZ (1976): A conceptual model of watershed response to rainfall, de- veloped through the use of oxygen-18 as a natural trac- er, Can. J. Earth Sci., 13, 271-283. SURBECK, H. (1996): A radon-in-water monitor based on fast gas transfer membranes, in Proceedings of the in- ternational conference on Technologically Enhanced Natural Radioactivity (TENR) caused by non-uranium mining, Szczyrk, Poland, 177-192.