Abstract. Flow visualization experiments were conducted on a transparent replica of a natural, rough-walled rock fracture from the Stripa Mine, Sweden, for inlet conditions of constant pressure and flow rate over a range of angles of inclination. The experiments demonstrated that infiltrating water proceeds through unsaturated rock fractures along nonuniform, localized preferential flow paths. Even though constant inlet conditions were maintained, pervasive unsteady or intermittent flow was also observed in these experiments, where portions of the flow channel underwent cycles of snapping and reforming. Experiments conducted on parallel plates with a sequence of apertures progressing from small to large to small reproduced intermittent flow. Measurements of the frequency of intermittent flow events and the volume of water metered between events were obtained from the fracture replica and parallel plate experiments and related to the Bond and capillary numbers to generalize the results. The frequency data from the fracture replica experiments did not follow the same trend as the data from the parallel plate experiments for similar Bond and capillary numbers, but the volume of water metered data was consistent in these experiments. IntroductionFractures in the unsaturated zone play an important role in water infiltration and contaminant transport. Field studies have provided considerable evidence that water proceeds along fast flow paths through partially saturated fractures. Bomb-pulse Chlorine 36 at Yucca Mountain, Nevada, was found at elevated concentrations several hundred meters deep, indicating that water had migrated from the land surface to these depths in a time period of only several decades [Liu et al., 1995]. Seeps collected in tunnels at Rainier Mesa, Nevada, were chemically distinct from matrix water, and their isotopic ratios indicated recent meteoric origin [Wang et al., 1993]. Recharge water was observed to flow quickly through a thick unsaturated fractured chalk zone in the Negev Desert in Israel [Nativ et al., 1995]. These observations were explained by water migrating rapidly along localized preferential flow paths through unsaturated rock fractures. Preferential flow channels reduce rock matrix-fracture interaction compared to models that predict spatially uniform flow, thereby accelerating groundwater travel in unsaturated fractured porous media [Glass et al., 1995]. Understanding the mechanisms controlling fast flow is important for developing conceptual models describing seepage of liquids through unsaturated, fractured porous media.Liquids in unsaturated porous media migrate under the combined action of gravity, pressure, capillary, and viscous forces. The relative magnitude of these forces can be quantified using the Bond and capillary numbers. The ratio between where A 9 is the density difference between the infiltrating liquid and the air, !7 is the gravitational acceleration constant, /3 is the angle of inclination of the fracture measured from the horizontal, b is the aperture, cr is ...

Heat and bromide were compared as tracers for examining stream/ground water exchanges along the middle reaches of the Santa Clara River, California, during a 10-hour surface water sodium bromide injection test. Three cross sections that comprise six shallow (<1 m) piezometers were installed at the upper, middle, and lower sections of a 17 km long study reach, to monitor temperatures and bromide concentrations in the shallow ground water beneath the stream. A heat and ground water transport simulation model and a closely related solute and ground water transport simulation model were matched up for comparison of simulated and observed temperatures and bromide concentrations in the streambed. Vertical, one-dimensional simulations of sediment temperature were fitted to observed temperature results, to yield apparent streambed hydraulic conductivities in each cross section. The temperature-based hydraulic conductivities were assigned to a solute and ground water transport model to predict sediment bromide concentrations, during the sodium bromide injection test. Vertical, one-dimensional simulations of bromide concentrations in the sediments yielded a good match to the observed bromide concentrations, without adjustment of any model parameters except solute dispersivities. This indicates that, for the spatial and temporal scales examined on the Santa Clara River, the use of heat and bromide as tracers provide comparable information with respect to apparent hydraulic conductivities and fluxes for sediments near streams. In other settings, caution should be used due to differences in the nature of conservative (bromide) versus nonconservative (heat) tracers, particularly when preferential flowpaths are present.

Commonly measured water quality parameters were compared to heat as tracers of stream water exchange with ground water. Temperature, specific conductance, and chloride were sampled at various frequencies in the stream and adjacent wells over a 2-year period. Strong seasonal variations in stream water were observed for temperature and specific conductance. In observation wells where the temperature response correlated to stream water, chloride and specific conductance values were similar to stream water values as well, indicating significant stream water exchange with ground water. At sites where ground water temperature fluctuations were negligible, chloride and/or specific conductance values did not correlate to stream water values, indicating that ground water was not significantly influenced by exchange with stream water. Best-fit simulation modeling was performed at two sites to derive temperature-based estimates of hydraulic conductivities of the alluvial sediments between the stream and wells. These estimates were used in solute transport simulations for a comparison of measured and simulated values for chloride and specific conductance. Simulation results showed that hydraulic conductivities vary seasonally and annually. This variability was a result of seasonal changes in temperature-dependent hydraulic conductivity and scouring or clogging of the streambed. Specific conductance fits were good, while chloride data were difficult to fit due to the infrequent (quarterly) stream water chloride measurements during the study period. Combined analyses of temperature, chloride, and specific conductance led to improved quantification of the spatial and temporal variability of stream water exchange with shallow ground water in an alluvial system.