H12B-01 INVITED 10:20h
Contaminant Transport Hydrology Of Surface/Ground Water Interactions
The crucial role of surface/ground water interactions in determining contaminant concentration distributions both in streams and ground water has only recently been recognized. In particular, in pool and riffle or meandering types of streams, long tails and small peaks are observed in the stream tracer data, which reflects the influence of surface/subsurface flow on solute concentration. Bencala and Walters (1983) first presented a comprehensive transient storage model for streams to explain the effects of surface storage and the hyporheic zone in stream concentration. We complement these surface/subsurface modeling efforts by incorporating the transient storage zone in a conjunctive stream-aquifer model. The core of this conjunctive model is MODFLOW which computes the ground water flow in the aquifer, while DAFLOW computes unsteady stream flow with diffusion wave routing technique and MOC3D computes the solute transport in ground water. In addition, an explicit finite difference package for solute transport in streams is developed, which solves one-dimensional transient storage equations. Results show that the conjunctive stream-aquifer model with transient storage can handle well the bank storage effect under a flooding event. The application of the model over a stream network shows that the stream-aquifer interaction acts as a strong source or sink along the stream. The existing transient storage model lumps the surface storage and hyporheic zone together in a single storage zone. Therefore, in order to further improve it, we develop a model that represents the movement of water through the hyporheic zone in order to explain the physics of water exchange between the surface water and the porous media in a mechanistic manner. For this purpose, we include the advection and dispersion processes into the transient storage zone, and we consider the hyporheic zone as a transient porous media from surface water to ground water. A new parameter, storage thickness, is introduced to represent the interface between the hyporheic zone and the channel, through which the water exchange takes place. This improved model not only includes the surface/subsurface water interaction directly through transient storage, but also represents the movement of stream water through the porous media with the advection and dispersion terms in the storage zone equation. We apply this new model to a hypothetical problem, and we also simulate the Uvas Creek experiment, comparing our results to observations reported by Bencala and Walters (1983).
H12B-02 10:35h
Diffuse Nitrogen Inputs Into Streams: Direct Experimental-Model Comparison and Up-Scaling of Loss Rates in Streams
Diffuse nitrogen input through the hyporheic zone along streams results from nitrogen leakage from sources distributed over the soil surface of catchments. Some nitrogen is then retained and lost from stream water due to biogeochemical and physical mass transfer and attenuation processes. Such losses are often modeled as first-order decay reactions, in both catchment-scale nitrogen transport models and interpretations of stream-reach-scale experiments. We find order of magnitude differences between reported loss rates in low order streams that have been calibrated by catchment-scale nitrogen budget modelling from those given by direct stream-reach-scale measurements. With a relatively simple model methodology, accounting for the spatially distributed character of solute inputs and transport in streams, we show that such discrepancies in loss rates can be explained and accounted for in process-based intrerpretation of catchment-scale nitrogen budget models.
H12B-03 10:50h
Incorporating coupled groundwater and surface water processes within a land surface model.
Traditional Land-Surface Models (LSM) used for numerical weather prediction, climate projection, and as inputs to water management decision support systems, often do not treat the subsurface in a fully process-based fashion. Conversely, models for fully and variably saturated groundwater flow, while addressing important features such as subsurface heterogeneity and three-dimensional flow, often have overly simplified upper boundary conditions that ignore soil heating, runoff, snow and root-zone uptake. In the present study, a distributed LSM (CLM2.0), a three-dimensional, variably saturated groundwater model (ParFlow) and a two-dimensional surface-water, overland flow model have been dynamically coupled as a single model. A set of simulations will be presented that are based on data from the Project for Intercomparison of Land-surface Parameterization Schemes (PILPS) Usadievskiy catchment in Valdai, Russia. These simulations using this distributed, coupled model investigate uncertainty, scale and the representation of important land surface processes such as runoff, infiltration and lateral subsurface flow. The coupled model will ultimately be used to assist in the management of an urban watershed in the United States. This work was conducted under the auspices of the U. S. Department of Energy by the University of California, Lawrence Livermore National Laboratory (LLNL) under contract W-7405-Eng-48 and Lawrence Berkeley National Laboratory (LBNL) under contract DE-AC03-76F00098.
H12B-04 11:05h
Simulating Stream-aquifer Interactions
Accurate simulation of the interaction between ground-water systems, which are typically fully three dimensional, and streams, which are typically linear features that are very narrow relative to the area of the ground-water system, is a challenge even for robust numerical models. This work examines how stream-aquifer interactions are represented in ground-water flow models using both coarse and highly refined grids and single and multi-layer models. Typically, the aquifer and the stream are assumed to be connected through a low-permeability streambed layer. Often, this is formulated as a streambed conductance term that connects the stream and the aquifer, where it is assumed that the dominant vertical head loss occurs in the streambed. This conceptualization is questionable in field situations where the fluxes between the two systems are large; however, it is these situations that are of most interest. For such situations, the assumption that most of the vertical head loss occurs through the streambed is not valid and we show that it leads to streambed conductances that have the unfortunate characteristic of depending on grid size. Alternative conceptualizations need to be considered that account for vertical resistance to flow in the streambed and the aquifer. Here they are explored and compared using numerical simulations. Grid size recommendations based on Dupuit assumptions for representing streams in groundwater models are evaluated. As the grids are refined both horizontally and vertically, the magnitude of stream-aquifer interactions changes. Vertical refinement changes the representation of vertical gradients, and the flux between the stream and the aquifer changes. In this work, simulations using single and multi-layer models produced differences in stream-aquifer flux of 80 percent. This suggests that simulated stream-aquifer fluxes have a significant grid-size dependency. This is problematic for using regional groundwater models, with typically coarse horizontal and vertical grid sizes, to model local-scale stream-aquifer fluxes. Here we use highly refined grids to investigate if there is a grid resolution that, when achieved, the grid-size dependency of stream-aquifer fluxes is eliminated. Local grid refinement is evaluated as a tool for achieving highly refined grids near streams while maintaining coarser discretization elsewhere.
H12B-05 11:20h
Comparison of Methods to Estimate Ephemeral Channel Recharge, Walnut Gulch, San Pedro River Basin, Arizona
Ephemeral channel transmission loss represents an important groundwater-surface water exchange in arid and semiarid regions and is potentially a significant source of recharge at the basin scale. However, identification of the processes and dynamics that control this exchange is a challenging problem. Specifically, data on the proportion of runoff transmission losses that escape from near-channel transpiration and wetted channel evaporation to become deep groundwater recharge are difficult to obtain. This issue was addressed through coordinated field research and modeling within the USDA-ARS Walnut Gulch Experimental Watershed (WGEW) located in the San Pedro River Basin of southeastern Arizona. Recharge was estimated using several independent methods which included a reach water balance approach, with near-channel ET estimated using sap flux and micrometeorological measurements; geochemical methods such as chloride mass balance; modeling of changes in groundwater level or microgravity measurements; and vadose zone water and temperature transport modeling. It was found that during the relatively wet 1999 and average 2000 monsoon seasons, the range of ephemeral channel recharge estimated from these methods differed by a factor of less than three. A rough scaling to the entire San Pedro Basin indicates that ephemeral channel recharge constitutes between approximately 15 percent and 40 percent of total annual recharge to the regional aquifer as estimated from a calibrated groundwater model. In contrast, during the weak monsoon seasons of 2001 and 2002 limited runoff and stream channel infiltration did occur but no discernable deep aquifer recharge was detected.
H12B-06 11:35h
Importance of the geosphere-biosphere interface on circulation of groundwater
This paper describes a hydrological and geochemical modelling approach for assessment of the migration pathways of reactive solutes in a watershed. The study shows that surface topography, stream network characteristics and thickness of quaternary deposits controls the circulation pattern of the deep groundwater. This issue has been found relevant for the safety evaluation of the final repository in bedrock of high-level nuclear waste. Reactive solute transport in a network of stream channels and sub-surface flow pathways is described by convoluting unit solutions based on a physical representation of transport and topographical information of the distributions of solute load as well as pathways. Topographical data and other geographical information are obtained for the mid part of Sweden. A 3D groundwater model is used to derive the residence time distribution of groundwater and the distribution of discharge areas in the watershed. The study shows that a relatively small fraction of the total water budget reaches as deep as 500 m were the Swedish nuclear waste repository is supposed to be placed. The water exchange is controlled by topography on both the continental scale as well as regional scale. The discharge areas are distributed in a relatively small part of the watershed in stream and lake bottoms and to some extent in wetlands. A minor fraction of migrating radionuclides are also taken up by the root system of trees.
H12B-07 11:50h
Systematic Groundwater/Stream Interactions Resulting From Regional Hydrogeologic Change
The analysis of groundwater/stream (gw/sw) interactions requires a clear understanding of the underlying hydrogeology. While much research has investigated the affects of more accessible hyporheic zone, there are still many unknowns regarding the influence of regional-scale gw/sw interactions on local hyporheic exchange fluxes. In an effort to isolate the effects of the regional hydrogeology we investigate gw/sw interactions at the edges of subcropping confining units, where unconfined aquifers transition confined conditions, using theoretical numerical modeling. The models predict substantial fluctuations in groundwater discharge to the streams in these regions. The predicted groundwater discharge patterns are tested by detailed downstream measurements of stream flow in the Loosahatchie River, TN, which traverses the gently dipping, confining unit subcrops of the eastern flank of the Mississippi Embayment. The theoretical model predicts two distinct zones of groundwater discharge fluctuation relative to the location of the subcrop. The first zone, immediately upstream of the subcrop, shows increasing groundwater discharge to the already gaining stream. In the second zone, immediately downstream of the subcrop, groundwater discharge to the stream drops abruptly to near zero, but the stream remains gaining. Downstream of this, the groundwater discharge begins to recover till it approaches its original rate. Measurements from the Loosahatchie River were taken during a period of low flow when flow in the stream was dominated by groundwater. They confirm the discharge increase in the upstream zone, and show a decrease in discharge in the second zone, but, much greater than the one predicted by the models; the stream becomes a loosing stream in this reach. This discrepancy is likely due to the simplified stream representation (a constant head boundary) in the models. The models and measurements show a clear region of significantly and systematically fluctuating gw/sw interactions that can serve as a laboratory for future research into the importance of hydrogeology on stream flow and geomorphology.
H12B-08 12:05h
Effects of River Discharge on Hyporheic Exchange Flows in Large Gravel-Bed Rivers: An Empirical Study
Current in situ understanding of exchange flows in riverbeds, and the ecological implications, has been largely derived from research in small streams. Comparatively little research has been conducted in large rivers, making it difficult to "scale-up" (spatially and temporally) our knowledge from small streams to understand observations and develop predictive models for large rivers. The studies conducted on small streams typically encompass too few sites, over too small of a longitudinal scale, with discrete sampling events (versus continuous sampling), resulting in an understanding of processes that is not directly transferrable to the study of much larger rivers. In addition, very few of these studies evaluate the effects of variable river discharge on riverbed exchange flows. We studied exchange flows between the river and riverbed at 14 sites distributed throughout 160km of the Snake River in Hells Canyon, Idaho, USA. Interactions between river water and pore water within the riverbed were quantified through the use of self-contained temperature and water level data loggers suspended inside of piezometers. The data were recorded at 20 min intervals over a period of 200 days when the mean daily discharge was 218-605 m$^{3}$ s$^{-1}$, with hourly stage changes as large as 1.5 m. The effects of discharge on vertical exchange between the river and riverbed were evaluated through measured hydraulic and temperature gradients, and the application of a numerical model. The vertical hydraulic gradients (VHG) between the river and the riverbed suggested the potential for predominantly small magnitude vertical exchange. The VHG also showed little relationship to changes in river discharge at most sites. Despite the relatively small VHG at most sites, the results from the numerical modeling suggested that there was significant vertical hydrologic exchange during all time periods. Our observations confirm the presence, and quantify the relative importance, of both advective and diffusive riverbed transport processes. The combined results of temperature monitoring and numerical modeling indicated that only two sites were significantly affected by short-term (hourly to daily) large magnitude changes in discharge. Although the two sites exhibited acute flux reversals between river water and hyporheic water resulting from short-term large magnitude changes in discharge, these flux reversals had minimal effect on mean daily riverbed temperatures.