4th International Conference on Integrating GIS and Environmental Modeling (GIS/EM4):
Problems, Prospects and Research Needs. Banff, Alberta, Canada, September 2 - 8, 2000.

A Linked GIS/Modeling Approach to Assessing the Influence of Flood-plain Structure on Surface- and Ground-water Routing in Rivers

GIS/EM4 No. 42

Geoffrey C. Poole
Jack A. Stanford
Steven W. Running
Christopher A. Frissell

Abstract

Development of a computerized modeling system used a GIS to perform spatial analysis followed by application of a dynamic water balance model to provide simulated rates and patterns of surface- and ground-water flow and exchange in a montaine alluvial flood plain. The modeling system was designed to incorporate the interacting effects of flood-plain morphology, surface-water flow patterns, heterogeneous/anisotropic aquifer structure, and hydrologic flow regime to simulate surface- and ground-water flow in three dimensions. The Nyack Flood Plain, Middle Fork Flathead River, Montana, USA, was the focus of study.

A simple classification of "flood-plain elements" (channels and vegetation patches) allowed the use of aerial photographs to describe quantitatively the three-dimensional surface and subsurface flood-plain morphology. The method associated element types with flood-plain inundation frequency and adjusted for localized channel incision to predict subtle (<1 m) differences in mean flood-plain element elevation and intra-element elevation variance. The method further associated element types with specific sediment patterns within the alluvial aquifer allowing inference of subsurface aquifer structure from aerial photography. Prior field observations and published literature values determined hydrologic model parameters for the aquifer and surface channels. During model validation, predicted patterns of surface-water flow, flood-plain water storage, and ground- and surface-water exchange agreed with field measurements.

The modeling systems is useful for comparisons between various hydrologic or flood-plain management scenarios and for understanding the role of flood-plain structure in determining patterns and pathways of water movement across the flood plain and within the associated alluvial aquifer. Similarly, since aerial photography is the basis for describing flood-plain structure within the modeling system, this method can be applied to simulate changes in surface and groundwater flux resulting from natural and anthropogenic changes in flood-plain structure captured in aerial photo archives.

Keywords

Flood Plain Morphology, Hyporheic Zone, Stream Ecology, Groundwater/Surface Water Exchange, Modeling, Geomorphology

Introduction

In spite of the increasingly recognized importance of ground-water/surface-water exchange in driving important fluvial ecosystem processes (Gibert et al. 1994; Grimm and Fisher 1984; Hynes 1983; Keller and Kondolf 1990; Stanford and Ward 1988; Stanford and Ward 1993; Stanford et al. 1994; Stanley and Boulton 1993; Triska et al. 1989; Ward 1998), holistic quantitative studies of the dominant controls on ground- and surface-water exchange are surprisingly few. Most typically, the processes of ground-water recharge, flow, and discharge in alluvial aquifers are simplified conceptually in an effort to model biological or chemical phenomena occurring in river systems (e.g., Bencala 1984; D'Angelo et al. 1993; Hill and Lymburner 1998). However, several studies have incorporated theoretical (Harvey and Bencala 1993; Vaux 1968), laboratory flume (Savant et al. 1987; Thibodeaux and Boyle 1987), and field (Harvey and Bencala 1993; White et al. 1987) techniques to investigate the effects of surface or subsurface physical structure on "hyporheic flow" (water from the stream infiltrating, flowing through, and re-emerging from the stream's alluvium). These studies have focused on the importance of characteristics such as streambed roughness, streambed topography, rate of ground-water recharge, and substratum homogeneity and particle size on hyporheic flow in the alluvium underlying the stream channel. Additionally, a few researchers have applied quantitative techniques on individual stream reaches to help understand morphologic controls on hyporheic flow within the flood-plain alluvium underlying the riparian zone. Wroblicky et al. (1994) and Wondzell and Swanson (1996) specifically incorporated variation of hydrologic properties into spatially explicit, two-dimensional ground-water models. While each study incorporated pathways of preferential ground-water flow, the two-dimensional constructs inherently limited their ability to describe more fully the complex flow dynamics associated with subsurface structural heterogeneity.

By combining a conceptual model of flood-plain structure, a GIS-based analysis of flood-plain surface features, the results from extensive field surveys, and a three-dimensional process model of ground- and surface-water flow, we have developed a coupled GIS/modeling system to assess the role of surface-water flow patterns and alluvial aquifer structure on the magnitude and pattern of ground-water flow and hyporheic exchange across an entire flood plain (Poole 2000).

Model Development and Application

In the modeling system, automated GIS routines perform spatial analysis of flood-plain morphology to create parameter files required by the Wetlands Dynamic Water Balance Model (WDWBM) (Walton et al. 1996; Walton et al. 1995) which then predicts pathways of ground-water flow and the magnitude of ground- and surface-water exchange on the flood plain. Specifically, aerial photograph interpretation of our study site - the Nyack Flood Plain, Middle Fork Flathead River, Montana, USA - yielded a simple cover-type classification scheme consisting of four channel-types and five vegetation-types that was used to delimit flood-plain "elements" (channel segments and vegetation patches) (Figure 1).

Figure 1. Photo-interpretation of cover types (left) on the Nyack Flood Plain from digital ortho quadrangle (right). 1:4000 scale color aerial photography was also used to assist in interpreting the ortho quadrangle.

In order to simulate water routing on the flood plain, the WDWBM requires estimates of mean elevation and elevation variance for each flood-plain element. Digital elevation models (DEMs) are not sufficient for deriving element elevations because DEMs are static and do not reflect changes in flood-plain morphology (e.g., channel avulsions) captured in aerial photograph time-series. Additionally, DEMs are not precise enough to capture hydrologically relevant variation (<1.0 m) in flood-plain elevation. Therefore, the GIS uses empirically derived relationships to estimate mean vegetation patch elevation (Figure 2) and minimum channel elevation (Figure 3) relative to elevation of the river's surface. The relative mean vegetation patch elevation varies within each cover type due to variation in the magnitude of channel incision into the flood plain. To correct for the effect of channel incision, the width of the scour zone (Figure 4) was used as a surrogate measure for channel incision (more incised channels have narrower scour zones) and is thus used as a covariate in Figure 2.

Figure 2. Plot of raw data with predictive regression lines from ANCOVA where scour zone width is a co-variate used to predict relative elevation for each of four cover types.

Figure 3. Box plot of minimum relative channel elevations for each channel-type.

Figure 4. Calculating scour zone width in proximity of a point on the main river channel. The scour area within the window is divided by the width of the window (in this case, 500 m) to obtain mean scour zone width within the window.

Elevation variation within each flood-plain element is also derived empirically. An analysis of spatial autocorrelation within vegetation patches provided relationships (Figure 5) to estimate intra-patch elevation variation; channel depth (a measure of elevation variation within channels) is estimated from channel type classification and width estimates derived from aerial photographs (Figure 6).

Figure 5. Variance in relative flood-plain elevation within cover type as a function of patch size. Relative elevation exhibits spatial autocorrelation on the flood plain; therefore, the intra-patch variance in relative flood-plain elevation increases as patch size increases.

Figure 6. Regression results from ANCOVA predicting channel depth from channel type using natural log of channel width as a co-variate. Adjusted channel depths reflect depth relative to expected mean elevation of a regeneration vegetation patch. Regression lines are for side-channels (top), springbrooks (middle), and high-flow/flood channels combined (bottom).

These elevation estimates are applied to a conceptual model of the subsurface flood-plain structure (Figure 7). Notably, the conceptual model assumes that, within a single flood-plain cross-section, the bed of the active river channel is at the same elevation as abandoned streambeds ("paleochannels") buried within the flood-plain alluvium. This assumption is supported by extensive elevation field surveys.

Figure 7. Idealized soil profiles based on field observations of subsurface structure underlying flood-plain elements. Letters associated with each stratum denotes hydraulic properties (from Table 2, below); lighter colored strata are more transmissive. Stratum thickness is shown below soil type. Precise thickness of variable strata is determined by elevation of vegetation patches.

When combined with the digitized mosaic of cover types (Figure 1) and empirical elevation estimates, the conceptual model yields a quantitative description of the surface and subsurface morphology of the flood plain. From this, the GIS provides requisite geomorphic model parameters required for the WDWBM. Other parameters required by the WDWBM include the hydrologic properties of sediments (Table 1) and "Manning's n" for channels. Since these properties are determined by the flood-plain element classification scheme, the GIS uses a look-up table to provide a complete set of parameter files necessary to run the WDWBM. All hydraulic parameters were derived from literature values or field measurements. No parameter "fitting" was employed because validation data were limited and because a single model run required roughly two weeks to complete, thus ruling out the possibility of iterating model runs while systematically varying different parameters.

Using parameter files created by the GIS, a slightly modified version of the WDWBM (see Poole (2000) for a description of the minor modifications) predicts the surface and subsurface water balance over time for each flood-plain element in response to the river flow regime. In aggregate, these individual water balances provide a detailed description of the magnitude, direction, and timing of ground-water flow and ground- and surface-water exchange on the flood plain. The WDWBM is uses a "link and node" approach to simulating water movement (Figure 8). Each flood-plain element is assigned a "node" where the model tracks water volumes, hydraulic head, and vertical ground-water movement between aquifer layers. Where flood-plain elements share a common boundary, the associated nodes are connected by "links" that track horizontal ground-water movement within aquifer layers and surface-water flow within stream channels. Together, the nodes and links provide a spatially explicit, asymmetric, three-dimensional wire-mesh representation of the flood plain (Figures 9 and 10). Parameters that describe flood-plain morphology (elevation, aquifer thickness, etc.) and hydraulic properties (channel roughness, hydraulic conductivity, etc.) are then associated with each "wire" (link) or intersection (node) in the mesh. Thus, the model is capable of representing the anisotropic, heterogeneous conditions in alluvial aquifers, including the presence of subsurface preferential flow pathways formed by buried stream channels ("paleochannels").

Figure 8. Schematic of "linked node" approach to ground-water modeling.

Figure 9. Model "mesh" resulting from node and link locations as determined by automated GIS algorithms. Grey background denotes boundaries of flood-plain elements digitized from aerial photo interpretation..

Figure 10. Model representation of subsurface aquifer structure (bottom) based on an idealized cross-sectional soil profile (top) and the conceptual model from Figure 7.

In order to validate the model, the river hydrograph from 1992 (Figure 11) was used along with the flood-plain structure derived from the 1992 aerial photographs. Field measurements of stream flow from 1992 were compared to model results (e.g., Figures 12 and 13; see Poole (2000) for additional comparisons). The comparisons between model results and field data suggest that the model's predictions are remarkably accurate, especially considering that none of the model parameters were fit to validation data.

Figure 11. Flow of the Middle Fork Flathead River during the simulation period measured by the USGS at West Glacier, Montana, USA, about 15 km downstream from the Nyack Flood Plain.

Figure 12. Comparison of observed rates of floodwater storage from field data (B. Ellis, W. Hansen, and J. Stanford, unpublished) vs. model predictions. Error bars represent ±5% of river flow as an estimate of potential error associated with stream gauge data.

Figure 13. Observed vs. predicted synoptic flood-plain surface-water flow at 11 locations on the flood plain. (See Figure 1 for site locations.)

Analysis of the model predictions suggest that the overall spatial pattern of aquifer recharge and discharge are, for the most part, independent of river stage and that flow within preferential flow pathways is not discrete, but that water exchanges readily between preferential flow pathways and the rest of the alluvial aquifer.

Conclusion

In summary, our methods are capable of simulating patterns and regimes of surface and hyporheic flow based on inference of three-dimensional flood-plain structure based on careful interpretation of aerial photographs. The modeling system can contrast ground- and surface-water flow patterns under different scenarios thereby allowing comparison of expected dynamics in water routing response to different river flow regimes or different flood-plain structures. Therefore, the system is useful for investigating potential impacts of flow regulation scenarios or for determining how flood-plain surface- and ground-water flow dynamics may have been altered by natural or anthropogenic morphologic changes documented in aerial photograph archives. Application of this system to real-world flood-plain management scenarios will help to further our understanding of natural and anthropogenic influences on hyporheic dynamics, an integral part fluvial ecosystem function.

Literature Cited

Bencala, K. E. 1984. Interactions of solutes and streambed sediment. II. A dynamic analysis of coupled hydrologic and chemical processes. Water Resources Research 20: 1804-1814.

D'Angelo, D. J., J. R. Webster, S. V. Gregory, and J. L. Meyer. 1993. Transient storage in Appalachian and Cascade mountain streams as related to hydraulic characteristics. Journal of the North American Benthological Society 12: 223-235.

DeBorde, D. C., W. W. Woessner, Q. T. Kiley, and P. Ball. 1999. Rapid transport of viruses in a floodplain aquifer. Water Research 33: 2229-38.

Fetter, C. W. 1994. Applied hydrogeology. MacMillan College Publishing Company, New York, New York.

Gibert, J., D. L. Danielopol, and J. A. Stanford, eds. 1994. Groundwater ecology. Academic Press, San Diego, California.

Grimm, N. B., and S. G. Fisher. 1984. Exchange between interstitial and survace water: implications for stream metabolism and nutrient cycling. Hydrobiologia 111: 219-228.

Harvey, J. W., and K. E. Bencala. 1993. The effect of streambed topography on surface-subsurface water exchange in mountain catchments. Water Resources Research 29: 89-98.

Hill, A. R., and D. J. Lymburner. 1998. Hyporheic zone chemistry and stream-subsurface exchange in two groundwater-fed streams. Canadian Journal of Fisheries and Aquatic Sciences 55: 495-506.

Hynes, H. B. N. 1983. Groundwater and stream ecology. Hydrobiologia 100: 93-99.

Keller, E. A., and G. M. Kondolf. 1990. Groundwater and fluvial processes; selected observations. Pages 319-340 in D. J. Hagerty and G. M. Kondolf, eds. Groundwater geomorphology; the role of subsurface water in earch-surface processes and landforms. Geological Society of America, Boulder, Colorado.

Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E. Sparks, and J. C. Stromberg. 1997. The natural flow regime. A paradigm for river conservation and restoration. Bioscience 47: 769-784.

Poole, G. C. 2000. Analysis and Dynamic Simulation of Morphologic Controls on Surface- and Ground-water Flux in a Large Alluvial Flood Plain. Ph.D. Dissertation. School of Forestry. The University of Montana, Missoula. 154 Pages.

Savant, S. A., D. D. Reible, and L. J. Thibodeaux. 1987. Convective transport within stable river sediments. Water Resources Research 23: 1763-1768.

Schroeder, P. R., B. M. McEnroe, R. L. Peyton, T. S. Dozier, and J. W. Sjostrom. 1992. The hydrologic evaluation of landfill performance (HELP) model, volume IV: documentation for version 2. US Army Corps of Engineers Envrionmental Laboratory, Waterways Experiment Station Internal Working Document #EL-92-1.

Schroeder, P. R., T. S. Dozier, P. A. Zappi, B. M. McEnroe, J. W. Sjostrom, and R. L. Peyton. 1994. The hydrologic evaluation of landfill performance (HELP) model: engineering documentation for version 3. US Environmental Protection Agency, Risk Reduction Engineering Laboratory #EPA/600/R-94/168b.

Shaver, R. B. 1998. The Determination of Glacial Till Specific Storage in North Dakota. Ground Water 63: 552-7.

Stanford, J. A., and J. V. Ward. 1988. The hyporheic habitat of river ecosystems. Nature 335: 64-66.

Stanford, J. A., and J. V. Ward. 1993. An ecosystem perspective of alluvial rivers: connectivity and the hyporheic corridor. Journal of the North American Benthological Society 12: 48-60.

Stanford, J. A., J. V. Ward, and B. K. Ellis. 1994. Ecology of the alluvial aquifers of the Flathead River, Montana. Pages 367-390 in J. Gibert, D. L. Danielopol, and J. A. Stanford, eds. Groundwater Ecology. Academic Press, San Diego, California.

Stanford, J. A., J. V. Ward, W. J. Liss, C. A. Frissell, R. N. Williams, J. A. Lichatowich, and C. C. Coutant. 1996. A general protocol for restoration of regulated rivers. Regulated Rivers: Research and Management 12: 391-413.

Stanley, E. H., and A. J. Boulton. 1993. Hydrology and the distribution of hyporheos: perspectives from a mesic river and a desert stream. Journal of the North American Benthological Society 12: 79-83.

Thibodeaux, L. J., and J. O. Boyle. 1987. Bed-form generated convective transport in bottom sediments. Nature 325: 341-343.

Triska, F. J., V. C. Kennedy, R. J. Avanzino, G. W. Zellwegner, and K. E. Bencala. 1989. Retention and transport of nutrients in a third-order stream in northwestern California: hyporheic processes. Ecology 70: 1893-1905.

Vaux, W. G. 1968. Intragravel flow and interchange of water in a streambed. Fishery Bulletin 66: 479-489.

Walton, R., R. S. Chapman, and J. E. Davis. 1996. Development and application of the wetlands dynamic water budget model. Wetlands 16: 347-357.

Walton, R., J. T.H. Martin, R. S. Chapman, and J. E. Davis. 1995. Investigation of wetlands hydraulic and hydrological processes, model development, and application. US Army Corps of Engineers, Waterways Experiment Station Technical Report #WRP-CP-6.

Ward, J. V. 1998. A running water perspective of ecotones, boundaries and connectivity. Internationale Vereinigung für theoretische und angewandte Limnologie, Verhandlungen 26: 1165-1168.

White, D. S., C. H. Elzinga, and S. P. Hendricks. 1987. Temperature patterns within the hyporheic zone of a northern Michigan river. Journal of the North American Benthological Society 6: 85-91.

Wondzell, S. M., and F. J. Swanson. 1996. Seasonal and storm dynamics of the hyporheic zone of a 4th-order mountain stream. I: Hydrologic processes. Journal of the North American Benthological Society 15: 3-19.

Wroblicky, G. J., M. E. Campana, C. N. Dahm, H. M. Valett, J. A. Morrice, K. S. Henry, and M. A. Baker. 1994. Simulation of stream-groundwater exchange and near-stream flow paths of two first-order mountain streams using MODFLOW. Pages 187-198 in J. A. Stanford and H. M. Valett, eds. The Second International Conference on Ground Water Ecology. American Water Resources Association, Atlanta, GA.

Authors

Geoffrey C. Poole, Systems Ecologist, Flathead Lake Biological Station and Numerical Terradynamics Simulation Group
University of Montana, 311 Biostation Lane, Polson, MT 59860
(Current address: 4051 Wildflower Lane, Tucker, GA 30084)
Email:gcp7@cornell.edu

Jack A. Stanford, Director and Bierman Professor of Ecology, Flathead Lake Biological Station
University of Montana, 311 Biostation Lane, Polson, MT 59860
Email: stanford@selway.umt.edu

Steven W. Running Professor, Numerical Terradynamics Simulation Group, School of Forestry
University of Montana, Missoula, MT 59812
Email: swr@ntsg.umt.edu

Christopher A. Frissell, Research Assistant Professor, Flathead Lake Biological Station
University of Montana, 311 Biostation Lane, Polson, MT 59860
Email: frissell@selway.umt.edu