GIS in Water Resources

Term Project Final Report

Fall 2005

Randy Goetz

Utah State University

Dept. of Aquatic, Watershed, and Earth Resources

TITLE

Longitudinal Bedform Classification for Comparison of Restored and Unrestored Reaches of the Provo River, Utah

Physical Controls on Hyporheic Flow

INTRODUCTION

Stream restoration efforts are increasing across the country, with the majority of projects seeking to enhance the quality of aquatic and riparian ecosystems inadvertently degraded through water resource development (National Research Council, 1990; Federal Interagency Stream Restoration Working Group, 1998; Graf, 2001). As part of Master’s project, I am currently investigating the relationship between restored channel morphology, and the function of the river’s hyporheic zone. The hyporheic zone is a crucial ecotone surrounding a stream where surface water enters the shallow subsurface and is subject to biogeochemical processes, and exchange with alluvial ground water. The physical controls on this process are comprised of a complex and heterogeneous combination of alluvial depositional processes, planform characteristics, channel topography, surface water stage, and regional groundwater gradients. In this study, I intend to take a simplified look at the longitudinal features known to exert control on hyporheic exchange, and compare restored reaches to unrestored reaches. Through this comparison, I hope to gain insight into the potential effects of restoration design in driving hyporheic flow.

Project Area

This study is focused on the effects of the Provo River Restoration Project (PRRP). Figure 1 displays the geographic setting of the PRRP, as well as the 9 reaches designated for restoration by the Utah Reclamation Mitigation and Conservation Commission, which oversees the project.

Figure 1. Geographic location of the Heber Valley within Utah, and the Provo River Restoration Project. For the inset, stream flow is from top to bottom.

The PRRP is intended to restore the geomorphic features and processes that will sustain a natural riverine ecosystem (Allred, 2003). We predict that the installation of geomorphic features intended to sustain ecological processes will increase hyporheic exchange as a physical consequence. We have designated three restored and three unrestored study reaches along the PRRP to test this hypothesis (Figure 2).

Figure 2. Map of the Heber Valley depicting the locations of three restored (red) and three unrestored (yellow) study reaches. Stream Flow is from top to bottom of the image.

The Hyporheic Zone

The hyporheic zone serves as the biological and hydrological intermediary between stream, riparian, and floodplain habitats (Biksey and Gross, 2001). This connection is vital for overall stream ecosystem health, and improved hyporheic function is emerging as a critical metric in evaluating the success of stream restoration (Brunke and Gonser, 1997; Fernald et. al., 2001; Sophocleus, 2002; Triska et.al., 1993).

Hyporheic flow occurs when stream water enters the shallow subsurface, flows some distance downstream, and returns to the surface (Figure 3).

Figure 3. Schematic depicting the hyporheic process. Arrows illustrate the flowpaths taken by streamflow during hyporheic exchange. Adapted from Reidy and Clinton, 2004.

Pool-riffle sequences can play a large part in creating the head gradients that drive surface water into, and out of, the subsurface along the longitudinal vector (Harvey and Bencala, 1993). In a simple conceptual model, the spacing and slope of riffles can yield a first-order approximation of the location, depth, and length of flowpaths in the hyporheic zone (Gooseff et. al., 2004). As such, I am interested in analyzing the spatial characteristics of these geomorphic features, and comparing restored and unrestored reaches of the PRRP. From this analysis I hope to understand the differences between restored and unrestored reaches in terms of those features that may have an influence on longitudinal hyporheic flowpaths.

Figure 4. Simple schematic depicting longitudinal slope breaks and the effects these features have on the location of subsurface flow.

METHODS

Adjusting Field Survey

The main source of data for this project was derived from total station surveys performed during the 2005 field season. The surveys were established in arbitrary coordinate systems due to a lack of geographic control on endpoints and temporary benchmarks. For the purpose of making visually informative maps, I needed to overlay the survey data on digital aerial photos of my study reaches. To do this I determined the UTM NAD 27 coordinates of the approximate top and bottom of my study reach. I then translated my arbitrary coordinates to these real world coordinates with survey software know as Foresight.

These translated coordinates were exported to a database file that was added to an Arc Map project along with the relevant aerial photos. By creating a xy event from the .dbf table, I displayed the survey as points on the photo. Figure 5 displays some of the problems that occurred in this process. This xy event had to be saved as a layer, an then converted to a shape file in order to visualize and edit it easily.

Figure 5. Survey points translated from arbitrary coordinates to UTM initially displayed backwards as shown in this figure.

These errors stem from the fact that I have no real world control for my survey. In choosing a reach top and bottom from the aerial photo, the x and y are transposed relative to those exported from Foresight. For the purpose of this exercise, I am forced to edit the points until they fit the photo in a meaningful way. Therefore, I mirrored the points to get them onto the stream, and then dragged and rotated them until they were in what appeared to be the correct location. The error associated with this could be on the order of meters. New coordinates derived from this process should not be used as control. For my final reach, I tried reversing the X and Y that I took from the aerial photo. When I brought the translated coordinates in from Foresight, they were no longer reversed. But they still had to be dragged and rotated to get them on the stream. This problem will be corrected when GPS coordinates of survey endpoints can be obtained.

Figure 6. The result of editing the location of survey coordinates in order to place them in their approximate location on the stream.

The result of this editing process is temporarily satisfactory despite potentially significant error. Verification of logical point location and slope breaks between line segments can be visually assessed as geomorphic units are later delineated.

Creating Slope Segments

The next step was to create a new line feature class. The feature class is composed of line segments whose endpoints are the survey coordinates.

Figure 7. Screen capture of the newly created line feature class with segment length and associated slope.

Using this feature class, geomorphic units were classified based on the slope of the line segment. The classification was done by creating a new field in the attribute table (titled GEO_UNITS in Figure 7) and calculating values using the advanced interface in the field calculator. A simple visual basic script was used to create an if-then statement that classified each segment:

dim tval as integer

tval = 2

if ([slope]) < .005 then

tval = 1

endif

if ([slope]) > .029 then

tval = 3

endif

GEO_UNITS = tval

This script classifies pools as 1, riffles as 2, and rapids as 3. The slope thresholds follow the classification scheme of Grant et al. (1990). The line feature class was then symbolized using the GEO_UNITS attribute, and a rough map of unit distribution began to form (Figure 8).

Figure 8. A close view of classified line segments. The pink represents pools, and the green riffles.

From this classified line feature class, a new polygon feature class was created. Each polygon was created around a continuous geomorphic unit. Holding to the classification outlined by Grant et al., line segments that were not on the order of the average stream width were not considered significant. These shorter segments would be included in the classification of the larger surrounding unit. Therefore, many units classified in the field calculator were not ultimately included in the final unit map. This is not error in slope delineation, but a product of the governing classification scheme (Figure 9).

Figure 9. Geomorphic units are defined by polygons that incorporate the dominant slope at reach width scale. Only the main channel is classified, and slope segments that are much smaller than the average stream width are disregarded, and that portion is classified in the same manner as the dominant slope for that unit.

Unit Length Determination

As part of the final spatial analysis of hyporheic controls, the spacing, and size of individual features must be known. Therefore, each unit was independently measured. By creating a new line feature class whose segments consisted of the centerline of each polygon, each unit polygon was measured. The attribute table of this new line feature class was joined with the attributes of the geomorphic unit polygons feature class, and a table with all the appropriate spatial information was created (Figure 10). This table was then exported, and the final analysis was carried out in excel.

Figure 10. Screen capture depicting the centerline segments of each geomorphic unit, and the final joined table including classification, and unit length.

Spatial Analysis

The table that was ultimately exported from the GIS was used to perform analysis and comparison in excel.  The analysis kept with the project goal of scrutinizing features in such a way as to highlight the characteristics that may influence longitudinal controls on the lengths, locations, and depths of penetration of hyporheic flowpaths.  The features of concern are those that create hydraulic head gradients, which drive surface water into the subsurface.   Along the longitudinal vector, these features are mainly the breaks in slope that occur at riffles, which cause hyporheic flow between adjacent pools (Figure 4). Therefore, the amount, spacing, and relative slope of riffles was determined.

RESULTS

Reach Maps

The GEO_UNITS polygon feature class was used to create unit maps for the six USU study reaches.  The maps provide visual representation of the final classification, and allow for qualitative assessment of the spatial relationships that exist for longitudinal hyporheic controls for each reach.  This visualization of my geomorphic classification was the most powerful product of this exercise.  While other steps were performed with greater speed through GIS, this is the one product that would have been impossible to achieve without this software.

Figure 11.  Six geomorphic maps produced through the slope classification and creation of associated polygons.  Pools are pink, riffles are green, and rapids are blue.  Each reach image is accompanied by the statistics of the reach including length, average slope, and the average lengths for each respective geomorphic unit.

Reach Composition and Unit Spacing

            The first component of the spatial analysis compared of the amount of each unit type found in a particular reach (Figure 12).  This is simply the ratio of the total unit length vs. the total reach length.  The point was to determine if a given reach is evenly divided between unit types, or if one unit is dominant.  From this, one can assess the longitudinal complexity of each reach, and gain insight into how important riffles are to the geomorphic composition of each reach.  The relative spacing of riffles was added to this comparison to illustrate how far apart riffles are relative to the reach length.  From Figure 12, one can begin to think about how upwelling and downwelling zones may be spaced in a given reach, and how long flowpaths might be.  For example, in Restored 1 we have a reach dominated by riffles, the feature that is assumed to drive flow.  We also see that the riffles are relatively far apart.  So we might form the hypothesis that in that reach hyporheic flow paths might be long, and far apart.  This would be in opposition to Unrestored 2, which is also dominated by riffles, but they are relatively closer together.  So perhaps we have several distinct regions of hyporheic flow in this reach, with shorter, more closely spaced flowpaths.

Figure 12.   Graph depicts the fraction of each reach contained in a certain geomorphic unit type as well as the spacing of riffles relative to the reach length.

Riffle Slope Analysis

The next component of this analysis was to focus in on the slope characteristics of the riffle units.  I normalized riffle slope by pool slope, and then reach average slope for each reach (Figure 13).  In this way I hoped to highlight how much riffle slopes departed from the pools in their respective reaches, and from the reach in general.  This should lend insight into how effective riffle units might be in setting up head gradients and driving flow.  Riffles that are very steep relative to associated pools might drive more hyporheic exchange in that reach.  This information, coupled with knowledge of the spatial distribution of riffles, should provide a robust picture of nature of physical controls for each reach.

Figure 13.  Graph displays the slope of riffles for each reach in relation to the slopes of pools, and the reach in general.  This is intended to suggest what magnitude of gradients might be present in each reach.

CONCLUSIONS

Effectiveness of GIS in Longitudinal Classification of Bedforms  

I found that using GIS in this analysis expedited the classification process.  The same computations could have been performed entirely in excel.  However, the ability to quickly visualize, and assess error in classification was orders of magnitude faster using GIS.  Overall, the visualization of the results was the most useful product.  The ability to produce a map of the unit classification and distribution that is overlain with aerial photos, and can be easily modified, is extremely useful in presentation and field verification. 

            Another important aspect of this project is the creation of a database.  I will be able to layer attributes for each reach, and enhance my ability to form and test hypothesis concerning the controls on hyporheic flow.  In my research, I am gathering data concerning streambed grainsize distribution and hydraulic conductivity for each unit type.  I will also make planform complexity measurements such as sinuosity, length of side channels, and radius of curvature of meanders.  Comparing all these attributes will create a full picture of multidimensional hyporheic controls.  Physical attributes can then be correlated to the results of hydrologic tracer tests and groundwater flow models in order to determine the overall effects of restoration on hyporheic exchange.

Longitudinal Controls on Hyporheic Exchange

Analysis of the spatial relationships of geomorphic units provides a foundation for conceptualizing the physical controls that exist in various reaches of the PRRP. This conceptual framework becomes important for testing assumptions concerning how restoration will affect hyporheic exchange. The larger research project that is being carried out on the PRRP operates under the working hypothesis that restored reaches will be geomorphically more complex that unrestored reaches. One might initially conclude that this increased complexity will result in increased hyporheic exchange.

It becomes apparent from this relatively cursory look at the morphology of the Provo that restored reaches may not have been constructed in a way that would result in increased volume of hyporheic flow. In general, unrestored reaches have steeper riffles relative to the pools in those reaches. This may be creating steeper head gradients and more exchange in those reaches. This conclusion needs to be weighed against the role that riffles play in the overall morphology of those reaches. We see that in general, unrestored reaches are composed of a higher fraction of riffles, and these are spaced closely together. The reach maps illustrate that the riffles are relatively short and alternate closely in unrestored reaches. From these characteristics, a general conclusion could be drawn that unrestored reaches are longitudinally configured to contain more distinct regions of hyporheic flow, and perhaps drive a higher volume of stream water into the subsurface. On the other hand, restored reaches may have longer flowpaths, with increased residence times, and higher probability of biogeochemical processing of nutrients.

Additional factors will need to be incorporated into this study in order to draw solid conclusions about the effects of the restoration in terms of hyporheic exchange. First, this study looks only at longitudinal features, and not the planform and floodplain characteristics that also control lateral exchanges between the hyporheic zone and the riparian zone. Also, these physical characteristics will need to be correlated to surface and groundwater models in order to determine the effects these physical controls actually have in this system.

REFERENCES

Allred, Tyler. 2003. Reaches 2 and 3-Propsed Design. Provided for UMRCC.

Biskey, T.M., and Gross, E.D. 2001. The Hyporheic Zone: Linking Groundwater and Surface Water-Understanding the Paradigm. Remediation: The Journal of Environmental Cleanup Costs, Technologies, and Techniques, Vol. 12, No. 1, pp. 55-62.

Brunke, M., and Gonser, T.G. 1997. The Ecological Significance of Exchange Processes Between Rivers and Groundwater. Freshwater Biology, 37, 1-33.

Federal Interagency Stream Restoration Working Group, 1998, Stream corridor restoration: principles, processes, and practices. Washington, D. C., National Technical Information Service.

Fernald, A.G., Wigington, P.J. Jr., and Landers, D.H. 2001. Transient Storage and Hyporheic Flow Along the Willamette River, Oregon: Field Measurments and Model Estimates. Water Resources Research, Vol. 37, No. 6, pp. 1681-1694.

Gooseff, M.N., Anderson, J.K., Wondzell, S.M., LaNiere, J., and Haggerty, R. 2004. A Modelling Study of Hyporheic Exchange Pattern and the Sequence, Size, and Spacing of Stream Bedforms in Mountain Stream Network, Oregon, USA. Hydrological Processes. Provided online by Wiley Interscience. DOI: 10.1002/hyp.5790.

Graf, W. L., 2001, Damage control: restoring the physical integrity of America’s rivers. Annals of the Association of American Geographers 91(1): 1-27.

Grant, G. E., Swanson, F.J., and Wolman, M.G. 1990. Patter nand origin of stepped-bed morphology in high-gradient streams, Western Cascades, Oregon. GSA Bulletin, v. 102, p. 340-352.

Harvey, J.W., and Bencala K.E. 1993. The Effect of Streambed Topography on Surface-Subsurface Water Exchange in Mountain Catchments. Water Resources Research, Vol. 29, No.1, pp. 89-98.

National Research Council, 1990, Restoration of aquatic ecosystems. Washington, D. C., National Academy Press, 552 p.

Sophocleous, M. 2002. Interactions Between Groundwater and Surface Water: the State of the Science. Hydrogeology Journal, 10: 52-67.

Triska, F.J., Duff, J.H., and Avanzino, R.J. 1993. Patterns of Hydrological Exchange and Nutrient Transformation in the Hyporheic Zone of a Gravel-Bottom Stream: Examining Terrestrial-Aquatic Linkages. Freshwater Biology, 29: 259-274.