Methods in Stream Restoration

Methods in Stream Restoration

As much of the literature often points out, stream restoration is a relatively new ecological approach to reversing anthropogenic degradation of the landscape. However, the number of stream restoration projects has increased dramatically within the past decade, with an annual cost exceeding one billion dollars (Bernhardt et. al 2005). Success rates vary, and much of the conditions of success depend upon the scope of the project. Unfortunately, it is quite rare for restoration projects to restore all aspects required for a complete project, with many being limited by the terms and qualifications of its funding.

Methods used in stream restoration can be subdivided in a number of ways. They can be spatial and temporal, theoretical and empirical, qualitative and quantitative, observational or conversational. All methods incorporate some aspect of spatiality or temporality. This can range from aerial photography and other forms of remote sensing data, to historical analysis garnered from aerial photographs or simply word of mouth: information transmitted by generations of primary witnesses. Stream restoration is a science, and naturally requires a lot of quantitative information acquired from data collection. Empirical data might then be sieved and sorted through mathematical and statistical analysis. Quantitative data can also be approached through modeling- by creating idealizations based on both observations and theory. Conversely, quantitative data can be used for qualitative approaches, which may arise in the form of broad classifications and grouping. Any of these methods can be applied to a variety of criteria required in assessing broad functions of streams, such as water quality, sediment transport, channel morphology, riparian vegetation, habitat heterogeneity, and stream bank stability.

This paper will examine five methodologies used to assess the criteria mentioned above, and use case-studies to illustrate how they work. Specifically, this paper will address:

1)      Historical Data Collection

2)      Numerical Modeling

3)      Airborne/Aerial methods

4)      Ground/ field- based methods

5)      Classificatory methods

Historical Data Collection Methodologies

Historical data collection is a vital component to stream restoration because it allows for the capture of long-term trends in river systems (Gurnell et. al. 2003). Rivers are dynamic, and the form and process characterizing a river reach at one point in time most likely will not be the same at another point in time. This is the reason why fluvial processes (meaning rivers and streams) are referred to as dynamic systems.

The use of stream gauges for real time data collection constructed at numerous reaches by the USGS allows for an unbelievable amount of data to be collected. More importantly, some of these gauges have been in commission for over fifty years. Thus, the stream gauges not only allow for collection of large amounts of data, but also allow for synthesis of long term trends.

Historical analysis from maps and aerial photographs are vital to stream restoration projects. Such documents can be used in a multitude of ways, which includes determining previous channel widths, erosion rates, land-use changes, and a variety of others (Cooke and Reeve 1976). Paleohydrology (ancient hydrologic processes such as flow rates), geoarchaeology (studying geomorphic processes that occur at archaeological sites), and historical geography are some other general examples of the usefulness of historical data.

Historical data is largely limited by extant and resolution. Photographs produced many years ago may be difficult to see certain details. Aerial photographs are limited in resolution and scale, and only a portion of the reach can be depicted (Gurnell et. al. 2003).

In Natural Streams and the Legacy of Water-Powered Mills, Walter and Merritts (2008) discuss a research project in Lancaster, PA addressing the role of colonial mill dams in stream bank erosion in mid-Atlantic stream. Colonial mill dams allow for the deposition of large amounts of sediments, which become immobilized within the stream bank. These sediments are referred to as legacy sediments (Schenk and Hupp 2009). During flood events, large portions of the stream bank erode, leading to high sediment yields into the Chesapeake Bay.

Using historical maps, Walter and Merritts were able to determine the extent of the mill dams. Using geochronology (geologic methods of age-dating rocks), they were also able to determine the age of the legacy sediments, which correlated to the same period of the construction of the dams. However, while some important historical characteristics of the stream were uncovered, large amounts of data were still undetermined. For example, the historical morphology (shape) of the stream was unknown. Riparian and in-stream data were limited to a small amount of uncovered pollen samples. Understanding prior stream functions required quite a bit of educated guesswork.

Numerical Modeling

Numerical modeling is a way of capturing the processes of streams and converting it to numbers, equations, and other forms of visualizations. Models also “provide the basis for aggregating from the scales of observation to the scales of interest” (Kirkby 1996). In addition, numerical modeling allows for the determination of thresholds: such as total maximum daily loads of sediment (TMDLs), sediment entrainment and competency, as well as erosion rates. Darby and Wiel (2003) list four different types of modeling: 1) conceptual models: providing qualitative descriptions to understand processes and forms; 2) statistical and empirical models: useful in determining relationships between variables; 3) analytical models: based on physical approaches that compensate for limitations in conceptual and empirical models; and 4) numerical models: multidimensional analysis of spatial and temporal features.

One of the drawbacks of numerical modeling is the inherent difficulty in converting complex systems to a simple model. Also, fluvial systems are dynamic and variable. Models may be limited in their applicability to streams. Models can be difficult to interpret depending on the type of mathematics and statistics utilized, and applicability will be limited to a specialized grouping of those that understand the formulae.

In the paper Development of an Ecohydraulics Model for Stream and River Restoration (Bockelmann et. al. 2004) a river modeling tool was developed using hydraulic, substrate, and ecological parameters. Ecohydrologic modeling is the coupling of hydraulics and ecology to develop an understanding of stream processes. Without getting bogged down with the mathematical specifics of the model, the project called for using macroinvertebrates as indicators of in-stream processes since they are relatively sensitive to both water quality and hydraulic processes. All macroinvertebrates sampled were given a binary code of 0 or 1 to categorize suitability within the stream (based on a variety of parameters that is too lengthy to discuss in this paper), which they referred to as a suitability index. A score of 0 meant unsuitable and a score of 1 meant suitable.  At each site sampled the average suitability index was calculated so that a spatial distribution could be developed along the reach to show variability and patchiness of scores.

It is easy to discern some of the inherent weaknesses and strengths in this approach. Macroinvertebrate suitability indices can be subjective since it incorporates human bias in determining which metrics to be used in the calculation. Macroinvertebrates are notoriously patchy, and abundances often vary from year to year. As mentioned previously, simplifying a complex process forces some aspects to be overlooked or ignored.

Aerial/Airborne Methods

Spatial data allows for analysis of topographic and other physical features of the landscape over a particular scale. In terms of stream restoration, this is often performed through remote sensing photogrammetry imaging, such as light detection and range (LiDAR), electromagnetic radiation (emr), and sensors. 

Aerial and airborne methods have many benefits, particularly in its ability to cover large spatial areas. When conducting LiDAR, absorption properties of water contrast highly with those of land and vegetation, thus making fluvial features easily distinguishable (Gilvear and Bryant 2003). It is much easier to store data in digital format compared to traditional cartographic methods. Data collected by remote sensing can then be used in geostatistical software and computations, or used in programs such as ArcGIS for further spatial analysis.

There are limitations to remote sensing that need to be considered. Some of the limitations include dense vegetation cover that prevents clear visualization of the stream channel, geometric accuracy in that the image may be distorted, and issues related to scale and size since many fluvial systems can be quite large. Also, it is difficult to get adequate resolution on very small reaches, or at the micro-scale (Gilvear and Bryant 2003). Once stream width becomes less than 3-5m, resolution becomes less accurate (Diakite et. al 1986). First order and second order ephemeral streams (those in which water flows less than 50% of the year) can also become problematic. Conversely, those that require large spatial scales can be costly and time-consuming. 

In Hauer et. al (2008), LiDAR is used to aid in reconnecting floodplains to allow for habitat for various species of fish. Furthermore, LiDar, in addition to some numerical modeling, is used to develop digital terrain models in order to take a snapshot of micro-scale bathymetric heterogeneity within the stream channel.  In other words, the goal was to be able to determine if the structure of the stream channel is suitable for fish habitat. The authors claimed that LiDAR is much more efficient than using historical aerial maps because the degree of accuracy is much higher. However, the margin of error for such analysis, even with high resolution remote sensing, becomes magnified as channel width/size becomes smaller. Therefore, as channel width decreases to a particular threshold the use of LiDAR begins to lose clarity.

Ground Based Methods

Process and field based methods cover a wide variety of techniques that generally fall under the same umbrella header of visual observations physically taken in the field, such as the stream channel, floodplain, or riparian corridor. All of these require taking direct measurements in the field and comparing these to standards and/or thresholds, i.e. water quality standards determined by the EPA, TMDLs determined by the state, sedimentation and erosion rates that are deemed detrimental to habitat quality by biologists, etc.

Field based methods can be for either geomorphic or biologic assessments. Some [rapid] geomorphic approaches include the pebble count for particle size distribution, sediment sampling to determine sediment load, discharge and other hydraulic geometry properties (such as channel width, water velocity, and water depth), and water chemistry. Some [rapid] biologic approaches include benthic invertebrate (mainly aquatic insects and other arthropods) sampling, habitat assessments, and descriptions of the riparian zone (the vegetated area directly adjacent to the stream bank). Regardless of the type of assessment, these are useful tools because they are quick, efficient, and relatively cheap. In terms of ecology, rapid assessment allows for quantification of diversity and community structure. In terms of geomorphology, rapid assessment allows for a depiction of the physical processes occurring within the stream.

Ground based methods are not without limitations. Primarily, these methods lack the ability of measuring large spatial scales. For example, time and money constraints would prevent for data collection of a stream reach that spans the range of kilometers. Therefore, this method is largely limited to small reaches. Additionally, this method can only provide information of the current condition of the stream, and largely ignores historical functions.

Most stream restoration projects incorporate some form of ground based methods, and so there is quite a bit of literature discussing it. This paper will address a study by Buchanan et. al. (2013) because it addresses both quantitative and qualitative features of ground based methods, and consists of several years of post-restoration data, oftentimes a rarity. The goal of the restoration was to reduce stream bank erosion, create benthic habitat for fish and other invertebrates, and adjust the sinuosity of the channel to allow for more efficient transportation of sediment and release along the floodplain. A variety of methods were used: Wolman Pebble Counts (counting 100 pebbles along the stream to determine the dominant size), hydraulic geometry measurements (stream width, depth, velocity, and discharge), sampling of the macroinvertebrates using rapid bioassessment, as well as some numerical modeling, historical analysis, and Rosgen methods of classification (discussed further in the next section). The stream was restored in 2005, and analysis and monitoring continued until 2012. Although the restoration was considered successful this statement does not come without its qualifications. Macroinvertebrate diversities did not improve due to a lack of proper habitat creation, and channel bed degradation continued immediately after restoration.

Classificatory – (Rosgen) Methods

With a form-based/classification approach, a stream is compared to that of a particular type. The Rosgen approach is used as a case study because this is the most popular method, containing an abundance of data and examples, as well as a ubiquity of criticism. In the Rosgen approach, formally known as Natural Channel Design (NCD), streams are classified alpha-numerically, such as C2, B3, F2a, for example (Rosgen 1994). This allows for a relatively easy way to determine how a stream should be behaving. However, this also allows for a relatively broad characterization of a stream. A broad generalization of classification is problematic. First, it assumes that all streams under one alpha-numeric heading are the same. Second, it ignores the possibility that the stream is in transition, and the current snapshot may not be indicative of its typical condition. Thirdly, it will encourage the forging of data such that it fits properly within the expected classification.

The Rosgen debates, often referred to as the Rosgen Wars (Lave 2009) are essentially arguments between an academic culture of understanding restoration (the fluvial geomorphologists with PhDs, and other geologists/geographers who have spent considerable time with rivers) versus engineers and those in private consulting firms who strictly employ the Rosgen methods. The Rosgen method of classification dominates the stream restoration industry. Those who employ the method do not necessarily have any academic training, but rather are educated by a series of short courses taught by Rosgen himself. Those who are trained in the academic field are largely shut out of participating in projects, simply because they are not trained in the Rosgen method.

A common example used by the opponents of Rosgen is the Uvas Creek restoration project (Simon et. al. 2007). The goal of the restoration of Uvas Creek in California was to return a very unstable urban stream into a stable “C4” channel (Kondolf 2006). A C4 channel is characterized by well-defined point bars and floodplains that are used to dissipate energy during high floodplains. However, within three months of the restoration project, a six-year storm completely washed out the channel. The ultimate explanation for the failure was the inability to incorporate dynamic, temporal trends of the river over time, which a classificatory method often overlooks.

Conclusion

There is no absolute method in stream restoration (nor ideal for that matter), but there are certainly methods that are better than others. Methods that strictly abide by one single paradigm (i.e. Rosgen) are doomed for failure because they ignore the dynamism inherent in fluvial systems. Therefore, it is imperative that restoration efforts synthesize a multitude of different methods that incorporate a diverse amount of perspectives. An effort that ignores historical land use will fail. An effort that focuses on only small scales will fail because it ignores the bigger picture. A project that emphasizes biotic without considering the abiotic is destined to fail because the two are intimately intertwined.

Stream restoration can be controversial, but it is important in an age of increasing environmental degradation. As the amount of restoration efforts continue to increase, so will understanding, and thus, success. Projects that incorporate the variety of methods presented will be the ones that achieve the most success.

Literature Cited

Bernhardt, E.S., Palmer M.A., Allan J.D., et. al. Synthesizing U.S River Restoration Efforts. Science 308: 636-637.

Bockelmann, B. N., Fenrich, E. K., Lin, B., & Falconer, R. A. (2004). Development of an ecohydraulics model for stream and river restoration. Ecological Engineering, 22(4–5), 227–235.

Buchanan, B. P., Nagle, G. N., & Walter, M. T. (2013). Long-Term Monitoring and Assessment of a Stream Restoration Project in Central New York. River Research and Applications

Cooke, R.U. and Reeves, R.W. (1976). Arroyos and Environmental Change in the American South-West, Oxford: Clarendon Press.

Darby, S.E., and Van De Wiel, M.J. (2003). Models in Fluvial Geomorphology. . In: G.M. Kondolf and H. Piegey, eds., Tools in Fluvial Geomorphology, Cinchester, England: John Wiley and Sons, pp. 503-537.

Diakite, M., Yerheau, M., Bonn, F. (1986). The Utility of Landsat TM Imagery in the Inland Delta Cartography of Mali. In: Proceedings of the 20^th International Symposium on Remote Sensing of Environment (ERIM), Nairobi, pp. 567-574).

Gilvear, D., and Bryant, R., (2003). Analysis of Aerial Photography and Other Remotely Sensed Data. In: G.M. Kondolf and H. Piegey, eds., Tools in Fluvial Geomorphology, Cinchester, England: John Wiley and Sons, pp. 135-170.

Gurnell, A.M., Peiry, J., Petts, G.E. (2003). Using Historical Data in Fluvial Geomorphology. In: G.M. Kondolf and H. Piegey, eds., Tools in Fluvial Geomorphology, Cinchester, England: John Wiley and Sons, pp. 77-101.

Hauer, C., Mandlburger, G., Schober, B., & Habersack, H. (2012). Morphologically Related Integrative Management Concept for Reconnecting Abandoned Channels Based on Airborne LiDAR Data and Habitat Monitoring. River Research and Applications

Kirkby, M.J. (1996). A Role for Theoretical Models in Geomorphology? In: Rhoads, B.L and Thorn, C.E. eds., The Scientific Nature of Geomorphology, Cinchester, England: John Wiley and Sons, pp. 257-272.

Kondolf, M.G. (2007). River Restoration and Meanders. Ecology and Society 11(2): 42.

Lave, R. (2009). The Controversey over Natural Channel Design: Substantive Explanations and Potential Avenues for Resolution. Journal of the American Water Resources Association 45(6) 1519-1532.

Muller, E., Décamps, H., & Dobson, M. K. (1993). Contribution of space remote sensing to river studies. Freshwater Biology, 29(2), 301–312. doi:10.1111/j.1365-2427.1993.tb00766.x

Schenk, E. R., & Hupp, C. R. (2009). Legacy Effects of Colonial Millponds on Floodplain Sedimentation, Bank Erosion, and Channel Morphology, Mid-Atlantic, USA1. JAWRA Journal of the American Water Resources Association, 45(3), 597–606.

Simon, A, Doyle, M., Kondolf, M., Shields Jr., F.D., Rhoads, B., McPhillips, M. (2007). Critical Evalutation of how the Rosgen Classification and Associated “Natural Channel Design” Methods Fail to Integrate and Quantify Fluvial Processes and Channel Response. Journal of the American Water Resources Association. 43(5).  

Sudduth, E. B., Meyer, J. L., & Bernhardt, E. S. (2007). Stream Restoration Practices in the Southeastern United States. Restoration Ecology, 15(3), 573–583.

Walter, R.C., Merritts, D.J. (2007) Natural Streams and the Legacy of Water-Powered Mills.    Science, Vol. 319.