The different sections of this article are written by three members of the Cobb Country, GA, Stream Monitoring Program. Three previous articles on the program have appeared in past issues of Stormwater: “Municipal In-Stream Monitoring” in the September 2008 issue, “Municipal In-Stream Chemical Monitoring” in the October 2008 issue, and “Municipal In-Stream Macroinvertebrate Sampling” in the November/December 2008 issue.
Introduction
Although I have summarized and then detailed the essential components of Cobb County’s Stream Monitoring Program in the last three issues of Stormwater, I want to conclude by presenting three important disciplines supporting the comprehensive nature of stream monitoring: geomorphology, riparian botany, and geographic information systems (GIS) applications for stream sampling. Robert Bourne, the program’s long-term biologist from 1994 to 1999, addresses fluvial geomorphology and riparian botany. Adam Sukenick and Erin Feichtner have administered the Stream Monitoring Program since 1999, and in this article Adam addresses GIS applications for in-stream monitoring. Whether considering geomorphological impacts of flow-sponsored sediment redeposition, riparian “buffering” of high-sheet-flow and absorption of overabundant plant nutrients, or GIS management of multilayered water-quality program assets and relevant land use features, it is not difficult to understand how geomorphology, riparian botany, and GIS applications directly support and aid in interpreting in-stream chemical and biological assessments.
Cobb County’s Stream Monitoring Program is mandated, firstly, by National Pollutant Discharge Elimination System (NPDES) legislation requiring a “monitoring program for representative data collection for the term of the permit that describes the location of outfalls or field screening points to be sampled (or the location of in-stream stations), why the location is representative, the frequency of sampling, parameters to be sampled, and a description of sampling equipment” (CFR 40 122.26). The program also fulfills biological sampling requirements for the NPDES permits of Cobb County’s Noonday Creek and other expanding wastewater reclamation facilities. Such sampling requirements were also recently added to sampling regimens emphasizing chemical sampling in Phase I communities such as Cobb in the Metropolitan North Georgia Water Planning District.
The California Stormwater Quality Association (CASQA) also addresses the call for enhanced water-quality assessment in NPDES programs, first defining general program assessment as providing managers “the feedback necessary to determine whether their programs are achieving intended outcomes (complying with permit requirements, increasing public awareness, changing behaviors, etc.), and ultimately whether continued implementation will result in water quality and/or habitat improvement” (CASQA 2008). Further distinguishing between the implementation outcomes of programs like street sweeping, erosion control, and education and the desired result of improved water quality, the document goes on to exhort more limited programs to “evaluate program implementation and water quality, and seek to find the relationship between the two.” The CASQA document concludes with some “Effectiveness Needs and Future Directions,” arguing for “development of cost-effective water quality assessment tools” and lamenting the “lack of readily available and understandable methods”; both deficiencies, it says, contribute to the preponderance of implementation assessment per se over water-quality (improvement) sampling. From California to Georgia, direct, comprehensive water-quality assessment is incumbent upon water resource managers and planners, as NPDES program implementation water-quality impact has been an issue for some time with entities ranging from nongovernmental organizations to Congress and the Office of Management and Budget.
Communities with opportunities to enhance their programs by initiating comprehensive water-quality assessments must become undaunted opportunists in preserving their most critical resource. Alternatively, communities already established in water-quality sampling must not ease into inertia and the easy chair of regulatory adequacy, as there are kinetic, complex, physical, and political pressures characterizing communal life journeys that are far from over. Indeed, environmental professionals and the communities they serve must consider what must be done in the tough terms of what had to be done.
-Lanse Norris
Establish GIS for the Future
Although water-quality analysis has advanced with technology, water-quality technicians have been using the same collection methods for some time. Although we have a better understanding of complex relationships between sampled water quality and biological populations, what has really changed is our capacity to store, analyze, and interpret the data.
Geographic information systems have the capacity to revolutionize what we do with whole, historical data sets. The most commonly recognized component of GIS is the final product, a graphical representation of the data–a map. However, to think of GIS as only a map would be tantamount to thinking of a water-quality study as only a final overall score of “good” or “poor” for a stream.
If you were tasked with the responsibility to conduct a study to determine baseline water-quality conditions, you would likely begin by selecting sampling sites–sites that represent average conditions within the study area. Considerations in choosing sites might include land use, industrial discharges, drainage area, topography, stream accessibility, hydrography, riparian composition, and soil type; the list is really limited only by the time and resources you have to invest in the project. To begin selection, you could find a land use map, determine major land use classes within the study area, and place a sampling site in each class. Next, you could look at industrial discharges to make sure sampling sites are not in a direct waste stream. Certainly, you would also evaluate drainage areas so that various sized watersheds were being sampled, as well as first-, second-, and third-order streams. After preliminary site selection, you would also pull out topographical maps and look at road coverage to see whether the sites selected are accessible.
Instead of using a drawer full of maps and countless hours of field reconnaissance, you could employ GIS. GIS can be useful in site selection by first defining the study area, then adding as many site-influencing “layers” as you have available. Think of each map in the above scenario as a layer on a GIS. Layers can be turned on and off to look at multiple images (or combination of images) at once. By using the hydrography, riparian, and soil layers simultaneously, you could, for example, determine that sites are likely situated in wetlands without having to go to the field and spar with aggressive wetland insects. To optimize time and efficiency, criteria for site selection are predetermined, and GIS is used to choose potential sites. Later, after sites are field-verified, you can create a whole new layer to geographically display your sites.
With site selection finalized, you must now consider which chemical parameters, for example, to analyze and how to manage data. Consider that every visual map is really only a graphical representation of some data set. Management of data is the second, less appreciated, component of a GIS. A road map is merely a list of road names geographically referenced and displayed. Water-quality data, like road names, can also be managed and later displayed using GIS. A database of results from water-quality analysis can be stored and used to “map” results of a water-quality study. Each site sampled for a study will be geographically referenced with a name, location, and dataset. Later, GIS can be used to visually display information contained within that dataset. For example, fecal coliform data stored within GIS can be queried and used to create a display that graphically displays each site that fails to meet the state standard. Beyond that, with proper planning, chemical and land use data can be cross referenced and used to display geographic relationships, such as displaying all sampled sites located downstream of agricultural land use areas that fail to meet state water-quality standards for fecal coliform. Even further, by analyzing several years of data, you may be able to determine trends in the data, such as when fecal coliform thresholds were exceeded and the rate at which they’re increasing.
With enough data entered into the system, the data relationships you can infer are quite comprehensive. If a water-quality study includes data from outside sources (such as industrial monitoring, sanitary, and storm sewer discharge data), as well as collected water chemistry, GIS may be useful in detecting sources of “hits,” or unusual results, found during routine sampling. Essentially, field investigation or reported problem inquiries often begin at your computer. For example, if lab analysis produces a metals result well above typical baseline levels, you can use GIS to first display topography and delineate the area draining to the point at which the sample was collected. Then, using land use coverage, the next step could involve highlighting all industrial land use (the area with the highest potential to produce metals contamination). Next, locate known industrial discharges within this selected area and query industrial monitoring data to determine which industry is likely to produce waste with metal identification and concentration close to the sample collected downstream. If this doesn’t immediately narrow your search or identify the source, it may mean that there is a new industry or previously unmonitored industry you are not aware is discharging.
Using GIS to expose complex relationships illustrates the third and final component of a GIS–its ability to analyze data and model or create unrealized or derived datasets. GIS has the ability to manage countless datasets, and its usefulness as an analytical tool is limited only by the stored data. The example applications above illustrate how chemical, land use, and industrial monitoring data can be used to detect trends and locate sources of pollution, but consider GIS coverage of nonpoint-source information. Biological and habitat data, for instance, are key elements of any water-quality program and are very susceptible to change and influence from external inputs. Storage of biological datasets gives a user the ability to graphically present sites with the best biological diversity or illustrate locations where rare or endangered fish were collected. However, by cross-referencing databases, we may discover relationships previously unknown. More obvious relationships between biological and chemical results have been explored, but using GIS might quickly identify subtle thresholds or parameters that limit diversity. For instance, by geographically mapping all sites where sensitive species, such as stoneflies, exist and analyzing concomitant water chemistry results versus water chemistry results where stoneflies are not found might demonstrate the possibility of a biological limiting factor like pH or low dissolved oxygen (DO). With enough chemical and biological sampling information, different existing datasets may be cross referenced, and they may already hold enough information to conclude that below a certain pH or DO level stonefly populations will vanish. With thresholds for stoneflies established, one can now filter the entire dataset to look for downward trends in pH or DO. If results are found to be approaching stonefly tolerance levels, the identified sites can be prioritized for additional study or protection.
GIS has broader applications that cover projects from implementation to assessment. From the initial question to the final answer, GIS streamlines each step. New components of compliance permits, for example, can be organized, managed, and evaluated using GIS. After identifying a problem–perhaps spill containment time–GIS can be used for initial analysis of the problem.
In this case, one would identify all streams and likely points of entry, such as storm drains or industrial discharges, then collect input flow data through field work or from local monitoring stations (such as USGS gages) to determine baseline flow rates at several points within each watershed. With these data, one can employ GIS to test spill scenarios and response times. With flow-rate data, and using spatial analysis to derive spill travel distance, one could determine the dispersion time of a spill from its point of entry to locations downstream, and, by adding traffic datasets for the same area, one could also determine response times to locations throughout the region. With the problem and response times now identified, one could focus on slow response times and implement strategies to expedite containment. Using GIS, one could create datasets to identify areas where current response plans are not adequate. The user could also map areas where the fastest response time is most critical.
With continued data input, GIS is better able to model real-world incidents and allow one to continually assess the effectiveness of a plan. After an occurrence of a spill, one can evaluate the implementation strategy and determine its effectiveness.
In short, GIS is the only tool that can illustrate the problem, analyze the data, model the impacts, and evaluate the results. The answers provided are limited only by the questions one can ask given the strength of the dataset, and Cobb County stream monitoring personnel often find themselves framing questions after data distribution answers have already emerged amid the comprehensive GIS data.
A geographic information system is a finely tuned, uniquely individualized, and highly adaptable analytical tool with the ability to store, retrieve, and graphically display any dataset. Information within the system is readily accessible and easily manipulated for tailor-made results. Investing time and effort in the beginning to establish a comprehensive GIS allows one to efficiently produce results for current permit monitoring requirements and also allows room for expansion as new regulations become effective. While sample analysis and collection protocols show little growth, the demands placed on those that impact, regulate, and evaluate water quality will continue to change with environmental demand-driven permit requirements. Incorporating GIS now establishes an efficient way to meet current reporting demands and provides the capacity to expand with future challenges.
-Adam Sukenick
Introducing Fluvial Geomorphology
Not long ago, fluvial geomorphology was an obscure discipline relegated mainly to collegiate research. Though great strides have been made by numerous researchers in our understanding of fluvial geomorphology, it is only recently that concepts of geomorphology have been applied. One of the major proponents of applied geomorphology is David Rosgen. His book, Applied River Morphology, is the bible of the discipline. In this book, Rosgen brings together and synthesizes vast amounts of research into a cohesive and comprehensive whole. He also standardizes and presents terms and nomenclature for expressing concepts that relate to fluvial geomorphology. This effort is not unlike what Eugene Odom did for the field of ecology, and the ramifications may be nearly as important. Needless to say, one cannot discuss the discipline of modern fluvial geomorphology with out evoking the name of Dave Rosgen. It is often said that all philosophy is a footnote to Plato; so too could it be said that all applied river geomorphology is a footnote to Rosgen.
The study of fluvial geomorphology unites the disciplines of geology, ecology, and hydrology. The principles of geomorphology were developed through persistent and meticulous study of the interactions between flowing waters and the land over which they flow. Applied geomorphology utilizes these principles developed by researchers to quantitatively evaluate how physical attributes of a stream impact the stream ecology and how the stream interacts with the physical environment. In Applied River Morphology, Rosgen takes much of the historical information on stream geomorphology along with his own research and unites them into a well-organized, systematic methodology for stream channel evaluation. Channels are initially classified and delineated by their most salient physical features. These features include bankfull, valley slope, sinuosity, belt width, slope, width-depth ratio, and entrenchment. Rosgen uses these morphological characteristics to establish a channel typing system that is very useful in understanding the major forces sculpting a given channel. The typing system uses four levels of resolution, with more specificity and parameters as one goes from one level to the next. For details on the Rosgen method of evaluating stream geomorphology (including definitions of terms), see Applied River Morphology.
The fundamental unit of most stream studies is the study reach. The study reach is a known length of stream segment where most of the measurements and observations will be made. Study reaches can be chosen for many reasons including establishing reference reaches, studying land use change, or as part of a watershed assessment. It is important to map the watershed upstream of the study reach and to gather information on land use and activities in the watershed. After establishing the study reach, one is ready to begin taking measurements. The first feature one should determine is bankfull. Bankfull is the most important feature in assessing channel morphology. The bankfull feature is created by a hydraulic event and is described as “the incipient elevation on the bank where flooding begins” (Rosgen 1996).
The bankfull event is now associated with the concept of competence, or the ability of the stream to transport sediment. This is an extremely important concept when assessing the impact of urbanization on stream channel morphology. In an undisturbed watershed, the largest amount of total sediment transport is accomplished by bankfull hydrologic events that occur roughly once every 1.5 years. These events are responsible for sculpting the channel. In an undisturbed state, most natural channels maintain equilibrium by aggrading and degrading at constant rates so that the forces of sedimentation and erosion balance each other. There is clear physical evidence in most undisturbed channels as to where these flows are transforming the channel. Typical features include flood planes, the tops of point bars, areas of wrested vegetation, and sedimentation. These features are used to determine bankfull. In an urbanizing watershed, however, bankfull can be a very problematic indicator. This is because the physical signs of bankfull may lose their consistency as the hydraulic regime of a stream begins to change in response to changes in land use. The well-documented changes in peak flows caused by changes in land use result in changes in point bars and wrested vegetation as well as erosion and deposition. During this state of flux, it may be all but impossible to determine bankfull. Field personnel must draw on their experience, and it is best to have two or more participate in the evaluation so they can discuss in depth the physical observations at the site.
Cross-sectional area is one of the fundamental dimensional measurements of geomorphology. A cross-sectional area is a slice of stream channel taken across the width of the channel. These measurements should be taken at a transect (a line drawn perpendicular to the flow of the stream from one bank to the opposite bank) located along the study reach. (The number and location of transects depends on the scope and purpose of the study.) To measure cross-sectional area, one needs a consistent, level line across the transect to be used as a reference against which changes in elevation can be noted (usually with a measuring rod) as one proceeds across the transect starting on one bank and ending on the opposite bank. Features such as bankfull, edge of water, and depth of water should be noted at each transection of the stream. Changes in bankfull lines and fresh scouring can be measured by stable physical markers like bank pins and scour chains driven into areas of the channel outside the cross section. Following the completion of the cross-section measurements, substrate studies are normally conducted at the transect using a pebble-count method. There are several competing methodologies, but all attempt to quantify the particle size of streambed material.
Mapping can also provide important information at a study site. The first measurement taken during mapping is channel length, as measured by the linear length of the thalweg, or deepest line of the channel. A tape is run the length of the thalweg, and lateral measurements are taken at specific intervals at right angles to the tape toward the banks. Features such as bankfull and edge of bank are noted. Other features such as point bars, riffles, and pools are also measured, providing detailed information on important physical features of the reach subject to change over time. Other important features, such as belt width and sinuosity, can also be established. Both measurements track the deviation of the stream channel from a straight line between two points on a channel. Sinuosity is the net difference in length between the meandering stream channel and a straight line between two points at each end of–usually–the study reach. Belt width is determined by the amplitude of the meanders. This information, combined with the transects, can then be compiled and used to plot a plan view map showing meanders, stream features, bed materials, and variations in the dimensions of the channel.
Geomorphology studies have a variety of uses, ranging from quick evaluations of stream condition to detailed long-term studies designed to trace changes over time. Geomorphology is very intertwined with the habitat component of biological assessment. The habitat score for the metrics used in biological assessment is in large part determined by geomorphological observations. Examples include particle size embeddedness, undercut banks, and bank stability. Bank stability is an indication of the capacity of streambanks to resist erosive forces. (Rosgen’s notes list bank height, bank angle, density of roots, soil stratification, and particle size as erodibility factors.) Many habitat assessment protocols address erodibility by addressing bank stability. Bank stability is defined as a measure of the potential for detachment of soil from the upper and lower streambanks. Habitat assessment protocols evaluate the bank’s erosive potential by examining salient signs of erosion along with bank slope. Steep slopes are considered more unstable than gradual slopes. Banks are rated inversely to percent slope and also to the amount of observable erosion. Other factors include soil stratification, particle size, and vegetative cover. Stratification increases erodibility; this is especially true when fine-grained material rests upon coarse material. Erosive potential also increases initially with particle size; for example, the stronger cohesive forces of consolidated clay particles resist erosion more than sand. This does not hold true as size increases further, such as when sand particles are compared to cobble and boulders.
Geomorphology studies are also important in tracking changes over time within a given watershed. This analysis can provide important information on the effectiveness of best management practices and land use ordinances designed to protect stream channels from the potentially negative hydrologic impacts of land use change, including flooding and erosion. Factors observed in geomorphology studies such as sediment transport, bank erosion, undercutting and deposition, width depth ratio, and sinuosity can all be very important indicators for hydraulic impact to a stream. In instances where it is desirable to restore or mitigate damage to eroded stream channels and damaged habitat, geomorphology techniques are essential. Without understanding the nature of the forces that caused the problem in the first place, success in mitigating these problems would be unlikely.
Traditional monitoring efforts have centered on the chemical–and, later, the biological–components of our flowing waters to assess water quality and stream health. The addition of stream geomorphology has greatly broadened our perspective. Geomorphology has played a central role in uniting all the disciplines involved in studying streams and rivers into a comprehensive whole. Geomorphology can help planners assess all aspects of flowing waters, including stream ecological health, erosion, and flooding. With these powerful tools available, it cannot be argued that we do not possess the means of both evaluating and mitigating the impact of our land use decisions on our streams and rivers, and that their quality is the direct result of our choices.
-Bob Bourne
Riparian Stream Buffers
Stream corridors are an important part of the terrestrial environment containing their own unique ecology. Plants and animals have historically migrated along the corridors because of the availability of water, the ease of movement in flatter terrain, and the moderating influence of water on air temperature. The interface between aquatic and terrestrial habitats produces many opportunities for plants and animals by providing a rich and diverse environment. The terrestrial region near a stream can have a profound impact on the quality of the stream itself. It is important that this region remain undisturbed and vegetated with a diversity of native plants in order to maintain both good water quality and healthy stream ecology.
Vegetation performs many essential functions in the riparian zone and is important to the stream environment even far away from the riparian zone. Woody shrubs and trees, though a biological component of the environment, have a profound effect on the physical integrity of the stream channel. For this reason, riparian vegetation is included in most habitat assessment protocols. The habitat assessment evaluates plants in terms of both vegetative cover on streambanks and riparian corridor width in order to evaluate streambank stability as well as potential for sedimentation, erosion, and water-quality-degrading runoff. Higher scores are given for woody plants such as trees and shrubs, and lower scores are given for grasses. This is because woody plants possess more substantial roots that are more effective at holding and stabilizing soil and provide more habitat and shading than do grasses. Most habitat assessments give lower scores when there are breaks in the natural vegetative zone, especially when the zone has been impacted by human activity.
As discussed, vegetation is essential in protecting the physical integrity of the streambanks and preventing erosion in the riparian zone. The root systems of plants play the most important role, with larger roots providing structural support and bracing and smaller roots providing a mesh, which holds soils and resists erosion. Exposed roots and woody debris reduce stream flow velocities by increasing the roughness of the banks and the streambed. Smaller plants, including grasses, mosses, and ferns, also help hold and cover soil, especially in lower regions of the banks. Ground cover, including leaf litter and humus stabilized by root mats, protects the soil in the upper flood plain and terraces, preventing sediment runoff into the stream from stormwater and flooding. All of these elements together–trees, shrubs, smaller plants, and natural ground cover–support the physical integrity of the streambank.
Vegetation reduces overland stormwater flows by increasing the absorptive efficiency of the soil. Leaves, whether on plants or on the ground, interrupt the kinetic force of raindrops before they can reach the soil. Leaf litter and humus act as a sponge, absorbing and holding moisture, then slowly releasing it to the soil. Roots penetrate deep into the soil, aid in mixing of the organic and inorganic fractions of the soil, and contribute organic matter when they die and rot. Burrowing forest animals also mix the organic and inorganic components of the soil. As a result, soils in the forest can more readily absorb moisture, resulting in a larger portion of the rainwater being stored as groundwater. Soils that have developed under hardwood forests usually have infiltration rates exceeding the annual maximum rate of rainfall. Groundwater migrates through the soil to low-lying areas, where it is discharged into a stream. Water entering a stream in this manner is rarely turbid, due to the low velocity of the water and the stability of the soil through which it passes. Because the water takes longer to reach the stream and is discharged over a long time, the stream is less impacted by rainfall and associated storm surges, mitigating the impact of peak flow events and thus reducing erosion.
Vegetated buffers also shade the stream and keep the water cool during hot periods. This, in turn, maintains higher and more stable dissolved oxygen levels. DO levels are crucial to stream organisms during the warm summer months. Shade also helps prevent excessive algal growth during warmer weather. Algal growth, as quantified in the general literature as chlorophyll a, is greatest in late winter and in the early spring, pre-leaf-out period, when temperatures have moderated and the leaf canopy is yet to be established. Although algal growth is documented as a natural phenomenon, it can be exacerbated in urban areas by application of lawn nutrients.
Information on species diversity, density, and age provides extremely important information about past land use and ecological health of the stream. In a mature, healthy piedmont forest, one would expect an upper canopy with larger trees of native lowland varieties and a lower canopy of understory trees, shrubs, and grasses. Beech, birch, maple, tulip, ash, hackberry, and elm are all desirable streambank trees. Hornbeams, alder bush, button bush, deciduous holly, swamp dogwood, elderberry, and native cane are desirable understory plants. These plants not only hold the soil well, but their leaves also provide a good food source for macroinvertebrates. Protocols exist to determine tree age, diversity, and density based on taxonomic keys and quantified field observations.
Riparian plants overhanging and entering the stream channel provide important habitat and are especially important as a source of cover and as a food source for macroinvertebrates and fish. As a result, plant species in the riparian zone may have an impact on macroinvertebrate and fish species diversity. Information on riparian condition and plant species can provide important insight when evaluating aquatic biodiversity. This is especially true in headwater streams. (There have been numerous studies published on riparian plant species and macroinvertebrate diversity, including several studies by the US Army Corps of Engineers and the University of Georgia.) Plant species composition can also provide information on the age of the forest. The presence of younger trees and early successional species, such as loblolly pine, sweet gum, and tulip, indicate more recent clearing of the banks. These past activities in the riparian buffer may have long-lasting effects on both fish and macroinvertebrate communities.
Exotic Plant Species. Many habitat assessment protocols give higher scores to native species than to non-native species. This is because some non-native species are able to outgrow and out-reproduce native species and thus can dominate the flora of some areas. By crowding out native plants, these invasives can alter the ecology of an area negatively impacting the biota dependence on the presence of native species. Thus, when non-native species dominate a buffer region, it can have a negative impact on the stream biota. Non-native plants such as kudzu or Chinese privet can inhibit the growth of native plants that are superior food sources and streambank stabilizers. Privet is shade tolerant and can survive high soil moisture as long as there is some drainage. As a result, it thrives on floodplains even beneath an established canopy and can form dense stands at the expense of native understory trees and shrubs. Kudzu will smother out all competing vegetation, and, as its root structure is weak, it offers little erosion protection during winter months when the plant dies back.
Large monocultures of non-native grasses are often planted along and near streams. Although these plants are usually not hardy enough to significantly naturalize or become aggressive pests, they may result in the exclusion of native species where they are aided and maintained by humans. These grasses, such as fescue, were cultivated in Europe and are adapted to a cool, moist climate with fairly rich soils. As a result, they have shallow roots and require extensive watering during dry periods, and they require chemical maintenance, especially when planted on graded fill dirt consisting mainly of clay with minimal topsoil. The hard clays beneath these grasses absorb very little water, and the roots are too shallow and weak to greatly enhance absorption. As a result, runoff containing lawn chemicals is a strong possibility when ornamental grass lawns descend into the riparian corridor. The shallow roots also do not secure soils nearly as well as native trees shrubs and grasses.
Stream buffers not only help to protect and maintain stream quality but also provide an important habitat for many aquatic and terrestrial organisms as well as an amenity for humans. The presence of streamside vegetation and stream buffers is not only aesthetically pleasing, but also vital to the stream and to all living things that depend on the stream for their survival.
-Bob Bourne
References
CASQA. 2008. An Introduction to Stormwater Program Effectiveness. California Stormwater Quality Association, Menlo Park, CA. June 2008.
Rosgen, David L. 1996. Applied River Morphology. Printed Media Companies, Minneapolis, MN.
