Handbook for Predicting Stream Meander Migration and Supporting Software (2004)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

7.1 INTRODUCTION This chapter defines the steps necessary to conduct a detailed meander migration analysis. The first method (Sec- tion 7.2) describes a manual overlay of historic maps and aer- ial photos for comparative purposes and applies the acquired data and information to predict the position of a bend in the future. The next method (Section 7.3) describes how certain types of common computer software can be used to assist in the historic assessment and prediction of bend migration. The last method (Section 7.4) describes the procedures necessary to use the Data Logger and Channel Migration Predictor soft- ware supplied with the Handbook to conduct the comparison and prediction steps. Care should be taken when using the following methodolo- gies in areas where geologic features, structures, erosion con- trol measures, and other “hardening” features may influence the characteristics of meander migration (see Section 6.4). 7.2 MANUAL OVERLAY AND PREDICTION The following steps provide direction in conducting a sim- ple overlay comparison of historic banklines and making pre- dictions on the potential future position of a bend based on the past channel migration characteristics. Step 1. The first step in conducting a meander migration analysis using an overlay technique is to obtain aerial photo- graphs and maps for the study area, as described in Section 4.4. Appendix A also provides general instructions on down- loading digital aerial photographs and topographic maps from Microsoft’s TerraServer Web site. Step 2. The maps and photos must be enlarged or reduced to a common working scale, as described in Section 5.2. The scale of the most recent map or photo should be used because it will be the basis for comparing historical meander pattern changes and predicting the position of a given bend in the future. Step 3. After a working scale is defined, the photos and maps are registered to a common base map or photo by iden- tifying several features or points that are common to each photo/map being compared. The registration points of a map or photo do not need to be common to all the maps and pho- tos, only to the subsequent map or photo to which it is being compared, because comparisons can be performed in pairs. For example, Figures 7.1 and 7.2 show the 1937 and 1966 aerial photos, respectively, for a reach of the White River in Indiana. Four registration points have been identified on the 1937 photo that are also on the 1966 photo. Two registration points are road intersections and two are isolated vegetation (trees or large shrubs). Registration points that bracket the site on both sides of the stream and at both ends of the reach are most useful because they reduce the amount of potential error within the bracketed area. Intermediate points between the end points are helpful in accurately registering the middle sec- tions of the reach. More than five or six registration points can make registration difficult because of the difficulty in align- ing all the registration points among the various aerial photos and maps used. However, there will be cases in which there will be very few identifiable common registration points, and these cases may have the potential for significant error. Step 4. After the registration points are identified, banklines and registration points for each year are traced from the aerial photo onto a transparent overlay. The method for identifying and tracing the banklines is described in Chapter 5 and Appen- dix B. Registration points are included on the overlay so that they can be easily plotted onto other aerial photos or maps for comparative purposes. The traced banklines and registration points of the White River for 1937 are plotted on the 1966 aer- ial photo in Figure 7.3 for comparative purposes. Because most meander bends are not simple loops, the loop classification of Brice (1975) can be used to characterize the shape of each bend that is to be analyzed (see Figure 7.4). Meander bends seldom form single symmetrical loops, but instead are composed of one or more arcs combined to form either symmetrical or asymmetrical loops. Brice (1975) derived the classification scheme for meander loops from a study of the meandering patterns of 125 alluvial streams. The scheme consists of four main categories of loops (simple, compound, symmetrical, and asymmetrical) comprising 16 form types. Although compound loops are regarded as aber- rant forms of indefinite radius and length, the meandering pat- terns can be divided into simple loops whose properties can be described, measured, and analyzed. The radius of curvature of most bends can be defined by fitting one or more circles or arcs to the bend centerline or outer bankline of a meander loop. The number of loops that describe a bend does not always remain constant as a bend evolves. For example, a bend at Time 1 may be defined by two loops, but as it evolves to Time 2 or Time 3, the bend may become a single loop. In this case, the bends from Time 1 and Time 2 should also be delineated by a single loop when conducting the meander migration CHAPTER 7 METHODOLOGY 30

31 analysis. The opposite case may also occur, in which the final bend is a double-headed bend (Figure 7.4-H, -I, -M, -N, and -O). In this case, an attempt should be made to delineate the Time 1 and Time 2 bends by two loops in similar positions as the Time 3 loops. Step 5. Once the banklines for each of the historic aerial photos have been traced, circles are best-fit to the outer bank of each bend to define the average bankline arc, the radius of curvature (RC) of the bend, and the bend centroid position (see Figure 7.5). The number of circles required to define the bend is based on the loop classification described above and shown in Figure 7.4. A detailed description of the method used to fit a circle to the outer bankline of a meander bend is provided in Appendix B. The radius of curvature and cen- troid position of the circle used to describe the bend will be Figure 7.1. Aerial photograph of a site on the White River in Indiana showing four registration points (circles designated a through d) common to the 1966 aerial photo. Figure 7.2. Aerial photo of a site on the White River in Indiana showing the four registration points common to the 1937 aerial photo in Figure 7.1. Figure 7.3. The 1966 aerial photo of the White River in Indiana with the 1937 bankline tracing and registration points. Levee Flow

32 and 1966. The vector arrow at each bend shows the direction and magnitude of movement of the bend centroid between 1937 and 1966. For each bend, this vector may be resolved into cross- and down-valley components to determine the rates of meander migration. The change in radius of curva- ture of each bend is defined by the difference between the magnitudes of the vectors for 1937 and 1966. Step 6. The position of the bend at a selected date in the future can be predicted by simple extrapolation if it is assumed that the bend will continue to move at the same rate and in approximately the same direction as it has in the past. To esti- mate the position of a bend centroid in 1998, for example, the distance the centroid would be expected to move during the 32 years between 1966 and 1998 can be determined by multi- plying the annual rate of movement for the 1937-to-1966 period by 32. This distance is plotted along a line starting at the 1966 centroid point and extending in the direction defined by the 1937-to-1966 migration vector. The radius of curvature of the bend in 1998 can be defined by determining the rate of change of the bend radius from 1937 to 1966 as a percentage relative to the 1966 radius and multiplying this value by num- ber of years from 1966 to 1998. A circle with that radius, cen- tered on the predicted location of the centroid, is plotted on the tracing to indicate the expected location and radius of the bend in 1998. Figure 7.7 shows the expected outer bank circles for each of the bends of the White River in 1998, based on simple extra- polation of the rates and directions of change from 1937 to 1966. Banklines for the 1998 channel can then be constructed on the tracing by joining the outer bank circles through inter- polation, with the 1937 and 1966 banklines used to indicate the reach-scale configuration of the channel. Figure 7.8 shows the 1966 aerial photo overlaid with the banklines observed in 1937 and estimated for 1998. Inspection of the estimated banklines reveals that Bend 1 would encroach into the levee to the north by 1998, while growth of Bend 5 would likely cut off Bends 6 and 7. Figure 7.4. Meander loop evolution and classification scheme proposed by Brice (1975). Flow is left to right. Figure 7.5. Inscribed circles that define the average outer banklines from the 1937 aerial photo of the White River site in Indiana. Also shown are the bend centroids and the radius of curvature (RC) for one of the bends. SOURCE: Brice, 1975 1937 RC used to make a comparison with the bend measurements of previous and subsequent years. These measurements can then be used to determine migration rates and direction and estimate future bend migration characteristics. Figure 7.6 compares the best-fit circles and bend centroids for each bend traced from the aerial photographs for 1937

33 Figure 7.6. Depiction of the bends from the 1937 and 1966 outer banklines as defined by best-fit circles. The movement of the bend centroids (arrows) defines migration of the bends. Figure 7.7. Depiction of the bends from the 1937 and 1966 outer banklines, as defined by best-fit circles, and the predicted location and radius of the 1998 outer bankline circle. Figure 7.8. Aerial photo of the White River in 1966 showing the actual 1937 banklines (white) and the predicted 1998 bankline positions (black). 1937 1966 1937 1966 1998 (estimated) Flow 1 2 3 4 5 6 7

In Figure 7.9, the banklines predicted for 1998 by extra- polation of trends of change between 1937 and 1966 are superimposed on an aerial photograph taken in 1998. Two of the registration points used for this comparison are dif- ferent because two of the original registration points from the previous aerial photos are no longer present on the 1998 aerial photo. Comparison of the actual and estimated banklines illus- trates that meander migration can be predicted relatively accurately using this simple approach. For example, the posi- tions of Bends 3 and 4 and the cutoff at Bend 5 are accurately predicted. Errors in the predicted banklines can be accounted for by (1) an artificial cutoff that affected Bends 1 and 2, (2) the natural cutoff at Bend 5 that led to Bends 6 and 7 being abandoned, and (3) construction of bank protection at Bends 3 and 5 during the period of 1966 to 1995. The artificial cut- off at Bend 1 may have been in response to the serious threat posed by bend migration toward the nearby levee. That cut- off caused Bend 2 to distort in a way that could not have been predicted from its previous behavior. Outer bank migration at Bends 3 and 5 appears to have been curtailed by bank revet- ments. The migration of Bends 3, 4, and 5, the cutoff of Bend 5, and the abandonment of Bends 6 and 7 were predicted rea- sonably well. It is likely that the positions of Bends 1 and 2, as well as the banklines in the revetted portions of Bends 3 and 4, would have been as predicted except for these engi- neering interventions. The case study of the White River used a single period (1937 to 1966) to predict the position of the banklines in 1998. To improve the reliability and accuracy of predictions, it may be desirable to use multiple pairs of aerial photographs to generate more than one period of analysis. Through evaluat- ing multiple periods, meander migration analysis can detect trends of change in the rate and direction of bend migration as well as time-averaged values. For example, the channel positions and circles inscribed on the outer banklines for a hypothetical channel in Years 1, 2, and 3 are shown in Figure 7.10. To predict the amount and direction of migration of a bend at some future date, the 34 Figure 7.9. Aerial photograph of the White River site in Indiana in 1998 comparing the predicted bankline positions with the actual banklines. 1 2 3 4 5 6 7 Levee amount and direction of migration for previous periods must first be determined. Figure 7.11 shows the best-fit circles for the outer bank positions in Years 1, 2, and 3; the outer bank radius of curva- ture for each year (RC1, RC2, and RC3); the angle defined by the change in direction of the bend centroid for each time period (θA and θB); and the amount of migration of the bend centroid (DA and DB). The subscripts A and B refer to Period A (Year 1 to Year 2) and Period B (Year 2 to Year 3), respectively. The aim is to predict the bankline position in Year 4 at the end of Period C (Year 3 to Year 4). The rate of change of the radius of curvature for the outer bank during Period A (Year 1 to Year 2) is defined by where ∆RCA = Rate of change in radius of curvature during Period A (ft/yr or m/yr), RC1 = Radius of curvature of outer bank in Year 1 (ft or m), RC2 = Radius of curvature of outer bank in Year 2 (ft or m), and YA = Number of years in Period A. The rate of change of the radius of curvature for the outer bank during Period B (Year 2 to Year 3) is defined by where ∆RCB = Rate of change in radius of curvature during Period B (ft/yr or m/yr), RC2 = Radius of curvature of outer bank in Year 2 (ft or m), RC3 = Radius of curvature of outer bank in Year 3 (ft or m), and YB = Number of years in Period B. To predict the rate of change of the radius of curvature for the outer bank during Period C (Year 3 to Year 4), it is neces- ∆R R R Y 7CB C3 C2 B= −( ) ( . )2 ∆R R R Y 7CA C2 C1 A= −( ) ( . )1

35 alous rate. Because of problems inherent in using a long-term average rate of change, it is recommended that the most recent rate, the rate of change for the previous period (Period B in this case), be used because it is more closely related to existing conditions and processes. Hence, the predicted radius of cur- vature of the outer bank in Year 4 is defined by where RC4 = Predicted radius of curvature in Year 4 (ft or m), RC3 = Radius of curvature of outer bank in Year 3 (ft or m), RC2 = Radius of curvature of outer bank in Year 2 (ft or m), YB = Number of years in Period B, and YC = Number of years in Period C. Judgment should also be used to evaluate the reason- ableness of the predicted radius; the predicted radius should not be significantly smaller than the average radius of small bends within the reach. Another method of determining whether the predicted radius is reasonable is to determine the ratio of the predicted bend radius to the average chan- nel width of the reach. If the ratio falls below 2, it is likely that the bend may cut off or become distorted prior to reach- ing Year 4. The angle of migration of the centroid (θ) of the circle that inscribes the outer bank and the amount of migration of the centroid (D) for each period are shown in Figure 7.11. There are two methods of defining the angle of bend migration for Period C. The first is to simply use the same migration angle for Period B (e.g., θC = θB). Using the migration angle from the previous period (i.e., Period B) is recommended where geomorphic and hydrologic conditions have not changed significantly. The second method uses the rate of change of the migration angle from the previous period to define the rate of change for the period being predicted such that: where θC = Predicted angle of outer bank migration for Period C, θA = Angle of outer bank migration for Period A, θB = Angle of outer bank migration for Period B, YB = Number of years in Period B, and YC = Number of years in Period C. The second method can be used where geomorphic, hydro- logic, and hydraulic conditions have changed significantly. It is recommended that both methods be used and compared for reasonableness. Again, judgment should be used to eval- uate the reasonableness of the predicted migration angle with respect to the general alignment of the bends or reaches immediately upstream and downstream. This is particu- larly important in reaches that may contain features such θ θ θ θC B A B C BY Y 7= −( )  ( )   + ( . )4 R R R R Y Y 7C4 C3 C3 C2 B C= + −( ) ( )   ( )   ( . )3 Figure 7.10. Banklines and circles drawn along outer bankline positions for a hypothetical channel in 3 different years. YEAR 1 YEAR 2 YEAR 3 Figure 7.11. Diagram defining the outer bank radius of curvature in Years 1, 2, and 3 (RC1, RC2, and RC3) and the amount (DA and DB) and direction (θA and θB) of migration of the bend centroid during Periods A and B. RC1 RC2 RC3 YEAR 1 YEAR 2 YEAR 3 BDA DB sary to decide if the long-term average or the most recent rate will be used. A long-term average rate of change of the radius of curvature is based on the entire period of record, which may be influenced by changes in land use, basin hydrology, or channel modifications. In addition, periods of bend expansion or contraction within the period of record can yield an anom-

as revetment, bridges, and levees that can have a significant influence on meander migration. The amount of migration of the bend centroid for Period A is DA, and for Period B it is DB. The rates of migration dur- ing Periods A and B are determined by dividing DA and DB by the number of years in the associated period. To predict the magnitude of centroid migration during Period C, it is more accurate to use the most recent Period B rate of centroid shift, which yields the following relationship: where DC = Magnitude of centroid migration for Period C (ft or m), DB = Magnitude of centroid migration for Period B (ft or m), YB = Number of years in Period B, and YC = Number of years in Period C. The centroid, radius of curvature, and predicted position of the circle that inscribes the outer bank in Year 4 are shown in Figure 7.12. Figure 7.13 shows an interpretation of the pre- dicted banklines for Year 4 based on the average channel width of the bend from previous periods. Using a CAD software package will serve to reduce the number of steps in this procedure and also provide more accurate answers for the prediction. The use of a CAD soft- ware package involves only two steps, and no change in scale takes place. The user can import a digital image into the CAD program with a given scale; trace the stream banklines; and delineate various features, such as the bend centroid location and the bend radius of curvature. From this CAD file, the user can import the traced banklines into the meander migration D D YCC B BY 7= ( ) ( . )5 36 prediction routine described in the next section, measure the various features required by the routine, and use the mea- sured data to predict the location of the bend at some point in the future. 7.3 COMPUTER-ASSISTED METHODOLOGY The methods described in the previous section can be con- ducted more easily and efficiently using common computer software such as Microsoft PowerPoint and Jasc’s Paint Shop Pro or more powerful and versatile programs like Autodesk’s AutoCAD and Bentley’s MicroStation. The photo-editing capabilities of Microsoft Word and PowerPoint were used to perform the comparison and pre- diction in the previous examples. In Word and PowerPoint, an aerial photo in the form of a JPEG or TIFF file, for exam- ple, can be imported into the software using the “Insert> Picture>From File” commands on the main menu. Right clicking on the main menu area pulls up a list of available toolbars, including the “Picture” and “Drawing” toolbars. The “Picture” toolbar is used to change the attributes of the picture, and the “Drawing” toolbar allows the user to mark the registration points and trace a line for the bank- lines using the “Autoshapes>Lines” or “Autoshapes>Basic Shapes” commands. Once the registration points and banklines for each aerial photo have been drawn, they should be grouped using the “Grouping” tool. This enables the user to move all the items together to another picture for comparative purposes. After Figure 7.12. Predicted position and radius of curvature of the circle that defines the outer bank of the hypothetical channel in Year 4. Figure 7.13. Predicted bankline position of the hypothetical channel in Year 4. YEAR 1 YEAR 2 YEAR 3 YEAR 4 (Predicted) θ CDC RC4 YEAR 1 YEAR 2 YEAR 3 YEAR 4 (Predicted)

37 the group of registration points and banklines associated with one aerial photo has been superimposed onto the second aer- ial photo, the group can be correctly registered on that photo by resizing and rotating the group using the “Format Picture> Size” and “Format Picture>Rotate” tools. Right clicking on the picture accesses these tools. Although the scale of the aerial photos is not provided by some graphics-editing software, the scale can be found by comparing distances between several points on the photos with measurements between the same points on a map or photo with a known scale. This allows use of the prediction portion of the comparison technique described in the previ- ous section by defining the change in bend radius of curva- ture and the direction of migration of the bend centroid. The user can then derive and plot the predicted radius and posi- tion of the best-fit circle (Figure 7.7) and interpret the future bankline position using the shape of the channel in previous photos (Figure 7.8). The prediction of the 1998 banklines for the White River shown in Figure 7.8 and the analysis leading up to the prediction were completed using the Microsoft PowerPoint tools described above. 7.4 DATA LOGGER AND CHANNEL MIGRATION PREDICTOR The Data Logger and Channel Migration Predictor are GIS- based, menu-driven ArcView extensions. These extensions were developed to assist in the measurement and analysis of bend migration data and aid in predicting channel migration (see Section 5.4 for basic description). The Data Logger and Channel Migration Predictor soft- ware are included on CRP-CD-48, provided at the back of the Handbook, and instructions for installing the Data Logger and Channel Migration Predictor are provided in Appendix C. Tips for delineating and registering banklines from historic aerial photos that are not georeferenced for use with the Channel Migration Predictor are provided in Appendix D. 7.4.1 Data Logger Bankline File Requirements Users of ArcView can choose from a variety of file types for importing visual representations of banklines into the Data Logger. Supported CAD file formats include MicroStation design (.dgn) files and two kinds of AutoCAD drawing files (.dwg and .dxf). ArcView supports 16 image file formats, including JPEG and TIFF. Digital elevation maps (DEMs) and Triangular Irregular Network (TIN) representations of terrain data can also be used to represent river banklines. 7.4.2 Data Logger Users Guide Prior to starting the Data Logger, the historic banklines for a given study site need to be digitized and saved. The set of banklines is defined here as a theme, which is just a set of geographic features in a view. Step 1. Start ArcView and add a theme containing (either two or three) historical channel banklines for a selected reach. Figure 7.14 is an example of a theme with two historical bankline records. Step 2. Clicking the green diamond, located at the top right of the ArcView button panel (Figure 7.14), launches the “Reach Data” dialog box (see Figure 7.15) of the Data Logger. Figure 7.14. ArcView project with a theme showing two historical bankline records. Flow

38 Figure 7.15. “Reach Data” dialog box of the Data Logger. Figure 7.16. Themes to store the bend registration points have been added to the ArcView project. Flow (“BigBlack” in Figure 7.14). Also the dates associated with historical banklines must be entered in the “m/d/yyyy” form illustrated in Figure 7.15. The Channel Migration Predictor extension will use these date objects to calculate migration rates and extrapolate bankline positions. After filling in the various text boxes in the “Reach Data” dialog box, the user then clicks on the “Create Data File” but- ton. The “Create Data File” button launches three ArcView Avenue scripts. The first script defines one theme for each year and each bend for the archiving of bankline registration points. This script also adds these bend-year registration point themes to the ArcView project shown in Figure 7.14. Figure 7.16 shows the ArcView project with the six new themes added to the view “BigBlack.” At this point, the user provides information about the particu- lar reach or sub-reach under consideration. The “Reach Data” dialog box enables the user to quickly enter information about a particular reach or sub-reach. For convenience, a reach with a large number of bends can be broken into sub-reaches to facilitate a modular approach to data archiving. For example, as indicated in Figure 7.14, the first sub-reach has been chosen to contain three bends. If the next sub-reach has five bends, then the next time the Reach Data dialog box is displayed, the user should enter 5 for the “Number of Bends” and 4 for the “First Bend Number.” It is important that the name entered in “View File Name” form (“BigBlack” in Figure 7.15) be an exact match of the name of the view containing the historical bankline contours

39 The second script prompts the user for a file name in which the Virtual Table (VTab) for this sub-reach is to be stored (see Figure 7.17). This script also formats the VTab to have one record per bend and fields in each record for storage of the data for each bend and each year. Finally, the second script creates a table for viewing the VTab during the data logging procedure. Figure 7.18 shows the initial contents of the VTab. Additional fields will be filled in the next phase of the data logging process. The third script launches the “Data Manager” dialog box, shown in Figure 7.19. This script also disables the “Create Data File” button on the “Reach Data” dialog box so the user cannot add duplicate bend-year themes to the view containing the bankline records for this sub-reach. The user can reactivate the “Create Data File” button by clicking the green diamond button in the ArcView button panel. Finally, the third script minimizes the “Reach Data” dialog box and places it at the bottom of the screen. For the remain- der of the data collection and archiving, the user interacts with the “Data Manager” dialog box. The “Data Manager” dialog box provides an efficient method for collecting and archiving the data necessary to predict channel migration. Step 3. This step describes how to locate delineation points along the outer bank of a selected bend. To begin, the user selects the “Enter Bend Delineation Points” tool at the top of the “Data Manager” dialog box. With this tool selected, the mouse pointer displays as crosshairs when over the cur- rent view. The user can move the crosshairs to a desired delin- eation point and then left-click to position the point on the bankline. Figure 7.20 shows seven delineation points on the outer bank of a bend. (Note: This procedure may also be used to define the channel centerline at a bend if the user is track- ing meander migration using centerlines.) If several of the bankline delineation points have been positioned incorrectly, the “Delete Current Bend Points” but- ton provides the user with the ability to clear all the bankline delineation points and start over. If only a few bankline delin- eation points appear incorrectly positioned, the “Select Bend Point(s) to Delete” tool enables the user to select one or more points (by holding down Shift key) to delete (by pressing the Delete key after the selection is complete). When the user is satisfied with the placement of the bank- line delineation points on a bend, the fitting of a circle to the bend is performed by pressing the “Fit Circle to Bend” but- ton on the “Data Manager” dialog box. This button runs an Avenue script that determines the center and radius of the cir- cle that best fits (using a least-squares approximation) the bankline delineation points the user has selected on the bend. The script also determines the orientation of the bend as the angle of the line joining the center of the circle to the mid- point of the bend. This angle is measured counterclockwise from a zero angle defined to be due east (Cartesian coordi- nate system). Figure 7.21 shows the fitted circle and the Figure 7.17. File dialog box for choosing a name for the database. Figure 7.18. Initial contents of the database.

40 • Upstream channel width (U-U), • Bend apex channel width (A-A), • Downstream channel width (D-D), • Wavelength of the bend (line between U-U and D-D), and • Amplitude of the bend. The upstream (U-U) and downstream (D-D) channel widths are measured perpendicular to flow from bankline to bankline at the crossing point between the measured bend and the next upstream or downstream bend. The width of the bend apex (A-A) is measured perpendicular to flow from bankline to bankline at the farthest outward extension of the bend. Wave- length is measured by placing a line between the middle of the channel at the upstream width measurement (U-U) to the middle of the channel at the downstream width measurement (D-D). The wavelength is calculated by the program by dou- bling the length of this line. The amplitude is determined by placing a perpendicular line from the wavelength line to the outer bank of the bend apex where the width (A-A) was mea- sured. The amplitude is calculated by the program by doubling the length of this line. A pull-down list dialog box (see Figure 7.23) is provided to assist the user in making and recording these five distance measurements. After making each distance measurement, the user is informed of the distance measured and asked to verify that it is acceptable for later entry in the database. Figure 7.24 shows an example in which the user is asked to verify a mea- surement of downstream channel width. Before moving to a new bend or a new historical record, the data for the current bend must be written to the database for this sub-reach. Clicking on the “Archive Bend Data” but- ton (see Figure 7.19) will prompt the user to confirm the bend number and the year number. After confirmation, a script writes the following information to the database: • Coordinates of the center of the circle, • Radius of the circle, Figure 7.20. Bankline delineation points on a selected bend. Flow Figure 7.19. “Data Manager” dialog box. orientation line for a selected bend (see Appendix C for the circle-fitting algorithm). To complete the data collection for this bend, the user selects the “Measure Distances” tool on the “Data Manager” dialog box. Pressing and holding the left mouse button at a given point on the view defines the beginning of a line seg- ment. Dragging the mouse pointer to a second point in the view and releasing the left mouse button defines a line segment whose length is calculated in an Avenue script. Following the procedure described above, the user must make five distance measurements (see Figure 7.22) at each bend representing the following:

41 • Orientation of the centerline of the bend, • Three channel width measurements, • Wavelength of the bend, and • Amplitude of the bend. Figure 7.25 illustrates the format in which the data are stored. If the user determines that the database contains incorrect values for some bend-year pair, the steps outlined above can be repeated. Archiving the data for a given bend and year overwrites the current field values for that record. Bankline delineation points for the current bend are stored in a theme attached to the ArcView project for this sub-reach. Because these points can be recovered at any time, use the “Delete Current Bend Points” and “Delete Circles” buttons on the “Data Manager” dialog box to clean the view up before moving to a new bend or a new historical record. Figure 7.21. Fitted circle and bend orientation line. Figure 7.22. Distance measurements used to describe each bend. Flow Flow Figure 7.23. “Distance Measurements” dialog box.

42 Figure 7.25. Some of the data fields for Bend 1 and Year 1. Figure 7.26. Bankline theme containing two historical bankline records. Flow When the user has performed the steps described above for each bend and each historical record, the ArcView project for this sub-reach can be closed and should be saved to provide a record of how the database for this sub-reach was created. 7.4.3 Channel Migration Predictor Users Guide The Channel Migration Predictor uses as input data two or three historical records of bankline position. From this input data, the Channel Migration Predictor calculates and records the extension and translation rates for each bend and each his- torical interval. Two historical bankline records result in one historical interval, whereas three historical bankline records result in two historical intervals. The Channel Migration Pre- dictor uses these calculated migration and extension rates to extrapolate and estimate future bankline locations. Step 1. Start ArcView and add a bankline theme containing the historical records that have been measured and archived using the Data Logger. Figure 7.26 shows an example of such a theme with two historical bankline records. The presence of the blue diamond button in the upper right corner of the proj- ect window of Figure 7.26 confirms that the Channel Migra- tion Predictor extension has been loaded into ArcView. Step 2. Ensure that the ArcView project contains the table of archived, measured data corresponding to the displayed banklines. Figure 7.27 shows the table named “bigblack.dbf” Figure 7.24. “Verify Measurement” dialog box.

43 Figure 7.29. Centroid, radius of curvature, and orientation of the predicted bend. Predicted Flow Figure 7.27. ArcView project tables window. Figure 7.28. Channel Migration Predictor interface. has been added to the project. If the “Tables” list does not con- tain the database file containing the archived data, then the user must press the “Add” button in Figure 7.27 and add it to the project before launching the Channel Migration Predictor. Step 3. Press the blue diamond button shown in Figure 7.26 to start the Channel Migration Predictor, which is illus- trated in Figure 7.28. The user must provide the name of the archived data file, the “view” name, the number of historical records, the bend number to analyze, and the date for which the bankline is to be predicted. After providing this information, the user presses the “Continue” button to initiate the calculation of the predicted bankline position. Comparing Figure 7.29 with Figure 7.26, it can be seen that, when the Channel Migration Predictor is started, new themes are added to the view “BigBlack.” Pressing the “Continue” button causes the Channel Migra- tion Predictor to read the database file of archived data and to calculate migration and extension rates. The Channel Migration Predictor then proceeds to create a theme for each historical record and a theme for the predicted bankline. To each of these themes is added a circle indicating the center

44 migrating at less than the specified rate and 100 minus “x” percentage were migrating faster than the specified rate. These figures show that the C sites migrate faster than the A, B1, and B2 sites. This result agrees with the premise of the screening and classification. These figures also show that rates of translation tend to be greater than rates of extension (bends tend to move faster in the downstream direction rela- tive to their orientation). Figures 7.31 and 7.32 show that 50 percent of C sites were observed to extend 0.008 channel widths per year and to translate 0.015 channel widths per year. This information can be used to provide an estimate of future bankline positions. For example, suppose an estimate of bankline position for 20 years into the future at a C site bend with a channel width of 140 ft (42.7 m) is needed. The data from Figures 7.31 and 7.32 suggest that half of other similar (C site) bends extended at least 22 ft (7 m) (0.008 channel widths/year × 20 years × 140 ft-width = 22 ft or 7 m) and translated at least 42 ft (13 m) (0.015 channel widths/year × 20 years × 140 ft-width = 42 ft or 13 m). Half of the bends also moved at slower rates than 0.008 (extension) and 0.015 (translation) channel widths per year. This approach requires judgment in selecting a “level of confidence” to use in making the estimate. The “level of confidence” refers to the percentage in the “Cumulative Per- cent” axis. The higher the percentage, the more likely it is that the bend will be moving less than the computed amount (shown on the “Extension” and “Translation” axes of Figures 7.31 and 7.32, respectively). Thus, selecting a higher per- centage will provide a more conservative estimate. More con- servative estimates may be desired for more important trans- portation structures. Returning to the example previously discussed, if a more conservative estimate were wanted, the bend could be assessed for a 75-percent level of movement rather than for a 50-percent level. Figures 7.31 and 7.32 show that 75 percent of the C site bends extended less than 0.018 channel widths per year and translated less than 0.031 chan- nel widths per year and that 25 percent of the C site bends were moving at higher rates. The 20-year prediction using 75-percent rates of extension and translation are 50 ft (15 m) and 87 ft (26 m), respectively. Using the frequency approach relies on identifying the channel classification and applying a rate based on a selected frequency or percentage. Higher percentages result in more conservative estimates of migration (i.e., the bend will likely move less than the estimated amount). In addition to Figures 7.31 and 7.32, the rates for the modified Brice classes (see Figure 3.2) and different probabilities are summarized in Table 7.1. The cumulative percent can be used as the proba- bility that a bend will migrate less than the given amount. One hundred minus the cumulative percent can be used as the probability that a bend will migrate more than that amount. Figure 7.33 is an illustration of the frequency analysis approach applied to an A site and a C site assuming similar starting conditions. For the initial condition, both banklines are shown, and for a 30-year future condition several potential Figure 7.30. Enabled check boxes allowing selective viewing of bend data. and radius of curvature of the bend; a line segment indicat- ing the orientation of the bend is also added (see Figure 7.29). As shown in Figure 7.30, the “Continue” button on the Channel Migration Predictor interface has been temporarily disabled, and check boxes allowing the user to selectively display bend data have been enabled. A change in any of the text fields above the “Continue” button will enable the “Con- tinue” button and disable the check boxes. 7.5 FREQUENCY ANALYSIS The GIS bend measurement and prediction tools are rec- ommended for predicting future bend migration because the individual bend conditions and history are considered to be the best information for making the prediction. The GIS pre- diction also computes the rate of migration (in channel widths per year) and provides a comparison of this rate to a database of measured bends. As part of the NCHRP 24-16 research project, a database was developed that includes nearly 2,500 measurements of bend migration. If the time sequence aerial photos are not available and a prediction of bend migration is still needed, the measured amounts of bend migration from the database can provide some guidance. For the modified Brice Class A, B1, B2 and C sites (see Figure 3.2), rates of extension and translation were measured and the cumulative percent, or frequency, of movement was cal- culated. The number of bends that were included in this analysis are 89, 249, 408, and 915 bends for the A, B1, B2 and C sites, respectively. The rates of movement are shown in Figures 7.31 and 7.32 as cumulative percent of extension and translation in channel widths per year. The figures should be interpreted such that, “x” percentage of the bends were

45 Figure 7.31. Cumulative percentage of extension in channel widths per year. Figure 7.32. Cumulative percentage of translation in channel widths per year. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 Extension (Channel Widths/yr) Cu m u la tiv e Pe rc e n t Brice A Sites Brice B1 Sites Brice B2 Sites Brice C Sites 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 Translation (Channel Widths/yr) Cu m ul at iv e Pe rc en t Brice A Sites Brice B1 Sites Brice B2 Sites Brice C Sites TABLE 7.1 Rates of extension and translation Extension (channel widths/yr) Translation (channel widths/yr) Cumulative % 50 75 90 95 50 75 90 95 A Sites 0.0015 0.008 0.015 0.018 0.0025 0.010 0.015 0.019 B1 Sites 0.004 0.010 0.015 0.026 0.0023 0.009 0.016 0.020 B2 Sites 0.004 0.009 0.016 0.020 0.007 0.016 0.026 0.033 C Sites 0.008 0.018 0.032 0.045 0.015 0.031 0.055 0.074

46 channel locations (outer bank only) are shown. At the 50-percent level, the A site shows almost no migration, whereas the C site shows the potential to migrate half the channel width. There is a 10-percent chance that the A site will migrate half a channel width in 30 years (90 percent of A site channels moved less than half a channel width in 30 years) and a 10-percent chance that the C site will migrate nearly two channel widths (90 percent moved less than two channel widths in 30 years). As an alternative to photo comparison, or as a check on the results of the photo comparison, this fre- quency analysis approach provides reasonable results. How- ever, it should only be considered as an alternative to photo comparison when no data are available for the extrapolation technique using photo comparison. Figure 7.33. Example movement percentages for a 30-year time period. 30 Years A site C site Initial Bankline 50 Percent 75 Percent 90 Percent