ScienceDirect® Home Skip Main Navigation Links
My Settings
 Quick Search
 Search tips (Opens new window)
    Clear all fields          Advanced Search
Volume 69, Issues 1-4, July 2005, Pages 1-13
Font Size: Decrease Font Size  Increase Font Size
 Article - selected
PDF (1676 K)
Thumbnails - selected | Full-Size Images

Related Articles in ScienceDirect
View More Related Articles
Find at UT
View Record in Scopus
doi:10.1016/j.geomorph.2004.09.034    How to Cite or Link Using DOI (Opens New Window)
Copyright © 2004 Elsevier B.V. All rights reserved.

Quantifying downstream impacts of impoundment on flow regime and channel planform, lower Trinity River, Texas

Jessica L. Wellmeyera, E-mail The Corresponding Author, Michael C. Slatterya, Corresponding Author Contact Information, E-mail The Corresponding Author and Jonathan D. Phillipsb, E-mail The Corresponding Author

aDepartment of Geology, Texas Christian University, Box 298830, Fort Worth, TX 76129, United States bTobacco Road Research Team, Department of Geography, University of Kentucky, Lexington, KY 40506-0027, United States

Received 1 June 2003; 
revised 17 August 2004; 
accepted 17 September 2004. 
Available online 1 February 2005.


As human population worldwide has grown, so has interest in harnessing and manipulating the flow of water for the benefit of humans. The Trinity River of eastern Texas is one such watershed greatly impacted by engineering and urbanization. Draining the Dallas–Fort Worth metroplex, just under 30 reservoirs are in operation in the basin, regulating flow while containing public supplies, supporting recreation, and providing flood control. Lake Livingston is the lowest, as well as largest, reservoir in the basin, a mere 95 km above the Trinity's outlet near Galveston Bay. This study seeks to describe and quantify channel activity and flow regime, identifying effects of the 1968 closure of Livingston dam. Using historic daily and peak discharge data from USGS gauging stations, flow duration curves are constructed, identifying pre- and post-dam flow conditions. A digital historic photo archive was also constructed using six sets of aerial photographs spanning from 1938 to 1995, and three measures of channel activity applied using a GIS.

Results show no changes in high flow conditions following impoundment, while low flows are elevated. However, the entire post-dam period is characterized by significantly higher rainfall, which may be obscuring the full impact of flow regulation. Channel activity rates do not indicate a more stabilized planform following dam closure; rather they suggest that the Trinity River is adjusting itself to the stress of Livingston dam in a slow, gradual process that may not be apparent in a modern time scale.

Keywords: Impoundment; Regulated discharge; Channel planform change; Trinity River

Article Outline

1. Introduction
2. Background
3. Purpose
4. Study area and methods
5. Results and discussion
5.1. Flow regime
5.2. Channel planform change
6. Conclusions

1. Introduction

The vast majority of US waterways are now affected by river engineering works, such as dams, canals, and artificial levees. Dams occur in a variety of sizes across the country with primary functions including hydropower, flood control, water supply, navigation assistance, and water-quality control. In Texas, the Trinity River basin provides drainage SE, from the Dallas–Fort Worth metroplex to the Gulf of Mexico (Fig. 1). Twenty-nine dams are in operation in the basin, the majority clustered around the metroplex on the four forks and tributaries of the Trinity River.

Full-size image (79K) - Opens new windowFull-size image (79K)

Fig. 1. The Trinity River Basin, NE Texas, showing the study area immediately below Lake Livingston between the cities of Romayor and Liberty. The channel shown here was digitized from the 1995 DOQQs.

Lake Livingston, located not, vert, similar100 km above the river's mouth, is the lowest as well as largest reservoir in the basin. Closed 28 September 1968, the lake has a conservation pool capacity of 2.2 billion m3; its primary purpose is water supply for Houston. Most of the upstream dams, however, are managed for flood control.

2. Background

Impoundment of river channels regulates discharge downstream, potentially affecting flooding patterns, flow regime, sediment transport, and the hydraulic gradient downstream (Williams and Wolman, 1984, Mangelsdorf et al., 1990, Brandt, 2000 and Graf, 2001; among others). Conceivably, planform size, form, and activity could be affected as a result of managed discharge. However, precisely how regulated flow differs from natural conditions varies significantly between individual dams as well as their operating methods. Regardless, the distribution of discharge will be affected by impoundment.

Whether in nature or a lab, meandering channels tend to follow a common evolutionary pattern over time (Friedkin, 1945). Initially straight reaches develop a series of pools and riffles where stream power is changing. Continued channelized flow, as well as periodic flooding events, allows a meandering pattern and its flood plain to establish and evolve. Ongoing debate exists regarding the precise discharge conditions required to effectively alter and change the path of a stream.

Meanders migrate through two major components. First, meanders migrate laterally, eroding their banks on the outside of bends and simultaneously depositing material on the inside. Growth of individual meanders continues until cutoff occurs, often the result of flow encountering more resistant material or shear being reduced by flattening of energy slope. The rate of erosion tends to be greatest on and below the apex of an individual bend. This erosion just below the apex of a meander bend drives the second component, down valley translation (Gilvear et al., 2000). As individual meanders migrate laterally, they are continuously tracking down the flood plain. Meander migration sculpts the flood plain with meander scars and natural levees, through a systematic process of depositing and eroding alluvium. Gillespie and Giardino (1997) hypothesized that bends act as suites, rather than independently, as they migrate.

When flow is impounded and regulated, the natural discharge regime may be altered, and hence the behavior or rate of meander evolution may be affected. Park (1977) and others noted that constructed reservoirs strip downstream flow of its sediment load, and regulation often dampens peak flow events (Petts, 1979). This may increase erosional capability and decrease channel capacity immediately downstream. Gregory and Park (1974) attributed decreased channel capacity to flow regulation upstream. Stabilization of planform is well documented to be expected downstream following impoundment and regulation. Lewin (1977) suggested that pattern change driven by altered discharge would stabilize a channel planform; this is frequently the very purpose of engineering works. Multiple studies in the U.S. confirm the theory of planform stabilization. Gillespie and Giardino (1997) observed an accelerated stabilization and equilibration on the Brazos River, TX, following impoundment. In Montana and Alberta on the Milk River, Bradley and Smith (1984) observed a reduction in channel width along with slowed meander migration initiated by dampened peak flows below the Fresno Dam. Shields et al. (2000) found channel migration on the Missouri River immediately below Fort Peck Dam to be nearly four times less than prior to impoundment.

Around the Gulf Coast of Texas, however, not all of the rivers regulated low in their profile exhibit depressed annual discharges or peaks. Phillips (2003) mentioned the Sabine and Trinity Rivers as two examples where flow regime on an annual scale is not strongly affected by regulation. Previous work suggests the Brazos River and Rio Grande may be added to the list of medium- to large-size rivers whose annual discharge is unaffected or only weakly affected by flow regulation (Hudson and Mossa, 1997). Alternatively, the Neches River in Texas has shown reduced peak discharges following closure of the Sam Rayburn reservoir whose primary purpose is flood control (Phillips, 2003).

Regardless of controlled discharge, the rate of channel activity via avulsion or migration is ultimately dependent on the flow regime. Three principle factors, magnitude, frequency, and duration will determine the efficiency of stream flow to scour its bed, erode its banks, or aggrade within the channel (Harvey, 1969 and Petts, 1979). Common effects of flow regulation on the hydrograph include a dampening of peak flows, an elevation of low flow magnitude, and a transition from “spiky” rises and falls over time to a more “step-like” delayed pattern. The end result is a new flow regime of generally less variable discharge from which a more stable channel arises.

Williams and Wolman (1984) suggested that the purposeful management of water releases varies greatly between dams and between operating methods. However, the distribution of discharge will most certainly be altered from the natural regime. Ultimately, the spatial and temporal extent to which regulation significantly impacts an individual river is frequently unknown (Petts, 1979 and Phillips, 2003).

3. Purpose

The purpose of this study is to quantitatively investigate the nature of channel planform change over a 75-km length of the lower Trinity River where Livingston Dam is currently regulating discharge. Historic air photos are used to establish a baseline study of planform change prior to dam construction. Photographs taken after commencement of flow regulation document the subsequent adjustment to the channel downstream. Downward et al. (1994), Winterbottom and Gilvear (2000), Simon et al. (2002), and others have conducted similar studies of rivers or streams using a GIS approach.

Analysis of daily discharge and flow duration data provides a record depicting any change of the flow regime that may correlate to spatial channel adjustments. Precipitation data complements the discharge-derived information, identifying periods of extreme rainfall events and high flow conditions unrelated to flow regulation. Six sets of air photos and digital orthographic quarter quadrangles (DOQQs) from 1938 to 1995 were used to determine rates of channel activity, including channel erosion and lateral migration. In conjunction, the temporal distribution of flows from the same period was analyzed in an effort to determine (i) the effects of impoundment on flow regime, (ii) the effects of impoundment on channel migration, and (iii) whether correlations truly exist between flow regime and channel activity.

4. Study area and methods

The study area is a 75-km reach of the Trinity downstream of Livingston Dam and is bound to the north by the FM-787 bridge near the community of Romayor and to the south by the US-90 bridge in the city of Liberty, TX (Fig. 1). The northern half is dominated by piney woods, while the southern by coastal marshlands. In general, the entire study area is rural-land adjacent to the channel and in the flood plain is used for logging, grazing, and some residential development.

Six individual years of photographic coverage were available over the entire study area: 1938, 1958, 1964, 1972, 1989, and 1995. For 1995, 16 DOQQs were obtained online from TNRIS (Texas Natural Resources Information System). These color-infrared images are the most recent for the area and were taken in January with a 1 m2 pixel resolution digitized from 1:40,000 aerial photography. Analog black and white aerial photographs obtained in 1989 and 1972 were at 1:40,000; those flown in 1964 and 1958 were at a 1:20,000 source scale. All were purchased from the US Department of Agriculture's Air Photo Field Office. The photos were scanned at high resolution (1200 dpi) and co-registered with the 1995 DOQQ using ground control points, such as road intersections and building corners. The images were then warped via quadratic rectification and finally projected as individual GeoTiff image files. The 1938 data comprised four photo-mosaics at 1:63,360 scale taken from an air photo run of undetermined origin, and were made available to us by the Liberty County NRCS (Natural Resource Conservation Service). These indexes for 1938 were processed similarly to the purchased photos, although their resolution is substantially lower than all following photographic years. Resulting root-mean-square (RMS) errors following rectification of photos are <8 m for 1958–1989 and do not exceed 25 m for the 1938 photos.

Daily stage-discharge data used to construct flow duration curves were obtained from USGS gauging station records. The Romayor gauging station, located at the northern boundary of the study area at FM-787, has been in operation since 1924 and has a complete record of daily and peak flows. Fig. 2 shows the mean annual discharge record coincident with the photographic coverage used in the study. Drainage above Romayor is 44,550 km2.

Full-size image (54K) - Opens new windowFull-size image (54K)

Fig. 2. Mean annual discharge at the Romayor gauging station, 1938–1995.

Southeast Texas, including Liberty County, traditionally records some of the highest annual rainfall totals in the state. Between 1939 and 1994, annual rainfalls ranged from 76 to 221 cm, with a mean of 127 cm (Fig. 3). Proximity to the Gulf Coast subjects this region to hurricanes, tropical storms, and warm fronts. While discharge is typically not strongly influenced by local precipitation this low in the basin, extreme events such as prolonged drought and heavy rains can often be tied to Romayor discharge measurements. In particular, the 1989–1995 photographic period experienced above average rainfall, including two of the five largest 24-h maxima (Fig. 4). Interestingly, the second largest 24-h maxima for 1989 would have ranked fourth overall in terms of rainfall depth.

Full-size image (59K) - Opens new windowFull-size image (59K)

Fig. 3. Annual precipitation record for Liberty County, 1939–1994.

Full-size image (49K) - Opens new windowFull-size image (49K)

Fig. 4. Twenty-four-hour rainfall maxima for the Liberty County precipitation record (n=57 years).

Planform channel changes and the derived rates were obtained by comparing the 6 years, or five intervals, of aerial photographs of the area. The georeferenced and spatially corrected images were assembled into the ArcView GIS. For each of the 6 years, a vector outline of the bankfull channel location was manually digitized (see Fig. 5).

Full-size image (121K) - Opens new windowFull-size image (121K)

Fig. 5. 1995 DOQQ showing a section of the Trinity River downstream of Romayor. Channel locations are mapped by individual photographic year and later overlain to permit computation of spatial change.

The first method for considering channel activity is based on the percentage of channel area occupying the same location between consecutive years of photography. To obtain this measure, all 6 years' bankfull locations were overlain in a raster analysis. Here we used a software package called MF Works. Sixty-four cell values result, and each one indicates a different combination of years in which an individual cell is occupied by the channel. These are reclassified into periods of continuous occupancy and a matrix of percent stability for an individual year's channel generated (Table 1). This and the following raster-based analysis are similar to that followed by Downward et al. (1994) and Leys and Werritty (1999).

Table 1.

Results of historic stability analysis showing percent channel occupied during subsequent photographic intervals

Percent channel stable through19381958196419721989

Rates of channel creation and abandonment occurring during individual photographic intervals were measured by raster overlay. As in the case of the historical stability components, layers of 2 consecutive years of channel occupancy are overlain. This results in three classes of cells; those occupied during both time periods, those occupied only in the first photograph, and those occupied only in the second photograph.

An aerial estimate for abandonment is derived from the total area of cells occupied only during the older photograph of the interval, converted to ground units. Created channel area is computed from cells occupied only during the more recent photograph, and the stable channel area based on the area of cells occupied in both photographs.

Rates of each process are computed by dividing the total area of process by the time elapsed between photographs from which the channel occupancy maps were derived (Fig. 6). By comparing the relative areas of abandonment and channel creation, a net change in channel area for a specific time period could be calculated.

Full-size image (61K) - Opens new windowFull-size image (61K)

Fig. 6. Rates of channel creation and abandonment by photographic interval. The 1938–1958 and 1989–1995 periods include data for both cutoff and non-cutoff conditions.

From the previously defined channel area polygons, channel centerlines were defined for the purpose of quantifying lateral migration. The rate and magnitude of lateral migration occurring along the main channel and on individual bends may be derived by combining previously existing data with newly derived data. To determine the area affected by lateral migration, a new raster dataset for each photographic interval is generated from superimposed channel centerlines of sequential years. In this study, the ArcView shapefiles of channel centerlines were imported into MF Works with a 2-m pixel resolution to maximize accuracy.

The two centerlines were superimposed to generate a new, single layer for each photographic interval. Polygons of various sizes, generally sliver shaped, may be defined in the area between centerlines, indicating the area of active migration for the interval. Using MF Works, these regions are classified as the active migration area, and the ground area computed. The total area of active migration is divided by the channel length measured during the first photograph of the interval, resulting in a linear measure of lateral migration. A migratory rate is established by comparing this linear measure to the period of time elapsed during the photo interval (Fig. 7).

Full-size image (52K) - Opens new windowFull-size image (52K)

Fig. 7. Rates of lateral migration by photographic interval.

5. Results and discussion

5.1. Flow regime

Mean annual discharge at the Romayor gauging station is highly variable, although no obvious changes following impoundment are apparent. Pre-dam and post-dam mean annual discharges are statistically indistinguishable (207 versus 249 m3/s, respectively). Flow duration distributions, generated from daily mean discharges, are a useful indicator of temporal variability in the record. The slope of these curves indicates how flashy or steady the discharge record is for a period of time.

The complete discharge record for Romayor shows a curve typical of humid environments (Fig. 8). Flow is not flashy, and discharge remains elevated for extended periods. Median (Qd50) discharge is 75.5 m3/s, and distribution is normal over three log cycles.

Full-size image (44K) - Opens new windowFull-size image (44K)

Fig. 8. Flow duration curves at the Romayor gauging station showing little change in flow regime following impoundment.

Flow regulation appears not to have altered the hydrologic regime of the lower Trinity in any significant way. Pre- and post-impoundment curves are of similar form, and managed release appears to closely mimic natural (i.e., pre-dam) conditions, except at low flow. Low flows are elevated following impoundment (Qd95 pre-dam=8.6 m3/s versus Qd95 post-dam=17.2 m3/s), and a break in duration curve slope occurs near the 50% exceedance level. However, high magnitude flows are unaffected.

5.2. Channel planform change

Overall, this 75-km long reach of the Trinity River displays a wide and varying range of activity both as a function of time and, to some extent, the cumulative effects of flow regulation.

During the first photographic period, 1938–1958, one major cutoff, near the top of the study reach, occurred (Fig. 9). Many bends in the upper one-third of the study section show a general straightening, and most other bends develop by increasing amplitude and migrating substantially during this 20-year period. Rates of channel creation (32.5 ha/year) and lateral migration (3.0 m/year), as well as the percent stability, are most likely facilitated by the simplification of so many complex bends. The flow regime is noticeably different during this period, as seen in the photographic period flow duration curves (Fig. 10), and suggests variable and quickly fluctuating flow with lack of sustained high or low flows. This would tend to lead to increased instability of bank materials, as only minor wetting events followed by rapidly falling discharge results in undercutting and bank collapse.

Full-size image (19K) - Opens new windowFull-size image (19K)

Fig. 9. Sequential channel centerlines of each of the 6 photographic years. One cutoff in the northern third of the study section occurred 1938–1958, and two cutoffs in the southern half 1989–1995, as indicated by arrows.

Full-size image (55K) - Opens new windowFull-size image (55K)

Fig. 10. Flow duration curves for each photographic interval. Data for the most recent period, 1995–2001, are included to further highlight the elevated nature of the 1989–1995 period.

One concern for this particular photographic interval involves the spatial accuracy of the 1938 channel location. As noted earlier, the 1938 photograph is of coarser resolution than following years; therefore, interpretation and digitization of the 1938 bankfull channel location are challenging and there is certainly a potential for compounding error propagating to the actual magnitudes of change occurring.

The second pre-impoundment period, 1958–1964, is characterized by a more rapid lateral migration rate (6.5 m/year), although precipitation and discharge data suggest no extreme events that might induce such behavior. A lack of high magnitude floods may have diminished the ability to “flush” coarse material through the system; however, the discharge is of sufficient magnitude to allow rearrangement of the sands accumulated on bars and dunes. No major cutoffs occur during the photographic period, and the lateral migration rate reflects a development of point bars and growth of some meanders. This induces a translation of the channel centerline without affecting the length of the active channel and increases the lateral migration rate. The rapid lateral migration may also have been accentuated by the short duration of this photographic interval. Even though lateral migration is normalized, a short-term event such as cutoff has a direct impact on the overall migration rates; in a short photographic interval <10 years, the weight of such an event will be much more pronounced than in a longer interval of 15 years or more.

The third study period, 1964–1972, is a period of dam construction, closure, and initial filling of Lake Livingston. Comparison to other photographic intervals is difficult because of the transitional nature of the period. In general, changes in lateral migration are due to development of meanders through translation and increased amplitude. The lower quarter of the overall study area appears much more stable than before. Construction of levees, installation of bank stability structures, or urbanization in the city of Liberty may be partially responsible for this phenomenon. Consequently, the process of dam construction has not had a significant impact on the characteristics studied during this period—at least not at this location downstream. Hydrologically, this period is consistent with means, and therefore gives the author a stable “baseline” period from which to examine pre- and post-dam dynamics.

The first period following closure of Livingston Dam spans from 1972 to 1989. This is a statistically wetter period than the photographic years prior to construction. Thus, the elevated flows (Fig. 10) are likely the result of increased precipitation and runoff in the local area rather than flow regulation. However, regulated rivers characteristically experience elevated low flows. Livingston is not intended to operate for flood control, and the lack of change in high flow duration and magnitude confirms this management strategy. Indeed, it is basically a flow-through reservoir.

Surprisingly, this is a period of low activity, despite higher-than-normal precipitation and discharge. The bankfull channel appears to have narrowed, resulting in a loss of active channel area. The rate of channel creation (10.7 ha/year) is remarkably low, and likely bar development is at least partially responsible for increasing sinuosity, though these were not visible on the imagery. Given high precipitation, discharge is sustained through a lack of low flows. Banks may not be as susceptible to failure as they are being continually saturated, and pore-water pressures are not fluctuating as frequently as during the drier periods prior to impoundment.

Minimal channel creation rates do not, however, imply that no erosion is occurring. As sediment is trapped behind Livingston Dam upstream, clear water released from the reservoir is scouring the bed immediately below and sediment is being entrained. The bulk of erosion occurring in the channel during this time frame appears to be occurring at depth, with larger adjustments to the channel geometry rather than planform taking place (see Phillips et al., in press). Low channel creation rates may also suggest that the section of channel between Livingston Dam and Romayor is not supply limited, and sediment discharge has recovered within the section immediately below the dam. This is in fact supported by recent suspended load sampling in tributaries such as Long King Creek (Dakshinamurthy, 2004).

The final photographic period, 1989–1995, is particularly challenging to interpret because of its extremely high precipitation and discharge. The 1995 photograph was obtained at a time where stage was above bankfull (see Fig. 5); thus, accuracy of bankfull channel location is partially compromised. This spatial accuracy problem may explain the substantial increase in active channel area, suggesting extremely rapid channel creation rates (42.0 ha/year) facilitating this change; however, delineation of the bankfull channel was estimated when possible using vegetative evidence, like large trees, rising above the water level. The rate of channel creation is explained primarily by increased channel area due to flooding rather than actual meander development taking place. These extreme precipitation and flooding events are intense enough to initiate two cutoffs and invoke substantial flooding (see Fig. 5). The impact of these cutoffs profoundly increased lateral migration rates as well (see Fig. 7).

Livingston Dam is not the sole engineering project affecting this stretch of the Trinity River. Withdrawals occur directly from the Trinity between Romayor and Liberty for irrigation and drinking water supply. Constructed levees are in place at several crucial locations. Repeated attempts at bank stabilization have been made near bridges and residential developments, although they have invariably failed. Point bars have been reinforced with rip-rap in some locations. And following impoundment of Lake Livingston, much new home construction has taken place adjacent to the channel under the false impression of flood security. This raises the issue of possible multiple causal factors for the river changes reported here, as it is not unusual for river hydrology and geomorphology to be impacted by a number of activities, including logging, grazing, and construction. However, there is no evidence or reason to suspect that land use, climate, hydrologic response, etc. have changed on a regional scale before and after the dam. Structures and land use have no doubt had local effects, but at the long reach scale considered here, they are unlikely to be significant. In addition, water withdrawals occur only downstream of Liberty, and amount to about 10% of flow there. They would not affect the channel down to that point. Another potential source for channel change is sea-level rise. This would likely increase meandering, but in addition to the relatively short response time in question here, there is no evidence of a change in the rate of SL rise since the late 1960s. Thus, by considering pre- and post-dam time periods, this leaves the effects of major flood events as the major potential non-dam channel changing factor, and these are considered.

In the process of evaluating channel activity rates, some shortfalls of currently published methodologies were uncovered. Lateral migration rates may be highly inflated when cutoffs occur, especially for short photographic periods; and these rates are not indicative of the true channel migration across the flood plain. Large areas may be contained between channel centerlines where a neck cutoff has occurred, and this area is incorrectly classified as “migration area” when no actual channel migration has taken place (see, for example, Fig. 5). This method of computing lateral migration rates is perhaps best suited for reaches lacking cutoffs and for those in which avulsion has occurred. This methodology is also highly affected by both spatial and temporal resolution. A high spatial resolution is very desirable to maximize aerial accuracy. Especially in instances of large magnitude planform change, time is of the essence—actual rates of change are not constant and long photographic intervals may minimize the rate while short periods will inflate the long-term average.

Comparing rates of channel creation and abandonment between photographic intervals incurs some risks similar to lateral migration calculation. As long as cutoff or avulsion effects on these processes are identified relative to the overall rate, they provide an acceptable estimate of periodic change. This method may be preferential to lateral migration as it incorporates the entire channel area and identifies areas of bank instability.

Most importantly, one should be aware that lateral migration and channel creation and abandonment are not interchangeable and do not describe the same changes taking place. Lateral migration, like sinuosity, is reflective of changes in channel form; while channel creation, abandonment, width, and length measures reflect changes occurring in channel size.

6. Conclusions

Lake Livingston is regulating discharge to the lowest reach of the Trinity River. However, this regulation is not strongly affecting frequency or magnitude of flood peaks at an annual scale. Low flow discharge subsequent to impoundment has been elevated, although this may not be entirely attributed to dam management given the local precipitation record.

On the Lower Trinity, the average lateral migration rates suggest closure of Lake Livingston has stabilized the planform, and activity has decreased by 42%. The magnitude of stabilization is less than that seen in the northern plains of the U.S., but consistent with the nearby Brazos River (Table 2). A possibility exists that there is a lag between closure and more substantial adjustment—perhaps the full effects of sediment trapping in the reservoir have not propagated to this reach. However, comparison of lateral migration rates by photographic period, as seen in Fig. 7, suggests such changes in channel activity are not abrupt; and the range in pre-impoundment migration rates illustrate a certain natural variation as lateral migration events are episodic.

Table 2.

Lateral migration rates for five US rivers all suggest some measure of depressed lateral migration following impoundment and flow regulation

Pre (m/year)Post (m/year)Transition (m/year)Decreased migration (%)RiverSource
4.952.853.642Trinity, TXThis study
1.750.453.474Milk, MTBradley and Smith (1984)
3.41.8N/A47Brazos, TXGillespie and Giardino (1997), MTShields et al. (2000), NDJohnson (1992)

Contrary to Gillespie and Giardino's (1997) findings in a similar study of the nearby Brazos River, the Trinity appears not to be in disequilibrium with its flow regime. The Trinity is a typical Gulf Coast river able to adjust to its environment in a slow, gradual process that may be imperceptible in a human's lifespan. Richards (1982, p. 252) noted that an individual channel's sensitivity to impact may vary geographically as “stable humid-zone streams absorb the effects of catastrophic floods, and measurable natural adjustment may require hundreds of years while in drier regions, channel change may be more instantaneous.” In this case, the lower Trinity is adjusting to the dam and other modern engineering impacts no differently than it accommodates other environmental changes and stresses through expected behaviors characteristic of alluvial rivers.

Furthermore, analysis of historical air photos for the purpose of determining rates of channel change and activity is of merit for short-term studies, but may not be indicating long-term trends accurately. Instantaneous changes like meander cutoff may only affect the short-term changes taking place, but if incorporated into a long-term study may imply exaggerated significance. This problem is exacerbated by irregular intervals between photographic runs. GIS platforms provide an ideal opportunity to investigate spatial and temporal variability in channels, increasing efficiency of analysis and aiding spatial accuracy as compared to older, manual methods of evaluation. However, comparison of results between multiple studies remains difficult at this time, given a lack of standardized, appropriate methods that account for dynamic processes accurately.


This study was funded by the Bay and Estuary Program of the Texas Water Development Board under contract 2002-483-440 that was funded by an Interagency Cooperation Contract between the General Land Office and Texas Development Board. The funds are made available to the state of Texas by the Department of Commerce, National Oceanic and Atmospheric Administration, under the Federal Coastal Zone Management Act of 1972. We gratefully acknowledge this support.


Bradley and Smith, 1984 C. Bradley and D.G. Smith, Meandering channel response to altered flow regime: Milk River, Alberta and Montana, Water Resour. Res. 20 (1984), pp. 1913–1920. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (19)

Brandt, 2000 S.A. Brandt, Classification of geomorphological effects downstream of dams, Catena 40 (2000), pp. 375–401. Article | PDF (502 K) | View Record in Scopus | Cited By in Scopus (47)

Dakshinamurthy, 2004 Dakshinamurthy, L., 2004. Quantifying the impact of impoundment on sediment transport in the lower Trinity River, Texas. MS thesis. Texas Christian University, Fort Worth.

Downward et al., 1994 S.R. Downward, A.M. Gurnell and A. Brookes, A methodology for quantifying river channel planform change using GIS. In: L.J. Olive, R.J. Loughran and J.A. Kesby, Editors, Variability in stream erosion and sediment transport, International Association of Hydrological Sciences Publication vol. 224 (1994), pp. 449–456.

Friedkin, 1945 J.F. Friedkin, A laboratory study of the meandering of alluvial rivers, U.S. Waterways Experimental Station, Vicksburg, MS (1945) 40 pp..

Gillespie and Giardino, 1997 B.M. Gillespie and J.R. Giardino, The nature of channel planform change: Brazos River, Texas, Tex. J. Sci. 49 (1997), pp. 109–142. View Record in Scopus | Cited By in Scopus (2)

Gilvear et al., 2000 D. Gilvear, S. Winterbottom and H. Sichingabula, Characters of channel planform change and meander development: Luangwa River, Zambia, Earth Surf. Processes Landf. 25 (2000), pp. 421–436. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (8)

Graf, 2001 W.L. Graf, Damage control: restoring the physical integrity of America's rivers, Ann. Assoc. Am. Geogr. 91 (2001), pp. 1–27. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (80)

Gregory and Park, 1974 K.J. Gregory and C.C. Park, Adjustment of river channel capacity downstream from a reservoir, Water Resour. Res. 10 (1974), pp. 870–973.

Harvey, 1969 A.M. Harvey, Channel capacity and the adjustment of streams to hydrologic regime, J. Hydrol. 8 (1969), pp. 82–98. Abstract | PDF (776 K) | View Record in Scopus | Cited By in Scopus (22)

Hudson and Mossa, 1997 P.F. Hudson and J. Mossa, Suspended sediment transport effectiveness of three large impounded rivers. U.S. Gulf Coastal Plain, Environ. Geol. 32 (1997), pp. 263–273. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (7)

Johnson, 1992 W.C. Johnson, Dams and riparian forests: case study from the upper Missouri River, Rivers 3 (1992), pp. 229–242. View Record in Scopus | Cited By in Scopus (66)

Lewin, 1977 J. Lewin, Channel pattern changes. In: K.J. Gregory, Editor, River channel changes, Wiley & Sons, Chichester, UK (1977), pp. 167–184.

Leys and Werritty, 1999 K.F. Leys and A. Werritty, River channel planform change: software for historical analysis, Geomorphology 29 (1999), pp. 107–120. Article | PDF (1039 K) | View Record in Scopus | Cited By in Scopus (13)

Mangelsdorf et al., 1990 J. Mangelsdorf, K. Scheurmann and F.H. Weiss, River morphology, Springer-Verlag, New York (1990), pp. 177–192.

Park, 1977 C.C. Park, Man-induced changes in stream channel capacity. In: K.J. Gregory, Editor, River channel changes, Wiley & Sons, Chichester, UK (1977), pp. 121–144.

Petts, 1979 G.E. Petts, Complex response of river channel morphology subsequent to reservoir construction, Prog. Phys. Geogr. 3 (1979), pp. 329–362. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (58)

Phillips, 2003 J.D. Phillips, Toledo Bend Reservoir and geomorphic response in the lower Sabine River, River Res. Appl. 19 (2003), pp. 137–159. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (15)

Phillips et al., in press Phillips, J.D., Slattery, M.C., Musselmann, Z., in press. Channel adjustments of the lower Trinity River, Texas downstream of Livingston Dam. Earth Surf. Processes Landf.

Richards, 1982 K. Richards, Rivers: form and process in alluvial channels, Methuen, New York (1982) 361 pp..

Shields et al., 2000 F.D. Shields, A. Simon and L.J. Steffen, Reservoir effects on downstream river channel migration, Environ. Conserv. 27 (2000), pp. 54–66. View Record in Scopus | Cited By in Scopus (27)

Simon et al., 2002 A. Simon, R.E. Thomas, A. Curini and F.D. Shields, Case study: channel stability of the Missouri River, eastern Montana, J. Hydraul. Eng. 128 (2002), pp. 880–890. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (10)

Williams and Wolman, 1984 G.P. Williams and M.G. Wolman, Downstream effects of dams on alluvial rivers, USGS Professional Paper 1286 (1984).

Winterbottom and Gilvear, 2000 S. Winterbottom and D. Gilvear, A GIS-based approach to mapping probabilities of river bank erosion: regulated river Tummel, Scotland, Regul. Rivers 16 (2000), pp. 127–140. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (13)

Corresponding Author Contact InformationCorresponding author.

Volume 69, Issues 1-4, July 2005, Pages 1-13
- selected
My Settings
Help (Opens new window)
About ScienceDirect  |  Contact Us  |  Information for Advertisers  |  Terms & Conditions  |  Privacy Policy
Copyright © 2008 Elsevier B.V. All rights reserved. ScienceDirect® is a registered trademark of Elsevier B.V.