ARGUMENTS AGAINST CONCRETE ARMORED CHANNELS IN

DOG RIVER WATERSHED

 

Philip R. Jones, Department of Earth Sciences, University of South Alabama, Mobile, AL 36688. E-mail: mrjonespr@yahoo.com. 

 

Dog River is an urban watershed with the urban problems of flooding and soil deposition due to high runoff rates.  Concrete, armored channels, that traverse many of the suburban, headwater regions of the River, have replaced the natural morphology of the streams.  These channels are engineered to quickly drain runoff water from an area, but they increase the runoff rate and the risk of flooding in areas farther downstream and at low bridge overpasses. This study evaluates the discharge generated from six, adjacent suburban watersheds for seven rainfall events.  The rainfall-runoff relationship shows that discharges increase exponentially with higher rainfall intensities and that runoff coefficients are higher and increase with longer concrete channel length.

 

Keyword: Dog River, runoff, concrete channels

 

 

Introduction

  Impermeable surfaces and or saturated soil prevent rainwater from infiltrating into the ground and cause overland flow into creeks and rivers. Under low infiltration circumstances, stream levels rise and larger flows carry more sediment and cause erosion and flooding.  Concrete channels are designed to prevent flooding and stabilize a stream bank, but do not offer a perfect solution to storm water runoff. 

Many concrete channels have been installed in the Dog River watershed, from small ditches to large creeks. These channels are relatively straight and impermeable, offer no resistance to an increase in water flow, and do not allow any water to infiltrate into the soil, thus increasing the speed of the stream. Concrete channels flush storm water from an area quickly, but discharge amounts accumulate with additional overland flow in high concentrations.    Water levels in concrete channels rise fast and fall sharp.   Higher discharges increase the risk of flooding and sediment deposition downstream.  Furthermore, concrete channels isolate creeks from their natural flood plains and riparian boundaries by steep concrete banks.  Over time, these channel walls are susceptible to cracking and collapsing as water erodes soil beneath them (Figs. 1 and 2 ).

Water quality in a creek is also affected by concrete channelization. The natural stream flow of Dog River tributaries is meandering and vegetated.   Shaded, deep basins keep water temperatures cool and dissolved oxygen contents high, which is important for aquatic life.  This type of stream course carries water slower than a straight lined channel. Flow in concrete channels tends to be shallow and warm during normal flow, if any flow exists. With the entire streambed layered with concrete, base flow from beneath the ground cannot feed the stream, so most of the headwater creeks become ephemeral streams - flowing only during a rain.

Large sediment deposits in Dog River create environmental problems and are costly.  Past and current, expensive dredging projects near the mouth of the river indicate a sediment problem.  Dredging that began in October of 2001 and scheduled to finish in April 2002 will have removed an estimated 134,00 cubic yards of sediment from the river for a six-foot deep, forty-foot wide channel (Blankenship 2002).  Dredging is only a temporary solution to the abundance of sediment deposition, and can suspend toxins and lower oxygen levels.  Much of the sediment in Dog River has been traced to Moore Creek tributary and Montlimar Canal.  Both basins have armored concrete channels and are in West Mobile.

Channel dimensions in Mobile are constructed to handle 100-year flood events (FEMA 1998).  One flood was witnessed by Mobile resident, Joe McDonald, who lives near Spencer Br. off Cottage Hill Rd (Personal Conversation 2002).  On the morning of January 7, 1998, he videotaped Spencer Branch over-flowing its channel and flooding some thirteen homes on one street.    Mobile climate records show a rain total of 6.16 inches for that day, with successive intensities of 1.02 and 2.80 inches an hour (NOAA 1998).  This storm duration intensity was below the 100-year flood event level.

Engineers use a value called a runoff coefficient to determine how much runoff occurs in a watershed for a given rainfall event.  It varies by land use and depends on the permeability of the surface.  This value ranges from 0.05 for flat grassy lawns to 0.95 for downtown business areas (Fetter, 2001).  It is similar to saying that 5% of water runs off a grass lawn and 95% of water runs off a paved business area.  The headwaters that feed into Montlimar Canal and Moore Creek run through mostly residential areas. The expected coefficient for a residential area ranges between 0.25 and 0.50 depending on soil type and the percent of impervious surfaces (Purdue, 2000). 

The best way to understand the characteristics of a watershed is to focus on small basins within that watershed.   In order to understand runoff in concrete channels and possibly prevent future dredging projects, a study must be conducted in upstream areas concerning the magnitude of peak flows and a comparison done on their measured runoff coefficients to typical values.

 

Research Question

It is evident that concrete lined channels increase the flow speed for sections of Dog River and may impair water quality.  How much runoff can be expected from a watershed with concrete channeled streams? Moreover, does increased flow yield a higher than expected runoff coefficient for the entire watershed?

 In the first part of this study, the peak discharge is calculated from seven rainfall events for six streams that feed into Montlimar Canal and Moore Creek.  The results from each stream are plotted on a graph as peak discharge vs. rainfall intensity.  Curves are interpolated and extended beyond the data set as to predict the expected discharge from a given rainfall event.  The second part is a comparison of calculated runoff coefficients, from measured peak discharges, to estimated ones based on land use.  This study regards the runoff coefficient as a watershed quality indicator.

 

Methods

Measurements were taken from six locations, on West Eslava Creek, Bolton Branch West and Spencer Branch for seven rainfall events.  The location on West Eslava Creek was at Scott Court north of Airport Blvd.  There were three tributaries of Bolton Branch West tested.  One was located across from Morrison Dr. below University Blvd. and two smaller branches were at University Professional Park one-quarter mile south.  There were two test sites for Spencer Br at the confluence of Spencer Branch and one of its tributaries east of Panorama Drive.  The six creek test sites are referred to as W Eslava Cr. (1), Bolton Br. (2), Little Bolton Br. (3), Bolton Drainpipe (4), Spencer Br. (5) and Little Spencer Br (6) (Fig. 3) .

 The values needed for calculations were: stream height at maximum discharge, local rainfall in inches per hour, slope of channel, dimensions of the channel where the data was collected, watershed areas, and land use data.  The required supplies included chalk, a measuring tape, ruler, level, heavy fishing line, 7.5 minute topographic map, pencil, notepad, calculator, graphing software, transportation, and a few raindrops. 

Velocity and area were needed to find a stream’s discharge.  Stream velocity was calculated by using the Manning equation for open-channel hydraulics (Fetter 2001):

(V) Velocity = 1.49 ´ ((R^2/3) ´ (S^1/2)) ¸ (n), Where,

            V = velocity in feet per second

R = the ratio of the cross-sectional area of flow (in square feet) to the

wetted perimeter (in feet) through the channel

S = the slope of the surface

n = the roughness coefficient                                                                                        

Deriving each of these values consisted of their own complications.  The roughness coefficient depends upon the amount of friction between the water and the channel.  Normally (n) is obtained from a table where it is 0.012 for smoothed concrete or 0.035 for a winding natural stream with weeds (Fetter 2001).  The surface of these concrete channels is rougher from weathering.  Therefore, a finite difference calculation of velocity was done using a floating object and a stopwatch.  This was only conducted on W Eslava Cr due to lack of water flow in the other streams.  With known depth and dimensions, the Manning equation was used to solve for the roughness coefficient (n).  The test runs indicate a roughness coefficient greater than smoothed concrete around 0.015.

A lengthy run was needed to accurately calculate the low slopes.  First, seven meters of line was secured uphill and connected to a level downhill.  The line was made tight and level, then the distance from downhill side to the channel surface was measured for the rise.

Velocity was needed at peak discharge, so the crest of the water in the channel was obtained to calculate (R).   Channel walls were chalked and then washed away by the stream flow up to the maximum height.  These marks were made beneath bridges, protected from falling rainwater.  Best judgment was used in discerning these erased marks.  Some washed away with a defining line, others eroded more gradually and plain chalk worked better than colored chalk.  Uneven left and right wall measurements were averaged because of stream currents, capillary fringe of water at the surface and or uneven channel floors (Appendix A).

With the channel dimensions and the height of peak stream flow known, R was determined for the Manning equation. Most streams were divided through two or three concrete frames under bridges, so calculations for each channel section were added to obtain the total ratio variable (R) for that stream (Appendix A).  A large sand slug had collected in the inside of a bend at Little Spencer Branch, and needed to be integrated out of the channel dimensions (Fig. 4) .  The sediment load occupied three quarters of the channel, so one quarter of the measurements were applied.

Measurements were taken from bridges with square concrete frames which conveniently allowed for calculation of exact stream dimensions through the channels (Area = Height ´ Width).  Bolton Drainpipe is an exception, where the arc measurement was the wetted perimeter and the cross sectional area was obtained by using the law of cosines and subtracting the triangle formed by the chord from the circle sector (Appendix A).

The discharge, Q, was obtained from multiplying the maximum cross sectional area through the channel by the velocity (Q =V ´ A).  The peak discharge values for each stream and corresponding rainfall intensities were listed in a data set (Appendix B).  These maximum flow rates were plotted against the rainfall intensity for that peak discharge using the computer software program Mathematica.  A function for a curve was applied to each data set depending on its shape.  Curves were extrapolated 75% beyond the sets of data to show expected trends at these locations for a given rainfall event.

In the second part of the study, a runoff coefficient was calculated for the streams using the Rational Method equation.

Q = K ´ C ´ I ´ A, where;

            (Q) = The peak discharge in cubic ft/ s

            (K) = A frequency correction factor

            (C) = The Runoff Coefficient (unit less)

(I) = The average rainfall intensity in inches per hour

(A) = The drainage area above the test location in acres

This 100-year-old equation is a simplified, practical approach to hydrological estimations.  It calculates discharge for a constant intensity rainfall but has many limitations due to its assumptions (Whipple 1983).  The equation was rearranged to find the runoff coefficient (C). The (K) variable will equal one for rain events in this study. Variable (I) was taken from National Weather Service measurements made at Mobile Regional Airport (Neilly 2002).  The heaviest duration of the rainfall usually lasted less than a full hour, so the rain total was divided by the fraction of the hour in which it occurred.  This approximated a storm total of the same concentration for a one hour period. The six watersheds were delineated on a 7.5 topographic map, and acreage was derived using a transparent, 1/16-inch dot grid over the map.

 Using the peak discharge and rain intensity for each creek from March 27, plus the land area, the runoff coefficient was solved for each basin.  Each calculated coefficient value was compared to an estimated one from land use and soil type. Based on data from the City of Mobile GIS office, soil types were predominately Alabama sandy loam or loamy sand which is Hydrologic soil type C (City of Mobile GIS 2002).  According to runoff coefficient tables, residential regions with soil type C had runoff coefficients close to 0.45, and neighborhood businesses had a runoff value of 0.70 (Purdue 2000).  A weighted-average was applied for the percentage of land covered by parking lots, which represented multi-family housing and businesses. These estimations represent the typical coefficient value expected for these type of suburban landscapes.

 

Results

Each creek had a degree of exponential growth in it's peak discharge with an increase in rainfall intensity (Fig. 5) , (Fig. 6) , (Fig. 7) , (Fig. 8) , (Fig. 9) , and (Fig. 10) .  This is expected as soils become saturated and more rain contributes to a stream's flow.  The stream with the largest watershed area (516.6 acres) and largest discharge per rainfall was Spencer Branch, and Bolton Drainpipe had the smallest watershed area (51.66 acres).  The watershed with the highest measured runoff coefficient was Spencer Br., followed by Little Spencer Br., then W Eslava Cr.  The lowest coefficient value was measured at Bolton Br. West.  See Table 1 .  The largest margin between measured and estimated runoff coefficients was for Little Spencer Br., then Spencer Br.  See Table 2 .

 

Discussion

Throughout this study, many educated assumptions were made pertaining to the measurements and equations.  The rational method technique is only valid for rainfall duration equal to the watershed’s time of concentration.  This is travel time from inlet to basin outlet or the time taken for the entire watershed to contribute to the flow of the stream (Whipple 1983).  For duration times less than this, the entire basin doesn’t contribute. For most of these observations, the rainfall duration exceeded the time of concentration.  Rainfall values for the rational method are usually obtained from an intensity-duration-curve based on recurrence interval for a particular region (Ferguson 1983). Rainfall amounts in this study were below the two-year recurrence interval.  Rainfall coverage was assumed uniform and justified by local radar when possible.  All test dates were preceded by a dry period of at least three days, which ensures no prior soil saturation.

 

Conclusions

The curves were only extended 75 percent beyond the data set to maintain minimal error.  Larger rainfalls are needed in order to show a more accurate trend and the equations used, including the rational method, are better suited to bigger rains. The shape of the curves, in Figure 5 to 10, was expected but the high numbers were not.  These large discharge values made an impact on the corresponding measured runoff coefficients.

Spencer Branch, Little Spencer Br and their tributaries have the longest reach of concrete channels than any other stream in this study. Spencer Br is channeled northward beyond Grelot Rd.  The extent of the concrete channels yielded a short time concentration and resulted in non-valid runoff coefficients (greater than one) for both creeks. In addition, the larger percentage of parking lot area in the upstream region of Spencer Br may have impacted the runoff results (City of Mobile 2002).  The large difference in runoff coefficients for Little Spencer Br may be due to this watershed’s steep terrain, which also lowers the time concentration.  The sediment deposited under a bridge at Little Spencer Cr, constricts the volume of water allowed under the bridge and could lead to channel overflow (See Fig. 4) .The lack of vegetation in the Spencer Br. channel may be attributed to the sharp peaks in water flow, not favorable to plant growth.

 The reason for vegetation in Eslava Cr. is due to a weak, steady flow of water that encourages growth.  The high runoff coefficient measured for W Eslava Cr. did not reflect the affects of the rough surface because of the many parking lots in the headwater region.

Parking lots were in the lower end of the Bolton Br West watershed, so runoff was more evenly drained.  Lack of parking lot area in the Little Bolton Br basin resulted in a low runoff coefficient.  Bolton Drainpipe drains an office complex nearby and may drain storm water from University Blvd., which would explain the large runoff coefficient there. The discharge from all three Bolton Br. tributaries is lessened since the slopes are lower than the other three sites and the length of the concrete channel upstream is small.  These lower slopes and a long bridge overpass at University Blvd., allowed for water to pool and aquatic life and vegetation to grow in all three.

Results show basins with impervious surfaces concentrated upstream contribute to high runoff rates.  Runoff in a watershed is focused along the drainage channel.  If this channel is concrete, stream flow increases and more discharge from upstream accumulates with inflow downstream resulting in high water levels. The variance of the roughness coefficient (n) in the Manning equation had a significant impact on stream velocity.  Vegetation and rougher concrete were witnessed to have an influence on a stream’s discharge amount. These observations could be implemented to retard water flow in upstream areas and lessen discharge totals downstream in concrete channels.  Along with more infiltration in the watershed and riparian boundaries, the risk of flooding would be reduced and the quality of water improved

 Future studies could observe runoff to rainfall relationships along the course of one channel and determine the impact due to the extent of the concrete channel.  More and larger rain events would aid in the accuracy of the findings.

 

References

 

City of Mobile GIS. 2002 Watershed Data Arcview shape files

 

Blankenship, Mike. 2001. DRCW JAN 2002 Newsletter Dog River Clearwater Revival.

 

FEMA July 6, 1998 Flood Insurance Study, Mobile County Alabama and Incorporated Areas . Volumes 1-3

 

Ferguson, Bruce K. 1998 Introduction to Stormwater John Wiley & Sons

 

Fetter, C. W. 2001. Applied Hydrology, 4th ed. Upper Saddle River, N.J. Prentice-Hall, Inc.

 

Neilly, Peter P. and UCAR 2001. http://www.met.fsu.edu/weather/

 

NOAA, National Climate Data Center 1998.  Mobile January 1998 Hourly Precipitation

 

Purdue University 2000. Runoff Coefficients for Rational Equation  http://abe.www.ecn.purdue.edu/~engelb/abe526/Runoff/C_table.html

 

Whipple, William and Randall, C.W. 1983.  Stormwater Management in Urbanizing Areas Englewood Cliffs, N.J. Prentice-Hall, Inc.

 

Acknowledgements

            Much thanks the following people for their unselfish time and assistance.  

John Crawford PE, City of Mobile Engineering Department, Scott Kearney and Sam Stutsman, City of Mobile GIS, Dr. Miriam Fearn, Dept of Earth Sciences USA.