Article Request Page ASABE Journal Article
Measuring the Erosion of an Irrigation Reservoir Levee
D. G. Wren, Y. Ozeren, M. L. Reba
Published in Transactions of the ASABE 59(1): 41-48 (doi: 10.13031/trans.59.10751). 2016 American Society of Agricultural and Biological Engineers.
Submitted for review in April 2014 as manuscript number NRES 10751; approved as a Technical Note for publication by the Natural Resources & Environmental Systems Community of ASABE in October 2015.
Mention of company or trade names is for description only and does not imply endorsement by the USDA. The USDA is an equal opportunity provider and employer.
The authors are Daniel G. Wren, Hydraulic Engineer, USDA-ARS National Sedimentation Laboratory, Oxford, Mississippi; Yavuz Ozeren, Research Scientist, National Center for Computational Hydroscience and Engineering, University of Mississippi, University, Mississippi; Michele L. Reba, ASABE Member, Research Hydrologist, USDA-ARS Delta Water Management Research Unit, Jonesboro, Arkansas. Corresponding author: Daniel G. Wren, USDA-ARS Watershed Physical Processes Research Unit, 598 McElroy Drive, Oxford, MS 38655; phone: 662-232-2926; e-mail: Daniel.Wren@ars.usda.gov.
Abstract. Increasing agricultural demands on limited groundwater resources in the Arkansas and Mississippi alluvial floodplain have created a growing need for the development of surface water resources. On-farm irrigation reservoirs are used to reduce dependence on groundwater supplies, but wind-induced wave erosion of levees creates an additional maintenance cost for producers. Measurements of levee erosion in a typical irrigation reservoir in east-central Arkansas were made with a ground-based LIDAR system and used to quantify erosion and estimate maintenance costs for irrigation reservoir levees. Approximately 1350 m3 of soil loss was measured from a 1530 m long levee during a seven-month period between December 2012 and July 2013. The damage was most severe in the northeast corner, where over 3 m of width were lost due to prevailing southerly winds blowing across the longest diagonal fetch length. Based on a typical cost (in 2014) for earth moving of $2.6 m-3, the cost to repair this loss would be approximately $3,500, resulting in a five-year maintenance cost of $17,500, which is about 16% of the initial cost of the levee. This research provides a basis for decision-making for producers who rely on irrigation reservoirs. Maintenance costs can be projected for a worst-case-scenario of levee erosion, and the cost of protection can be balanced against expected erosion and repair costs.
Keywords.Groundwater conservation, Irrigation reservoir, Levee erosion, LIDAR, Surface water storage, Wave erosion.
The need for water to irrigate crops in the Arkansas and Mississippi alluvial floodplain, which is traditionally called the Delta region, has led to steep declines in groundwater levels in both states (YMD, 2009; ANRC, 2012). Average annual reductions in groundwater elevation of 0.3 m per year have been documented for the Mississippi River Valley alluvial aquifer in Arkansas (ANRC, 2012). In spite of average annual rainfall of approximately 1.27 m, irrigated acreage steadily increased over the 25 years prior to 2008 (USDA, 2008). The limited duration and extent of the convective rainfall that dominates the July-August growing season mean that rainfall cannot be relied upon to provide sufficient water to maximize the growth of crops (Dyer and Mercer, 2013). Construction of on-farm reservoirs to store water has increased in recent years, with approximately 111 reservoirs constructed between 2000 and 2009 and 54 constructed in 2013 in Arkansas (Bowie, 2014). The levees that bound the reservoirs are constructed from local soils that may be susceptible to erosion because they are often low in clay content. Levee protection, because of its cost, is rarely included in the construction process. The costs of traditional protection methods, such as rock lining, are high. In a project completed in Arkansas in 2013, approximately $80,000 was spent to protect a 2.4 km levee using rock (Huber, 2013). This cost varies by location and is a function of the distance between the rock source and reservoir.
Little research has focused on waves and erosion in small bodies of water, such as irrigation reservoirs. Previous research on erosion by waves has typically concentrated on coastal erosion of sand beaches (e.g., van Rijn, 2011; van Rijn et al., 2011). In spite of research into the related problem of coastal bluff erosion, there is still no clear consensus on the failure mechanisms (Collins and Sitar, 2008). Ozeren et al. (2008, 2011) described the testing of floating wave barriers for levee protection, and Ozeren and Wren (2009) provided a method for predicting wave characteristics for small water bodies as a function of wind speed and fetch length.
In spite of the need to increase surface water use, the susceptibility of the reservoir levees to erosion, and the high cost of levee protection, a review of relevant literature revealed no direct measurements of erosion of an irrigation reservoir levee. These measurements are needed in order to provide an economic basis for decisions about levee protection and for long-term planning for reservoir management. Both of these factors impact the producers who rely on irrigation reservoirs for their livelihood. In this article, measurements of erosion from an irrigation levee are presented along with a basic economic analysis of levee repair and protection costs. The results also include a summary of wind and wave measurements collected from the irrigation reservoir over a 13-month period. This study is not intended to be an exhaustive exploration of the general problem of wave erosion; the results should be viewed as an initial assessment of earthen levee erosion that can be used to help quantify the cost of levee protection and to stimulate research relevant to the problem.
Materials and Methods
Erosion measurements were collected from an irrigation reservoir near Fisher, Arkansas (fig. 1), which will be referred to as Fisher Reservoir. The reservoir had a surface area of approximately 14 ha and a total interior levee length of 1530 m. The levees were approximately 2.7 m in height and 8 m in top width. The reservoir was in active use for irrigation, so fluctuations in water level between approximately 1 and 3 m were encountered during the period of data collection. The soil types in the area are Calloway silt loam (11% sand, 69% silt, and 20% clay) and Henry silt loam (10% sand, 70% silt, and 20% clay) (soil data obtained from the USDA-NRCS Web Soil Survey at http://websoilsurvey.sc.egov.usda.gov/App/HomePage.htm).
Figure 1. Location of Fisher Reservoir in northeast Arkansas.
Topography data for a large portion of the levee was collected with a ground-based LIDAR system. LIDAR uses time-of-flight or phase-difference measurements to obtain the distance from the laser source to a target using an automated system to scan over areas while recording the range, allowing a point cloud made up of the position of each scanned point to be constructed (Wehr and Lohr, 1999). This technology was previously limited to expensive airborne systems that made it impractical for small-scale studies. The availability of ground-based LIDAR systems has led to measurements of gullies (e.g., Perroy et al., 2010; Hancock et al., 2008), stream channel erosion (e.g., Thoma et al., 2005), and geomorphic change on archaeological sites (e.g., Collins and Kayen, 2006). LIDAR was chosen for this study because of its ability to quickly scan over a large area while providing sufficient detail to resolve local changes in topography caused by soil erosion. Additionally, since local soils were used for levee construction, and the soil type could vary, it was expected that the amount of erosion would not be uniform around the reservoir.
The measurements were collected with a Topcon GLS 1500 ground-based LIDAR system that used the phase-difference technique to achieve a 30 kHz scan rate. The device has a maximum range of 330 m and a single point accuracy of 4 mm at 150 m. The resolution is 1 mm at 100 m. Multiple setups were needed to scan the levees. It was possible to scan across the shorter axis of the reservoir, but not the longer one. A benchmark established through a static session with a GPS system was used to ensure that each survey would have a consistent point of reference. The location of the fill pipe and the walkway leading to the drain valve on the north end of the reservoir were used to make small (on the order of <10 cm) adjustments to align the LIDAR point clouds collected in 2012 and 2013.
The original LIDAR survey was performed on 19 December 2012, and the second survey was completed on 23 July 2013. The data were preprocessed using Topcon Scanmaster (ver. 220.127.116.11, Topcon, Livermore, Cal.) for georeferencing. LASTools (www.cs.unc.edu/~isenburg/ lastools/) and ArcMap (ver. 10.1, ESRI, Redlands, Cal.) were used prior to final analysis in Matlab (ver. R2013a, Mathworks, Natick, Mass.). The point clouds were stored and manipulated using the American Society of Photogrammetry and Remote Sensing LASer file format (LAS). Some operations, such as removal of selected point classes and combining LAS data sets, were accomplished using the libLAS library (http://liblas.org/).
As a supplement to the LIDAR survey, pin erosion measurements were collected on four different dates: 29 September 2012, 10 November 2012, 2 March 2013, and 18 July 2013. A total of 108 brass pins, approximately 1 cm in diameter and 60 cm long, were driven into the levees. The pins were driven parallel to the water surface and extended from the levee toward the reservoir. The pins were placed at approximately one-fourth and one-half of the levee elevation. The protrusion of the pins from the levee face was measured when they were installed and then four more times over the next ten months. The difference in the measured lengths yielded the amount of levee retreat between survey dates.
The levees were divided into 96 sections that were 10 m in length along the straight sections and 18 sections that were 2 m long in the corners. Elevation data along lines parallel to each levee axis were averaged together to produce one line of elevations normal to the levee axis for each 10 m section. These lines were used for a section-by-section comparison between the 2012 and 2013 LIDAR surveys. This approach minimized issues caused by slight misalignments in the surveys and noise in the data. Two sections of the levee were not measured: approximately 200 m (approx. 40%) of the west levee and 275 m (approximately 55%) of the east levee. Loss volumes for these sections were estimated by linearly interpolating between measured levee sections that bounded the unmeasured regions. This approach was adopted in lieu of adding no erosion in the unmeasured areas, since there was erosion from every point around the levee.
Data points representing vegetation growing on the levees had to be removed from the LIDAR data in order to achieve an accurate representation of the levee surface. The dense vegetation consisted mainly of wheat and green leafy plants similar in appearance to lamb’s quarters (Chenopodium album). The vegetation was mainly a problem for the initial scan in 2012 and ranged from approximately 0.1 to 1 m in height. The point cloud was imported into ArcMap to allow reclassifying of the data that resulted from vegetation. Sections about 5 m wide were reclassified by hand by selecting a profile of the levee. Points above the levee that were classified as vegetation were removed in a later step using LASTools.
Wind and Wave Measurements
Wave growth is proportional to fetch length; therefore, the largest waves are expected at the location with the longest fetch along the wind direction. Based on discussions with the landowner and others familiar with the area, the wind and wave measurement stations were placed in the northeast corner of Fisher Reservoir in order to maximize the fetch length for prevailing winds from the south and southwest. The fetch was approximately 470 m at the location of the measurement station for a south wind (fig. 2). Four 1 m long capacitance water level sensors were used to record the water surface displacement at a rate of 30 Hz. The water surface elevation signal was divided into 5 min segments for analysis. For each segment, spectral significant wave height (Hm0), maximum wave height, (Hmax), and peak spectral period (Tp) were calculated using frequency domain analysis (Goda, 2010).
A wind speed and direction sensor recorded data at 5 min intervals and was mounted 10 m above the water surface. Wind speed and direction were averaged in 5 min, 1 h, and 1 d intervals. The average wind speed for each interval is the arithmetic mean of the time series data in that interval. Wind direction was averaged by unit vector averaging in which lateral and longitudinal components of the observations were averaged separately for each time segment, and the resulting direction was calculated by vector summation.
Figures 3 and 4 show the severity of the erosion on the north levee of Fisher Reservoir. Block failure was the primary mode of failure, as evidenced by the vertical banks of the eroded levees (fig. 3). Figure 4 illustrates the qualitative change in the north levee from June 2012 to April 2013.
Figure 2. Aerial view of Fisher Reservoir and locations of the wind and wave instrumentation towers.
The amount of erosion from one 10 m section of the north levee is shown in figure 5 to illustrate how levee losses were calculated from the LIDAR data. Figure 5a shows the surface of the levee on 19 December 2012, and figure 5b shows the same section of the levee on 23 July 2013. Figure 5c shows that a large volume of soil was lost from this short section of levee. The plots shown in figures 6 through 8 are all based on the same type of analysis that produced figure 5. Figure 6 shows the volume of loss that occurred between the two LIDAR scans from all measured sections. The entire levee was not scanned, but based on observations of erosion, the most relevant sections for erosion were covered. The largest volume loss was measured on the eastern side of the north levee. Large volume loss was measured in small sections of the east, southwest, and south levees, while loss of all measured sections on the north end of the west levee was less than 10 m3 per section. The high erosion of sections of the east levee may be due to variations in soil type and in soil compaction during construction. Based on the data shown in figure 6, approximately 1350 m3 of soil was lost from the levee in seven months.
Figure 3. Wave damage on Fisher Reservoir in Arkansas (April 2013). Figure 4. Soil loss from Fisher Reservoir levees between June 2012 and April 2013.
Lateral retreat is the amount by which the width of the top of the levee was reduced. While loss volume can be readily used to estimate levee repair cost, lateral retreat may be a more intuitive measure of levee loss. The percentage of 10 m sections of the measured part of each levee with a given amount of lateral retreat is shown in figure 7. The north levee (fig. 7a) had the highest percentage of high-loss sections, followed by the south levee (fig. 7b). The east (fig. 7c) and west (fig. 7d) levees had only a few sections with lateral retreats over 2 m. The maximum lateral retreat for each 10 m section of each levee indicates the degree of risk of levee failure (fig. 8). Since it only takes a breach at one point for the levee to fail and the stored water to be lost, the maximum lateral retreat is a useful index. Based on this approach, all of the levees had at least one section with >2 m of lateral loss. Figure 8a shows that the north levee had 13 of the 26 surveyed sections with >2 m of lateral loss and that retreat was greatest on the east end of the north levee. The south levee (fig. 8b) and the north end of the east levee (fig. 8c) also had substantial losses, but they were still lower than the north levee. The west levee (fig. 8d) had ten sections with retreat >1 m.
Figure 5. Example of one 10 m section of LIDAR data showing 2012 and 2013 scans and soil loss on the north levee of Fisher Reservoir for the (a) 2012 scan, (b) 2013 scan, and (c) profile view of both sections. Figure 6. Volume loss (m3) in 96 straight 10 m and 18 2 m corner sections from levees at Fisher Reservoir from December 2012 to July 2013.
The pin erosion measurements from 29 September 2012 and 10 November 2012 (fig. 9) showed a surprising amount of erosion before the reservoir was filled. This early erosion was caused by runoff from rain falling directly on the levees, and rill erosion was the primary mechanism. When a pin was in a rill, it showed as much as 20 cm of loss due to runoff. After the reservoir was filled, only two more pin measurements were completed (2 March 2013 and 18 July 2013), and they showed large amounts of erosion. It can be seen in figure 9 that there were far fewer pin measurements in 2013. This is because the pins were lost as large blocks of soil collapsed into the reservoir. If the pin method is to be used successfully, much longer pins will be required. A technique for measuring the pins while the water level is above the pins would also be needed. Based on the large amount of soil loss and the difficulty in measuring the pins when submerged, it is likely that survey techniques, such as LIDAR or photogrammetry, are better suited for quantification of soil loss from irrigation reservoir levees.
Figure 8. Maximum lateral retreat of the (a) north, (b) south, (c) east, and (d) west levees. Figure 9. Pin erosion measurements from Fisher Reservoir. Levee retreat is indicated by exposed length of pins after different time periods.
Figure 10 shows a directional histogram of winds measured between 15 June 2012 and 15 February 2013. The slices show the frequency of winds from each of 16 directions during the eight-month period, and the gray-scale gradient indicates the distribution of wind speed for each direction. Figure 10 shows that the prevailing winds were from the south. Figure 11 shows histograms of 30 min average wave period, wave height, and wind speed measured between June and September 2012. The data presented in these histograms are filtered for south winds so that the fetch length at the measurement station would be approximately the 470 m line depicted in figure 2. Table 1 shows a comparison of the measured wind and wave parameters during sustained high wind events from the present study and from two older studies (Ozeren and Wren, 2009). The data show that winds of up to 14 m s-1 were observed at the studied reservoirs, generating waves with significant heights of up to 0.25 m at a 0.6 km fetch. The wind and waves observed in Fisher Reservoir were not exceptionally severe; they were both lower than previously observed in other reservoirs.
Figure 10. Wind speed and direction histogram for 15 June 2012 to 15 February 2013 at the northeast corner of the study site.
Figure 11. Wave period, wave height, and wind speed histograms for the time period between June 2012 and October 2012.
Based on a total loss of 1350 m3 of soil from the levee of Fisher Reservoir, and a typical cost (in 2014) for earth moving of $2.6 m-3, the cost to repair the levee would be approximately $3,500. If no protection is put in place and erosion continues at this rate, the five-year maintenance will be $17,500. At a cost of approximately $72 m-1 to build reservoir levees (Delp, 2014), Fisher Reservoir would have an initial cost of approximately $110,000. Based on
this figure, the five-year maintenance cost is approximately 16% of the initial cost of the reservoir. These data provide a target for mitigation strategies, which, to be successful, must be able to reduce the overall cost of using the levee to an amount that is below the maintenance cost.
Table 1. Comparison of wave parameters measured in reservoirs. Characteristic
Wind speed (U10, m s-1) 14 (NE) 13 (SW) 9.5 (S) Fetch length 586 450 470 Wave height (Hm0, m) 0.25 0.23 0.14 Maximum wave height (Hmax, m) 0.38 0.39 0.24 Peak period (Tp, s) 1.7 1.6 1.4
As mentioned earlier, the cost of using rock to protect a levee was approximately $80,000 (Huber, 2013). Note that this is only an example, since the cost of rock is dependent on location and the cost of hauling. The levee was 2.4 km around the perimeter, and rock of approximately 10 cm diameter was placed 0.3 m deep in 2.4 m. This is approximately $32.80 per running meter of levee. The total of $80,000 for rock equates to about 20 years of annual maintenance cost, based on the assumptions stated previously. Since it is likely that levee maintenance costs per unit of levee length increase in proportion to reservoir size, and since longer fetch lengths will result in higher amplitude waves, the maintenance cost discussed here may not accurately represent what will be needed on other reservoirs.
An optional strategy for reducing the potential for erosion of irrigation reservoir levees is to maintain the lowest water level possible until just before the water will be needed for irrigation; however, there must be a balance between filling the reservoir in a timely fashion during rain events and minimizing erosion. One strategy is to fill the reservoir to 50% full in December, 75% full in February, and then 100% full in April in anticipation of irrigating production fields in May (Wimpy, 2013). This strategy requires balancing the risk of levee damage with the risk of insufficient rain to fill the reservoir in April. Further study will be required to assess the viability of this method and how it might be used in conjunction with techniques for reducing fetch length.
Summary and Conclusions
Measurements of soil erosion from an irrigation reservoir levee between 19 December 2012 and 23 July 2013 were presented. The width of the levee was reduced by up to 3 m, and more than 1350 m3 of soil was lost over the measured sections of the levee. Based on conservatively estimated costs, this damage would require approximately $3,500 for repairs. If similar amounts of erosion continue, levee integrity may be compromised and accumulated maintenance costs may reach a significant fraction of the original cost of the levee system. The damage was concentrated on the northeast corner of the reservoir, which coincides with the dominant southerly winds and the longest diagonal fetch length. There are numerous possible methods for reducing the erosion of these levees, such as geotextiles or lining with rock, but the cost of such methods has limited their adoption. This research should help in providing a basis for decision-making for producers who rely on irrigation reservoirs for their livelihood. By using the information here as a worst-case scenario, maintenance costs can be projected and the cost of protection can be balanced against expected erosion and the resulting repair costs.
The results provide a benchmark for estimating levee maintenance costs that can be used as a guide for planning the protection of an irrigation reservoir. They also highlight the need for continued research into an important component of systems aimed at conserving declining groundwater sources. The main needs for future research are:
- Determine relationships between wave energy and erosion for cohesive soils.
- Long-term economic impact of investment in levee protection.
- Maximum fetch length and optimized reservoir shape guidelines that provide resistance to erosion from waves.
- Comprehensive design guidance, including material specifications, for techniques, such as floating barriers or interior levees, that may be used to reduce wave energy.
We would like to thank Keith Admire at the USDA-NRCS National Water Management Office in Little Rock, Arkansas, for supporting this work. Glenn Gray provided invaluable technical support for all phases of the work. Jacob Ferguson and Alan Barger provided technical support during the field station installation and maintenance. Scott Treece, W. Jonathon Delp, and Leonard Rawlings provided help during field data collection.
ANRC. (2012). Arkansas Groundwater Protection and Management Report for 2012. Retrieved from http://anrc.ark.org/news-publications/annual-reports/.
Bowie, C. (2014). NRCS State Irrigation Engineer for Arkansas. Personal communication.
Collins, B. D., & Kayen, R. (2006). Applicability of terrestrial LIDAR scanning for scientific studies in Grand Canyon National Park, Arizona. USGS Open-File Report 2006-1198. Reston, Va.: U.S. Geological Survey.
Collins, B. D., & Sitar, N. (2008). Processes of coastal bluff erosion in weakly lithified sands, Pacifica, California. Geomorphology, 97(3-4), 483-501. http://dx.doi.org/10.1016/j.geomorph.2007.09.004.
Delp, W. (2014). Personal communication. NRCS State Conservation Engineer for Arkansas.
Dyer, J., & Mercer, A. (2013). Assessment of spatial rainfall variability over the lower Mississippi River alluvial valley. J. Hydrometeorol., 14(6), 1826-1843. http://dx.doi.org/10.1175/JHM-D-12-0163.1.
Goda, Y. (2010). Random Seas and Design of Maritime Structures. Advanced Series on Ocean Engineering, Vol. 33. Singapore: World Scientific.
Hancock, G. R., Crawter, D., Fityus, S. G., Chandler, J., & Wells, T. (2008). The measurement and modelling of rill erosion at angle of repose slopes in mine spoil. Earth Surf. Proc. Landforms, 33(7), 1006-1020. http://dx.doi.org/10.1002/esp.1585.
Huber, B. (2013). Personal communication. Agricultural producer, Weiner, Arkansas.
Ozeren, Y., & Wren, D. G. (2009). Predicting wind-driven waves in small reservoirs. Trans. ASABE, 54(2), 1213-1221. http://dx.doi.org/10.13031/2013.27793.
Ozeren, Y., Wren, D. G., & Alonso, C. V. (2008). Development of floating wave barriers for cost-effective protection of irrigation pond levees. Trans. ASABE, 51(5), 1599-1612. http://dx.doi.org/10.13031/2013.25317.
Ozeren, Y., Wren, D. G., Altinakar, M., & Work, P. A. (2011). Experimental investigation of cylindrical floating breakwater performance with different mooring configurations. J. Waterway Port Coastal Ocean Eng. ASCE, 137(6), 300-309.
Perroy, R., Bookhagen, B., Asner, G., & Chadwick, O. (2010). Comparison of gully erosion estimates using airborne and ground-based LiDAR on Santa Cruz Island, California. Geomorphology, 118(3-4), 288-300.
Thoma, D. P., Gupta, S. C., Bauer, M. E., & Kirchoff, C. E. (2005). Airborne laser scanning for riverbank erosion assessment. Remote Sensing Environ. 95(4), 493-501. http://dx.doi.org/10.1016/j.rse.2005.01.012.
USDA. (2008). Farm and ranch irrigation survey (2008). AC-07-SS-1. Washington, D.C.: USDA National Agricultural Statistics Service. Retrieved from www.agcensus.usda.gov/Publications/2007/Online_Highlights/Farm_and_Ranch_Irrigation_Survey/fris08.pdf.
Van Rijn, L. C. (2011). Coastal erosion and control. Ocean Coastal Mgmt., 54(12), 867-887. http://dx.doi.org/10.1016/j.ocecoaman.2011.05.004.
Van Rijn, L. C., Tonnon, P. K., & Walstra, D. J. R. (2011). Numerical modelling of erosion and accretion of plane sloping beaches at different scales. Coastal Eng., 58(7), 637-655. http://dx.doi.org/10.1016/j.coastaleng.2011.01.009.
Wehr, A., & Lohr, U. (1999). Airborne laser scanning: An introduction and overview. ISPRS J. Photogram. Remote Sensing, 54(2-3), 68-82. http://dx.doi.org/10.1016/S0924-2716(99)00011-8.
Wimpy, M. (2013). Personal communication. Agricultural producer, Jonesboro, Arkansas.
YMD. (2009). Irrigation water use in the Mississippi Delta: 2009 report. Stoneville, Miss.: Yazoo Mississippi Water Management District. Retrieved from www.ymd.org/pdfs/wateruse/2009%20Mississippi%20Delta%20Irrigation%20Water%20Use%20Report.pdf.