Hydrologic modification and channel evolution degrades connectivity on the Atchafalaya River floodplain
Funding information: Audubon Louisiana; Louisiana Department of Natural Resources, Atchafalaya Basin Program; US Army Corps of Engineers
Abstract
The Atchafalaya River Basin is the largest remaining forested wetland in the contiguous United States. Since 1960, dredging and channel erosion in the Basin have resulted in changes to the hydrologic connectivity that have not been quantified. Analyses were conducted to determine the hydraulic and geomorphic factors that have changed since discharge became controlled that may have decreased river/floodplain connectivity. We examined: (1) stage/discharge relationships from 1960 to 2014; (2) hydroperiods across the floodplain; (3) discharge distribution to the floodplain by comparing discharge measurements from 1959–1968 to 2005–2012; and (4) channel cross-sections and floodplain elevations. Our results indicate that much of the floodplain no longer receives headwater discharge (upstream to downstream, > 200 km2) or receives too little discharge to alleviate stagnancy and hypoxia in the forested wetland at lower stages. Large portions of the Basin (400 km2) have low water levels controlled by channel geomorphology and sea-level rise that inundate the forested floodplain for more than 50% of the calendar year. This extended duration of inundation contributes to hypoxia and likely reduces nutrient retention. The confinement of discharge to a large efficient channel compromises the ability of this system to respond to sea-level rise and subsidence. This study provides insight to the effects of flood management projects along Coastal Plain rivers and deltas.
1 INTRODUCTION
Large river deltas worldwide are ecologically challenged by the interacting forces of sea-level rise, subsidence, and human development (Brown & Nicholls, 2015; Nicholls et al., 2020). These deltas are formed by large rivers such as the Amazon, Ganges/Brahmaputra, Indus, Mekong, Nile, Volta, and the Mississippi/Atchafalaya that carry and deposit high loads of sediment. Deltas are typically flat, vast, and sit upon hundreds or thousands of meters of deposited sediment. Delta formation is driven by prograding channels, changing and abandoning courses regularly as hydraulic gradients dictate (Kolb & Van Lopik, 1958).
During the initial stages of channel avulsion, an inefficient, undersized channel is incapable of transmitting increasing discharge. This channel progrades, eroding and depositing through lower areas. During the initial erosion of the channel, overbank flooding occurs during relatively minor discharges with a high percentage of discharge crossing the floodplain until the channel cross-section becomes competent and efficient (Roberts, 1997). At this stage of increased floodplain interaction, widespread portions of the stream valley experience heavy sedimentation both from the eroding channel and the watershed (Kroes & Hupp, 2010; Pierce & King, 2007; Roberts, 1997). As channels are abandoned, sediment delivery to areas of the floodplain is reduced and subsidence processes become dominant over large areas until hydraulic gradients cause channels and distributaries to switch and once again deliver sediment (Nicholls et al., 2020).
Subsidence, sea-level rise, levees, and channel dredging have all affected the hydrology of the Atchafalaya River (AR) Basin in south central United States. Subsidence can result in higher water levels relative to the land surface (increased hydroperiod and flooding) changing the ecology and human utilization of the area. An alternative, common scenario occurs when people want to use the floodplains for farming and living. River flooding is an impediment to those uses. Sea-level rise and climate change are compounding issues for these scenarios (Le et al., 2007; Nicholls et al., 2020; Schmidt, 2015). Common engineering solutions to fight flooding and drainage issues are to construct levees to keep water out of agricultural and developed areas, and dredge channels to move water quickly downstream (Le et al., 2007; Kroes & Hupp, 2010; Schmidt, 2015).
The AR Basin is the largest remaining bottomland hardwood wetland of the contiguous United States (Figure 1). The Basin formed geomorphically as a backswamp distributary channel floodplain between relict channels of the Mississippi River (MR) within the MR delta. The western and southern boundaries were formed by natural levee deposits of the MR's Teche channel, and the eastern boundary was formed by levee deposits of the Salé-Cypermort and LaFourche channels (Kolb & Van Lopik, 1958). The geomorphic boundary of this forested wetland encompasses an area of approximately 5700 km2. Half of this area has had its hydrology dominated by the AR since completion of the protective levees in 1955.
The AR has been heavily modified over the past 170 years. Channels have been created, straightened, and enlarged in cross-section by the US Army Corps of Engineers (USACE) to provide efficient flood-control for large cities along the lower MR and to facilitate navigational commerce. Channel training is the engineering process of forcing a poorly defined, shallow, braided, or multiple channel waterway into a single channel and/or concentrating discharge into a channel while reducing off channel discharge. In the Basin, a preferred channel was dredged through existing segments of lakes and channels and by placing spoil along the channel (Hardee, 1870). Other large and small channels were partially or completely blocked with earth and rock dams to the bank level as part of USACE channel training (Reuss, 1998; USACE, 1979). During modifications to the AR, it began to draw an increasing portion of the MR discharge (see section Study Area for modification details and references). The risk of the MR switching channels became evident, and the decision was made to regulate the AR discharge to approximately 30% of the MR and Red River combined discharge (P.L. 780, 83rd Congress).
These modifications have, over time, nearly eliminated direct overbank flooding from the river to the floodplain and concentrated discharge to the floodplain into distributary channels. Overbank or bankfull stage in this study indicates the stage required to inundate the top of the primary river channel bank resulting in non-channelized sheet-flow over the bank and is classically defined as occurring with a 1.5-year flood interval discharge (Leopold et al., 1964). The lack of overbank flooding results in numerous effects. Following dredging, the remaining reduced discharges across the floodplain have been shown to favor sedimentation in backswamp areas (Hupp et al., 2008, 2015), and to promote floodplain channel and lake filling (Hupp et al., 2019; Kroes & Kraemer, 2013). Nutrient trapping is reduced (Bennett et al., 2014; Jones et al., 2014) when compared to periods when bankfull stage is exceeded because the amount of sediment moving over the floodplain is reduced (Scott et al., 2014). Oxygen concentrations in water over the floodplain may be reduced due to slow moving waters and a buildup of organic material (Pasco et al., 2015).
Sediments deposited during the previous hydrologic regime (pre-1969) and spoil placement have left more than 200 km2 of the 2014 floodplain ringed by high-elevation sediment deposits with a singular input/output channel during much of the year causing water residence times that can exceed 6 months (Allen et al., 2008; USGS, 2014; BLR gage data). These units are essentially backwater lakes and only rarely experience flow-through hydrologic connectivity (Allen et al., 2008; BLR gage data). These long residence times combined with high water temperatures (> 30°C) and very high organic carbon deposition (340 g C/m2/yr) result in a hypoxic water column (Hupp et al., 2019; Mehring et al., 2014; Todd et al., 2007; Utley et al., 2008). In addition to limiting fisheries composition (Justus et al., 2014), hypoxia can cause the export of previously deposited nitrogen from sediments as ammonia (Bason et al., 2017; Jones et al., 2014; Zhang et al., 2011).
There have been limited analyses to date of channel erosion, stage/discharge relationships, or the effect of distributary channel blockages or openings on the hydrology of the Basin (Newman, 2020; Tang et al., 2021). The objectives of this study were to examine the changes in stage/discharge relationships, channel cross-sections, discharge distribution, and stage exceedance since 1960 so that the hydrologic effects of channel training and the subsequent channel adjustments can be assessed in the context of water moving from the AR and over the floodplain. These analyses will help to determine the hydrologic and geomorphic factors that have changed since discharge regulation that could result in decreased dissolved oxygen concentrations in the water column and promote conversion of forested wetland to open water. While the study area is a freshwater swamp delta lobe contained within the Holocene MR deltaic plain (Fisk, 1952), the issues of vast area, low hydraulic gradients, low elevations, sea-level rise, high subsidence rates, and sediment deposition (both high and low rates) are common among many large deltas worldwide. Long-term data on most of the world's deltas are rare if they exist (Lehner & Grill, 2013) and management choices often must be made without information as human populations struggle with increasing flooding and sea levels in this changing climate. Here we compare this system as it existed in its previously altered state beginning with discharge and water-level monitoring from 1960 through 2014. This study provides information for modeling and management of the hydrologic effects of channel training on a vast floodplain after the projects are completed.
2 STUDY AREA
2.1 Atchafalaya hydrologic characteristics
The hydrologic characteristics of the Basin have changed dramatically over the last 170 years from an interfluvial backswamp lake with distributary inland deltas and periodic discharge from the MR, to an active floodplain of the AR (Fisk, 1952). During this conversion, an inland delta filled 470 km2 of lake area before 1950 (Hale et al., 1999; Piazza, 2014; Roberts, 1998). Many of the changes occurred prior to the monitoring of water levels and discharge conditions, thus, it is difficult to define the system prior to modern monitoring (since about 1955). Further, the Basin continues to geomorphically evolve through manmade and natural alterations toward a state that will likely capture the MR, as has occurred several times over the last 5000 years (Latimer & Schweizer, 1951).
The history of the Basin modification has been summarized by Reuss (1998) and Mossa (2013) (Figure 2). In the context of this study, the Basin was recognized as a navigation route and a possible solution to flooding in southern Louisiana in 1850 after the New Orleans flood of 1849 (Wooldridge, 1850). The connection between the AR and MR required dredging for navigation throughout the 1890s when levee construction began along the upper AR (SLBSE, 1900). Following the placement of the river levees (on both sides of the main channel with a narrow set-back from the channel banks), channel erosion began (Ockerson, 1907) and was likely enhanced by several years of flooding during 1911–1922.
2.2 Measures since the 1928 Flood Control Act
The 1927 MR flood was one of the highest recorded stage floods on the lower MR resulting in catastrophic flooding. This flood prompted the 1928 Flood Control Act designating the Project Flood (hypothetical maximum MR flood) distribution of discharge through the Basin as 42,500 m3/s to prevent the MR levees downstream of the AR divergence from being overtopped by flood waters (Kemper, 1929; USACE, 2008). To enable this capacity, levees were erected along the main stem of the AR to 88 km downstream of the MR divergence (1890–1948, Herbert, 1967), and a singular main channel was dredged (1932–1969; Fisk, 1952; USACE, 1979, 1982). USACE contracts indicate that for the mainstem of the AR from 1932 to 1961, 132 × 106 m3 of sediment was removed by dredging (Lamar Hale, USACE, written communication, 2001). Contracts for 1962–1969 indicate 125 × 106 m3 of sediment was removed by dredging upstream of Morgan City (Lamar Hale, USACE, written communication, 2001). No dredging contracts were issued for the river mainstem, upstream of Morgan City (MC), from 1970 to 2015 except for work near the Old River Control Structure (ORCS) (Supporting Information Table S1; Lamar Hale, USACE, written communication, 2001). Protection levees (placed on either side of a floodway with a large distance between the levee and the channel banks) were created (1935–1955) forming a floodway approximately 190-km long and approximately 25-km wide (Reuss, 1998; USACE, 1979). In the lower half of the Basin, the east Gulf Intracoastal Waterway (GIWW) and west Freshwater Distribution Canals (FWDC) were dredged to help convey water to the floodplain. A secondary outlet, the Wax Lake Outlet (WLO), was dredged through the relict Teche channel levee deposits in 1941 (Latimer & Schweizer, 1951).
| Gage name (partial) | Abbreviation | USACE gage number | USACE period of record | USGS station number | USGS period of record | Data type | % missing days |
|---|---|---|---|---|---|---|---|
| Simmesport | SP | 03045 | 1930–2014 | 07381490 | 2009–2014 | S, Q | 0 |
| Butte La Rose | BLR | 03120 | 1930–2014 | 07381515 | 1996–2014 | S | 0 |
| Arm of Grand Lake | AGL | 49,197 | 1976–2014 | 300312091320000 | 2003–2014 | S | 28 |
| Buffalo Cove | BC | 49,235 | 1976–2014 | 07381567 | 1996–2014 | S | 18 |
| Keel Boat | KB | 03615 | 1957–2014 | — | — | S | 23 |
| Bayou Sorrel Lock | BSL | 49,630 | 1942–2014 | — | — | S | 0 |
| Chicot Pass at Myette Point | MP | 03540 | 1963–2014 | 073815450 | 1996–2014 | S | 18 |
| Little Bayou Sorrel | LBS | 49,725 | 1970–2014 | — | — | S | 22 |
| Morgan City | MC | 03780 | 1932–2014 | 07381600 | 1992–2014 | S, Q | 1 |
| Wax Lake Outlet | WLO | 03720 | 1942–2014 | 07381590 | 1995–2014 | S, Q | 10 |
- Abbreviated notation and gage numbers, bold typeface names = mainstem sites, italic typeface names = interior, boat access sites. Some gages are currently maintained by the US Geological Survey (USGS), — = no USGS station (USGS, 2020). Data type S = stage, Q = discharge. % missing days = percentage of days missing from gage record from 1960 or start of gage record through 2014.
Numerous modifications to the AR resulted in increased hydraulic efficiency that began to capture discharge from the MR. To prevent the capture, the Flood Control Act of 1954 (P.L. 780, 83rd Congress) designated two flow-control structures that were completed in 1962. The percentage of latitudinal discharge (the combined total discharge of the MR and Red River) going down the AR was designated to be 30% (P.L. 780, 83rd Congress). Since 1962, discharge has been at approximately that distribution except during floods and Louisiana-requested deviations (LADNR, 2013). An additional structure was completed in 1986, and a hydropower plant was completed in 1992. Collectively, these structures comprise the ORCS (USACE, 2009).
The combined changes resulted in substantial channel erosion (Mossa, 2016; Tang et al., 2021) that likely changed the hydrologic interaction between the river and floodplain. However, there have been numerous other processes, projects, and distributary channel blockages within the Basin since that time. USACE rangeline (RL) surveys and records indicate that the majority of spoil placement and dredging volumes within the study area occurred between 1957 and 1969 with final channel dredging dimensions attained between 1962 and 1969 (Nelson & Gaea, 2003; USACE, 1979). Spoil placement occurred from the head of Whiskey Bay to 3 km upstream of MC (Mossa et al., 2019; USACE, 1979) and was still visible in the 2012 light detecting and ranging (LiDAR) dataset (USGS, 2014) downstream to approximately 9 km upstream of MC. The majority of distributary channels (hereafter defined as any channel that distributes water from the primary river channel to the floodplain at sub-bankfull stages) were blocked from 1955 to 1965 (USACE, 1979) with additional blockages occurring through 2021 (CPRA, 2019). The remaining distributary channels required for navigational access were realigned (straightened and connection to river relocated; Nelson & Gaea, 2003; USACE, 1979). The portion of the AR from Myette Point (MP) to MC was dredged through an existing lake from 1960 to 1966 and spoil placed adjacent to the dredged channel to force the AR to form a single main channel in that area, although discharge was still distributed though several lakes and channels as of 1974 (USACE, 1979). In 1987, a weir was placed across the WLO to maintain the MC channel of the AR but was removed in 1995 because of increased flooding near MC and likely increased water levels throughout much of the study area during moderate to higher discharges (Majersky et al., 1997; Figure 1). There have been multiple projects constructed to increase water inputs onto portions of the floodplain (CPRA, 2019; USACE, 2004). Throughout the floodplain, there has been extensive development of canals and placement of spoil banks for pipelines, petrochemicals, and timber removal (Kroes & Kraemer, 2013).
2.3 Modern Atchafalaya floodplain
Long-term observations by environmental groups indicate that there is “less water” in the Basin than was historically observed and navigation across the floodplain is hindered by infilling lakes and channels (LADNR, ABP, 2017a). Restoration efforts by the USACE and the State of Louisiana have been to provide “more” water and to reduce sediment in that water (LADNR, ABP, 2017b). While it is unclear what “less” water means, “more” water focuses on increasing discharge and flow velocities across the floodplain. Widespread hypoxic conditions exist that limit fisheries habitats (Justus et al., 2014; Kaller et al., 2011; Pasco et al., 2015; Sabo et al., 1999a, 1999b), and in some areas, permanent hydroperiods and stagnancy limit tree regeneration and growth (Faulkner et al., 2009; Keim & Amos, 2012). Further, the river has only reached bankfull stage (6.8 m) once at the Butte La Rose (BLR) gage (in 2011) since 1973 (USACE, 2022b). Inundation of the floodplain along this river normally occurs at a stage several meters below bankfull stage at BLR via channeled distributary discharge (Allen et al., 2008; Hupp et al., 2008; USGS, 2014). Primary distributaries have high spoil banks and natural levees that were formed during a previous flow regime and are no longer submerged in a 1.5-year flood in 2014 (Allen et al., 2008; Hupp et al., 2008; USGS, 2014). Because the majority of water does not enter the floodplain directly from the river or primary distributaries, secondary and tertiary distributary channels are the main conduit of water to the floodplain, causing backwater flooding to large areas and inducing complex flow patterns across the floodplain (Allen et al., 2008; Hupp et al., 2008).
The AR received on average 23% of the discharge of the MR from 1991 to 2010 (Heath et al., 2015) as well as the entire discharge of the Red River to approximate an annual 30% of the latitudinal discharge. The Basin traps 12.5 billion kg of sediment annually, of which 695 million kg is organic carbon, and thus considered a globally important carbon sink (Hupp et al., 2019). Deltas forming in the Basin (lacustrine) and from the two Atchafalaya outlets (estuarine) to the Gulf of Mexico are the primary locations of land growth along the eroding and subsiding coastline of Louisiana (Couvillion et al., 2011).
The upstream extent of the study area was the old AR and the Whiskey Bay Pilot channel divergence (the head of the Whiskey Bay channel). The downstream extent was defined by the two outlets, MC and WLO. The sides of the study area were defined by the east and west protective levees (Figure 1).
3 MATERIALS
Water levels (gage data), topography/bathymetry (surveyed RLs and hydrographic surveys), and discharge distribution (synoptic discharge surveys) of the Basin inside the protective levees have been well documented (USACE, 1979, 2022b; USGS, 2020). The Simmesport (SP) gage, upstream of the study area, provides recorded river discharge data with a mean annual discharge of 6500 m3/s (1960–2013; USACE, 2022a). Gages with records starting before 1960 were selected on the main stem of the AR. Gages on the floodplain had various start dates (Table 1, Figure 1).
Discharge records for the AR at SP were used to recreate discharge exceedance-probability curves (USACE, 2022a). The USACE (1979) previously designated a 1.5-year flood (hereafter 1.5-year high) on the AR to be 11,500 m3/s based on the long-term discharge records of the MR and 30% AR distribution with 10% of days exceeding this discharge. The 1.5-year low water discharge (hereafter 1.5-year low) was determined to be 2700 m3/s with 90% of days exceeding this discharge (USACE, 1979). This 90% exceedance has been defined as a critical boundary for instream biological habitat (Bovee, 1982). Both the 1.5-year high and 1.5-year low occurred in 40 of 56 years of record (USACE, 2022a).
Numerous topographic RL and hydrographic surveys were completed from 1932 to 1974. We compared the elevations of the eastern half of the 1974 RL18 (Figure 1; USACE, 1979) and the 2012 LiDAR for an area previously identified as being hydrologically isolated (Allen et al., 2008; Hupp et al., 2019). Elevations were adjusted for datum changes as determined by the National Geodetic Survey (NGS) Coordinate Conversion and Transformation Tool (NGS, 2020).
Hydrographic surveys used for this analysis were conducted from 1962 to 1964 (hereafter 1962; USACE, 1967), 1974–1976 (hereafter 1974; USACE, 1977) and 2010 (USACE, 2012). For analyses, 40 cross-sections were selected longitudinally along a reach of channel from the head of Whiskey Bay to MC with an interval of 2 to 3 km avoiding bends and abnormal holes. Earlier surveys were point-type surveys that were measured across the channel and over the top of the banks. The 2010 survey was a multibeam sound navigation and ranging (SONAR) survey that traversed from water edge to water edge (USACE, 2012). For comparison between survey types, the 2010 survey sections were point-measured at 20 equal-width intervals across the channel to top of banks. The 2010 bank height and distance from the hydrographic survey edge were measured from the 2012 LiDAR (USGS, 2014). The 2010 survey did not include the old AR channel or the Six Mile Lake (SML) (Figure 1), which were surveyed in early 2015 during high water, using the 2012 LiDAR to determine bank height (Kroes, 2022, details in methods).
Discharge distribution divergent from the main channel was calculated using discharge measurements made by USACE and US Geological Survey (USGS) in coordinated synoptic sampling missions to capture discharge distribution for a narrow range of AR discharge, similar in concept to the synoptic measurements of Hiatt and Passalacqua (2015). USACE measurements were used for the period 1959–1968 (USACE, 1979) and USGS measurements were used for the period 2005–2012 for a range of discharges from 2700 m3/s to 11,500 m3/s (1.5-year low to 1.5-year high; USGS, 2020). These calculations allowed for comparison of discharge distribution during channel training (1959–1969) with discharge distribution as it existed during 2005–2012. USACE 1959–1968 measurements were likely done using a Price AA mechanical flow measuring device. The accuracy of the historical USACE measurements is unknown but should be expected to have an error of 5% to 10% (Buchanan & Somers, 1969). USGS discharge measurements were made using an acoustic Doppler current profiler (ADCP) and software using USGS standard procedures for ADCP measurements (Simpson, 2001). For these measurements, tolerance of variation within a measurement was designated to be acceptable if ≤ ±5% discharge variation was observed between a minimum of four repeated transects in paired left/right bank starting sequence.
4 METHODS
- Stages of discharge were examined for change over time.
- Stage exceedance curves were calculated to determine if stages changed, to identify the elevation of obstacles to flow on the floodplain, and to identify the stage to which the floodplain will drain.
- Discharge distribution off the AR channel was measured to determine the effect of channel changes and training on discharge across the floodplain.
- Changes in the main channel cross-section and elevations of the spoil banks and floodplain were calculated to determine causes of stage and hydroperiod changes.
4.1 Changes in stage/discharge relationships
The long-term AR Basin gages (Table 1) were examined for stages measured during 1.5-year high, low, and median discharges (USACE, 2022b; USGS, 2012, 2020). These gages were in locations that allowed for the calculation of hydraulic gradients across the Basin. Because there is a large water storage capacity on the floodplain of the Basin, rising and falling limbs of the hydrograph may have different stages for a specific discharge. To minimize this stage/discharge variability with floodplain storage, only the first annual occurrence of the target discharge on the rising limb of the hydrograph was considered for stage/discharge analysis.
4.2 Stage exceedance
Stage exceedance was developed for long-term gages of the Basin (Table 1). Data were compiled for each gage from 1960, or from the start of each gage's record, through 2014. Stage data were sorted by elevation, and then exceedance was calculated by the proportion of daily stage values greater than or equal to any specified stage. Data were subdivided into 5-year periods, and exceedance was calculated for each 5-year period (i.e., 1960–1964, 1965–1969, etc.) to help reduce normal interannual variance while maintaining short-term resolution. The same method used to determine stage exceedance was used to calculate discharge exceedance at the SP gage. MC stages with greater than 60% exceedance (i.e., lower elevation stages) were analyzed between the years 1970–1979 and 2005–2014 for mean and significance of difference using a Mann–Whitney test (IBM Corporation, 2017) for hydroperiod comparison with the 1974 RL18 survey (Figure 1) and the LiDAR. Interior gages, accessed by boat, typically were missing approximately 21% of the record for many reasons (Table 1). These data were omitted from analyses except at the Arm of Grand Lake (AGL) gage where missing data were estimated for short time periods (days to weeks) using stage trends from the nearby Buffalo Cove (BC) gage. Estimates of missing data using stage trends from nearby gages on the same water body is a common data management practice when distances are short and stage at the nearby gage was likely similar in pattern (Corbett, 1943).
Prior to the gage record, some high-water mark elevations were recorded in the AR Basin RL surveys (USACE, 1979) for discharges comparable to a 1.5-year flood, specifically for 1932 (11,500 m3/s) and 1951 (12,000 m3/s). These high-water elevations were used to calculate the hydraulic gradient through the floodplain before flow regulation.
4.3 Flow distribution
Discharge entering the study area for the 1959–1968 period was the combined measured discharge for Whiskey Bay and old AR (Figure 3A, sites b and c; USACE, 1979). For the 2005–2012 period, entering discharge was measured at the SP gage (Figure 3A, site a; USGS, 2020). Measurements from both sets of discharge data were made on distributary channels and on the main stem of the AR and the distribution was calculated. For 2005–2012 measurements, discharge for two distributaries were measured at downstream locations and have possible discharge gains or losses from storage (sites 4 and 5 approximate site e, site 9 approximates site h; Figure 3A,B). Overbank non-point flow from the river to the floodplain was calculated by subtracting distributary channel flow and the flow in the AR at MP (Figure 3A,B, site o) from the entering discharge (Whiskey Bay + old AR [Figure 3A, sites b and c] 1959–1968, and SP [Figure 3B, site a] 2005–2012). The discharge bypassing the east and west side of the Basin was calculated from the discharge entering the side divided by the discharge in the GIWW and FWDC at downstream locations, that is, west percent bypassing = [(sites j + 4 + 5)/site 13] × 100; east [(sites i + q + 9)/site 14] × 100 (Figure 3B, Table S5). Downstream of sites 13 and 14, most of the channels discharge into the bypassing channels (USGS, 2012, 2020).
4.4 Topography
Channel geomorphology (cross-sectional areas, bankfull elevations, thalweg elevation, mean depth from bankfull, and bankfull width) were analyzed from hydrographic surveys and LiDAR. The 2010 hydrographic survey did not include the SML channel or the old AR channel (Figure 1). These two channels were surveyed at high water in 2015 using a SONAR chart plotter (±0.1 m) with water-surface elevations surveyed at both ends of the reaches using real-time kinematic global positioning system equipment (±0.02 m; Kroes, 2022; Rydlund & Densmore, 2012). The 2015 surveys were matched with the 2012 LiDAR to calculate bankfull cross-sections and compared with earlier surveys (USACE, 1967, 1977, 2012). The significance of geomorphic changes within river reaches between survey dates were determined using t-tests.
Mean elevation from the eastern half of the 1974 RL18 was determined using a best fit line of the survey elevation plot, averaging endpoints of the line segment. Mean elevation of the LiDAR of the same segment of RL18 was determined using ArcPro functions (Esri, 2019). The derived elevations were compared with stage records from MC. Three-dimensional (3D) areas and volumes of inundated floodplain for the entire Basin were calculated for a range of elevations from 0.1 to 2.3 m North American Vertical Datum 1988 (NAVD88) from the LiDAR with permanent open water bodies (National Hydrography Dataset, USGS, 2011) removed from the analysis. Inundation was calculated as a planar water surface over the floodplain using ArcPro 3D Analyst (Esri, 2019). Floodplain inundated surface and volume below the target elevation was reported as total and stepwise changes. The upper limit of 2.3 m was selected because it was the 1.5-year high stage for a large portion of the study area floodplain and exceeds 1.5-year flood levels downstream of MP.
Historic elevation loss rates were analyzed for spoil bank heights along RL18 and compared with the same line in the LiDAR to determine the long-term effect of spoil banks in the Basin. Analysis of LiDAR-derived spoil bank elevations shows a range of values, but field observations indicated that widespread spoil bank inundation for the east side of the Basin occurred when stage at the Keelboat (KB) stage was 2.0 m and was referenced to the stages at the BLR (4.0 m), Little Bayou Sorrel (LBS) (1.6 m), and MC (1.4 m) gages for the same dates and times. Mean spoil bank and floodplain stage exceedances were calculated from the 5-year stage exceedance curves. Stages were examined for exceedance for 0.5, 0.6 and 0.7 m NAVD88 (0.7 m NAVD88 at KB = 0.98, LBS = 0.97, and MC = 1.0 m). Stages of 0.9 and 1.1 m were also examined for the KB gage to compensate for hydraulic gradient (site locations in Figure 1).
5 RESULTS
5.1 Changes in discharge and stage/discharge relationships
During the study period, the AR mean discharge at SP was 6310 m3/s. The duration of discharge between the 1.5-year high and 1.5-year low represented 80% of daily discharges from 1960 to 2014 on the AR. The greatest 5-year discharge period was 1970–1974 (8040 m3/s, 27% above the study period mean, Figure 4), and the lowest discharge for a 5-year period was 1960–1964 (4890 m3/s, 22% below the study period mean, Figure 4). All other periods had less than a 14% deviation from the long-term mean discharge (Table S2, SP discharge).
River stage for the 1.5-year high upstream of BLR dropped rapidly from 1960 until approximately 1988 when stage began to stabilize. Stage at mean discharge began to stabilize around 1978, and stage for the 1.5-year low stabilized in 1968. Stages at all gages during high and mean discharges were increased by the SML weir that was in place from 1988 to 1995 except at BLR and WLO (Figure 5).
Stage on the east side of the Basin decreased sharply from 1960 to 1971 at Bayou Sorrel Lock (BSL) and KB as distributaries were blocked and the river stages dropped (USACE, 1979). By 1965, stages at all gages on the east side of the Basin were similar and began to reflect the water level at MC (Figure 5). During the period of study, stages at MC increased 0.54 m (Figure 5, 1960–1969 to 2005–2014: Z = −52, sig < 0.00) because of the combined effects of sea-level rise (4 to 9.7 mm/yr or 0.22 to 0.53 m, from 1960 to 2014 [NOAA, 2021; NOAA, gage # 8764311, Eugene Island, LA]), channel and tidal dynamics, and likely by the formation of deltas in the northern Gulf of Mexico that lengthened the channel by > 12 km (Couvillion et al., 2011). The reductions in low discharge stage on the upstream end of the east side of the Basin (KB) and the increasing stage at the downstream end (LBS, Figure 5) have changed the hydraulic gradient from 0.12 m/km in 1960 to 0.011 m/km in 1971 and to 0.003 m/km in 2014. At high AR discharge, the hydraulic gradient on the east side floodplain was greater than at low AR discharge, 0.22 m/km in 1961 to 0.019 m/km in 1971 decreasing to 0.018 m/km in 2014, but, still exceptionally low at the end of the study period. The AR in 2015 at high flow had a slope of 0.035 m/km, half of the 1971 river gradient (0.073 m/km). Stages at the 1.5-year low discharge over large portions of the east side of the Basin are now controlled by the level of water in the northern Gulf of Mexico, and thus are becoming increasingly permanently inundated (Figure 5; NOAA, 2018a). In comparison to highwater marks from an approximate 1.5-year high from 1932 (pre-levee) and 1951 (post-levee), stages in the lower portion of the east side (1.98 m) were slightly higher than the 2013 stages of the 1.5-year high at LBS (1.80 m) and MC (1.62 m).
Changes in the hydrology of the west side of the Basin were more difficult to determine because of a lack of gages before 1976. A hydraulic gradient of 0.043 m/km was reconstructed from high water marks measured in 1932 near AGL and BC during which time the area between these locations was dominated by a lake. After 1976, the AGL and BC gages showed similar hydrographs to the MP gage, and hydraulic gradient for the 1.5-year high from AGL to MP was calculated to be 0.053 m/km (1976–2003). The hydraulic gradient then increased to 0.12 m/km from 2004 through 2014 indicating a possible increase in discharge entering the unit. The gages on the west side of the Basin have not yet reflected the increasing MC water stages (Figure 5).
5.2 Channel cross-section
Changes to the channel cross-sectional geomorphology occurred over the study period. Bank elevation at BLR was not exceeded by the AR between 1974 and 2010, thus no fluvial deposition occurred on top of the banks for 60 km downstream of the head of Whiskey Bay (Figure 6, BLR and MP gage data; USGS, 2014). The lack of deposition allowed subsidence and erosional processes to decrease bankfull elevations by 0.79 m (mean, p < 0.01), producing a 22 mm/yr rate of loss. This decreased bank elevation to 0.28 m (mean, p = 0.67) above 1962 elevations. Bankfull elevation approached maximum measured elevation in 1974 for the 68–98 km reach from SML to MC. Mean elevation slightly increased by 2010 (0.05 m) as a result of frequent overbank flow and deposition despite subsidence rates that are likely comparable to the upstream (0–60 km) portion of the study reach (Table S3, Figure 6).
The AR (Whiskey Bay to MC) increased in cross-sectional area by 181% between 1962 and 2010, 65% of which occurred prior to 1974 due to dredging and erosion (Reuss, 1998; USACE, 1979; Table S3, Figure 6E). After 1969, all changes in channel cross-section in the AR study area were from erosion or deposition (Table S1, Figure 7). From 1962 to 2010, the Whiskey Bay channel increased in cross-section while the old AR channel decreased from 1962 to 2015 (Figure 6E,F). The AR channel from MP to SML has maintained a constant cross-section from 1975 to 2010 indicating that it may have been near equilibrium (Figure 6E). The AR channel from SML to MC showed insignificant reductions in cross-section while the SML channel to the WLO channel cross-section increased by 195% from 1962 to 2015 (Table S3, Figure 6E,G) indicating a flow preference to the SML channel because of its greater hydraulic gradient.
5.3 Flow distribution
Analyses of the Basin flow distribution from 1959 to 1961 indicated that discharge leaving the main channel ranged from 2700 m3/s to 12,700 m3/s and totaled 51% of the total discharge (24% to the west, 27% to the east side). The total amount of discharge leaving the main channel above MP by all routes including channel and overbank non-channelized flow and groundwater was on average 73%, leaving 27% of the incoming discharge in the main channel at MP (Table S4).
Channel dredging and most of the distributary blockages were completed by 1969. Analyses of the April 1968 synoptic discharge survey at an AR discharge of 11,800 m3/s indicated a substantial reduction in distributary and overbank flow because of spoil placement and blockage of distributaries. Discharge loss from the river to distributaries was reduced to 25% (15% to the west side, 10% to the east). Of the incoming discharge, 59% remained in the main channel at MP (Table S4).
During the 2005–2012 synoptic measurements, mean total discharge loss to distributaries was 14% (8% to the west side, 6% to the east). The total surface water discharge leaving the main channel was solely due to distributary discharge up to about 8700 m3/s when additional losses of discharge from the main channel other than by distributaries was observed. Of the incoming discharge, 73% remained in the main channel at MP, losses of flow at MP were 9% more than accounted for by distributaries (Table S5).
It is important to note that water leaving the main channel does not always indicate movement of water across the floodplain unless the water stages are over the primary distributary banks. During the USGS 2005–2012 synoptic surveys, 25% of the water leaving to the east side stayed in the GIWW, and 65% of the water leaving to the west side stayed in the FWDC (sites 13 and 14, Table S5, Figures 1 and 4). The volume of water passing over the floodplain has not been calculated, but the volume leaving the AR channel has been substantially reduced. Because of this reduction in flow, the mass of sediment and nutrients interacting with and depositing on the floodplain has been reduced.
5.4 Stage exceedance relative to topographic changes
The highest 5-year mean stages on the main channel of the river at BLR and MP occurred during 1970–1974, coinciding with the period of greatest discharge (Figure 4). In general, mean stage has decreased at BLR since then. Mean stage rose at MP from 1960 to 1965 as spoil was placed and flow was consolidated into the main channel (USACE, 1979). The gages at the MC and WLO outlets have risen in mean stage. During the placement of the SML weir (1988–1995), MP and MC exhibited mean stages 25% and 39% higher than the long-term mean while WLO only showed a 4% increase. During 1995–1999 MP, MC, and WLO (Table S2) continued to have higher stages relative to the long-term mean (12%, 19%, and 13%, Figure 5) despite similar 5-year mean discharge values. Comparison of MC river stages between the periods of 1970–1979 and 2005–2014 showed significant differences for the lower 60% of stages (Z = −29, p-value < 0.00, 1970–1979 mean 0.15 m and 2005–2014 mean 0.32 m).
The gages on the east side of the Basin show mixed patterns in stage. KB had its highest average stages from 1960 to 1964. All other gages (BSL and LBS) had their highest mean stage during the period of 1990–1994 while the SML weir was in place. On the west side, AGL and BC also had their highest average stages during 1990–1994 (Table S2, Figure 5). During channel training (1960–2021), channel blockages were placed that reduced flow to KB and prevented drainage at the AGL gage, resulting in perched water levels (USACE, 2022b). Minimum stages at AGL indicated that the conditions within the west side that prevent drainage have not changed since 1990–1995.
RL surveys indicate large-scale dredging of canals and placement of spoil banks during the late 1960s and early 1970s (USACE, 1979). The average canal spoil bank height from RL18 in 1974 was 3.04 m NAVD88 (USACE, 1979; n = 28). The mean elevation of those spoil banks in 2010 was 1.58 m indicating a net elevation loss of 41 mm/yr, likely because of subsidence and erosion. Analysis of the eastern half of RL18 indicates that the mean elevation of that section of swamp floor was 0.81 m in 1974 and 0.61 m in 2010. This comparison indicates a net elevation loss of 0.2 m over that time with a rate of −4.3 mm/yr.
Two stages were investigated for exceedance to determine the duration of time that spoil banks obstruct flow over portions of the floodplain (0.7 and 1.97 m NAVD88). The analyses indicated that from 1995 to 2014 the spoil banks obstructed flow between 22% and 33% of the year, assuming flow across the floodplain begins to occur above a stage of 0.7 m NAVD88 (Table 2). If only the duration of time where flow over the floodplain could occur (> 0.7 m) is considered, spoil banks obstructed the flow for 54% to 73% of the possible days (1995–2014).
| KB | LBS | MC | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1.97 | 1.1 | 0.9 | 0.7 | 0.6 | 0.5 | 1.97 | 0.7 | 0.6 | 0.5 | 1.97 | 0.7 | 0.6 | 0.5 | |
| 1960–2014 | 23 | 43 | 52 | 56 | 70 | 77 | 9 | 47 | 55 | 64 | 2 | 30 | 37 | 44 |
| 1960–1964 | 39 | 67 | 73 | 79 | 82 | 86 | — | — | — | — | 0 | 4 | 7 | 10 |
| 1965–1969 | — | — | — | — | — | — | — | — | — | — | 0 | 6 | 9 | 15 |
| 1970–1974 | 29 | 50 | 60 | 68 | 74 | 84 | 19 | 45 | 52 | 64 | 15 | 31 | 35 | 42 |
| 1975–1979 | 3 | 17 | 28 | 43 | 49 | 56 | 16 | 38 | 44 | 52 | 11 | 23 | 27 | 32 |
| 1980–1984 | 14 | 36 | 43 | 52 | 59 | 67 | 14 | 43 | 48 | 53 | 11 | 27 | 35 | 42 |
| 1985–1989 | 12 | 41 | 48 | 56 | 65 | 73 | 12 | 45 | 54 | 64 | 10 | 33 | 41 | 48 |
| 1990–1994 | 26 | 55 | 68 | 79 | 83 | 89 | 43 | 70 | 79 | 89 | 37 | 56 | 63 | 71 |
| 1995–1999 | 16 | 38 | 45 | 58 | 64 | 70 | 23 | 52 | 61 | 70 | 23 | 46 | 53 | 61 |
| 2000–2004 | 18 | 39 | 48 | 58 | 66 | 77 | 8 | 37 | 46 | 55 | 12 | 33 | 41 | 53 |
| 2005–2009 | 12 | 36 | 51 | 69 | 81 | 84 | — | — | — | — | 18 | 34 | 41 | 51 |
| 2010–2014 | 20 | 48 | 56 | 67 | 75 | 85 | 6 | 46 | 54 | 65 | 15 | 42 | 51 | 60 |
- Period of record exceeding was calculated from 1960 or the start of the gage's continuous record. Spoil banks are widely submerged at 1.97 m at the KB gage. Example, for spoil bank obstruction Little Bayou Sorrel (LBS) 2010–2014: (exc of 0.7 m [46%] to exc. of 1.97 m [6%] = flow obstruction (40%). Percent of year where flow is possible but is obstructed = 100 × (exc 1.97 − exc 0.7)/exc 1.97. MC = Morgan City. — = no data.
Geographic information system (GIS) analyses of the LiDAR showed slight increases in the inundated surface area from 0.1 to 0.3 m water surface elevation. Above 0.3 m, the amount of inundated surface area change increased rapidly, with the greatest increase in area for the elevation interval of 0.5 to 0.6 m, and large increases up to 0.9 to 1 m (Figures 8 and 9). Above 1 m, the inundated surface area change became more even. Below 1 m where the inundated surface area increased the most, relatively small increases in volume occurred (Figure 8C–F) This pattern and relationship of area and volume indicates that water is primarily in channels below an elevation of 0.3 m. Large areas of floodplain are inundated between 0.3 and 0.6 m, and the majority of swamp floor is inundated below 1 m (Figure 9). Higher swamp floors (Figure 8B,D, 1.6–1.7 m), natural levees, and other high elevation features were inundated with increasing water elevations (Figures 8 and 9). Using the RL18 elevation comparison for reference, the hydroperiod from MC stage exceedance of 0.81 m indicates 21% of 1970–1979 days were inundated, for 2005–2014 using exceedance for 0.61 m indicates 41% of days were inundated. If these elevation changes are extended to the inundation pattern in Figure 8(B), the MC stage where water leaves the channel (0.3 m + 0.2 m net elevation loss since 1974 = 0.5 m) was exceeded 37% of days during 1970–1979 and 74% during 2005–2014 (0.3 m). Examination of the ratio of surface area change to volume change indicates that the stages between 0.3 to 0.7 m provide the highest ratios (Figures 8F and 9) with 395 km2 being inundated to a depth of less than 0.4 m by 82 ha3 of water for 34% of the year.
6 DISCUSSION
6.1 Hydrologic effects
The hydrology of the Basin has changed dramatically because of increasing discharge, dredging, channel erosion, and the blockage of distributaries. The AR channel increased in cross-section by 181% from 1962 to 2010 increasing the discharge capacity of the main channel while reducing discharge across the floodplain. Discharge from the AR by distributary and overbank flow to the floodplain was reduced by almost 54% between 1959 and 1969 when dredging and large distributary blockages were completed and reduced by 63% by 2012. Placement of spoil along the main channel reduced overbank discharge into the forested wetland (USACE, 1977). Water levels on the floodplain saw the biggest reduction in stages relative to distributary blockage and channel dredging before 1970, and to a lesser degree, reflect the channel erosion that has occurred since. For gages on the east side of the Basin, the hydraulic gradient has remained constant since 1970 and mimics the increasing water levels at MC.
It seems likely that the ecosystem functioning would also have changed dramatically with the changes in hydrology. The trend in stage for the east side of the Basin is indicative of increased areas of isolated backwater conditions since the WLO weir placement in 1988, and although briefly alleviated after removal in 1995, have continued. Water levels in the Gulf are rising (NOAA, 2018a, 2018b, 2021), and a sediment deficit continues on the lower east side of the AR Basin, creating a situation where increasing portions of the floodplain are no longer dry during low water. The survey line segment investigated in this study has doubled its hydroperiod since 1974.
When historic spoil bank height is taken in context with the stage exceedance as compared with 2010 spoil bank height, flow blockage caused by spoil banks has decreased from 1974 to that observed in 2010. As the spoil subsided and eroded, the duration of time that these spoil banks are inundated has increased allowing discharge to pass over them in normal years (Table 2) although still obstructing sheet flow across the floodplain. Spoil banks, even when gapped at regular intervals and at the intersections of channels, create pockets of non-moving water typically with dissolved oxygen levels near 0 mg/L because of high water temperatures (> 30°C) and high carbon inputs (> 340 g C/m2/yr) from submerged, floating, and rooted vegetation (Hupp et al., 2019; Kaller et al., 2015; Pasco et al., 2015; Todd et al., 2007; Utley et al., 2008). Increased hypoxia on the floodplain changes the chemistry of nutrient processing, and with high deposition of organic material, could result in the remobilization of previously deposited nutrients (Noe et al., 2013).
The reduction in stage associated with channel enlargement, in conjunction with spoil banks, relict natural levees that formed under previous hydrologic regimes, and dense bank vegetation (Hiatt & Passalacqua, 2015) inhibit flow over the forested wetland. Discharge is focused into channels resulting in high sediment deposition rates within open water and along channel banks further promoting the isolation of the floodplain from flowing water.
6.2 Ecosystem effects
A decline in species richness was observed by Bennett and Kozak (2016) in fish populations including small-bodied fishes (not typically commercially or recreationally harvested) from 1990 to 2010. While many factors likely play into this decline, including possible overfishing and hurricane induced hypoxia/anoxia (Allan et al., 2005; Stevens et al., 2006), sea-level rise, subsidence, and the reduction in dry time on the floodplain may contribute to this species shift. A rise in northern Gulf of Mexico water levels between 0.16 to 0.44 m since 1974 (NOAA, 2021; NOAA, gage # 8764311) is reflected at the east side floodplain gages beginning around 1990 (Figure 5). The rise observed in this study was also observed in a reduction in unflooded days during the growing season by Keim et al. (2006). The processes that spawn from that long-term inundation, that is, high-water temperatures, unburied organic material build up, high biological oxygen demand, high sediment oxygen demand, anoxia, and hypoxia are inhospitable to many fisheries (Justus et al., 2014). These processes are likely enhanced by the obstruction to flow caused by spoil banks.
The floodplain surface has been subsiding at a net rate of 4 to 41 mm/yr (derived from floodplain, bank, and spoil bank elevation loss). Preliminary measurements of deposition within the hydrologically isolated area indicate a deposition rate of 2 to 8 mm/yr of highly organic sediment (Hupp et al., 2019). This scenario could result in a 2 to 39 mm/yr sediment deficit without considering sea level rise. This deficit has over time contributed to permanent inundation in many areas and is consistent with observations of permanent water in areas that were once regenerative for trees. Faulkner et al. (2009) estimated the area permanently inundated and non-regenerative even by planting to be 245 km2, the majority of which is on the east side of the Basin.
6.3 Management considerations
- Reduction of obstructions to flow. Numerous high spoil banks run perpendicular to dominant flow across the floodplain, and parallel spoil banks reduce lateral interactions with adjacent floodplains. Many spoil banks are gapped where channels exist but decreasing the elevation of spoil banks at locations that align with floodplain and not channels could be considered to promote flow over the floodplain to alleviate stagnant pockets of water, although complete removal of spoil banks may be impossible. In some locations, spoil and the subsequent deposition of sediment at gaps prevents water from draining, resulting in extended hydroperiods that may not be represented by gage data alone.
- Introduction of more discharge. Additional water is inseparable from sediment because of the dominance of suspended fine-grained sediment (silts, clays) present in the Basin (SP gage, Field/Laboratory water-quality samples; USGS, 2020). To prevent introduced water from filling open water bodies with sediment, the water could be directed onto an intermediary floodplain and allowed to form deltas of silts while clays continue down valley. This option could be desirable if floodplain resiliency to sea-level rise and subsidence is desired.
- Introduction of discharge into large lakes at the up-river valley end of water management units that could then be used to trap and harvest silt sediment. However, dredging sediment from lakes has not been shown to be economically viable because of the type and quantity of sediment, as well as the cost and distance of removal and shipping (DDG, 2021). This option would likely reduce open water habitat and may also be less desirable for sea-level rise and subsidence resiliency.
7 CONCLUSIONS
The AR Basin has been hydrologically affected by the creation and maintenance of a single channel. The main channel has almost doubled in cross-sectional area since 1962. The increase in cross-section has lowered the stage of a 1.5-year flood at BLR (11,500 m3/s) from 1 m above to 2.4 m below bank height. Bankfull discharge has increased from 6400 m3/s (1960) to more than 17,500 m3/s (2011).
While channel training provided the ability to move the “Project Flood” discharge and extend the life of deep swamp habitats, it has also created issues. The blockage of distributaries, placement of dredge spoil from the preferred channel, and subsequent erosion of the channel have reduced the amount of discharge leaving the river and crossing the floodplain. Stage reduction and distributary blockage reduced the volume of water leaving the main channel by 33% during a 1.5-year flood since 1968. This reduction from the river means that the off main channel discharge was reduced by 55% since 1968, although discharge leaving the river does not necessarily indicate discharge across the forested wetland floor. Widespread creation of access canal spoil banks that exceeded the elevation of the annual hydrograph peak resulted in blockages to flow over much of the floodplain. These blockages led to areas of floodplain that have little or no throughflow depending on stages and a sediment deficit on large portions of the floodplain. Rising sea-level and off-shore delta formation resulted in decreasing hydraulic gradients across much of the floodplain yielding higher water temperatures and lower dissolved oxygen. Large portions of the floodplain have shown an increasing hydroperiod from 37% to 74% of the year when floodplain elevation loss is factored into the calculation. In the absence of the reintroduction of discharge and sediment, it seems likely that as hydroperiod increases, large portions of the Basin will become shallow open water as existing trees die and tree species fail to regenerate. The changes in fisheries species composition to species more tolerant of hypoxia will likely also continue.
Large areas of the AR Basin, like many other Coastal Plain rivers, deltas, and forested wetlands, are succumbing to sea-level rise and subsidence. In the grand scale, managing this system for flood relief has created conditions that may result in the conversion of thousands of square kilometers of the geomorphic AR Basin, inside and outside of the protective levees, to open water. While ample sediment is deposited near or in the channels, the hydrologic and geomorphic mechanisms for the broad, heavy deposition of sediment required for long-term delta ecosystem resiliency against subsidence and sea-level rise were bypassed by the dredging and channel training of the main channel. The depth and distribution of flood waters over the floodplain have been greatly reduced. Now this efficient, trained channel transports 73% to 95% of the discharge and the associated sediment entering the Basin to estuarine deltas without crossing the surface of the floodplain. These modifications occurred before the low-lying portions of floodplain filled to a level that could stay above sea-level for decades or centuries. If channel training efforts were ceased, delta lobe building sedimentation (although lower in maximum potential elevation) could still occur in some areas farther away from the main channel. For the 3000 km2 of the geomorphic Basin outside of the levees, almost none of the AR borne sediment is being deposited while subsidence and sea-level processes continue at likely similar rates.
Long-term stage and discharge data on most of the world's deltas are uncommon, if they exist, and management choices often must be made without information as human populations struggle with increasing flooding and sea levels in this changing climate. This study provides information for modeling and management of the hydrologic effects of channel training on the vast AR floodplain after hydraulic modification projects were completed.
ACKNOWLEDGMENTS
Thanks to David Walther with U.S. Fish and Wildlife Service for bringing the Atchafalaya Basin Floodway Draft Environmental Statement (USACE, 1979) to my attention. Without these historical data sets, the comparisons made in this article would not be possible.
This study was funded in part by cooperative agreements with Audubon Louisiana; Louisiana Department of Natural Resources, Atchafalaya Basin Program; and US Army Corps of Engineers, New Orleans District. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government.
Open Research
DATA AVAILABILITY STATEMENT
Data derived from public domain resources.
Data not wholly publicly available were included in Supporting Information tables per reviewer request.
