Search

Large, low-lying rivers have a variety of societal uses and show a wide range of disturbances. One such example, the Apalachicola River in the Florida Panhandle, experienced anthropogenic activities that altered its hydrologic, geomorphic, and ecologic condition. The river's sand bars enlarged more than threefold in area because of dredge disposal and localized incision from anthropogenic activities during a navigation project. Once dredging ended, the river experienced passive recovery through the slow regrowth of riparian vegetation, with sand bars reducing in area ~18% from 2005 to 2015. The outcomes of tree-line change analyses from 1999 to 2019 indicate a higher recovery rate in the first decade (1999-2010), migrating fastest in locations where vegetation islands stabilize sediments. To help accelerate vegetation regrowth, during the growing seasons of 2018 and 2019, we planted willow stakes at various positions and microhabitats covering ~3.6 ha of three enlarged sand bars near Blountstown, FL. Our mapping shows some synergies between passive recovery and active restoration as tree-line expansion had the fastest rates between 2019 and 2021. Our study gives insights and guidance into passive recovery and restoration projects involving plantings on large, low-lying rivers. [ABSTRACT FROM AUTHOR]

Floodplain characteristics contribute to the natural regeneration of vegetation and help facilitate their ecosystem recovery but vary within and between river systems and more so in a river system recovering from anthropogenic disturbances. Here, we explored the floodplain characteristics that contribute to the woody vegetation regeneration on point bars of the middle reach of the Apalachicola River in Florida, USA, which has been passively recovering for over two decades after dredging and disposal activities ended. Woody vegetation regeneration on point bars was analysed using 1999 and 2019 aerial imagery and divided into zones and sections. Data were extracted from a digitized geodatabase regarding several floodplain characteristics with potential to influence vegetation regenetation. Later, we analysed noncollinear floodplain characteristics using a random forest (RF) regression model. The top characteristics that contributed to the RF model include vegetation density, distance‐to‐seeding source and bend curvature. Findings suggest that high vegetation density contributes substantially to woody vegetation regeneration. The negative relationship between distance‐to‐seeding source and vegetation regeneration shows the importance of seeding availability. The bend curvature corresponds with the highest regeneration in the lowermost section of point bars. Our results signify the importance of maintaining healthy vegetated corridors as a buffer to reinforce the vegetation community and potential seed source for passive recovery. These findings regarding floodplain characteristics can be applied to future active restoration of anthropogenically disturbed floodplains. [ABSTRACT FROM AUTHOR]

Hydrological drought forecasting is essential for effective water resource management planning. Innovations in computer science and artificial intelligence (AI) have been incorporated into Earth science research domains to improve predictive performance for water resource planning and disaster management. Forecasting of future hydrological drought can assist with mitigation strategies for various stakeholders. This study uses the transformer deep learning model to forecast hydrological drought, with a benchmark comparison with the long short-term memory (LSTM) model. These models were applied to the Apalachicola River, Florida, with two gauging stations located at Chattahoochee and Blountstown. Daily stage-height data from the period 1928–2022 were collected from these two stations. The two deep learning models were used to predict stage data for five different time steps: 30, 60, 90, 120, and 180 days. A drought series was created from the forecasted values using a monthly fixed threshold of the 75th percentile (75Q). The transformer model outperformed the LSTM model for all of the timescales at both locations when considering the following averages: M S E = 0.11 , M A E = 0.21 , R S M E = 0.31 , and R 2 = 0.92 for the Chattahoochee station, and M S E = 0.06 , M A E = 0.19 , R S M E = 0.23 , and R 2 = 0.93 for the Blountstown station. The transformer model exhibited greater accuracy in generating the same drought series as the observed data after applying the 75Q threshold, with few exceptions. Considering the evaluation criteria, the transformer deep learning model accurately forecasts hydrological drought in the Apalachicola River, which could be helpful for drought planning and mitigation in this area of contested water resources, and likely has broad applicability elsewhere. [ABSTRACT FROM AUTHOR]

Lowland coastal plain rivers are important for navigation and ecosystems, having unique concerns regarding water and sediment management. The Apalachicola River in Florida, the lower end of the ACF (Apalachicola–Chattahoochee–Flint) system, had historical modifications beginning in the 1800s. Most impacts are attributable to the Apalachicola Navigation Project in the latter half of the 20th century, which included dredging, disposal, rock and snag removal, artificial cutoffs, and wing‐dike building. As one of few studies using paired hydrographic surveys over a long reach and time (170 km, 50 years), we defined four zones of bed elevation change between 1960 and 2010 based on the cumulative thalweg elevation plots. The river has four bed change zones: (1) uppermost, marked degradation from river mile (RM) 106 to 85 because of dam construction; (2) middle, minimal degradation from RM 85 to 34; (3) artificial cutoff zone, highest degradation, from RM 34 to 27; and (4) lowermost zone, aggradation from RM 27 to Apalachicola Bay because of upstream bed and bank sediment disturbances. Artificial cutoffs and lithology influence lateral migration, with rates being highest in cutoff zones and least in areas with some bedrock upstream and finer sediments downstream. Length changes from 1941 to 2019 are stable, except for the middle reach, which lengthened nearly 10%, and the lower non‐tidal reach, where length shortened with cutoffs then regained in subsequent decades. Minimizing disturbance has led to 18% of the sand bar area revegetating since a decade of dredging ended. This framework helps in understanding historic and ongoing human activities, notably upstream water consumption, connectivity changes, sediment inputs, and climate change, that will affect the water and sediment management of the river, its floodplain, and bay, thus varying approaches for restoration. [ABSTRACT FROM AUTHOR]

Anthropogenic impacts in large rivers are widely studied, but studies of recovery once a disturbance has stopped are uncommon. This study examines the biogeomorphic recovery of a 40‐km river corridor on the mid‐Apalachicola River, Florida following the cessation of dredging, disposal, and snag removal in 2002. This failed navigation project resulted in vegetation losses (~166 ha between 1941 and 2004), river widening, and increased point bar areas. We used paired sets of imagery for a 10‐year period during the recovery process at two different flow levels to assess sand bar change, land cover change, and their spatial variations. Most large sand bars decreased significantly in area due to growth of pioneer species, typically from the bankside of the bar. Mean bar area shrank 0.17 and 0.20 ha for the 30th and 1st percentile flows, respectively. For the entire study area, both water‐level comparisons showed gains in vegetation (23.36 and 15.83 ha), compensated by losses in the extent of water (16.83 and 8.55 ha) and sand bar losses (6.53 and 7.28 ha). Overall, these gains during the 10‐year passive recovery period are equivalent to ~15% of the vegetation losses that resulted from the navigational dredging. As found in other studies, most of the pioneer vegetation grew approximately 2 m relative elevation above the low‐water surface. The initial length of the tree line and the area of herbaceous growth both had a significant and positive relationship with the area of new vegetation growth over the study interval. As parts of the river are healing, reduced channel capacity from narrowing and tree growth will benefit the floodplain. As elsewhere, understanding of a river's biogeomorphology, hydrology, and disturbance history can help in selecting appropriate recovery metrics to further advance the understanding and management of disturbed floodplains. © 2020 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Bankfull stage and flow are considered important thresholds in geomorphology. Bankfull indicators were collected at 17 gaged and 28 ungaged relatively undisturbed peninsular Florida streams, ranging in drainage area from 0.52 to 805 km2 (0.2 to 311 square miles) with valley slopes from 0.02 to 2.27 percent. Based on prevalence among sites, comparisons of indicator slopes with water slopes, and correspondence of hydrologic data with morphologic data, the elevation of the flat floodplain was determined to be the most reliable bankfull indicator at sites with a wetland floodplain, while the elevation of the inflection on the bank was the most reliable indicator at sites with an upland floodplain. Small streams and rivers in peninsular Florida achieve bankfull stage relatively frequently compared to the modal values elsewhere, with flow durations of about 25 percent. Results have implications to river restoration projects, and regional differences in relations between hydrology and channel dimensions. [ABSTRACT FROM AUTHOR]

A number of large rivers across the world are dredged for navigation. Dredging removes sediment from the channel, creates spoil mounds in the adjoining floodplain, and disposes sediment along the channel margin on a bar, or in open water. Retrospective studies that examine the placement and fate of this disposed material can provide guidance for other localities where dredging is practiced, and better elucidate the impacts associated with disposal at different types of sites. Many studies exist for impacts and benefits of dredging in the coastal environment, particularly to ecosystems and water quality. Few studies have examined disposal site selection and the fate of dredge spoil placed along a river corridor from a geomorphic and retrospective perspective. The Apalachicola River in Florida was dredged from the 1950s for several decades, with disposal on sand bars, in open water on the channel margins, and on the floodplain. Numerous large and small floodplain dredge-spoil mounds are a legacy from this navigation project. One of the larger mounds, disposal site 40 (DS 40), which locals call "Sand Mountain," stands ~22 m high and extends ~10 ha in area. This disposal site originated in an artificial cutoff created in 1968–69 and grew to hold >625,000 m3 of sediment. Because this section of river was a problem for navigation when dredging was being conducted, detailed historical survey sheets with topographic and bathymetric data exist. The purpose of this paper was to analyze the local floodplain and channel geospatially, coupled with dredging and disposal data, to interpret the geomorphic changes of DS 40 and the adjoining river. Poor placement has resulted in this and other dredge spoil mounds returning sediment to the river through lateral erosion and mass wasting. Intermittent sediment pulses have altered channel morphology adjacent to the spoil mound with the channel becoming wider and shallower than originally, mean bed and thalweg elevations increasing by ~1.5 to 2 m, and width-depth ratio increasing three-fold. Although dredging of this river has not been conducted for nearly two decades, study of historical dredge spoil mounds in this reach suggest poor placement of spoil can continue to stress and alter a river in the decades that follow. Based on our findings, dredging downstream of distributaries will have poor outcomes. Dredge spoil placement in cut-off channels close to the river can be problematic because the river will regain sinuosity. Further, sites on the outer bend of meanders are more subject to erosion, and sites immediately upstream of problem or shallow areas will feed sediment to the problem reach. Findings of this study have implications to other settings, such as intensively mined rivers where unused unconsolidated material is often left on floodplains. • Dredging downstream of distributaries is difficult due to riverbed elevation increases. • Artificial cutoffs regain sinuosity making nearby disposal sites susceptible to erosion. • Outer banks are inferior disposal sites compared to inner bends of meanders. • Sediment from eroding disposal sites affects many channel-form measures for decades. • Placement of the dredge material upstream of shallow areas exacerbates navigation. [ABSTRACT FROM AUTHOR]

The natural flood pulse maintains river-floodplain ecosystems through the exchange of freshwater resources between the main-stem and floodplain habitats. Few prior studies have quantified the relationship between flows and floodplain response including estimating inundation area, floodwater volumes, and slough connectivity. Floodplain modeling typically uses the flow-stage height relationship at river gauge stations. In this study, we compared a relative elevation model (REM) and the Hydrologic Engineering Center River Analysis System (HEC-RAS 1D) model using events from 2015 and 2016 covering a range of flows from the 1st to the 99th percentile for the Apalachicola River, Florida. Because digital elevation models (DEM) from LiDAR (light detection and ranging) data lack details of riverbed topography, we compared a LiDAR-alone and LiDAR-sonar combined DEM to assess their differences. Estimates from the REM and HEC-RAS models were compared to maps based on Landsat imagery-derived water and vegetation indices. In this river, we found a non-linear relationship between the inundated area and flow, increasing markedly through the 90th flow percentile after which increases are minimal. Inundated areas from both REM and HEC-RAS models were similar for all selected flow levels except at the 74th percentile (708 m3/s) flow at which the REM produced 11% higher inundated area than HEC-RAS. Near the median flows, major sloughs were fully connected with backswamps and low-lying patches being inundated. At the higher flows, only a few anthropogenic features were exposed. Floodplain inundation estimates from Landsat performed poorly, detecting 9% with the modified normalized difference water index (mNDWI) and 41% with the open water likelihood index (OWL). These estimates were much lower than the HEC-RAS model (96% flooded), largely because the satellite is unable to penetrate dense forests and examine the floodplain surface, and the Landsat pixel size is twice the width of floodplain sloughs. The LiDAR-sonar combined DEM produced a higher floodwater volume estimate with the HEC-RAS model than using LiDAR-alone. The difference of 1,368,000 m3 at the 1st percentile (142 m3/s) and 2,825,000 m3 at the 89th percentile (1133 m3/s) demonstrate the limitation of using a LiDAR-alone DEM, which cannot penetrate the water surface, and the importance of surveying floodplains using sonar. Modeling results under predict historical wetting of the floodplain because Corps dredging made the main channel approximately 13% wider from its historical width in 1941. Further, in the past few decades, droughts and low flows have become more common because of varied upstream water uses, resulting in less inundation than in the past. Frequent high flows are required to maintain river-floodplain connectivity, floodplain forests, and other hydroecological functions in the Apalachicola River floodplain. Our findings present a basis to assess the legacy of past and ongoing disturbances, inform potential policy decisions for water and floodplain management, and provide a baseline for further research. • Increasing flow levels produced a non-linear relationship with inundation area. • Moderate flows maintained slough connectivity enabling mainstem-floodplain exchanges. • Landsat-derived indices underestimated inundation compared with REM and HEC-RAS. • LiDAR-sonar combined DEMs better quantify floodplain storage than LiDAR alone. • REM models provide sound initial results that help assess HEC-RAS performance. [ABSTRACT FROM AUTHOR]