Wetland Ecosystems

Much of the area east of the Suffolk scarp originally included Lake Drummond, abundant wetland-forest ecosystems, and limited marshes. The original forest ecosystems consisted of bald cypress/tupelo/black gum, Atlantic white cedar, and pocosins for which the swamp was known (Shaler, 1890; Dean, 1969). These communities included a few, scattered, small red maple trees. The original forest communities have been replaced by predominantly red maple/gum communities having limited cypress, cedar, and pocosin communities, which have abundant populations of large red-maple trees (Carter and Gammon, 1976; U.S. Fish and Wildlife Service, 2006).

The distribution of the original communities was controlled partly by subtle differences in topography that affected the wetness of the peat (Shaler, 1890). Cypress/gum communities occupied the slightly lower-altitude, wetter areas, whereas cedar occupied the slightly higher-altitude, less wet areas (Shaler, 1890). Pocosin communities (shrub communities having an open, pine-tree canopy) typically occupy slightly drier conditions.

With the construction of the Dismal Swamp Canal, spoil banks on both sides of the canal inhibited the natural flow of water from the west into the canal and from the canal to parts of the swamp east of the canal. Inhibition of flow made conditions wetter west of the canal and drier east of the canal than before canal construction, altering the ecosystems (Shaler, 1890). Similar hydrologic conditions appear to exist along many ditch/spoil-pile combinations within the swamp. Tree-ring analyses indicate that drier than natural conditions from the drainage have stressed existing communities, likely contributing to this shift in community types (Phipps, and others, 1978).

The swamp originally was targeted for drainage and conversion to agricultural use in the 1760s. Although agriculture succeeded outside the current swamp, attempts within the current swamp failed because the soil was unsuitable for agriculture. Consequently, land use across the swamp focused on timber harvesting. Bald cypress and Atlantic white cedar were the principle species harvested because of their resistance to rot and other valuable characteristics. Most of the swamp reportedly has been harvested at least 2 or 3 times; only a few, scattered, old-growth trees remain. Because selective harvesting left red maple and other species, these harvesting preferences can be partly responsible for the current forest-community distribution. Carter and Gammon (1976) published the first map of the forest and understory distribution of the entire swamp after commercial harvesting ended.

Since then, harvesting has been only for refuge-management purposes, such as the removal of maple and some pine trees from parts of the Blocks to reduce the maple community and re-establish the more typical, open pine-tree-canopy pocosin. Spraying of herbicides augmented the harvesting to limit regeneration of maples from seeds and sprouting from the trunks and roots of harvested trees. Water-control structures are used to manage the hydrology in a manner to support the growth of desired, resilient, forest communities.

Appreciable change likely has occurred in the wetland-forest communities of the swamp from that mapped by Carter and Gammon (1976) to the current communities. Less than 5 years passed from the end of the extensive ditch construction and associated timber harvesting across the swamp (1953–70) to the acquisition of the data interpreted for publication of the Carter and Gammon map, whereas nearly 50 years have passed since collection of data for that map. The map was based on imagery collected between September 1970 and March 1975 (Gammon, 1977). Interpretation of imagery from flights over the swamp in 2015 under a contract between the FWS and a private company provides an updated map of the wetland-forest canopy (fig. 6), but this map does not include understory as does the map of Carter and Gammon (1976).

The one easily identifiable change in the forest community since Carter and Gammon (1976) occurred southwest of Lake Drummond where winds from Hurricane Isabel blew down a large cedar forest in 2003. Subsequent fires in that area in 2008 and 2011 burned into the peat, leaving the area as open water having limited emergent peat and trees. This area is identified as “burn scar” in figure 1 and figure 2 and “Early Successional” in figure 6. Otherwise, a general visual comparison of the recent interpretations with those of Carter and Gammon (1976) indicates similar major patterns but a greater resolution of the forest communities with the recent interpretations. An accurate, detailed comparison of these maps can be difficult because of differences in methods including the type of imagery, classification categories, interpretations, and the apparent resolution. As an example of the classification differences, Carter and Gammon (1976) classify an area along Juniper Ditch (fig. 2, northeastern quadrant) as pine/maple/cedar, whereas the recent map (fig. 6) classifies the area as pond pine pocosin. Anecdotally, although cedar trees remain standing in this area, many appear to have blown down during Hurricane Isabel in 2003 without subsequent regeneration. To better track future changes in the forest community, the FWS has established a network of vegetation-monitoring plots coupled with groundwater wells.

Fire

Fire can also change forest communities. Fire typically is caused by lightning strikes during the dry growing season from late spring into the early fall when evapotranspiration (ET) is high, but sometimes fire is caused by human activity. The available history of fires across the swamp is incomplete and cannot be thoroughly evaluated, although fire occurrence from 1980 through 2012 reflects monthly and interannual distributions (fig. 7). These distributions indicate “fires starting” so that fires that span multiple months or years are not counted more than once.

The number of fires starting each month for all years combined shows a seasonal distribution. Fires typically are less common during the wet season from late fall into spring (fig. 7A) when they typically are small, quickly extinguished, and have minimal effect on the forest and peat. November was the only month having no fires starting; each month from December through March had only one fire starting. During the dry growing season, however, fires are more frequent, more likely to be difficult to extinguish so that they become large and intense, and have a greater effect on the forest and peat. Every month from May through September had more than five fires starting (fig. 7A).

The annual distribution for the 33 years from 1980 through 2012 shows no fires starting in 10 years. Eleven years had only one or two fires starting (fig. 7B). Twelve years had more than two fires starting. The 7-year period from 2006 through 2012 was the worst period; more than two fires started in all years except 2009. The 17 fires in 2007 was the most fires starting in any year. Five and seven fires started the years of the South One (2008) and Lateral West (2011) fires, respectively.

Drainage and fire-suppression techniques likely affect fire frequency, intensity, and spatial distribution across the swamp, but because of the lack of a more complete fire record, these effects cannot be evaluated and can only be assumed. Large, intense fires can destroy mature forest canopy, leading to regeneration of entirely different communities and habitats. In drained peatlands, they can become ground fires that burn through the peat to the water table and create depressions in the peat that change the hydroperiod (depth, duration, and timing of inundation and (or) groundwater levels; Frost, 1989; Turetsky and others, 2011). In contrast, the ability to identify and extinguish fires while they remain small likely has increased because spoil-pile roads adjacent to the ditches improve access.

Although fire can be catastrophic in wetland-forest communities, it also is one of the most important factors contributing to the healthy pocosin and Atlantic white cedar communities (Frost, 1989). The location, size, intensity, timing, and duration of fire, however, is critical. Although pocosins are fire-dependent ecosystems, fire severity must be limited, and return intervals typically must be long enough to allow peat accretion and forest-stand development (Frost, 1989; Richardson, 1991). Atlantic white cedar forests, on the other hand, depend on even less frequent fires but fires of greater intensity than those needed by pocosins (Frost, 1989). Such intense fires destroy competing vegetation and burn debris that normally limit the sunlight availability at ground level needed for the growth of cedar seedlings. Such fires promote establishment of the typical dense, monoculture, cedar forests. Because of the improved ability to identify, access, and extinguish fires, fires required by pocosin and cedar communities likely are less common.

Large, intense fires in the swamp occurred in 1806, 1839, 1923–26, 1930, 1941–42, 1955, 1967, (Simpson, 1990; U.S. Fish and Wildlife Service, 2006) and include the more recent 2008 South One fire and 2011 Lateral West fire south of Lake Drummond (fig. 2; fig. 8). Because of the potential for ground fires, fire-suppression techniques developed for peatlands focus on flooding the burning area to extinguish and reduce the spread of peat fires beneath land surface. Even by using these techniques, however, ground fires can smolder for months. The fires of 2008 and 2011 provide excellent examples of the difficulty in fighting such fires and their effects on the swamp and surrounding communities. The 2008 and 2011 fires burned 4,884 and 6,200 acres, respectively. Both started during dry growing seasons in an area being prepared in 2008 for regeneration of the cedar forest blown down by Hurricane Isabel. The 2011 fire burned throughout and beyond the footprint of the 2008 fire. The 2011 fire likely became intense and difficult to fight because of abundant fuel resulting from the herbaceous growth throughout the 2008 burn scar. Because of the depth and spatial extent of the burn scar that remained after these fires, the hydrology near the area was appreciably altered, and an emergent marsh replaced the forest community previously covering the area.

These two fires required techniques not commonly used in fighting typical forest fires because groundwater and ditch-water levels were low, and the fires burned into the peat. Large pumps lifted water above temporary dams and existing water-control structures on the ditches to push the water upstream through the ditches from low-altitude sources into the higher-altitude burning areas. Because groundwater levels were low and conditions were dry, much of this water infiltrated the peat along the ditch banks reducing the success of this technique and allowing the fires to burn deeply into the peat and continue smoldering for 3–4 months.

Agricultural and Residential Development

Land use surrounding the swamp is important because areas west of the scarp drain to the swamp and affect its water budget and water quality, and the swamp drains to areas north and south of the swamp, affecting water quality and contributing to flooding. Colonists and early residents cleared areas surrounding the swamp for agricultural, residential, and other uses. Uplands on the adjacent Isle of Wight Plain were naturally drier than other areas, making them easily cleared and converted by the early 19th century. Early residents likely converted areas north, south, and east of the swamp during the same period. Spoil material deposited on both sides of the Dismal Swamp Canal when it was constructed inhibited the natural flow of water from the west, dried the land east of the canal and facilitated its conversion to agriculture (Shaler, 1890).

Areas draining the Isle of Wight Plain generally are rural, containing forested, agricultural, and rural-residential areas throughout Virginia and North Carolina; land-use types from the National Land Cover Database (https://coast.noaa.gov/digitalcoast/data/nlcd.html) are listed in table 1, by percent. Forests typically are pine and (or) mixed hardwoods. Flood plains of streams draining surrounding areas commonly contain bottomland hardwood wetlands. Agricultural areas typically contain row crops and pasture. Row crops include corn, soybeans, cotton, peanuts, and sorghum. Wheat is a common winter cover crop that is either plowed into the soil in the spring or allowed to mature, and then harvested.

Residential development is increasing around the swamp and consists of a combination of dense urban and scattered rural residential types. Residential and other urban uses are expanding toward the swamp from Suffolk to the northwest and the City of Norfolk, the City of Virginia Beach, and the City of Chesapeake to the northeast. Agricultural fields east of Portsmouth Ditch and north of Big Entry Ditch (fig. 2, northeastern quadrant) that directly border the northeastern part of the swamp are being converted to residential subdivisions. Rural residential type development is spreading in other areas and consists of homes built on land parcels along roads, at the edge of fields, and in woodlands.

Wells

Water-table wells having screens open to the peat and (or) sand were constructed at 67 sites. Wells consisted of 2-in.-diameter polyvinyl chloride well casings and screens (0.010-in.-wide openings) installed in 4-in.-diameter holes hand bored into the peat and (or) sand. After each hole was bored, water was baled or pumped from the borehole until the water was free of peat particles to minimize plugging of well screens and well-filter sand when the wells were installed. Boreholes were pumped until the water was clear using a centrifugal “trash” pump having a rated pumping capacity of 36 gallons per minute until water was clear. In holes where the water table was in the upper part of the peat, water cleared within 2 minutes when pumped at the maximum rate. During pumping, the amount of drawdown was estimated but was not measured because of the difficulty in establishing a stable measuring point. In these holes, the water table typically declined less than 0.1 ft when the borehole was pumped continuously at the maximum rate. In holes where the water table was in the lower part of the peat or sand, however, the hole was dry within about 10 seconds and briefly spurted water about every 20–120 seconds during pumping.

Screened intervals extended from the bottom of the hole to land surface. The annular space between the well screen and borehole wall was filled with #2 well-filter sand. Wells were bailed or pumped to remove peat or sediment plugging the well screens. Details on construction methods and resolution of problems encountered because of the peat and other natural characteristics of the system are documented in detail in appendix 1 to assist in future efforts in the swamp.

Data on wells constructed at 17 sites in transects along the north-south and east-west firebreaks in Block C1 (fig. 9; table 3) are used in this report. The section for the two-dimensional flow model in Eggleston and others (2018) coincides with the north-south transect and is identified as “X–X′” in fig. 9. Wells are located about 5 ft, 100 ft, 300 ft, 1,300 ft, and 2,600 ft from both ends of each transect. Well C00W at the intersection of the transects is the 2,600-ft well for each transect. Data from 9 (table 4) of the 25 wells at the nine (2 sites are adjoining) Land Carbon study sites (fig. 1) are used in this report. Data from wells constructed at 25 sites (fig. 10; table 5) in the northeastern quadrant of the swamp (fig. 1) also are used in this report. In general, wells in the northeastern quadrant are at the corners of the eastern and the western blocks in the quadrant and along north-south and east-west transects across each block (fig. 10). Transects extend from the midpoint on one side of a block to the midpoint on the opposite side of that block. Corner wells generally are 100 ft from each ditch forming that corner. Transect wells generally are about 100 ft and 1,300 ft from each end of the transect. Details on locations of all wells and well nomenclature are presented in appendix 1.

Groundwater levels were measured periodically in all wells and continuously in selected wells, depending on their location and purpose. Periodic water levels were measured at approximately 2-month intervals to the nearest 0.01 ft using an electrical or steel water-level tape. All continuously measured groundwater levels were measured at 15- or 30-minute intervals to the nearest 0.01 ft. At sites where transmission to satellites was possible, levels in wells were measured using KPSI, 0- to 35-ft, vented pressure transducers and recorded using Sutron SatLink (SAT) or SatLink2 (SAT2) data-collection platforms (DCPs). Data from these sites were transmitted through satellites to the USGS National Water Information System (NWIS) database (U.S. Geological Survey, 2019) as a part of the “real-time” network. For wells where the forest-canopy density prevented transmission to satellites, In Situ Level Troll 700H, 0- to 35-ft, vented pressure transducers having internal logging capability measured and recorded groundwater levels continuously. Data were downloaded from internal logging pressure transducers and loaded into the USGS NWIS database when water levels were measured periodically. Periodic measurements were used to verify and adjust continuous water-level data.

Ditches

In the Block C1 study, ditch-monitoring sites were at both ends of each transect (table 3; fig. 9). In the northeastern quadrant of the swamp, ditch-monitoring sites were (1) at 3 of the 4 ends of the two transects in the western block, (2) upstream and downstream from one water-control structure constructed during the study, and (3) upstream and downstream from one structure repaired during the study (table 5; fig. 10). Sites at the end of transects are at East Ditch at the western end of the east-west transect (station number 364259076262300), Juniper Ditch at the southern end of the north-south transect (station number 364145076245400), and Big Entry Ditch west of Portsmouth Ditch at the north end of the north-south transect (station number 364355076245000). One paired set of pressure transducers is at the water-control structure constructed at Rosemary Ditch above Big Entry Ditch. One transducer (364312076211800) is above the structure, and one (station number 364314076211900) is below the structure. The other paired set of pressure transducers at the control structure repaired during the study is at Portsmouth Ditch above and below the structure at Big Entry Ditch, station numbers 364336076231400 and 364336076231300, respectively. Water levels also are measured continuously with paired pressure transducers operated by the FWS upstream and downstream from structures at the Camp Ditch inflow into East Ditch and Portsmouth Ditch upstream from its confluence with Rosemary Ditch (table 2). The altitude of the measuring point for each ditch site was measured in the same manner as for wells, usually from the same reference point as used for nearby wells.

All continuously measured ditch-water levels were measured at 15- or 30-minute intervals to the nearest 0.01 ft. At sites where transmission to satellites was possible, levels in wells and ditches were measured using KPSI, 0- to 35-ft, vented pressure transducers and recorded and transmitted using SAT or SAT2 DCPs. Ditch levels measured continuously were referenced to staff plates at each site referenced to NAVD 88.

Precipitation

Precipitation was measured to the nearest 0.01 in. at 15- or 30-minute intervals using tipping-bucket rain gages, recorded, and transmitted using the Sat or Sat2 DCPs at 5 sites in Block C1 (table 3) and 2 sites in the northeastern quadrant of the swamp (table 5). Precipitation was not measured at Land Carbon study sites because sites were spread across the swamp, and forest canopy at the sites prevented measurement of reliable precipitation data. In Block C1, precipitation was measured at the center of the Block and near each end of each transect near the ditches (table 3). Precipitation sites in the northeastern quadrant of the swamp were near the intersection of the east-west transect across the western block with East Ditch and at the confluence of Portsmouth and Big Entry Ditches (table 5).

Geology

The geology of the swamp consists primarily of the Dismal Swamp peat at land surface and underlying formations, consisting primarily of mineral sediment. In some areas, the peat is absent, and the mineral sediment is at land surface.

Peat

The Dismal Swamp peat first formed in stream channels (Harrison and others, 1965) approximately 8,000–9,000 years ago (Whitehead, 1965; Stevens and Patterson, 1998). As in other peatlands, prolonged saturation of organic material by a shallow water table and (or) standing water limited the decomposition of the organic material and promoted conversion to peat (Oaks and Coch, 1973; Oaks and Whitehead, 1974). The upper clay and gentle slope of the Yorktown Formation and clayey facies of the overlying Sand Bridge and Londonbridge Formations undoubtedly contributed to these conditions (Oaks and Coch, 1973). As the peat accumulated, it impeded flow in the streams, filled the valleys, and expanded across the uplands to create the flat, present-day surface (Heath, 1975). Thus, the thickest peat overlies ancient valleys, whereas the thinnest peat overlies ancient uplands. Consequently, the topography of the top of the underlying sand, silt, and clay surface reflects the ancient surface-drainage system (Oaks and Coch, 1973). This surface represents early forms of the present-day drainage system where streams flowed east from the Isle of Wight Plain across the present-day swamp to the Northwest River to the east and the Pasquotank River to the south (Oaks and Coch, 1973).

Across the forested wetlands of the Great Dismal Swamp, leaves and other plant material (the duff layer) are underlain by a root mat of varying thickness (Kearney, 1901; Osbon, 1920; Lewis and Cocke, 1929; Henry, 1970; Oaks and Coch, 1973). Peat commonly fills space within the root mat and extends to the top of the sand, silt, and clay (Oaks and Coch, 1973; Oaks and Whitehead, 1974). Peat thickness can exceed 12 ft but commonly is 3 to 5 ft. The contact between the peat and mineral sediment is distinct in some areas, whereas it is a gradational transition in other areas where the sand, silt, and clay content of the peat increases with depth.

Two types of peat that reflected the dominant, overlying forest ecosystem originally were identified across the swamp: (1) black gum peat, a dark brown to black degraded peat and (2) juniper (cedar) peat, a dark to light brown, less degraded, fibrous peat (Kearney, 1901). Except in the peat and mineral-sediment transition, the peat has little mineral content, probably because few streams flowed through the system (Whitehead, 1965; Stevens and Patterson, 1998). Various sizes of partly decomposed tree and shrub parts are present throughout most peat. Large roots, branches, and trunks are so dense that when hand boring holes to depths of 5 ft or more, 3–6 attempts often were required at separate locations to achieve desired depths.

Field observations indicate that the consistency of the peat might have changed from that identified by Kearney (1901), possibly because of subsequent decomposition and other processes caused by fluctuating water levels from the extensive drainage through ditches since 1901. Kearney (1901) describes the highly degraded peat as mucky, which is the common description of the peat across the swamp by the 1960s and 1970s (Henry, 1970; Oaks and Coch, 1973; Oaks and Whitehead, 1974). The descriptions typically do not include characteristics typical of juniper peat.

As part of an analysis of groundwater levels in well C00W in the center of Block C1 (discussed in section “Hydraulic Characteristics”), a different response in groundwater levels to precipitation was noted between the upper and lower parts of the peat with a zone of transition between them. Throughout the remainder of this report, these are referred to as the “upper peat,” the “lower peat,” and the “upper/lower peat transition,” respectively. Close visual inspection of peat removed from subsequent boreholes revealed that the consistency of the mucky peat varies with depth in a manner not described in previous lithologic descriptions. Below the duff layer and root mat, (sometimes within the root mat), the upper degraded peat typically has a granular consistency that is particularly evident when drained and results from larger organic particles within the muck. The lower peat has a more clay-like, mucky consistency as described by previous researchers. A mixed transition zone lies between these two intervals.

Because the peat is similarly degraded throughout its vertical extent, these differences in consistency commonly are not identified, partly because the grains can be destroyed by drilling or boring holes; also, the material is pushed aside when holes are hand bored. Differences in the structure of the upper and lower peat are characterized as “sub-angular blocky” and “massive,” respectively, using the U.S. Department of Agriculture, Natural Resources Conservation Service Field Book (Natural Resources Conservation Service, 2012). Gently rubbing these particles between one’s fingers readily crumbles them into a consistency like that of the muck, demonstrating their fragile structure. Based on discussions with other scientists studying peatlands of eastern Virginia and North Carolina, this consistency is common elsewhere with the granular upper peat commonly described as “coffee-grounds.”

Once this difference in consistency was identified, field crews that were hand-boring holes to construct additional wells would reach into the boreholes to feel the sides of each borehole to identify and measure the depths of contacts between the different layers. This method provided greater accuracy than the retrieval of sediments using the auger because the auger commonly collapsed the duff layer and root mat after cutting through them and contained little peat when retrieved from the borehole. While feeling the sides of boreholes, 1- to 2-in. and larger solid “chunks” of peat not previously identified were retrieved from the sides of many boreholes. When this material was gently squeezed by hand, the material crumbled into the granular consistency typical of the upper peat. The particles became the consistency of the muck when gently rubbed between one’s fingers. This fragile physical structure of the chunks likely is a reason they were not identified previously.

The depth and thickness of the root mat and peat varies in lithologic logs from boreholes hand bored for the construction of project wells in the northeastern quadrant of the swamp, probably because this area is a limit of the extent of the peat. This variability partly results from a fire in 1930 that burned to depths of 6 ft into the peat (Oaks and Coch, 1973). Variability also can result from the decreasing wetness from west to east as the extent of standing water decreases and (or) the water-table depth increases. The depth of the bottom of the combined duff layer and root mat increases to the south and west from 0.0 to 0.2 ft along Big Entry Ditch to more than 0.5 ft along East Ditch, Juniper Ditch, and the southern arm of Rosemary Ditch and to 1.2 ft at well HE01 west of East Ditch (fig. 12).

The depth of the bottom of the peat (excluding the peat and mineral-sediment transition) also increases to the south and west. It ranges from 0 ft east of the confluence of Rosemary and Big Entry Ditches to more than 5 ft near East Ditch in the southwest corner of the study area (fig. 13). The thickness of the peat to mineral-sediment transition (depth of the transition minus the depth of the peat) is as great as 1.5 ft near East Ditch (fig. 13).

Across the northeastern part of the swamp, the depth to the bottom of the upper peat increases toward the south and west (fig. 14). Where the root mat lies on the mineral sediment, space around the roots typically contains fibric or granular upper peat so that the bottom of the root mat is indicated as the bottom of the upper peat and is common along Big Entry Ditch. The depth to the bottom of the upper peat generally increases to 1.5–2.5 ft but can extend to 3.4 ft. Across the northeastern quadrant of the swamp, the upper peat is directly in contact with the mineral sediment or the transition from the peat to the mineral sediment at 19 of 25 sites (fig. 14, open circles); no upper/lower peat transition or lower peat was identified at these sites. The upper peat overlies the upper/lower peat transition or the lower peat at only six sites (fig. 14, solid circles).

Hydraulic Characteristics

Lateral and vertical heterogeneities in the specific yield, hydraulic conductivity, and thickness of the peat and mineral sediment affect where groundwater is recharged, is stored, and flows throughout the swamp. These characteristics are critical to delineating the aquifers and confining units of the shallow aquifer system and understanding their hydrologic response.

A difference between the specific yield of upper peat and lower peat in the swamp was first identified in the analysis of the response of groundwater levels to precipitation during seven selected events from November 2009 through September 2011 in well C00W at the center of Block C1. In this analysis, about 0.5–1.0 in. of precipitation was incorporated in the interception and rewetting of unsaturated peat before recharge began and groundwater levels rose during each event as highlighted by the logarithmic scale of the X axis in fig. 15A. Groundwater response to precipitation after interception and rewetting was rapid because of the limited open pine-forest canopy. Once recharge began, three zones characterized the rate of rise in groundwater levels: levels rose slowly within the upper peat (above a depth of 1.5 ft), rapidly within the lower peat (below a depth of about 2.0 ft), and at a changing rate in the upper/lower peat transition between depths of 1.5 and 2.0 ft, as highlighted by the arithmetic scale of the X axis in fig. 15B. Based on the rates of rise, specific yield of the upper peat averaged about 59 percent, whereas that of the lower peat averaged about 7.5 percent (table 7). Because the rate of rise was similar from land surface to the depth of 1.5 ft and the well was in a hollow, specific yield estimates appear to be minimally affected by nearby standing water. Typically, little standing water was observed when the water table was below land surface at the well. The specific yield of 59 percent for the upper peat is within the 56–63 percent identified for fibric peat in a Picea mariana (Black Spruce) bog in Canada (Petrone, and others, 2008). The specific yield of 7.5 percent for the lower peat is similar to the 8 percent of well-decomposed peat (Boelter, 1968).

Results of a similar analysis of nine wells across the northeastern quadrant of the swamp (HE01, WW13, WS01, WN01, WE13, WNE01, EW13, ENW01, and EE13) having adequate continuously measured groundwater levels shows possible vertical and spatial variations in the specific yield of the peat and sand (tables 7, 8). Precipitation events used for this analysis were limited to those having (1) groundwater levels below land surface, (2) appreciable precipitation after interception and rewetting unsaturated peat to create adequate groundwater recharge and a rise in groundwater levels, and (3) enough precipitation intensity. Because the canopy was denser in the northeastern part of the swamp than in Block C1, recharge appeared to be delayed as precipitation was intercepted by the canopy and then fell through the canopy or drained down the trunks of the trees. This delay caused periods of appreciable rise in groundwater levels during low-intensity precipitation periods following periods of less than about 1 hour of high-intensity precipitation so that the method was modified as discussed in the methods. These response characteristics also made numerous precipitation events unusable. Additionally, evaporation of the fallen precipitation, probably resulting from the tall dense canopy, excluded low-intensity precipitation periods.

Differences in specific yield of the peat likely reflect effects of differences in the formation and subsequent decomposition of the peat. Differences in specific yield of the sand likely reflect differences in its sand-grain size and silt and clay content. Although spatial variations might reflect true lateral variations within the peat, they also might reflect differences with depth. The number of values of specific yield varied among wells because specific yield was calculated only when groundwater levels changed through an appreciable interval of a material type during an event.

Standing water typically reflects an emergent water table where the root mat and peat are present at land surface because of their high permeability. Because this method averages effects of water levels over an unknown area, and the specific yield of standing water is 100 percent (has no media), standing water near wells limits the events useable for specific-yield calculations, even if no standing water is at the well because the presence of standing water near a well is caused by microtopography. Water levels typically were near or above land surface in the western part of the northeastern quadrant of the swamp because it is a wet part of the swamp. Most wells were in hollows so that standing water near the well was limited or absent when the water level in the well was near land surface. The one exception is well WW13; this well was on a hummock because standing water covered much of the surrounding land surface during well construction.

Water-level responses are depicted in four wells in fig. 16A–D as examples of responses observed. Specific yield calculated at well WW13 decreased from 100 percent (indicating standing water) at groundwater depths of 0.3–0.5 ft to 70 percent at depths of 0.75–0.95 ft (fig. 16A). This high, decreasing-with-depth specific yield is consistent with a decrease in the amount of standing water and decreasing groundwater levels from a qualitative assessment of the site. Between depths of about 1.0 and 1.5 ft, specific yield was 39 and 42 percent, consistent with that of other wells across the study area.

At well HE01 to the west of East Ditch, useable precipitation events were limited to those during dry, summer periods because standing water was present at the well at other times. When the water table was below land surface, little standing water was near the well because the well was in a hollow. Specific yield at this well ranged from 57 (shallowest interval) to 67 percent (deepest interval) and averaged 63 percent at depths to 0.8 ft (table 7; fig. 16B). Because specific yield increased with depth, the change represents peat characteristics, not standing water. The average is similar to the 59 percent observed in well C00W in Block C1 (fig. 15). At other wells throughout the northeastern quadrant of the swamp, specific yield reflects a general, although not fully consistent, downward trend with depth as observed in well EW13 (tables 7, 8; fig. 16C). The lithologic log of well EW13 was the only log having the upper peat, lower peat, and sand.

For all wells having groundwater levels that fluctuated 1 ft or more through the sand during precipitation (WS01, WN01, WNE01, EW13, ENW01, and EE13), specific yield of the sand ranged from 1 to 33 percent and averaged 17 percent (table 8). Well EE13 reflects differences in the response between the peat and sand and over a 2.5-ft interval of sand (fig. 16D). The cause of the range in the sand is difficult to determine because the method integrates the effects of an unknown area where sediment characteristics likely differ from those observed in the well borehole. In comparison, Johnson (1967) reported specific yield averages of 8 percent for silt, 7 percent for sandy clay, 21 percent for fine-grained sand, and 26 percent for medium-grained sand.

Porosity of the upper peat averaged 87–91 percent at the nine Land Carbon sites (Drexler and others, 2017), which is consistent with the 92 percent of Morris and Johnson (1967). Although porosity of sand beneath the swamp was not measured, Morris and Johnson (1967) identified 42, 46, 43, and 39 percent as the porosity of clay, silt, fine-grained sand, and medium-grained sand, respectively. Thus, porosity of the sand, like its specific yield, likely is substantially less than that of the upper peat, meaning that the sand stores, transmits, and yields less water than the upper peat. Part of the water in saturated peat is in primary pore space (space between peat particles), and part is in secondary pore space (space within peat particles). Because sand has little secondary porosity, most water saturating sand is between sand grains.

Although specific retention was not measured, using the difference between a calculated range of 40–60 percent for the specific yield of the upper peat and the porosity identified by Drexler and others (2017), the water remaining in the peat after draining is between 27 and 51 percent of the total peat volume. This compares with a specific retention of 49 percent for peat reported by Morris and Johnson (1967) and equals between 30 and 59 percent of the water originally in the saturated peat. Thus, substantial water remains in unsaturated peat immediately after draining.

Because high specific yield typically accompanies high hydraulic conductivity; low specific yield typically accompanies low hydraulic conductivity, the upper peat likely has a high hydraulic conductivity and the lower peat and fine-grained sand likely have low hydraulic conductivities. Flow modeling calculated a single calibrated hydraulic conductivity for each modeled layer to include 13,220 feet per day (ft/d) for the upper peat, 24 ft/d for the lower peat, and 100 ft/d for the sand (Eggleston and others, 2018). Modeled hydraulic conductivity of the spoil-pile roads was 23 ft/d. These values compare to representative hydraulic conductivities of 8 ft/d for fine-grained sand, 40 ft/d for medium-grained sand, and 1,500 ft/d for gravel (Todd, 1980). Consequently, the hydraulic conductivity of the lower peat is between those of fine- and medium-grained sand; that of the upper peat is nearly an order of magnitude greater than that of gravel and more than three orders of magnitude greater than that of a fine-grained sand.

Differences among drawdowns when pumping water from open boreholes having the water table in the upper peat, lower peat, and sand reflect these differences in the hydraulic conductivity. The limited drawdown when the water table was in the upper peat reflects its high hydraulic conductivity. The draining of the borehole when the water table was in the lower peat or sand reflects their substantially lower hydraulic conductivity. Although use of pumping or slug tests of finished wells was considered for obtaining hydraulic properties of the peat and sand, finished wells having the water table within the upper peat that had been developed as completely as possible went dry within 10–15 seconds when similarly pumped. This indicates that hydraulic characteristics of well material, not aquifer material, controlled hydraulic responses of wells. These relative hydraulic characteristics of the peat layers are consistent with those of the acrotelm and catotelm, the distinct upper and lower peat layers identified by Ingram (1978) and differ greatly from the characteristics in the conceptual model of Lichtler and Walker (1974). These relative hydraulic characteristics were applied to the flow model of Eggleston and others (2018) before calibration and are substantiated by the calibration.

Aquifers and Confining Units/Zones

The Coastal Plain aquifer system beneath the swamp extends laterally into adjacent States and vertically from land surface to depths of about 1,500–2,300 ft beneath the swamp. The aquifer system consists of sediment of Cretaceous age and younger (McFarland and Bruce, 2006). Concern exists that large groundwater withdrawals from the Potomac aquifer (the deepest aquifer) and other deep aquifers that draw down groundwater levels within the aquifers by as much as 100 ft or more also drain appreciable water from the shallowest overlying aquifers and the swamp. Numerical model simulations were used to compare groundwater levels and flow in these aquifers prior to appreciable groundwater withdrawals, with levels and flow in 2003 (a year of large withdrawals and normal precipitation and groundwater recharge) and 2001 (a year of large withdrawals and reduced precipitation and groundwater recharge) (Heywood and Pope, 2009). Simulated levels in the surficial aquifer (the shallowest aquifer) for 2003 showed negligible change from levels prior to large withdrawals. Simulated declines of 0.5–1.0 ft in the surficial aquifer for conditions in 2001, however, resulted from the decreased precipitation that reduced groundwater recharge, not from the large withdrawals (Heywood and Pope, 2009). This implies that annual amounts of precipitation and resulting groundwater recharge have substantially greater effects on groundwater levels in the swamp than groundwater withdrawals from deep aquifers. Only aquifers shallower than depths of about 110–175 ft beneath the swamp, above the St. Marys confining unit, likely affect the swamp appreciably.

This shallow part of the aquifer system consists of the surficial aquifer, the underlying Yorktown-Eastover aquifer, and the intervening Yorktown confining zone (McFarland and Bruce, 2006). Local heterogeneities in the aquifer system that might locally control hydrologic processes and responses across the swamp, however, are not identified in the aquifer-system framework of McFarland and Bruce (2006) because of the scale of this framework that covers the entire Coastal Plain of Virginia and parts of adjacent States. For example, this framework does not identify the formations that make up the surficial aquifer and Yorktown confining zone beneath the swamp. The framework, however, identifies a general area where the Yorktown confining zone is absent beneath the swamp. By applying the geology of Oaks and Coch (1973) and hydraulic and other information described in this report to the framework of McFarland and Bruce (2006), the spatial distribution and other characteristics of the surficial aquifer and Yorktown confining zone can be described in greater detail to better understand their role in hydrologic controls and responses across the swamp.

The surficial aquifer covers the entire swamp and consists of the upper peat, lower peat, and underlying sandy sediment. Where the peat is absent, and the sand is at land surface, only sand forms the surficial aquifer. The shallow sand identified in boreholes likely is the medium-grained sand of the Tabb Formation that lies directly beneath the peat, as identified by Oaks and Coch (1973). The upper peat, however, forms the primary part of the aquifer because of its higher specific yield (fig. 15, fig. 16; tables 7, 8) and hydraulic conductivity (Eggleston and others, 2018) compared to the lower peat and sand. Beneath the Isle of Wight Plain, the surficial aquifer consists of the Windsor and Tabb Formations near the scarp (Oaks and Coch, 1973). The depth to the bottom of the surficial aquifer is about 30–40 ft beneath the Isle of Wright Plain near the scarp.

McFarland and Bruce (2006) describe the Yorktown confining zone as consisting of clay interbedded with sand and silt that appear to be absent beneath parts of the swamp. This unit is termed a “confining zone” rather than the more commonly used “confining unit” because it has substantial vertical hydraulic conductivity that allows more groundwater flow between the surficial and Yorktown-Eastover aquifers than a typical confining unit. McFarland and Bruce (2006) identify the extent of the confining zone as beneath the Isle of Wight Plain and swamp, except in an area that extends from about 1 to 10 mi north of North Carolina along the toe of the scarp to about 2 to 8 mi east of the toe. The thickness of the confining zone ranges from 0 to about 15 ft. Where present, the depth of the top of the confining zone ranges from about 10 to 15 ft (McFarland and Bruce, 2006).

Based on Oaks and Coch (1973), the confining zone appears to consist of the silty and clayey sediments of the Sedley Formation along the scarp and the Londonbridge and Sand Bridge Formations also along the scarp but extending across the northern part of the swamp. These sediments are absent across the southern part of the swamp east of the toe of the scarp in a pattern consistent with the confining zone depicted in McFarland and Bruce (2006). Upper clayey parts of the Yorktown Formation that were partly responsible for formation of the peat (Oaks and Coch, 1973) likely are part of the Yorktown confining zone.

The Yorktown-Eastover aquifer consists of sandy, permeable parts of the Yorktown and Eastover Formations (McFarland and Bruce, 2006). The upper part of the Yorktown-Eastover aquifer likely affects the swamp more than the deeper part (Eggleston and others, 2018). Because of the somewhat permeable nature of the Yorktown confining zone and its absence beneath the southern part of the swamp, the surficial and Yorktown-Eastover aquifers locally form an interconnected system that can be classified as either separate aquifers or a combined single aquifer beneath much of the swamp. The depth of the top of the Yorktown-Eastover aquifer ranges from about 10 to 35 ft beneath the swamp. The approximate thickness of this aquifer ranges from 60 to 140 ft.

The surficial aquifer is hydrologically the most important part of the shallow aquifer system to the swamp because of its shallow depth, connection with the ditch network, and the hydraulic characteristics of the upper peat. The upper peat transmits substantially more groundwater laterally beneath the swamp because of its much higher hydraulic conductivity than the lower peat and sand (Eggleston and others, 2018). Because of the low hydraulic conductivity and shallow depth of the lower peat, and the typical range in levels of the water table, the lower peat likely has a dual function: (1) it inhibits flow between the peat and sand aquifers when the water table is within the upper peat, exhibiting semi-confining unit or confining-zone characteristics, and (2) it stores, transmits, and discharges water when the water table declines into the lower peat. The exact role depends on location and hydrologic conditions. When the water table is within the lower peat or sand, groundwater primarily flows laterally through the sand because hydraulic conductivity of the sand is greater than that of the lower peat (Eggleston and others, 2018). Consequently, in this report the upper peat is referred to as the “peat aquifer” and the sand underlying the lower peat as the “sand aquifer.” The peat/mineral-sediment transition is part of the sand aquifer.

The northeastern quadrant of the swamp demonstrates the effects of thinning peat on the relative hydrologic importance of the peat and sand aquifers because they exist in distinct spatial patterns in this area (fig. 17). From west to east, land-surface altitude declines about 8 ft, from about 20 ft to 12 ft above NAVD 88 (fig. 17A), and accompanies a decrease in the thickness of the peat aquifer from about 3 ft to being present only within the root mat. Land-surface altitude in the swamp and peat thickness also decrease near Portsmouth Ditch. From south to north, the land-surface altitude declines about 2 ft across the western block (fig. 17B) and varies about 2 ft in no distinct pattern across the eastern block (fig. 17C). The thickness of the peat aquifer also decreases from south to north with a greater decrease across the eastern block (fig. 17B and C).

The decrease in thickness of the peat aquifer indicates the increased hydrologic importance of the sand aquifer to the east and north. Toward the south and west where the upper peat is sufficiently thick, the water table commonly remains within the upper peat, and most groundwater likely flows through the peat aquifer. Where groundwater flows through the peat aquifer, it can flow more easily between the peat and sand aquifers than in most other parts of the swamp because the lower peat is thin or absent in much of the area, and the upper peat lies directly upon the underlying sand or the peat/mineral-sediment transition throughout much of the area. Where the peat is not sufficiently thick, the water table declines to a great enough depth that most groundwater flows through the sand aquifer.

Ditches, Spoil-Pile Roads, and Railroad Grades

Ditches, spoil-pile roads, and railroad grades are the legacy of timber harvesting across the swamp prior to its transfer to conservation status. The ditch network can be considered the primary component of the surface-water system because of its extent and interconnection with the groundwater system and other parts of the surface-water system.

The timing of ditch construction affects the duration and extent that an area has been drained. The duration of this drainage can affect the extent of the subsequent subsidence and changes in the forest communities. Although George Washington and co-investors constructed the first ditch in the 1760s, only 55 mi of the 200-mi-long ditch network was completed by 1953 (Shaler, 1890, plate VI; U.S. Geological Survey, 1919; 1940; 1954 a–g) (fig. 2, table 9). Timber companies, however, constructed approximately 145 miles of ditches and adjacent spoil-pile roads between 1953 and 1970 (U.S. Geological Survey, 1954a–g; 1977a–e) (fig. 2; table 9). Ditch and road construction peaked during the 1953–70 era because of improved construction techniques and the post-World War II increased demand for forest products. Previously existing ditches such as Washington Ditch (northwestern quadrant), Jericho Ditch (northwestern quadrant), and Cross Canal Ditch (South of Lake Drummond) were deepened and widened during this time to further improve drainage (Brown, 1970). The 1953–70 era was designated as such because 1953 imagery having limited existing ditches, was used for the maps published in 1954, and because timber companies discontinued ditch construction in about 1970 as they planned to transfer the land to conservation status.

Ditches constructed in the late 1700s and early 1800s are limited in number, width, and depth because they were dug by hand (Brown, 1970). Consequently, these ditches likely had only modest effects on the hydrology of the swamp. Techniques evolved through the years until the 1950s and 1960s when the ditches were excavated through the surficial peat and into underlying sand, silt, and clay using modern draglines and trucks (Brown, 1970). Removed material was deposited as spoil banks, typically on one side of each ditch (fig. 5). Roads were constructed on the spoil banks to provide access to the swamp and transport timber from the swamp. Some roads, termed “corduroy roads,” have spoil material deposited upon downed timber laid upon the peat. Corduroy roads typically allow flow through voids in the permeable timber base, especially as the logs decomposed and created additional voids. Spoil material was deposited directly upon the peat for other roads, compressing the peat and reducing its permeability. In some cases, fill material, such as sand and bricks, was hauled from offsite, spread, and compacted onto the spoil material to improve road stability and drivability. Ditches typically have about a 30-ft width and 6-ft depth; spoil-pile roads normally have a 30- to 40-ft width and are elevated 1 to 3 ft above the adjacent peat surface. Roads constructed using the two techniques cannot be readily differentiated.

Ditches and spoil-pile roads having spoil material deposited directly upon the peat more profoundly alter the hydrology of the swamp than ditches and corduroy roads because of the low permeability of the compressed peat, spoil material, and fill. Where ditches directly adjoin the swamp, surface water and groundwater readily drain into the ditches, making the swamp drier than it naturally was (fig. 5). Where spoil-pile roads intervene between ditches and the swamp, spoil piles inhibit flow, raising water levels and potentially creating standing water that makes the swamp wetter than it was naturally.

Railroads commonly were used to remove timber from the swamp between the mid-1800s and the 1960s (Trout, 1998; Simpson, 1990). Prior to 1940, short railroad lines constructed on downed timber transported harvested timber short distances to the limited ditch network through which logs were floated from the swamp (Shaler, 1890; U.S. Geological Survey, 1940). More recently, railroad grades were constructed either of downed timber or fill material hauled from off site. Grades constructed of downed timber have not persisted on the landscape. Grades constructed of fill material were elevated 1 to 4 ft above the adjacent peat, and many were integrated into the road network constructed in the 1950s and 1960s (Trout, 1998). Grades constructed of fill material and not integrated into the road network remain identifiable in the field. One of these grades, the abandoned Richmond Cedar Works Railroad, is evident in the swamp across the Blocks (fig. 1) and to the southwest. The only currently (2019) active railroad is the CSX rail line (fig. 1), crossing the swamp about 0.5 mi north of Williamson Ditch (fig. 2).

Lake Drummond, the Feeder Ditch, and the Dismal Swamp Canal

Lake Drummond, the Feeder Ditch, and the Dismal Swamp Canal (fig. 2) function as a single system maintained and managed for navigation by the USACE. This system has a long history of change beginning in 1787, when legislation authorized construction of the canal and prioritized use of all water in Lake Drummond for supporting navigation in the canal (Brown, 1970). Lake Drummond, a 3,108-acre natural lake at the center of the swamp, has a maximum depth of about 6 ft and stores a maximum of about 15,000 acre-feet of water (U.S. Army Corps of Engineers, 1996).

Between 1793 and 1805, a private company constructed the original 22-mi-long Dismal Swamp Canal to commercially transport goods between Chesapeake Bay to the north and Albemarle Sound to the south (Brown, 1970). Initially, the canal allowed only shallow-draft boats to pass because of its shallow depth. Shortly after completion of the canal, owners and operators realized the volume of water naturally entering the canal was insufficient to maintain navigation for even these boats during dry seasons. Consequently, the Feeder Ditch was constructed in 1812 to augment the supply of water to the canal with water from Lake Drummond (Brown, 1970; Trout, 1998). The Feeder Ditch is the primary surface-water outlet from the lake and is the largest surface-water input to the canal (Eggleston and others, 2018). Multiple navigation locks were added, and the canal was deepened between 1805 and 1826 to better control water and allow the canal to pass deeper-draft vessels. A stone lock completed in 1830, replaced a timber lock and provided greater control over the release of water from Lake Drummond (Trout, 1998). Additional modifications further deepened and widened the canal in the 1890s creating its modern footprint (Brown, 1970). The width of the current canal is approximately 100 ft, and water depth is maintained at about 6 ft (U.S. Army Corps of Engineers, 1996).

In 1925, the Federal Government purchased the Dismal Swamp Canal to incorporate it into the Atlantic Intracoastal Waterway (AICW). The AICW is managed by the USACE to facilitate the commercial transportation of goods along the Atlantic Coast in inland waters protected from coastal storms. The 1974 legislation establishing the refuge re-prioritized the use of lake water to refuge-habitat management; navigation became a secondary use (U.S. Army Corps of Engineers, 1996). Lake operations currently follow the 1996 Dismal Swamp Canal Operations Manual using standards for lake operation based on a 1976 agreement between the FWS and the USACE (U.S. Army Corps of Engineers, 1996). This agreement establishes a maximum operating lake level of 16.08 ft above NAVD 88 (5.3 ft above the Dismal Swamp Canal Datum [DSCD]) and a minimum operating lake level of 14.28 ft above NAVD 88 (3.5-ft DSCD; U.S. Army Corps of Engineers, 1996). Consequently, lake and canal operations strive to maintain lake levels within these targets by controlling releases from Lake Drummond. As long as lake levels remain above the minimum operating level, water is released from the lake to maintain navigation. Once lake levels reach the minimum target, the USACE closes the canal to navigation until lake levels rise again. Because the draft of vessels currently (2019) used for transporting goods is deeper than can pass through the canal, the canal now is used primarily for recreational boat traffic; commercial transportation of goods is through a part of the AICW farther east.

Streams

No natural streams flow across the swamp. Two types of streams, however, surround the swamp: (1) those that flow from the Isle of Wight Plain across the Suffolk scarp and drain into the western part of the swamp and (2) headwater streams into which the swamp drains that flow north into Chesapeake Bay or south into Albemarle Sound (fig. 1).

Most streams flowing across the Suffolk scarp are termed swamps because they form black-water, bottomland, hardwood swamps typical of streams in lower parts of the mid-Atlantic and Southeastern Coastal Plains; peat extends part of the way up these stream valleys from the Great Dismal Swamp. These streams drain about 50,000 acres of gently rolling terrain. The average slope of their watersheds ranges from 1.0 to 1.5 percent; the maximum range in watershed altitudes is about 55 ft. The largest streams are Pocosin Swamp and Cypress Swamp (fig. 1), which flow directly into Washington Ditch (West of Lake Drummond), and Corapeake Swamp, known locally as Daniels Road Swamp, which flows directly into Cross Canal Ditch (formerly Hamburg Ditch; fig. 1; fig. 2). Smaller streams include Moss Swamp, Folly Swamp, and Goose Creek, which cross the Suffolk scarp and discharge diffusely across the surface of the swamp (fig. 1).

The Great Dismal Swamp forms the watershed divide between Chesapeake Bay and Albemarle Sound, serving as the headwaters of several freshwater tidal tributaries to these estuaries. In the Chesapeake Bay watershed, the swamp drains primarily into Shingle Creek, a tributary of the Nansemond River, and Deep Creek, a tributary of the Southern Branch of the Elizabeth River. In the Albemarle Sound watershed, the swamp drains primarily into the Pasquotank River. Historically, the swamp also drained into the Northwest River and the Perquimans River, flowing into Currituck Sound and Albemarle Sound, respectively. Construction of the Dismal Swamp Canal hydrologically isolated the swamp from the Northwest River; construction of U.S. Highway 158 in the 1930s hydrologically isolated the Perquimans River.

Water-Control Structures

Managing water levels and flow through the swamp has been a priority of resource managers for generations (U.S. Fish and Wildlife Service, 2006). Structures on the canal system and the ditches have been constructed to achieve these goals. Current structures designed to manage flow through the canal system consist of a dam having an adjustable spillway at Lake Drummond and a navigation lock, dam, and spillway at the north and south ends of the canal in Deep Creek, Va., and South Mills, N.C., respectively (fig. 1). Manually adjustable gates, or wickets, on the spillways control the discharge of water from the system (U.S. Army Corps of Engineers, 1996). Widths of 40–60 ft allow spillways to discharge more than 1,000 cubic feet per second (ft³/s).

Smaller water-control structures have been used historically to manage levels and flow through ditches throughout the swamp (fig. 2). Crude timber structures were erected on Jericho Ditch (fig. 2, northwestern quadrant) in the 1800s to help maintain water levels for transporting shingles from the swamp through Shingle Creek to the harbor on the Nansemond River in Suffolk, Va. (Brown, 1970; Trout, 1998). Although evidence of such structures no longer exists, corrugated metal culverts and concrete structures fitted with hand-operated slide gates dating from the 1950s and 1960s remain (fig. 18A and B). Since 1974, the FWS and North Carolina State Parks have installed 59 of the 64 structures currently present across the swamp (fig. 2).

Although some structures have fixed-height weirs, most have adjustable-height weirs that provide a flexibility to retain or release water in a manner not afforded by fixed-height weirs. The most common structure design is like that used on ditched agricultural land and managed forests across the Virginia and North Carolina Coastal Plain (Dukes and others, 2003: Evans and others, 2007). This design consists of a 4- to 6-ft-diameter, horizontal, corrugated-aluminum culvert having a riser (fig. 18C). Each riser consists of a 6- to 9-ft-diameter, vertically oriented, half culvert having two vertical channels designed for addition or removal of 6-in.-tall horizontally oriented boards, also called stop logs, to adjust weir height. To accommodate the maximum-anticipated flow, 1 to 3 culverts typically are installed beneath one of the spoil-pile roads at two intersecting ditches; risers are on the upstream ends of the culverts (fig. 18D).

Fixed- and adjustable-height weirs of vinyl or steel sheet piles are another common structure design (fig. 18E). Weirs typically have widths of 15–20 ft to accommodate the maximum anticipated flows. Staff manually adjusts weir altitudes using 6- to 12-in.-tall stop logs like those used with corrugated aluminum structures. The largest adjustable-height weirs were constructed on South Martha Washington Ditch (southeastern quadrant, fig. 2) to control larger flows (fig. 18F).

The Burn Scar

The area of the 6,200-acre burn scar (southwest of Lake Drummond, fig. 1) created by the 2008 South One and 2011 Lateral West fires (fig. 8) was primarily part of the groundwater system controlled by the hydraulic characteristics of the upper peat and its connection with the ditches before the fires. The loss of an unknown depth of peat from the 2008 fire, and an average of 1.5 ft (Reddy and others, 2015) and a maximum of 4.5 ft (Hawbaker and others, 2016) of peat from the 2011 fire, created a shallow, open-water reservoir having an extended hydroperiod that supports an emergent freshwater-marsh habitat. The loss of land-surface altitude and upper peat through which water flowed have altered water flow paths, levels, and storage, which affects management decisions.

Water Sources

The swamp has three water sources: streams and groundwater that flow from the Isle of Wight Plain across the Suffolk scarp and discharge to the western part of the swamp and precipitation. Precipitation is the original source of all water, falling across the Isle of Wight Plain and providing the water to these stream and groundwater sources. Flow of water from ditches into the peat or sand aquifers is an exchange of water within the swamp and not a source.

Stream and Groundwater Flow across the Suffolk Scarp

Streams and groundwater that flow from the Isle of Wight Plain and across the Suffolk scarp are two additional water sources to the western part of the swamp. A water budget partly based on chemical separation of flow in streams across Virginia estimated that 25.5 percent of the precipitation falling on the City of Suffolk, which includes the Isle of Wight Plain, discharges from groundwater into streams, and 10.1 percent discharges into streams as surface runoff (Sanford and others, 2012). Based on model-simulated flow during average springtime conditions, streams flowing across the scarp are the source of 5 percent of the water flowing through the swamp (Eggleston and others, 2018). Three larger streams discharge directly into Washington Ditch (West of Lake Drummond) and Cross Canal Ditch (South of Lake Drummond, fig. 1), whereas the three smaller streams discharge diffusely across land surface into the swamp (fig. 1).

Part of the groundwater discharged into streams in the analysis of Sanford and others (2012) likely does not discharge to streams but is the groundwater that flows through the surficial and Yorktown-Eastover aquifers, across the scarp, and discharges into the swamp. Thus, groundwater discharge into the swamp is part of the 25.5 percent of the precipitation falling on the City of Suffolk indicated as groundwater discharging to streams in Sanford and others (2012). Model-simulated flow during average springtime conditions indicates groundwater flowing across the scarp provides 3 percent of the water that flows through the swamp (Eggleston and others, 2018). A limit on the groundwater and stream sources appears to be temporal effects of pumping groundwater from the surficial and Yorktown-Eastover aquifers beneath the Isle of Wight Plain for domestic, livestock, and irrigation uses (Lichtler and Walker, 1974). This lowers groundwater levels, locally reversing gradients during parts of the year and reducing groundwater and stream flow to the swamp.

Seasonal and annual differences in streamflow in Cypress Swamp, one of the two largest streams that cross the scarp, help characterize variability in these stream sources and in the groundwater source because streamflow includes surface runoff and groundwater discharge; relatively low flows consist primarily of groundwater discharge. Streamflow was monitored by the USGS at Cypress Swamp at Cypress Chapel, Va. (station number 02043500; fig. 1; table 2) from 1953 to 1996 with a gap between 1971 and 1978. This stream has a 24-square-mile (mi²) watershed that represents 30 percent of the total watersheds draining from the Isle of Wight Plain into the swamp.

Mean monthly streamflow in Cypress Swamp varies seasonally, increasing through the winter non-growing season and peaking from January through March (fig. 19; fig. 20). Streamflow decreases as ET increases during the growing season, reaching seasonal lows through the summer into early fall. Mean monthly streamflow in August and September is higher than that in June and July, partly because large precipitation events from tropical low-pressure systems occur in some years and raise the average. In years without such systems, streamflow in August and September can be equal to or less than that in June and July, having zero flow much of the time (fig. 20C).

Mean daily streamflow in Cypress Swamp was episodic and flashy (fig. 20), depending on precipitation. Following large precipitation events, instantaneous streamflow approached 1,300 ft³/s, although mean daily streamflow remained less than 800 ft³/s. Outside the direct effects of precipitation, streamflow typically was less than 100 ft³/s (fig. 20). Eighty percent of the total annual streamflow typically occurs from October to early April. As streamflow decreases, the percent of groundwater discharge into streams increases, although the actual amount of groundwater discharge decreases as water levels and gradients decrease and riparian ET increases (Sanford and others, 2012). This causes streamflow to decrease eventually to 0 ft³/s, partly because of the high ET during much of the summer growing season. Although streamflow decreases to zero, groundwater flow to the swamp likely continues longer than streamflow as indicated by water levels in the ditches parallel to the scarp that typically remain stable into the growing season. Total annual runoff ranged from 5.8 in. (1995) to 20.0 in. (1957) and averaged 13.1 in.

Flow

Water flows through the swamp in a combination of local, intermediate, and regional or swamp-wide patterns. Local flow patterns are superimposed upon intermediate flow patterns that, in turn, are superimposed upon swamp-wide flow patterns as reflected in groundwater and surface-water flow paths (fig. 21). Swamp-wide flow is to the east in the primary direction of regionally decreasing land surface (fig. 2) and decreasing water-level altitudes from water sources along the Suffolk scarp to the Dismal Swamp Canal, the primary surface-water discharge. This pattern is most evident in the northeastern quadrant of the swamp and in the water budget developed using model-simulated flow during average springtime conditions (Eggleston and others, 2018). In the southern part of the swamp, an intermediate, southerly/southeasterly flow pattern toward the Pasquotank River is superimposed upon the easterly flow pattern (fig. 21). In the northern part of the swamp, two intermediate flow patterns are superimposed upon the easterly flow pattern: a southerly flow pattern toward Lake Drummond and the Feeder Ditch, and a northerly flow pattern toward Shingle Creek and Deep Creek (fig. 21). Local and intermediate flow patterns partly mask the swamp-wide flow patterns.

Local Flow

Groundwater recharged by precipitation flows through the peat aquifer to ditches predominately in local flow patterns. Understanding controls and characteristics of this flow also helps to understand characteristics and controls on intermediate and swamp-wide flow. Results of water-level measurements in Block C1 reveal major aspects of local controls and the resulting local flow paths.

The configuration of the water table in Block C1 reflects the effects of precipitation recharge and subsequent discharge as ET, flow through the peat and sand aquifers, flow inhibition by a spoil-pile road, and discharge into ditches (fig. 22). Groundwater levels varied seasonally, having high levels during wet seasons from late winter into early spring as on March 8–10, 2010 (fig. 22A), and low levels during dry seasons through the summer as on July 27–29, 2010 (fig. 22B). (Note that the contour intervals for these periods are 0.1 ft and 0.25 ft, respectively, because use of the same intervals would make one figure cluttered or the other would have limited definition of the water table.) Dry seasons are caused by the onset of ET, not decreased precipitation. The average monthly precipitation generally increases from its low in February to the maximum in August. Similarly, average monthly PET increases from a low in January to a high in July, exceeding average monthly precipitation from May through August to cause the dry conditions.

During the wet season, groundwater levels were highest in the southwest quadrant of Block C1 and lowest along Sycamore Ditch and at the northern end of Myrtle Ditch (fig. 22A). Levels ranged from a high of 17.48 ft above NAVD 88 in well W13W, west of the center of the block, to a low of 15.99 ft above NAVD 88 in well N00W next to Sycamore Ditch. This gradient indicates flow was north toward Sycamore Ditch and northeast toward Myrtle Ditch into which groundwater discharged. Much of the water discharged into Sycamore Ditch despite the low hydraulic-conductivity, spoil-pile road intervening between the swamp and the ditch because the water level in Sycamore Ditch was about 2 ft lower than that in Corapeake Ditch where no spoil-pile road intervened. The low hydraulic conductivity of the road is evident in the differences in water-level gradients from the center of the block to the road compared to those across the road to Sycamore Ditch. Water levels declined only 0.36 ft across the more than 2,500 ft (nearly half a mile) from the center of the block to well N01W, which is about 100 ft from Sycamore Ditch (horizontal gradient of 1.4 × 10^-4; see fig. 9 for well names). Levels declined 0.91 ft in only 100 ft from well N01W, across the road to the ditch, about 2.5 times the decline from the center of the block to well N01W. Because the horizontal gradient from well N01W to the road would be similar to that from the center of the block and the width of the road is about 30 ft, the horizontal gradient beneath the road was approximately 3.0 × 10^-2 or 210 times that from the center of the block.

The actual flow path from the swamp into Sycamore Ditch, however, remains uncertain based on these data. Results of two-dimensional, cross-sectional model simulations of flow between Corapeake and Sycamore Ditches (Eggleston and others, 2018), however, reveal this flow path and critical controls on groundwater flow. In a simulation that replicated the greater than 2-ft difference in water levels between Corapeake and Sycamore Ditches during the wet season of 2016, 84 percent of the groundwater flowed toward and discharged into Sycamore Ditch, despite the intervening spoil-pile road. Most of this water flowed through the upper peat until it was near the spoil-pile road and ditch where it flowed downward through the lower peat into the sand and flowed laterally under the spoil-pile road before discharging into Sycamore Ditch. Most groundwater flowing toward Corapeake Ditch, however, flowed through the upper peat and discharged directly into the ditch. The combination of field observations and model simulations indicate that most groundwater flows through the upper peat and that, despite the intervening spoil-pile road, the greater than 2-ft lower water level in Sycamore Ditch caused most of the water to flow toward and discharge into Sycamore Ditch.

The pattern during the dry season differed somewhat from that during the wet season. During the dry season, water levels declined 1.11 ft from the center of the block to well N01W, 100 ft from Sycamore Ditch (fig. 22B). This is more than three times the decline during the wet season. Levels declined 0.58 ft between well N01W and the ditch, about two-thirds of the decline of the wet season, indicating a lower rate of groundwater discharge during the dry season. Consequently, horizontal gradients increased to 4.4 × 10^-4 from the center of the block to the road and decreased to 1.9 × 10^-2 across the road. The horizontal gradient across the road was only 43 times that from the center of the block during the dry season.

Differences in groundwater levels, flow paths, and horizontal hydraulic gradients between the wet and dry seasons largely reflect the onset of ET coupled with the hydraulic characteristics of the upper peat, lower peat, sand, and roadbed. During the wet season (fig. 22A), the water table likely was in the upper peat throughout the block except between the road and Sycamore Ditch, based on the 16.0 ft above NAVD 88 altitude of the bottom of the upper peat and the 15.5 ft above NAVD 88 altitude of the bottom of the upper/lower peat transition at well C00W. Consequently, groundwater primarily flowed through the upper peat during the wet season as indicated by Eggleston and others (2018).

With the onset of ET that caused the dry season, the water table likely declined into the upper/lower peat transition or lower peat along and south of the east-west transect and into the lower peat to the north. Because the hydraulic conductivity of the sand is about four times that of the lower peat (100 ft/d compared to 24 ft/d (Eggleston and others (2018)), lateral groundwater flow beneath the northern part of the block likely was primarily through the sand aquifer.

Lateral flow within the upper/lower peat transition in the southern part of the block likely was limited because so little of the upper/lower peat transition was saturated, a distinct lateral gradient in groundwater levels was not present, and the contact between the lower peat and upper/lower peat transition likely varied. Groundwater in the southern part of the block likely flowed downward through the lower peat into the sand aquifer and then laterally through the sand aquifer toward the ditches. Because the upper peat was not saturated and most flow was through the sand aquifer during the dry season, discharge was directly from the sand aquifer into the ditches. Flow through the lower hydraulic conductivity sand aquifer would account for the greater hydraulic gradient from the center of the block to Sycamore Ditch during the dry season.

Characteristics of the response of groundwater levels from the center of the Block C1 (well C00W) east to Myrtle Ditch to precipitation on August 22, 2010, differed by distance from the ditch (fig. 23). Groundwater levels were stable near the altitude of the bottom of the upper/lower peat transition until after the approximately 0.5 in. of precipitation needed for interception and rewetting unsaturated peat; levels rose rapidly as precipitation continued. The rate of rise differed among wells but was rather uniform within each well when the precipitation rate was uniform. The rise was greatest at the center of the block (well C00W), less at 100 ft from the ditch (well E01W), and even less at the well adjacent to the ditch (well E00W) (fig. 23).

Responses to changes in precipitation rates after interception and rewetting were almost immediate away from the ditch (fig. 23). When the precipitation rate briefly slowed after about 1.5 in. of cumulative precipitation about midway through the event, the water level declined slightly in well C00W and rose more slowly in well E01W within the same 30-minute interval when the precipitation rate decreased (precipitation was recorded at 30-minute intervals). The water level again rose more rapidly at both sites as the precipitation rate increased. The rate of water-level rise was slower and nearly uniform throughout the event in the ditch and adjacent well E00W with the ditch water levels controlling the level in the well. Control of groundwater levels by ditch water levels decreased with distance from the ditch and demonstrates how groundwater-level response depends on location. The rapid response of groundwater levels to changes in precipitation rates away from the ditch reflect effects of the high hydraulic conductivity of the peat and open canopy of Block C1.

Intermediate and Swamp-Wide Flow

Flow through the swamp is complex because of the size of the swamp and the varied hydrologic controls. Eggleston and others (2018) identified 15 watersheds across the swamp, based on combined groundwater and surface-water flow paths derived from a model simulation during average springtime conditions (fig. 21). Throughout the swamp, water flows from sources and groundwater divides to ditches, Lake Drummond, and other surface-water features to which the water discharges (fig. 21). Watershed boundaries and flow paths from these boundaries illustrate the complexity of the flow caused by these hydrologic controls. These boundaries and flow paths can change appreciably along some ditch intervals because of manipulation of water-control structures and other changing controls. Although field-observed flow through ditches and watershed boundaries (fig. 2) might not fully match modeled ones (fig. 21), they appear to be close in most locations.

One of the major controls on flow paths is the surface-water and groundwater sources from the Isle of Wight Plain. Model-simulated flow during average springtime conditions (Eggleston and others, 2018) indicates that groundwater flowing from the Isle of Wight Plain and across the Suffolk scarp into the swamp discharges into the north-south oriented ditches nearest and parallel to the scarp and to the western end of east-west oriented ditches originating at the base of the scarp, or connecting to or originating immediately east of, the north-south ditches (fig. 2; fig. 21). North-south oriented ditches from the north are Lynn, West, Sherrill, and Weyerhaeuser Ditches. Thus, these ditches function as barriers by intercepting the eastward flow of water that crosses the scarp (fig. 21; fig. 24). Spoil-pile roads adjacent to the north-south ditches further impede flow, acting like dams to that flow (fig. 5). Spoil-pile roads are east of Lynn, West, and Weyerhaeuser Ditches; these roads impede flow past these ditches and help force the discharge of water into the ditches. Sherrill Ditch is the only ditch where the spoil-pile road is between the scarp and the ditch. Each ditch except West Ditch has a reach characterized by minimal flow between northward and southward flowing water (fig. 21) and water-control structures at each end of the ditch that retain water in the ditch (fig. 2). These splits in flow directions form watershed boundaries that can shift north or south depending on hydrologic conditions and water-control-structure settings. West Ditch has no structures (west of Lake Drummond, fig. 2) and only connects directly to South Ditch at its southern end, not to Railroad Ditch to the north or Interior or Lateral Ditches in between. Consequently, most flow through this ditch is to the south into South Ditch. The modeled watershed boundary around West, South, and Riddick Ditches likely does not fully depict flow because some groundwater from the scarp flows past West Ditch into South Ditch. Some of the water in South Ditch then flows south through the burn scar to Corapeake Ditch as depicted in figure 2, and some flows east through the South Ditch into Riddick Ditch and then flows north into Lake Drummond.

Four different flow patterns occur from this source (fig. 21; fig. 24). Where east-west ditches originate near the scarp, some of the groundwater discharges into these ditches before reaching the north-south ditches. Most groundwater flowing from the scarp west of Lynn and Weyerhaeuser Ditches discharges directly into these ditches and does not flow past them. At Sherrill Ditch, however, groundwater from the scarp west of the ditch flows past the ditch and splits into flow to the north and south (fig. 21; fig. 24). This groundwater then flows west where it discharges into Sherrill Ditch through its eastern side. Near West Ditch, part of the groundwater flows past the ditch and discharges into the parts of the east-west ditches immediately east of West Ditch (fig. 21; fig. 24). The decrease in land-surface and water-table altitudes across the Suffolk scarp combined with low gradients in the land-surface altitude and hydraulic gradients across the swamp cause this discharge. Thus, this combination of ditches limits the eastward extent of groundwater flow from the scarp.

These discharge patterns appreciably alter the conceptualization of Lichtler and Walker (1974), which describes how groundwater flows beneath the swamp through the Norfolk aquifer (surficial and Yorktown-Eastover aquifers) and discharges upward into the peat across the entire swamp and into Lake Drummond. Although Lichtler and Walker (1974) state that groundwater discharges in this manner, they also present groundwater levels from Charles T. Main, Inc. (1971) for three dates in late 1971 and early 1972 at five sites located in a line from near the scarp to Lake Drummond. Each site had paired wells individually open to the peat or underlying sand at different depths. These levels show a downward hydraulic gradient indicating downward, not upward, flow at all sites except the one next to Lake Drummond where the vertical gradient was negligible (Lichtler and Walker, 1974, fig. 13).

Across the swamp, modeled groundwater divides include watershed boundaries and other linear bands having no flow lines but from which flow lines originate where watershed boundaries are not depicted (fig. 21). Some divides are close to evenly spaced between ditches on opposite sides of the local swamp, whereas many are much closer to the ditch on one side than the other because of differences in water levels between these ditches, as observed in the local study in Block C1. Areas where the divide is nearly evenly spaced between ditches include parts of the northwestern quadrant and the Blocks. Several areas where the divide is shifted toward one of the ditches are (1) the northeastern quadrant between Portsmouth Ditch and the Dismal Swamp Canal, (2) many of the areas draining south to Lake Drummond and the Feeder Ditch, and (3) the southern part of the swamp that drains to the Pasquotank River (fig. 21).

Responses of water levels and flow in ditches to precipitation (fig. 25) depend on local flow controls and the location on the ditch relative to the controls. Because the permeability of sand, silt, and clay is less than that of the peat, precipitation falling on sand, silt, and clay infiltrates more slowly and creates more surface runoff than precipitation falling on the peat. Additionally, flow as surface runoff is more rapid than flow through the peat. Consequently, ditches receiving flow directly from streams originating on the Isle of Wight Plain contain more surface runoff and respond more rapidly to precipitation than ditches for which flow originates in the peat. Thus, the response to precipitation is muted in ditches in which flow originates in the peat but is more pronounced in Washington Ditch and Cross Canal Ditch because of the direct connections at their head of flow with streams that drain across the scarp from the Isle of Wight Plain (fig. 1). These ditches rapidly transport water to the east to discharge into Lake Drummond and the Pasquotank River, respectively. Such pronounced effects are apparent for the Corapeake Swamp inflow into Cross Canal Ditch (fig. 25A).

The effects of flow across the scarp become less pronounced eastward toward the Dismal Swamp Canal whether the ditch connects directly to inflowing streams or entirely receives groundwater discharge and diffuse streamflow from across the scarp. Discharge of groundwater recharged in the peat by precipitation, manipulation of water-control structures, and dampening effects of the ditch on flow increasingly control the flow and water levels in ditches to the east.

The less pronounced effects on water levels to the east in a ditch receiving flow from a stream flowing across the scarp is evident in a comparison of levels in the inflow of Cross Canal Ditch (fig. 25A) with downstream levels in Cross Canal Ditch at Weyerhaeuser and County Line Ditches (southwestern quadrant, fig. 2) 3 mi into the swamp (fig. 25B). Hydrographs of flow from February 1, 2012, to May 9, 2012, at these sites show these effects on flow (fig. 26). Flow increases rapidly at the Cross Canal Ditch inflow site following a precipitation event, typically peaking about 24 hours after precipitation ceases. The downstream response at the intersection with Weyerhaeuser and County Line Ditches is dampened and lags that of the inflow, having peak flow downstream increasing less than one-half that at the inflow site and peaking about 3-4 days later for large events.

Ditches hydrologically isolated from streams that flow across the scarp are even more stable. One such site is the Camp Ditch outflow that drains water from the wet, northwestern quadrant and primarily receives groundwater that flowed across the scarp and precipitation that recharged the peat aquifer (fig. 25B). Yet another site isolated by spoil-pile roads, ditches, and water-control structures is Portsmouth Ditch at Rosemary Ditch (fig. 25B). This site primarily receives precipitation that recharges the peat aquifer and little water from either source that crosses the scarp.

Knowledge of the extent of the effects of a water-control structure on upstream levels in a ditch is critical for management decisions. The extent of effects along a ditch depends on the slope of the land surface adjacent to the ditch. Structures along high-relief sections (midstream, fig. 27A; fig. 27B) affects gradients along relatively short reaches of ditches but the short reaches have greater gradients than reaches having structures in gently sloping sections (upstream, fig. 27A) based on U.S. Fish and Wildlife Service data. Levels are closest to land surface immediately upstream from water-control structures and farthest below land surface immediately downstream from structures, the farthest point upstream from the next downstream structure. In the Corapeake Ditch profile (South of Lake Drummond, fig. 2; fig. 27A), the water level is near land surface along the entire segment west of the structure at Western Boundary Ditch and above land surface through the burn scar. Between Western Boundary and Laurel Ditches, the water surface is farther below land surface because of the greater slope and variable land-surface altitude. No structure was present downstream from Laurel Ditch at that time (November 23, 2011) so that the level of the Dismal Swamp Canal and the gradient in the ditch bed-controlled water levels. This accounts for the 4-ft to 5-ft difference between the water and land-surface altitudes and clearly shows the adverse effects of the absence of a downstream structure. A structure constructed on South Martha Washington Ditch between the confluence of South Martha Washington Ditch with Corapeake Ditch and the canal (fig. 18F) after November 23, 2011, maintains levels close to land surface along this ditch interval.

Like the interval on Corapeake Ditch between Western Boundary and Laurel Ditches, water levels along County Line Ditch between Weyerhaeuser and Insurance Ditches (South of Lake Drummond, fig. 2) are nearly 1 ft below land surface immediately downstream from Weyerhaeuser Ditch because of the slope of the land surface (fig. 27B). Even with water control in place, ditch water levels drop several feet below the land surface during dry climatic conditions. Control structures, however, raise the altitude and decrease the duration of low levels.

Adding water-control capability, either through installation of new water-control structures or repair of damaged and leaking structures effectively raises ditch water levels as observed at station number 363342076261100 (table 3) on Sycamore Ditch at the midpoint of the north side of Block C1 (fig. 2; fig. 28). Before structures were constructed in March 2013, flow in the ditch originated as groundwater discharged from the peat aquifer on both sides of the ditch into the 0.5-mi interval upstream from the site. The spoil-pile road along Western Boundary Ditch separated Western Boundary Ditch from Sycamore Ditch. The only structures on the ditch were more than 3 mi downstream. These structures leaked so that they retained little water in the ditch.

In 2013, the leaking structures were repaired, and two new structures were installed. One new structure connected the head of flow in the Sycamore Ditch with Western Boundary Ditch; this is a culvert and riser installed under the spoil-pile road to provide an additional water source to Sycamore Ditch. The other structure is at Laurel Ditch, about 1.5 mi downstream from the water-level monitoring station; a culvert and riser replaced a culvert having no control under Laurel Road. The addition of an upstream inflow and a downstream control on outflow raised average winter high levels by 1 to 2 ft and summer extreme low water levels by more than 2 ft, decreasing the range in levels. These changes demonstrate the combined effects of the connections of ditches with the peat and water control on water levels in ditches.

Discharge from the Swamp

Water discharges from the swamp as ET, through the ditches, through the canal system, into streams north and south of the swamp, and through groundwater flow to adjacent areas. Discharge from the peat and sand aquifers into the ditches within the swamp are not considered discharge from the swamp but exchanges within the swamp. Based on model-simulated flow during average spring conditions, discharge as ET is 55 percent of the total discharge, through the canal system is 23 percent of the total discharge, into streams north and south of the swamp is 18 percent of the total discharge (stream location not differentiated), and through groundwater to adjacent areas is 4 percent of the total discharge (Eggleston and others, 2018).

During the growing season, groundwater discharges as ET causing groundwater levels to decline in 1 of 2 diurnal patterns that reflect local discharge into ditches and discharge from the swamp as ET (fig. 29). The diurnal pattern most commonly observed in groundwater levels is “sinusoidal,” having a daytime decline followed by a smaller nighttime rise, generally resulting in a net day-to-day decline (fig. 29A). White (1932) first described this pattern, which many wetland (for example, Mitsch and Gosselink, 2007) and groundwater-hydrology textbooks (for example, Freeze and Cherry, 1979) have long referenced. This pattern occurs in systems where appreciable groundwater flows laterally and (or) vertically from an external source, down hydraulic gradient to partly replenish lateral outflow and discharge as ET. The daytime decline occurs because discharge as ET and as lateral outflow exceeds inflow from the external source; levels rise during the nighttime because this inflow exceeds outflow when ET is minimal. The day-to-day decline reflects a net discharge during the 24-hour period. In some instances, the decline and rise can be of similar magnitude so that levels change little day-to-day (well ENE01, fig. 29A), indicating that most of the discharged groundwater is replaced by the source.

The second and apparently less common diurnal pattern is “step-type” in which levels remain stable or decline gradually during the nighttime and decline more rapidly during the daytime (fig. 29A and B). The nighttime decline is almost entirely from net lateral outflow and discharge into the ditches. The level declines rapidly in the daytime because discharge into the ditches continues, but ET becomes the dominant discharge pathway. This pattern occurred in wells throughout Block C1 and was greater at the center of the block (well C00W) than near the ditches (wells N01W) because levels in ditches partly control groundwater levels near the ditches.

The small decline in the absence of a rise during the nighttime identified in Block C1 indicates the presence of a minimal, or the complete absence of an external, source of groundwater so that precipitation is the primary or only source of groundwater. Thus, groundwater levels rise only during recharge by precipitation (fig. 23) but decline between recharge events because of a net outflow from discharge into the ditches combined with discharge as ET.

The daily cycle in groundwater levels commonly increases in magnitude as the rate of ET increases from the spring into summer and as groundwater levels decline. Such increases have been identified similarly in surficial sand aquifers elsewhere in the Coastal Plain of Virginia where the daily amplitude of the diurnal, sinusoidal cycle increased to as much as 0.75 ft at a depth of about 3 ft (Speiran, 1996; Speiran, 2010). Across most of the swamp, the increased amplitude has three possible causes: (1) decreasing specific yield of the peat with depth, (2) a shift in ET from soil water (moist unsaturated peat) derived from precipitation to the groundwater source, and (3) an increasing percentage of fine and small roots (roots that take up water and nutrients) of trees and other vegetation become exposed to unsaturated peat as water levels decline. For the first cause, decrease in specific yield is unlikely in Block C1 while water levels remain in the upper peat because specific yield of the upper peat appears to be uniform (fig. 15). The specific yield decreases, however, through the upper/lower peat transition into the lower peat so that the amplitude of the cycle increases as the water table declines through that interval.

For the second cause, most of the precipitation intercepted by vegetation and rewetting unsaturated peat discharges as ET. As water remaining in unsaturated peat becomes depleted, increasing amounts of groundwater are used by the vegetation, increasing the amplitude of the diurnal groundwater-level cycles. In Block C1, discharge from interception and rewetted, unsaturated peat likely persists for only a short period because the shallow water table limits the precipitation stored in this reservoir. Thus, an increase in the amplitude of the diurnal cycle because of this shift is likely but probably is of short duration.

For the third cause, when the water table is shallow, most fine and small roots are in saturated peat. Fine- and small-root density is greatest near the peat surface and decreases with depth. The 0.0- to 0.1-m (0.0- to 0.3-ft) depth interval at the swamp contains 67, 57, and 57 percent of the fine and small roots present in the upper 0.4 m (1.3 ft) of peat in cedar, bald cypress, and maple/gum forest communities, respectively (Powell and Day, 1991). This 1.3-ft depth approaches the depth of the bottom of the upper peat in Block C1 (fig. 15B). Saturated soil typically becomes anoxic, especially in organic soil such as peat, interfering with normal metabolic processes. The anoxic conditions in saturated soil can stress and possibly kill upland species (Armstrong and others, 1994). Wetland species, however, employ adaptations for tolerating these conditions. These are metabolically inefficient “survival” adaptations. As groundwater levels decline, more fine and small roots are exposed to unsaturated, oxygenated peat, allowing the vegetation to shift to more metabolically efficient processes that allow them to thrive and not just survive. Because trees take up groundwater through unsaturated flow from the water table to the roots (Sánchez-Pérez and others, 2008), more roots are in unsaturated peat as the water table declines. This increases the amount of groundwater uptake and, consequently, increases the rate of water-level decline as groundwater levels become deeper within the upper peat.

As groundwater levels decline into the lower peat, the low hydraulic conductivity of the peat limits the flow of groundwater to the roots. This decreases the rate of groundwater discharge as ET, potentially stressing the forest and affecting growth. Phipps and others (1978) noted stress attributed to the “dry” conditions in their analysis of tree growth rings across the swamp and identified this stress as a possible factor in the change in forest types. Exposure of more roots to unsaturated peat accompanied by decreased specific yield and hydraulic conductivity creates competing effects on plant uptake of groundwater and synergistic effects on the decline of groundwater levels.

Characteristics of the canal system and ditches and how they are managed in response to precipitation also cause seasonal and annual variability in discharge from the swamp. The infrastructure on Lake Drummond and the Dismal Swamp Canal at Deep Creek, Va., and South Mills, N.C., provide the USACE with the appreciable capability for managing discharge from the swamp though the canal systems. Water-control structures on the ditch network provide the FWS with the ability to manage discharge from the swamp into surrounding rivers and the canal system that increases as more structures are constructed and structures are better managed.

A large part of the management of discharge from the swamp through the canal system depends on how the USACE manages levels in Lake Drummond by controlling releases from the lake through the Feeder Ditch. Prior to establishing the minimum, operating level of 14.38 ft above NAVD 88 for Lake Drummond in 1976, levels commonly dropped below this level (fig. 30). Since 1976, the USACE has maintained lake levels above or near this minimum level. This change in operation has reduced the monthly range in water levels by an average of 2.0 ft from the pre-1976 agreement water levels (fig. 30) and increased the frequency of closing the canal to navigation (U.S. Army Corps of Engineers, 1996).

Releases from Lake Drummond to the Dismal Swamp Canal create a flow divide at the mouth of the Feeder Ditch. South of the divide, water predominantly flows south toward South Mills. North of the divide, water predominantly flows north toward Deep Creek. Based on data provided by the USACE, estimated mean monthly releases from Lake Drummond and the Deep Creek and South Mills locks and spillways were of similar volume from 2010 through 2015 (fig. 31), averaging 141, 136, and 150 ft³/s, respectively. Model simulation of flow during average spring conditions estimated that 56 percent of the surface-water discharge from the swamp was through the canal system (Eggleston and others, 2018). Approximately 80 percent of the annual release was from October through May and typically was greatest from January through April, corresponding with the highest water levels in the lake and surrounding swamp. The lowest releases were during dry summer months of June through August. Because releases from each lock and spillway combination are about equal to releases from Lake Drummond, about one-half of the releases from the locks and spillways enters the canal as discharge directly from ditches, groundwater discharge through spoil banks, or precipitation falling directly on the canal.

Flow characteristics of the three streams that drain the swamp also affect management of discharge from the swamp. These characteristics are not well documented, partly because these are small tidal streams, and continuously recording flow in such systems was difficult using earlier technologies. The only streamflow-gaging station on these streams is the Pasquotank River near South Mills, N.C. (USGS station number 0204382800 fig. 1; table 2). The watershed for this station covers 64 mi² of the southern part of the swamp and agricultural land south of the swamp. The period of record for the station extends from 2004 to the present (2019). Tidal conditions derived from Albemarle Sound affect water levels and flow, causing periodic flow reversals. Although the general seasonal distribution of flow in the Pasquotank River (fig. 32) in water year¹ 2013 is consistent with that of Cypress Swamp (fig. 20), responses to precipitation were less flashy in the Pasquotank River, and flow persisted through dry seasons (fig. 32). These differences likely result from the combination of tidal effects, differences in the periods of record, effects of peat compared to mineral sediment, extent of wetlands, and watershed size. The Pasquotank River watershed area is more than 2.5 times the area of the Cypress Swamp watershed.

¹

A water year is the 12-month period from October 1 to September 30 of the following year. It is designated by the year in which it ends.

Groundwater typically discharges to areas to the north and south of the swamp, varies seasonally like other flows, and is similar in magnitude to groundwater flow across the scarp. Groundwater does not discharge to the west because the scarp is a water source where groundwater flows into the swamp. Groundwater discharging into the Feeder Ditch and the Dismal Swamp Canal through intervening spoil banks is part of the discharge through the canal system and is not included as part of the groundwater discharge from the swamp. This groundwater discharge is part of the doubling of flow between the outflow from the Feeder Ditch and the locks and spillways at Deep Creek and South Mills.

The Northeastern Quadrant of the Swamp

The northeastern quadrant of the swamp provides a snapshot of the hydrologic response to numerous controls likely to be encountered elsewhere across the swamp and several controls seldom encountered, creating differing responses across the area. Discussion of the resulting responses and their controls helps to identify the likely controls when similar responses are observed elsewhere across the swamp. Flow across the area reflects local patterns superimposed upon the intermediate, northerly pattern and the swamp-wide, easterly pattern as reflected in the configuration of the water table during wet (fig. 33A) and dry seasons (fig. 33B). (Note: the red values in fig. 33 followed by “INT” are ditch levels linearly interpolated between upstream and downstream ditch levels.) The area has two main water sources: precipitation and flow from East Ditch along the western boundary of the area (fig. 33). Flow like that from East Ditch is not common elsewhere, but like groundwater flow across the scarp into the swamp, it provides a generally stable water source. Discharge is as ET across the entire area, as well as from the swamp through local ditches to Deep Creek to the north (the northerly, intermediate pattern) and toward the Dismal Swamp Canal to the east (the swamp-wide, easterly pattern). From southwest to northeast, groundwater flow increasingly shifts from the peat into the sand aquifer (fig. 17).

Because precipitation recharge is a main water source to this area, precipitation amounts, and temporal patterns provide important controls. In 2015 and 2016, annual and monthly precipitation at the gage at East Ditch varied appreciably and differed from mean annual and monthly amounts at the National Oceanic and Atmospheric Administration Lake Kilby station (fig. 34). Measured annual precipitation of 53.32 in. in 2015 and 79.66 in. in 2016 exceeded the mean annual precipitation of 49.49 in. (7.7 percent greater and 61 percent greater, respectively) at the Lake Kilby station. In contrast to the general seasonal pattern in the mean monthly precipitation for the Lake Kilby station, monthly precipitation in 2015 and 2016 followed no seasonal pattern with large month-to-month variations. Mean monthly precipitation at the Lake Kilby station ranged from a low of 3.23 in. in February to a high of 5.71 in. in August, whereas measured monthly precipitation in 2015 ranged from a low of 2.22 in. in May to a high of 11.55 in. in June. Precipitation in 2016 ranged from a low of 0.57 in. in November to a high of 17.37 in. 2 months earlier in September. The highest monthly precipitation for 2015 and 2016 was more than 2 and 3 times the highest mean monthly values at the Lake Kilby station, respectively. The large amounts of precipitation in September (17.37 in.) and October (13.33 in.) 2016 resulted primarily from two tropical weather systems. More than one-half the precipitation in September (more than 9 in.) fell in just over 36 hours during September 20–22 from Tropical Storm Julia. Hurricane Matthew produced an even greater 12.21 in. in 24 hours on October 8–9. Precipitation for each of 6 months in 2016 exceeded the maximum mean monthly amount at the Lake Kilby station.

The heavy precipitation from Tropical Storm Julia and Hurricane Matthew caused the worst flooding across the swamp and surrounding areas since Hurricane Floyd in 1999. Flooding of the swamp from these events inundated spoil-pile roads, making them impassable. It also short-circuited cable connections of pressure transducers at the Big Entry, Portsmouth, and Rosemary Ditch sites. This caused missing record from September 22, 2016, 2 days before the peak level in Juniper Ditch from Tropical Storm Julia, until about October 25, more than 2 weeks after the peak level in Juniper Ditch from Hurricane Matthew. The condition of spoil-pile roads prevented access to repair the equipment throughout this period. Additional factors, including damage from black bears, contributed to missing record that ranged from no missing record at Juniper Ditch to more than 36 percent missing record at Rosemary Ditch.

Water-level gradients among ditches and wells were similar across the northeastern part of the swamp (fig. 35; fig. 36). Water levels in ditches decreased about 10 ft from west (East Ditch) to east (Rosemary Ditch) and about 2–3 ft from south (Juniper Ditch) to north (Big Entry Ditch) across the western block (fig. 35). Similarly, groundwater levels decreased about 7 to 10 ft from west (well WW13) to east (well ENE01) and less than 1 ft to about 3 ft south (well WS01) to north (well WN01) across the western block (fig. 36).

Ditch water-level and nearby groundwater-level responses to controls are discussed using hydrographs having different temporal resolutions and periods at East Ditch (fig. 37), Portsmouth Ditch (fig. 38), and Rosemary Ditch (fig. 39). The same three periods are used for each ditch-water/groundwater combination to facilitate comparisons among sites. The periods are (1) January 2015 to January 2017, (2) the early spring wet season (represented by March 12 to April 2, 2015), and (3) the late summer dry season (represented by August 16 to September 6, 2015). Data for the Rosemary Ditch site, however, did not begin until March 18, 2015 (fig. 39). The 2-year hydrographs depict the large variability in levels throughout the year. The wet-season hydrographs reflect recharge and flow during minimal ET. The dry-season hydrographs reflect recharge and flow during high ET. Hydrographs for both the wet and dry seasons cover 3 weeks to facilitate comparisons between seasons and among sites. These seasons and sites depict a range in the types of responses that occur with no intention of reflecting all types of responses.

The magnitudes of, and rates of change in, water levels differed among ditches (fig. 35), because of the effects of water sources and other controls from within and outside of the area. East Ditch, which forms the western boundary of the northeastern quadrant of the swamp, exchanges water with North and Williamson Ditches to the north, and Hudnell and Camp Ditches to the south. These ditches interconnect with other ditches across the wet, northwestern quadrant of the swamp. Connections of East Ditch with these ditches and the persistent wetness of the area provided a relatively stable water source to East Ditch. Where East Ditch forms the western boundary of the area, lateral flow in the ditch was minimal, sometimes visibly imperceptible, and changed direction. Not only was flow minimal in East Ditch, but water-level fluctuations were muted compared to other ditches (fig. 35) and nearby well HE01 to the west (fig. 36, fig. 37) because of the stable water source. In this interval, East Ditch generally supplements precipitation as a water source to the peat aquifer to the east, based on the gradient between ditch and groundwater levels (fig. 33; fig. 37).

The hydrographs for East Ditch and the wells 100 ft west (well HE01) and 1,300 ft east (well WW13) of the ditch (fig. 37) reveal responses to several controls. Water levels in East Ditch were stable throughout the wet season but decreased and fluctuated more during the dry season than during the wet season (fig. 37A). Levels appeared the least stable as the altitude of levels declined below about 19.2 to 19.4 ft above NAVD 88; fluctuations below this level resembled those of well HE01. The difference in response above and below this altitude was demonstrated by the relatively rapid rise in levels of about 0.5 ft owing to the more than 9 in. of precipitation from Tropical Storm Julia in late September 2016 with levels rapidly dropping less than 0.2 ft for a net rise of 0.3 ft when levels were below this altitude. In contrast, levels rose less than 0.4 ft owing to more than 12 in. of precipitation from Hurricane Matthew in early October and dropped about 0.3 ft within 24 hours of the peak for a net rise of 0.1 ft when levels were above this altitude. The initial rise was less than one-half of what would result from the precipitation falling directly on the ditch. The rise in East Ditch in response to these tropical systems was substantially less than the rise of about 3  ft at Juniper Ditch (fig. 35) and the more than 1.5-ft rise (record was lost near the peak) in well HE01 (fig. 37A).

The stability of levels in East Ditch and recharge of the peat aquifer to the east from the ditch maintain similarly stable levels generally at a slightly lower altitude in well WW13, 1,300 ft to the east (fig. 37). However, levels in well HE01, only 100 ft to the west fluctuated more and were about 1 ft or more above ditch levels throughout much of the year. Differences were, in part, due to the presence of the spoil-pile road between well HE01 and the ditch but not between well WW13 and the ditch. The higher and less stable levels in HE01 are a strong indication of the flow-inhibiting effects of spoil-pile roads, also identified in water-level differences between well N01W and Sycamore Ditch in Block C1 (fig. 22). Consequently, the swamp west of the spoil-pile road is part of the stable water source to East Ditch with the low hydraulic conductivity of the spoil-pile road partly contributing to this stability. Well HE01 also is upgradient from the ditch and well WW13 in the swamp-wide easterly flow pattern.

East Ditch and the two nearby wells show only a limited, stair-step ET cycle during the dry season (fig. 37C). This includes well WW13, despite the inflow from East Ditch, which commonly causes a sinusoidal cycle. A similar rate of ET and the low hydraulic gradient from the ditch to the well and across the entire northwestern quadrant source area could account for the absence of a sinusoidal ET cycle. The stable nighttime water levels indicate that inflow and outflow are similar. ET might in part contribute to less stable conditions as water levels decline below 19.2 ft NAVD 88 and more fine and small roots are in unsaturated peat.

Water levels in Portsmouth Ditch and wells ENW01 and WNE01 on either side of the ditch responded differently than East Ditch and its adjacent wells but similarly to each other (nearly coincident throughout the period), changing more than 6 ft over the 2-year period (fig. 38A). Water-level fluctuations in Portsmouth Ditch were among the greatest of all ditches in the area (fig. 35). Water levels in Portsmouth Ditch rose more in response to precipitation (fig. 38) than those in East Ditch (fig. 37), possibly because Portsmouth Ditch receives a large part of its flow from its watershed south of the study area without a stable water source (fig. 21).

During the wet season, the ditch and wells at the Portsmouth Ditch site responded similarly to precipitation (fig. 38B). Before precipitation events, a small water-level gradient was from the wells toward the ditch, indicating discharge from the peat into the ditch as water levels steadily declined. As water levels rose, the gradient remained small and from well WNE01 to the ditch, but it briefly reversed from the ditch to well ENW01 in the eastern block, indicating flow was from the ditch toward that well. Gradients from both wells again were toward the ditch as water levels declined. The direction of the gradient between well ENW01 and the ditch likely changed during the rise because the spoil-pile road intervened between the peat aquifer at well ENW01 and the ditch, impeding flow and delaying equilibration of levels between the well and the ditch, whereas no road separated the peat aquifer at well WNE01from the ditch.

During the dry season (fig. 38C), responses and relations were different than during the wet season (fig. 38B). Water levels in the ditch declined at a steady but decreasing rate with no clear ET cycle. Water levels in wells WNE01 and ENW01, however, reflected the effects of ET. As water levels declined, the gradient indicated a shift from flow from the wells toward the ditch to flow from the ditch toward both wells. The magnitude of this gradient increased as the characteristic of the ET cycle shifted from a stair-step to somewhat of a sinusoidal pattern as water levels declined, reflecting the gradient reversal. This shift was more evident in well ENW01 than WNE01, probably because of the effects of the intervening roadbed on the response in well ENW01. This shift to the sinusoidal pattern having an increased gradient shows that the ditch functioned increasingly as a local water source to the peat aquifer.

The response to precipitation during the dry season (fig. 38C) also differed from that of the wet season (fig. 38B). When precipitation began, groundwater levels rose rapidly at first, then slowed, and then rose more rapidly, whereas the response in the ditch was slower and more uniform. The rise in the ditch water level was largely flow from throughout the upper watershed (fig. 21) because the gradient locally remained from the ditch to the wells so that the ditch continued to contribute water to the peat throughout the event. The initial rapid rise in groundwater levels appears to be local recharge by precipitation, which slowed after the initial high intensity precipitation. As ditch water levels rose, flow into the peat from the ditch increased, and flow through the peat from the south and west also contributed to flow (fig. 33), causing the second rapid rise in groundwater levels as the gradient from the ditch to the wells increased.

Responses at the Rosemary Ditch site differed appreciably from responses at the two other sites in several ways. Water-level responses in the ditch were muted compared to those in the two nearby wells (fig. 39A) and Portsmouth Ditch (fig. 38A) but not as muted as that of East Ditch (fig. 37A). When comparing water levels, note that well ENE01 is about 100 ft from the Rosemary Ditch monitoring site, but well EE13 is about 0.75 mi upstream from the site so that water-level altitudes in that well cannot be compared directly to water-level altitudes in the ditch. Levels in the ditch typically ranged from 8 to 9 ft above NAVD 88 but increased to about 12 ft above NAVD 88 during the brief response to Hurricane Matthew. Levels in well ENE01 ranged from about 6.5 to more than 12.5 ft above NAVD 88 and in well EE13 ranged from about 8 to more than 13 ft above NAVD 88. Although the water-level ranges in the two wells were similar to those observed in wells by Portsmouth Ditch, the changes were more rapid in EE13 and ENE01 (fig. 38A; fig. 39A). Part of this difference in response likely was due to the limited peat toward the east causing groundwater levels to fluctuate more within the sand aquifer than within the peat aquifer. Fluctuations of the shallowest levels at well EE13 were smaller because the thickness of the peat is about 2 ft in that well (fig. 13). Because the sand has a lower specific yield than the peat, levels would rise more for the same amount of precipitation recharge and fall more for the same ET or ditch flow. This likely was part of the reason groundwater levels rose more in well ENE01 than in well EE13 in response to precipitation during the wet season (fig. 39B). During the wet season, levels in well ENE01 were 1 to 2 ft higher than the ditch water level with no spoil pile between the ditch and the well so that groundwater readily discharged into the ditch.

The nearby Dismal Swamp Canal likely contributes to muting changing levels in Rosemary Ditch in two ways. First, the canal is much wider than Rosemary and Big Entry Ditches, allowing a large amount of flow to be stored in the canal and transported through it with only a small change in levels. Secondly, operation of the canal for navigation attempts to maintain stable levels that can extend into Rosemary Ditch. The one exception to this was when flooding from Hurricane Matthew overwhelmed the response of the canal and ditch systems (fig. 35).

During the dry season, several changing responses to controls occurred (fig. 39C). Ditch levels remained stable compared to groundwater levels because of the effects of the canal. Groundwater levels in well ENE01 declined below ditch-water levels because of ET, which reversed the gradient as the ditch became a water source to the sand aquifer near the well. Levels in well ENE01 declined in a sinusoidal pattern that reflected the ditch as a water source to the sand aquifer. As groundwater levels declined, the rate of decline decreased and the amplitude of the cycle decreased in contrast to the increase observed in sand of wetlands elsewhere in the Coastal Plain of Virginia (Speiran, 1996; Speiran, 2010). The decrease in rate and amplitude with depth could be because the water table was about 5 ft below land surface, so that fine and small roots that typically are near land surface were substantially above the water table. Unsaturated flow toward the roots and land surface, therefore, would decrease as the water-table depth increased, indicating a decrease in the groundwater ET rate.

For well EE13, available data could not be used to identify the level at which the ditch became a water source, if it did. Because well EE13 is upstream from the Rosemary Ditch monitoring site, the reversal would occur at a higher altitude than the ditch water level at the monitoring site. Levels in well EE13 declined in a step pattern in mid-August, indicating discharge as ET with the ditch as a minimal water source near the well. Response to conditions in the ditch likely are limited because Well EE13 is 1,300 ft from the ditch whereas well ENE01 is only 100 ft from the ditch. As groundwater levels declined below 8.5 ft above NAVD 88 in well EE13 (below the ditch water level at the ditch monitoring site), the amplitude of step-type cycle decreased and shifted to a steady but decreasing rate of decline. Because groundwater levels declined to more than 5 ft below land surface, a decrease in the rate of groundwater discharge as ET is likely, as identified for well ENE01. Although an increase in specific yield could explain these changes, such a change is unlikely because the rate of decline slowed by a factor of 4 to 5, meaning specific yield would have to increase by that factor. Because the level in well EE13 continued to decline after the ET cycle ceased and until the recharge by precipitation on September 2, it is unlikely that the ditch became a source of water to the sand aquifer at that well.

The water-table configuration and altitudes across the northeastern quadrant of the swamp reflect the effects of controls at the East, Portsmouth, and Rosemary Ditch sites and additional controls. During the wet season in March 2015, the water table was high, and gradients in the western block indicate flow was from the East Ditch water source toward Big Entry and Portsmouth Ditches. The magnitude of the gradient changed little across the southwestern part of the western block but increased toward Big Entry and Portsmouth Ditches (fig. 33A). This increase resulted from the combination of the thinning of the peat aquifer to the north and east (fig. 14), the lower land surface near the ditches (fig. 17A; fig. 17B), and the presence of a spoil-pile road between the swamp and Big Entry Ditch. The gradient also was sharp in the northwestern corner because the East Ditch water source is close to the head of flow of Big Entry Ditch to which groundwater discharges (fig. 33). Although a spoil-pile road separates the swamp from Juniper Ditch, the gradient toward Juniper Ditch was small, indicating limited discharge into the ditch.

In the eastern block, the gradient was gradual from Portsmouth Ditch toward Rosemary Ditch to the east during the wet season having the highest level in the southwestern corner of the block (fig. 33A). The gradients toward Rosemary Ditch to the south and Big Entry Ditch to the north were small, indicating limited discharge into these ditches, probably because spoil-pile roads separate the swamp from the ditches; no spoil-pile road separates the swamp from Rosemary Ditch to the east. Because the level was highest in the southwestern corner of the eastern block during the wet season, Portsmouth Ditch possibly recharged the groundwater to the east. Such recharge likely is small because the gradient from the ditch to the east is small along most of the boundary and the spoil-pile road separates the ditch and the eastern block.

The water table was lower during the dry season in September 2015 (fig. 33B) than during the wet season in March 2015 (fig. 33A), and the pattern was similar to that of the March 2016 wet season (fig. 33A) with one major exception. The water table in the northeastern corner of the eastern block in September had a depression from ET with groundwater levels below levels in Big Entry and Rosemary Ditches, indicating the flow was from the ditches into the aquifer. The flow reversal from Portsmouth Ditch to wells WNE01 and ENW01 (fig. 38C) is not depicted in the dry-season water-level map (fig. 33B) because the gradient was small at the time of the periodic measurement, no more than a 0.04-ft difference, which is less than the resolution of the water-level map.

Input and Input Storage

Precipitation followed no distinct seasonal pattern in 2010; month-to-month amounts quickly decreased from the annual high of 19.71 in. in September to 3.55 in. in October, and then the annual low of 0.68 in. in November (table 10). Precipitation totaled 62.84 in. annually and averaged 5.24 in. per month. Most precipitation in September resulted from the 18.78 in. (about 30 percent of the annual total) from the remnants of Tropical Storm Nicole during September 26–29. Unlike the large amounts of surface runoff from Tropical Storm Julia and Hurricane Matthew, effects of surface runoff were not evident after this event, even with the larger amount of precipitation, because of dry antecedent conditions.

Approximately 9.47 in. of the annual precipitation (15 percent of the annual total) was intercepted by vegetation and land surface or rewetted unsaturated peat in no distinct seasonal pattern (table 10). Amounts likely depended on the frequency and duration of precipitation events and antecedent conditions. Many precipitation events were not large enough to produce recharge, which varied monthly from 0.14 in. in April to 15.95 in. in September and totaled 54.89 in. annually (85 percent of the calculated annual input) and showed no distinct seasonal pattern.

Discharge and Change in Storage

Groundwater discharged by two pathways: (1)  lateral flow that discharged into ditches and (2)  ET to the atmosphere. Monthly lateral flow decreased from 6.05 in. in January to only 0.85 in. in September as groundwater ET became appreciable and groundwater levels declined into the low-hydraulic-conductivity, low-yielding lower peat. Lateral flow to the ditches then increased to a monthly high of 7.34 in. in October after groundwater levels rose to near land surface at the end of September. Lateral groundwater flow to ditches averaged 3.60 in. monthly and totaled 43.21 in. annually, which was 79 percent of the groundwater recharge or 69 percent of the annual precipitation (table 10).

As indicated by PET, actual ET typically increased from the winter through the spring, peaked in June or July, and remained high through August before it decreased into the winter. Water from interception and rewetted unsaturated peat discharged as ET throughout the year. ET discharged moisture either directly from the peat into the atmosphere or from the peat through the vegetation to the atmosphere. During many periods in the growing season, groundwater became the major or sole water source for vegetation because discharge as ET depleted soil moisture after precipitation events. During the non-growing season, groundwater discharge as ET appeared to be minimal as indicated by the uniform, 24-hour decline in groundwater. The moisture content of unsaturated peat appeared to supply most of the need for vegetation ET during the non-growing.

Monthly groundwater ET increased from 0.0 in. in February to 5.69 in. in June. Contrary to what was typically expected based on average monthly PET, ET then decreased by about 50 percent to 2.88 in. in July, 2.13 in. in August, and 1.56 in. in September (table 10). Total annual ET was 28.34 in. (45 percent of the measured precipitation) with 9.47 in. from interception and rewetted, unsaturated peat (15 percent of precipitation) and 18.87 in. from groundwater. Total ET was about 87 percent of the 32.44 in/yr average PET for the Lake Kilby weather station. Groundwater ET was 44 percent of the groundwater discharged laterally toward ditches, 34 percent of the annual recharge, and 30 percent of the annual precipitation.

Total groundwater discharge by lateral discharge (43.21 in.) and groundwater ET (18.87 in,) was 62.08 in., exceeding groundwater recharge by 7.19 in. Discharge greater than annual recharge causes a decrease in groundwater storage that manifests itself as a decline in the groundwater levels over the year. The change in groundwater storage based on groundwater levels was 9.29 in., which is 2.10 in. or 29 percent more than the change in storage calculated from the difference between groundwater recharge and discharge. Subtracting the calculated annual change in groundwater storage from total discharge, the net 2010 discharge of precipitation recharged in 2010 was 62.27 in., 0.57 in. (0.9 percent) less than measured precipitation.

Monthly changes in the water budget illustrate the attributes of seasonal variations in budget components (table 10). Although precipitation and recharge do not follow a seasonal pattern, ET showed the expected increase into the growing season (April–October), which causes a distinct seasonal pattern in discharge. Increased ET causes groundwater levels to decline and reduces flow toward and discharge into the ditches. During the non-growing season (November–March), the primary discharge pathway was lateral flow to the ditches. During the growing season, discharge was a combination of lateral flow and ET (table 10). Discharge by lateral flow to ditches and as ET decreased during July and August because the water table dropped from the upper peat (depth above 1.5 ft) into the lower peat (depth greater than 2.0 ft). This decrease in ET indicates that the low hydraulic conductivity of the lower peat limits the unsaturated flow of water toward fine roots, thereby limiting water availability and accounting for the “dry” conditions identified by Phipps and others (1978) in their tree-ring analysis. This limitation on flow stresses the trees and potentially alters habitats. Consequently, groundwater ET is reduced as the water table approaches land surface in the upper peat because more roots are in saturated peat and as the water table drops into the lower peat where the low specific yield and hydraulic conductivity limit water availability.

When the water table declines into the lower peat, most of the lateral flow and discharge into ditches likely is through the sand. Although the volume and rate of groundwater discharge was less when the water table was in the lower peat than in the upper peat, the rate of water-table decline nearly doubled due to low specific yield. The combination of lower flow and discharge combined with the greater hydraulic gradient during low water levels (fig. 22) further supports the conclusion of the low hydraulic conductivity of the lower peat. The average water-table decline was 0.08 foot per day (ft/d) above a depth of 1.5 ft and increased to 0.16 ft/d below a depth of 2.0 ft. Despite doubling the rate of decline and tripling the horizontal hydraulic gradient, the daily discharge volume decreased by more than 75 percent from 0.18 inch per day (in/d) to 0.04 in/d.

Comparison of Input, Input Storage, and Discharge

Annual input storage (interception and rewetting unsaturated peat plus recharge) was 1.52 in. (2.4 percent) more than measured precipitation (table 10). Annual input storage minus discharge from annual recharge (discharge minus the change in storage) is only 2.10 in. or 3.4 percent of the measured precipitation. The close agreement of these values indicates that specific yield estimates likely are close to actual values and that appreciable other sources such as surface runoff, other flow paths, and other storage reservoirs are unlikely.

Relation Between pH and Specific Conductance

The relation between pH and specific conductance was derived from data from the synoptic field survey of March 15–23, 2016, and groundwater collected at Land Carbon sites March 29–31, 2016 (fig. 11). To better depict the effects of sources, mixing of the waters, and chemical evolution across the swamp, a symbol convention is used in fig. 40 to track water sources and the subsequent evolution in water chemistry. Samples from the same ditch are indicated by the same solid symbol and color. Different shaped solid symbols of the same color represent other ditches flowing from one into the other or located in the same vicinity. Open symbols of the same shape and color represent water in the adjacent swamp. An asterisk (*), plus (+), or “X” represent groundwater; symbols having the same color as a ditch represent a well near that ditch.

The pH of water derived from the Isle of Wight Plain ranged from about 5.5 to 6.4 standard units; the specific conductance ranged from about 70 to 150 microsiemens per centimeter at 25 degrees Celsius (μS/cm) (light-blue-shaded area, fig. 40). The pH of groundwater from sites where precipitation infiltrated into and flowed through peat was more acidic, ranging from about 3.1 to 4.0 standard units; specific conductance of those waters ranged from 42 to 188 μS/cm (brown-shaded area, fig. 40). The pH of groundwater recharged primarily through the sand aquifer from two wells near Big Entry Ditch (fig. 2, northeastern quadrant; fig. 11) was about 4.9 standard units because minimal peat is present (blue-shaded area, fig. 40). Most waters not within these shaded areas but within the shaded transition are mixtures from these sources and (or) were chemically altered as the water flowed through the peat or sand aquifers (fig. 40).

Like spatial changes in hydrologic responses, water quality evolved in ditches as the water flowed from the scarp, and the effects of the peat chemistry increased as the ditch contained more groundwater discharged from the peat aquifer. The quality evolved to a lower pH and an initial decline in specific conductance followed by an increase. The nature and degree of this evolution, like hydrologic response, partly depends on differences in the characteristics of hydrologic and chemical controls. Differences in evolution of the water quality in Washington, Lynn, and Corapeake Ditches (fig. 2) highlight some major effects.

Washington Ditch (West of Lake Drummond, fig. 2) is 1 of only 2 ditches receiving water discharging directly from streams that cross the scarp. The pH and specific conductance of water in Washington Ditch changed little on March 23, 2016, relative to the overall differences in water quality across the swamp, remaining similar in characteristics along the entire ditch (fig. 40). Data are from six sites along a 4-mi interval of the ditch from the inflow of Pocosin Swamp at the scarp to where the road became impassable near Lake Drummond. The pH ranged from 5.9 to 6.4 standard units; specific conductance ranged from 70 to 86 μS/cm (fig. 40). This quality was like that of Moss Creek, a stream flowing across the scarp and into the swamp to the south (fig. 40). The water quality of Washington Ditch changed little because the large inflows from Pocosin Swamp and Cypress Swamp appeared to be much greater than groundwater discharge from the peat and, consequently, dominated the water quality.

The quality of water from Lynn Ditch on the same date reflected the evolution of the chemistry of groundwater discharged across the scarp that mixed with precipitation recharged through the peat (fig. 40). Lynn Ditch is 1 of the 4 ditches close to and parallel to the scarp and is nearly perpendicular to Washington Ditch about 1.9 mi from the scarp. A water-control structure under Washington Ditch road controls flow from Lynn Ditch into Washington Ditch. The quality of water at seven sites along the length of Lynn Ditch differed little from each other but differed substantially from that of Washington Ditch. Its chemistry was characteristic of that affected by peat; pH ranged from 3.7 to 3.9 standard units, and specific conductance ranged from 82 to 106 μS/cm (fig. 40). The quality differed little because Lynn Ditch is a similar distance from the scarp along its length, causing similar relative distributions in sources and chemical evolution of the water along its length.

Badger Ditch water originates as groundwater discharging from the base of Suffolk scarp and flows east past the approximate midpoint of Lynn Ditch. This flow mixes with groundwater recharged through the peat. In contrast to the quality of Lynn and Washington Ditches, the pH was 4.6 standard units and specific conductance was 58 μS/cm in Badger Ditch near Lynn Ditch (fig. 40). This quality was within the relation for the mixing of water from the scarp (Washington Ditch) with that dominated by the peat (Lynn Ditch), indicating a greater effect of chemical evolution from the peat than in Washington Ditch but less than in Lynn Ditch (fig. 40).

Flow in Corapeake Ditch (South of Lake Drummond, fig. 2) also originated as groundwater discharging from the base of Suffolk scarp. The quality of water at its head of flow on March 16, 2016, differed from that of Washington Ditch and evolved substantially downstream. Two sites at the head of flow at Corapeake Ditch had a similar quality to each other but appear as outliers from the pH/specific conductance relation (fig. 40). One site was standing water in the swamp across the road from the ditch; the other was directly in the ditch. Water at both sites likely was derived from the combination of groundwater discharging from the scarp as indicated by flow-model results of Eggleston and others (2018) (fig. 24) and recharge within the swamp. Unlike Washington Ditch, the head of flow in Corapeake Ditch had no direct inflow from a stream crossing the scarp. Consequently, a difference in quality of at least one source, likely groundwater flow across the scarp, or a different, likely short-lived evolution in the chemistry of the water from the two sources might have caused these outliers. The quality of the next downstream point on Corapeake Ditch is within the relation, like that of Washington Ditch. The quality continued to evolve through the mixing-zone relation toward the quality of the peat groundwater having a decreasing pH and variable specific conductance (fig. 40).

The cause of other apparent outliers, except groundwater from well SEW01 near Portsmouth Ditch (fig. 2, northeastern quadrant; fig. 11), can be explained by the water sources and the likely subsequent chemistry (fig. 40). The quality of groundwater from well SEW01 is different from that of Portsmouth Ditch and other nearby wells and is most like that of the two samples from the source of flow in Corapeake Ditch. This well, however, is near the mid-point of the east side of Portsmouth Ditch within the northeastern quadrant of the swamp. It is several miles from the scarp with intervening ditches so that flow across the scarp is not a likely water source to this well. No explanation of the quality of water from this well is readily evident, based on known conditions in this area.

The outlier from Interior Ditch is from the upstream end of the ditch adjacent to, but not directly connected to, West Ditch where groundwater from the scarp discharges into the ditches (fig. 24). Part of the groundwater likely flows through the peat before discharging to the ditch. The next downstream value is within the mixing relation.

Additional outliers, although outlying only a little, are from Big Entry and Rosemary Ditches in the northeastern quadrant of the swamp (fig. 2; fig. 40). The water source to Big Entry Ditch is groundwater discharging from the swamp along the entire length of the ditch, except at its intersection with Portsmouth Ditch where flows from the two ditches converge, mix, separate, and flow into both ditches. Groundwater discharge along Big Entry Ditch is a combination of peat- and mineral-sediment-derived groundwater as indicated by the thickness of the upper peat and thickness of the full peat (fig. 13; fig. 14). Near the head of flow, water has the lowest specific conductance, about 40 μS/cm, and appears to be appreciable mineral-sediment-derived groundwater. The two outliers from Big Entry Ditch were for samples collected immediately downstream from Portsmouth Ditch and farthest downstream, immediately above the confluence with Rosemary Ditch. The Rosemary Ditch outlier is for a sample collected immediately above the confluence with Big Entry Ditch.

Another aspect of the effects of flow on the pH/specific conductance relation is of note for two of the Land Carbon wells: Maple 2 and Maple 3. These two wells were the only Land Carbon wells to consistently have water samples with a pH greater than 4.0 standard units. The pH at first was thought to be caused by the forest-type land cover because these were 2 of the 3 maple forest sites. As the hydrology of the swamp became better understood, these pH values were considered to result possibly from water sources and flow paths, not forest type. This theory was a large part of the reason for conducting the synoptic survey. The relation (fig. 40) indicates that the quality of water from these wells is in the transition of water discharging across the scarp and groundwater derived from the peat, which is supported by model-simulated flow during average springtime conditions (Eggleston and others, 2018). Maple 3 is east of West Ditch where part of the groundwater flowing across the scarp passes under West Ditch as it flows to Lateral and Railroad Ditches (fig. 24). Maple 2 is between Weyerhaeuser and County Line Ditches (fig. 2) in a small area near their confluence with Cross Canal Ditch where water flows from Weyerhaeuser Ditch through the swamp to County Line Ditch (fig. 21). The pH of the water in Weyerhaeuser Ditch was greater than 4.0 standard units, like that of Maple 2. Like Lynn Ditch, Weyerhaeuser Ditch is 1 of the 4 ditches parallel to the scarp and receives groundwater discharge from across the scarp. Unlike Lynn Ditch, Weyerhaeuser Ditch indirectly receives water from streams that flow across the scarp and discharge diffusely across the surface of the swamp, likely accounting for the difference in the quality.

Dissolved Organic Carbon, Nutrients, and Major Ions

Decomposition of organic carbon in the peat produced dissolved organic carbon (DOC) in the groundwater that created a different quality than that typically encountered. At the nine Land Carbon sites, DOC concentrations ranged from 55 to 195 mg/L with a median concentration of 121 mg/L and an average concentration of 120 mg/L (fig. 41; table 11). Numerous other studies included DOC concentration for various periods. Maximum DOC concentrations in groundwater of only 2 of the 11 countries worldwide cited in Mostofa and others (2013) exceeded this median of 121 mg/L (183 mg/L in Botswana and 170 mg/L in Germany). DOC concentrations in groundwater of the Great Dismal Swamp were substantially higher than concentrations in swamps elsewhere in the eastern United States. In five swamps from eastern North Carolina, for example, DOC concentrations of samples ranged from 10 to 20 mg/L and averaged 15 mg/L (Mulholland and Kuenzler, 1979). The Chickahominy River is a black-water stream (has high DOC concentrations) that flows through a bottomland hardwood forest in the Coastal Plain of Virginia having DOC concentrations that ranged from 1 to 20 mg/L (Speiran, 2000). In the Okefenokee Swamp and Big Cypress Swamp to the south, DOC concentrations averaged 30–40 mg/L (Thurman, 1985). DOC concentrations ranged from 20 to 400 mg/L (average 55 mg/L) in the Great Heath, a sphagnum bog in Maine (Thurman, 1985). Although the maximum DOC concentration in the Great Heath exceeded the maximum measured in the Great Dismal Swamp, the average equaled the minimum measured in the Great Dismal Swamp. The cited references were published from 7 to more than 40 years before this report so that reported analyses likely cover a longer time period. Possible temporal effects on concentrations of DOC are uncertain and likely are different in the different systems.

Peat decomposition also releases various nutrients, making them available to aquatic systems. Effects of decomposition, as reflected in dissolved nitrogen, result partly in ammonia but mostly in dissolved organic nitrogen (table 11). Ammonia concentrations ranged from 0.05 to 1.79 mg/L as nitrogen (N), whereas concentrations of dissolved organic nitrogen ranged from 1.3 to 5.7 mg/L as N. Although the dissolved organic nitrogen is incorporated in the DOC, the relation between DOC and dissolved organic nitrogen (not depicted) was poor, having a correlation coefficient (R²) of 0.09. The nitrate plus nitrite concentrations were low, having a median concentration of less than the 0.02 mg/L as N minimum reporting level and the maximum concentration of 0.28 mg/L as N. The 95th percentile of the concentration was 0.009 mg/L as N (table 11). The low nitrate plus nitrite and low to moderately high ammonia concentrations are typical of systems having a high organic content where low dissolved oxygen concentrations limit the oxidation of ammonia to nitrate after the nitrogen is mineralized and released from the carbon. Concentrations of dissolved (water filtered) orthophosphate and dissolved phosphorus were less than the minimum reporting level in 67 and 59 of the 91 samples, respectively. The minimum reporting level for the orthophosphate was 0.004 mg/L as phosphorus (P). The minimum reporting level for the dissolved phosphorus varied, ranging from 0.008 to 0.184 mg/L as P. Maximum concentrations were 0.076 and 0.098 for orthophosphate and dissolved phosphorus, respectively.

The pH of groundwater at the Land Carbon sites ranged from 3.3 to 5.1 standard units with a median of 3.6 standard units (table 11). The low pH of the water in the Great Dismal Swamp (fig. 40; table 11), results from the humic and fulvic organic acids that partly compose the DOC, peat, and other organic material (Thurman, 1985). When the pH is less than 4.5 standard units, as it was in about 95 percent of the samples (table 11), the organic acids rather than carbonate and bicarbonate buffer the waters against changes in pH. Organic acids also can provide part of the buffering at pH greater than 4.5 standard units in these systems.

Ionization of the organic acids and other changes in chemistry create negative charges on the DOC. Negative charges on the DOC help the groundwater to conduct an electrical current, potentially contributing to a moderate relation between specific conductance and DOC (R² of 0.49, fig. 41). The contribution of the DOC to specific conductance appears to be disproportional to the anionic composition required to balance the cation concentrations. As part of a check on water-quality analyses that is commonly used, the USGS NWQL issues alerts for results having 100 times the milliequivalence of cations divided by the specific conductance not within a targeted range of 0.92 to 1.24. The ratios were much lower than that range for all samples from all Land Carbon sites, except most samples from Maple 3 (fig. 42). Although the ratios for samples from Maple 2 are outside of the target range (about 0.75), they are closer to this range than remaining sites. The ratios for all other sites are less than 0.5 with those of many samples less than 0.25. The Maple 2 and Maple 3 sites are the only Land Carbon sites identified as mixtures that include groundwater discharging through the sand aquifer from the Isle of Wight Plain and across the scarp, further demonstrating the role of the peat and the DOC in affecting the groundwater chemistry in the swamp and the differences in the chemistry of the different water sources.

The negative charge of the DOC is a part of the anionic composition of the groundwater that balances the cationic composition. A comparison of anionic and cationic charges, termed an ion balance, typically is used to identify possible analytical errors or to determine whether all ions are accounted for in the analysis of water samples. Concentrations of major cations (calcium, magnesium, sodium, potassium, iron, and manganese) and major anions (chloride, sulfate, and fluoride) typically were less than 10 mg/L in most samples from Land Carbon sites (table 11). Because the anionic content of the DOC cannot be quantified, it is reflected at least as part of the cation excess in the ion balance.

Another factor affecting the ion balance in water having high DOC is the exchange of cations onto the DOC. When the cations are present on the DOC in this form, they are not dissolved in water as ions but will be analyzed as if they are because the ions pass through the filter on the DOC and are measured in the analysis. Because these ions are reported as dissolved and because the associated anionic characteristics of the DOC cannot be quantified, the ion balance displays an excess of cations (Thurman, 1985). For groundwater from the Land Carbon sites, the excess of cations ranged from about 10 to 70 percent (fig. 43). Although the Maple 2 and Maple 3 sites were closest to the range for the specific conductance/cation relation (fig. 42), these two sites, particularly Maple 3, tend to have the greatest percent difference between anion and cation concentrations (fig. 43), reflecting a likely high anionic charge on the DOC.