An observational study was conducted to identify mechanisms of suspended sediment flux and turbidity maintenance in the Delaware River estuary. From March through October 2005, instrumented moorings were deployed to obtain continuous measurements of currents and suspended sediment concentration at sites along the estuarine channel and on flanking subtidal flats. Data time series were analyzed to determine the relative influence of nontidal advection and tidal pumping on residual fluxes of sediment. Results indicate that the estuarine channel is a strongly advective transport environment with residual sediment fluxes driven mostly by gravitational circulation. Tidal pumping is a contributing process of residual sediment flux in the channel near the estuarine null point and turbidity maximum, though the magnitude and direction of pumping vary with river flow and resident sediment inventory in the upper estuary. Sediment pumping in the channel is driven by tidal asymmetries in velocity and particle settling and perhaps by tidal variations in internal mixing in the stratified lower estuary. In contrast to the estuarine channel, residual sediment fluxes over the subtidal flats are weak and dominated by tidal pumping. Landward advective fluxes of sediment in bottom waters of the lower estuarine channel are strongest during neap tides; during large spring tides sediment is mixed high in the water column and the advective flux reverses to seaward under the residual surface outflow. Despite these transient seaward fluxes, the estuary has an enormous capacity to buffer extreme freshwater discharges and suppress export of suspended sediment to Delaware Bay.
 Understanding the movement of suspended matter from sources to sinks in estuaries is a standing priority in coastal ocean research with broad implications to water quality, biogeochemistry, and morphodynamics. In particular, the origin of estuarine turbidity maxima has been an enduring research theme following the pioneering works of Postma  and Schubel . An estuarine turbidity maximum (ETM) is a feature of elevated suspended sediment concentration created by a particular combination of tidal and nontidal currents together with continual cycles of particle settling, deposition, and resuspension. A large quantity of suspended matter can become entrapped and temporarily stored within the ETM, perhaps as much as a year's worth of river input; thus, the dynamics of turbidity maxima are fundamental to the sediment mass balance of an estuary.
 Estuarine processes research conducted over the past several decades has challenged the traditional notion that turbidity maxima arise solely from convergent gravitational circulation characterized by seaward barotropic flow and landward baroclinic flow near the landward limit of salt penetration [e.g., Schubel, 1968; Festa and Hansen, 1978]. It is now well-established that turbidity maxima can exist in estuaries even when there is no tidally averaged (residual) movement of water induced by gravitational circulation, and there is broad agreement that tidal processes can be on par with gravitational circulation in maintaining ETM in stratified estuaries [Uncles et al., 1985; Wolanski et al., 1995; Burchard and Baumert, 1998; Brenon and Le Hir, 1999; Lin and Kuo, 2003; Park et al., 2008]. Phase differences between water velocity and suspended sediment concentration produce tide-induced residual fluxes of sediment (tidal pumping fluxes) that contribute to the formation of turbidity maxima in some estuaries [Allen et al., 1980]. These phase differences arise from flood-ebb tidal velocity asymmetry in response to tide-topography interactions, in conjunction with tidal variations in sediment resuspension and settling rates [Dronkers, 1986; Dyer, 1995].
 Independent of tide topography effects, flood-ebb differences in density stratification and vertical mixing have been shown to generate tidal pumping fluxes of significance in turbidity maxima formation [Jay and Musiak, 1994; Geyer et al., 2001; Sanford et al., 2001; Scully and Friedrichs, 2003, 2007]. In a tidally strained density field, tidal pumping fluxes of dissolved and suspended matter arise from flood-ebb differences in vertical mixing (eddy diffusivity) and velocity shear [Stacey et al., 2001; Scully and Friedrichs, 2003]. During the ebb tide, vertical mixing is suppressed by density stratification and suspended matter is concentrated near the bed. However, during the flood tide, higher salinity water is transported over fresher water and the increased turbulent stresses induce vertical mixing of fine sediment high within the water column. On the tidal average this form of internal tidal asymmetry produces an up-estuary sediment flux. Because tidal pumping fluxes of sediment are superimposed advective fluxes driven by the residual gravitational circulation [Scully and Friedrichs, 2007], elucidating the mechanisms of turbidity maxima formation is fraught with observational challenges.
 Estuaries of the U.S. Mid-Atlantic region have long been recognized as traps for fine-grained river sediments [Meade, 1969, 1982], but the underlying processes of entrapment have been identified in only a handful of well-studied systems [Geyer et al., 2001; Lin and Kuo, 2001; Sanford et al., 2001; Chant and Stoner, 2001; Scully and Friedrichs, 2007]. The available observations make clear that classical gravitational circulation is not the sole mechanism of sediment entrapment as originally suggested by Meade  and that sediment pumping related to tidal asymmetries in velocity and (or) turbulent mixing are important to consider in turbidity formation and maintenance. In this paper we report results of an observational study of suspended sediment flux in the Delaware River and Bay estuary (Delaware Estuary), the second largest estuarine basin on the U.S. Atlantic coast. The existence of a broad ETM zone in this estuary has been known for decades [Biggs et al., 1983], yet despite having a demonstrated influence on primary production, nutrient distributions, and biogeochemical cycling [Sharp et al., 2009], the nature of this feature has eluded investigation until the present study. Here we show that the Delaware ETM is maintained by a combination of gravitational circulation and tidal pumping and that the magnitude and direction of residual sediment fluxes are time dependent and spatially variable. Significantly, an extreme river discharge event occurred during the observational period in 2005 and provided a rare opportunity to examine the influence of external hydrologic forcing on estuarine sediment transport.
2. The Delaware Estuary
 The 215 km long Delaware Estuary can be subdivided into a tidal freshwater reach (Trenton to Philadelphia), a well-mixed, oligohaline upper estuary (Philadelphia to Artificial Island), a partially stratified, mesohaline lower estuary (Artificial Island to Bombay Hook), and a well-mixed, polyhaline bay (Figure 1). On the basis of U.S. Geological Survey streamflow data for the period 1898–2007, the main stem Delaware River, the Schuylkill River, and the Brandywine-Christina River have mean annual discharges of 330, 77, and 19 m3/s, respectively, and together supply ∼80% of the total freshwater inflow. These tributaries contribute over 80% of the total annual suspended sediment load delivered to the estuary, which has been estimated at 1–2 × 109 kg/yr [Mansue and Commings, 1974]. Bottom sediments of the tidal river and upper estuary consist of mostly sand and gravel eroded from relict channel strata admixed with modern river mud [Biggs and Beasley, 1988]. The upper and lower estuary floor consists of interbedded silts and clays, whereas the bay floor is mantled by sand eroded from the bay shoreline and Atlantic coast. The ETM zone typically extends from the mouth of the Christina River to just seaward of Artificial Island (Figure 1).
 Prior research in the estuary includes studies of tidal and subtidal circulation [Pape and Garvine, 1982; Parker, 1991; Wong and Sommerfield, 2009], local and remote wind forcing [Wong and Garvine, 1984; Wong, 1991; Janzen and Wong, 2002], axial salinity variation [Garvine et al., 1992; Wong, 1995], and the lateral structure of salinity and currents [Wong, 1994; Wong and Moses-Hall, 1998]. On the basis of axial surveys of salinity and suspended sediment concentration [Biggs et al., 1983; Sharp et al., 2009], it has long been speculated that the estuarine null point, where the tidally averaged bottom current is zero, falls roughly 100 km landward of the Delaware Bay mouth in the vicinity of the C&D Canal (see Figure 1 for location). However, prior to the present study, the location and variability of the null point was never determined directly through flow measurements. Suspended sediment transported in the tidal Delaware River is delivered to the ETM zone by the river-induced mean current together with a tidally rectified current in compensation for Stokes Drift [Cook et al., 2007; Wong and Sommerfield, 2009]. As the width of the estuary increases toward Delaware Bay, lateral circulations become increasingly important in the axial dispersion of salt [Wong, 1994; Wong and Moses-Hall, 1998]. Although lateral transport processes are bound to play a role in moderating turbidity in the estuary, the present study was designed specifically to identify mechanisms of along-estuary sediment flux.
3.1. Instrumentation and Measurements
 During the spring and summer of 2005, instrumented moorings were deployed along three cross-estuary transects designated New Castle (NC1), Blackbird Creek (BC1−3), and Bombay Hook (BH1−5) (Figure 1). Station locations were selected to bracket the ETM zone as mapped during prior shipboard surveys (reported by Cook et al. ). Two mooring deployments extending from 8 March to 2 June (yearday 67–153) and from 2 June to 27 October (yearday 153–300) were completed. During the second deployment, mooring disturbances by ship traffic compromised data recorded at Station BC3, and at Stations BH1 and BH5 biological fouling reduced the data quality. For these reasons, data collected only during the first deployment at these stations are reported herein. RD Instruments 1200 kHz (NC1) and 600 kHz (BC2 and BH3) acoustic Doppler current profilers (ADCP) mounted on bottom frames were deployed at three channel stations along the estuary in 10−12 m water depths. These moorings were sited just adjacent to the channel axis to avoid the heavily trafficked midchannel shipping lane. The lowermost ADCP bin at the channel mooring sites fell 1−1.5 m above the seabed, but no attempt was made to extrapolate the velocity and SSC profiles to the bottom. We estimate that loss of the lowermost profile underestimated the depth-averaged current and sediment flux by <5% of the reported values.
 At Stations BH2 and BH3 the frames were outfitted with a SBE-37 conductivity and temperature (CT) sensor and an OBS-3 optical backscatter sensor mounted 0.5 m above the bottom (mab). At all of the midchannel stations SBE-37 and OBS-3 sensors were mounted on a mooring cable approximately 5−6 mab. On transects BC and BH, InterOcean S4 current meters with integrated CT sensors were deployed on the channel flanks to obtain additional information on flow and salinity, but the data are not discussed in this paper. Instrument packages consisting of a Sontek Ocean Probe acoustic Doppler velocimeter (ADV) with integrated OBS-3 turbidity and SBE-37 CT sensors were deployed at the both ends of transect BH on subtidal flats (2−3 m water depths). The ADV probe and OBS-3 and SBE-37 sensors were mounted on a frame and positioned at 1 mab. The velocity and OBS-3 turbidity sensors were similarly programmed to sample in 2-minute bursts at sampling rates of 1−6 Hz every 15 min, whereas the SBE-37 CT sensors sampled in 3 s bursts every 5 min.
 To generate profiles of suspended sediment concentration (SSC), ADCP echo intensity was calibrated against optical backscatter data recorded by the moored OBS-3 sensor, which was oriented to sample the water mass insonified by the ADCP beam. The calibration was performed in two steps. First, OBS-3 voltages were calibrated against gravimetrically determined SSC using water samples collected and filtered during two hydrographic surveys of the study area in 2005. In the laboratory, OBS-3 sensors were immersed in water baths of known SSC determined gravimetrically, the baths gently stirred to maintain a suspension, and a 2 min average OBS-3 voltage was recorded. This procedure was repeated for a suite of SSC standards over a 10−200 mg/L concentration range. The voltages were linearly regressed against gravimetric SSC to generate a calibration curve for each of the OBS-3 units used in the study. OBS-3 voltage was strongly correlated with gravimetric SSC (r2 = 0.97−0.99) and exhibited a linear relationship over the full range of concentrations.
 Second, ADCP echo intensity was converted to SSC following a procedure adapted from Holdaway et al.  and Gartner . The sonar equation was used to compute SSC from acoustic backscatter as
where SSC is the mass concentration of suspended solids, RB is the relative backscatter, and A and B are constants determined by regression of ADCP relative backscatter against SSC as determined by optical backscatter (SSCobs). ADCP relative backscatter was computed as
where Kc is the signal strength scale factor used to convert backscatter counts to dB (0.45 for our ADCPs), E is the received ADCP echo intensity, Er is the background echo intensity (46−47 counts), R is the slant range of the insonified volume (m), and α is the water absorption coefficient (dB/m). The second and third terms on the right-hand side of equation (2) comprise the two-way transmission loss of the acoustic signal. The slant range was determined from the ADCP bin depth and beam angle relative to vertical, and α was computed following Schulkin and Marsh . Although we included a correction for the near-field nonspherical spreading, it did not significantly improve the correlation between SSCobs and ADCP echo intensity.
 Correlations between relative backscatter and SSCobs were not as strong as those between optical backscatter and gravimetric SSC, perhaps because of the different ways in which acoustical and optical sensors sense suspended particles [Holdaway et al., 1999]. Even so, good correlations (r2 = 0.75−0.81) were obtained for the ADCPs used in this study, and the resulting regression equations for relative backscatter versus SSCobs and SSCobs versus gravimetric SSC were used to compute time series of SSC for each midchannel station. A more direct calibration for SSC would have involved filtering water samples collected directly from the water volume insonified by the ADCP for regression against the corresponding acoustic backscatter intensity [e.g., Fugate and Friedrichs, 2002].
 Time series of SSC where also derived from point velocity ADV data obtained at Stations BH1 and BH5. The ADV backscatter signal was used as surrogate for optical backscatter because the OBS-3 sensors had a tendency to foul rapidly by biological growth. Backscatter intensity was computed from the ADV signal amplitude and calibrated against gravimetrically determined SSC using filtered water samples collected periodically at these stations [Fugate and Friedrichs, 2002; Ha et al., 2009]. Backscatter intensity was well correlated with gravimetric SSC (r2 = 0.90−0.95) on a log-log plot and exhibited a linear relationship over the full range of measured concentrations.
3.2. Sediment Flux Decomposition
 Time series of current velocity and derived SSC were used to compute continuous records of suspended sediment flux at the observational sites. For the midchannel ADCP stations (NC1, BC2, and BH3), sediment fluxes were computed as the product of velocity and SSC for each 50 cm ADCP bin and then integrated over the mean depth of the water column to determine total sediment flux per unit width of flow (mass/length/time). For the subtidal flat sites (BH1 and BH5), the total sediment flux at 1 mab was computed from time series of ADV point velocity and SSC (in mass/area/time). The ADCP and ADV current data were rotated to the dominant axis of tidal flow prior to filtering; thus, the computed sediment fluxes approximate the streamwise flux.
 The instantaneous sediment flux was computed as
where U and SSC are instantaneous values of along-channel velocity and sediment concentration, respectively, and z is depth. The instantaneous velocity is the sum of the tidally varying velocity (U′) and the tidally averaged velocity (), and the instantaneous SSC has a similar composition. Integrating equation (3) over z and averaging over the tidal cycle gives the total residual sediment flux per unit width of flow (FT). Tidal averaging was accomplished by filtering the time series using a 36 h Lanczos low-pass filter; FT was computed as the low-pass filtered product of U and SSC.
 To identify tidal and nontidal mechanisms of residual sediment flux, FT was decomposed into advective (FA) and tidal pumping (FP) components. The advective sediment flux is driven by the tidally averaged (residual) velocity and the tidally averaged SSC and is given by
where the overbars denote tidally averaged (low-pass filtered) values; FA was computed as the product of low-pass filtered U and low-pass filtered SSC. The main source of the advective residual current and sediment flux in the estuary is gravitational circulation, i.e., seaward barotropic flow and landward baroclinic flow. The Stokes Drift component of the residual current was determined to be on the order of 2–3 cm/s at Stations BH3 and BC3 and 3−6 cm/s at NC1. This component of the residual current was not removed from the velocity time series; thus, the sediment flux on the Stokes Drift compensation flow is contained within the record of FA.
 The tidal pumping flux is driven by correlated fluctuations in tidally varying SSC and U and is given by
where the primes denote tidal fluctuations around the tidally averaged values. Tidal values were computed by high-pass filtering timeseries of U and SSC, and FP was computed as the low-pass filtered product of U′ and SSC′. Here it is important to point out that sediment pumping fluxes related to tidal velocity asymmetry and tidal mixing asymmetry are both contained within the record of Fp. In this paper positive and negative values denote seaward (down-estuary) and landward (up-estuary) transports, respectively.
4. Results and Interpretation
4.1. River Flow and Sediment Load
 On yearday 94 (4 April) 2005, the instantaneous discharge of the Delaware River at Trenton (USGS 1463500) reached 6776 m3/s, the third largest peakflow recorded at the gauging station since 1898 (Figure 2a). By comparison, spring season peakflows of the Delaware at Trenton are typically 2500 m3/s (1 year recurrence interval). Discharge of the Schuylkill River at Philadelphia (USGS 1473800) and the Brandywine-Christina river at Wilmington (USGS 1481500) also peaked on yearday 94 for cumulative freshwater inflow of 7250 m3/s (Figure 2a). River discharge remained elevated for ∼20 days before decreasing to flows more typical for the spring season (500−600 m3/s). The flood had an immediate impact on salinity throughout the estuary; surface salinity in the lower estuary decreased by 10−15 PSU for a period of ∼20 days (Figure 2b). The flood-produced salinity excursion (∼50 km seaward) exceeded the M2 tidal excursion in the lower estuary (∼9 km) and was considerably larger than excursions reported for the ordinary range of freshwater discharge [Garvine et al., 1992].
 Suspended sediment loads for the three main tributaries were estimated using time duration rating curves developed from sediment concentration and water discharge data archived by the U.S. Geological Survey (http://co.water.usgs.gov/sediment/). The gauging stations are located above the head of tides, so the estimated sediment loads represent net transport from the watershed to the tidal freshwater reach of the estuary. The estimated three-tributary sediment load averaged over the study period (yeardays 67−300) was 3.4 × 109 kg. By comparison, 2.6 × 109 kg was delivered during the period of river flooding from yeardays 88 to 107, approximately 60% of which was supplied by the Delaware River alone. Impressively, this 20 day sediment load exceeds the long-term mean annual supply of suspended sediment delivered to the estuary (1−2 × 109 kg/yr).
4.2. Profiles of Mean Velocity and SSC
 Along-estuary variations in residual velocity and mean SSC are captured by vertical profiles averaged over the first and second mooring deployment periods. Averaged over the first deployment, the null point fell near Station BC2 and the null level extended seaward to middepth at BH3 (Figure 3a). Velocities at NC1 in the upper estuary were seaward throughout the water column (10−20 cm/s), whereas two-layer flow was present at BH3 in the lower estuary. Maximal seaward and landward velocities at BH3 were on the order of 10−15 cm/s. Profiles of mean SSC exhibited the characteristic increase with depth in the water column; values ranged from 8 to 20 mg/L at the surface and 40 to 100 mg/L at 0.5 mab (Figure 3b). As expected, the highest values of mean SSC were observed at Station BC2 near the estuarine null point. At Station NC1, landward of the null point, SSC values were only marginally higher than those measured at Station BH3.
 Effects of the 2005 Delaware River flood on the measured properties are captured by vertical profiles averaged over the peak freshwater flow between yeardays 88 and 107 (9 March to 17 April). At Stations NC1 and BC2, the residual velocity was higher than that averaged over the first deployment period by nearly a factor of two, and the surface velocity at BH3 was also higher (Figure 3c). Interestingly, the landward velocity of bottom water at BH3 was largely unchanged during the river flood, and the null level increased by only one unit of relative depth (z/h). Near-bottom SSC at Station BC2 increased during the period of high river flow, but SSC profiles at NC1 and BH3 were largely unchanged from the first deployment average (Figure 3d).
 Averaged over the second mooring deployment period, residual velocity and mean SSC throughout the water column were lower than during the first deployment. At Station NC1, surface velocities of 10−12 cm/s decreased to zero near the bottom, indicating that the null point migrated landward from its position earlier in the year. Two-layer residual flow was again present at BH3, and the structure of the velocity profile was nearly identical to that observed for the first deployment. At NC1 and BH3, mean SSC ranged from 4 mg/L at the surface to 35 mg/L at the bottom.
4.3. Time Series of Tidal and Residual Properties
4.3.1. Channel Axis
 Time series of depth-averaged currents, SSC, and depth-integrated sediment flux for the axial channel stations shed light on mechanisms and pathways of sediment transport in the estuary. During the period of high river discharge the residual depth-averaged velocity at Station NC1 was strongly down-estuary (Figure 4a). There was no apparent increase in depth-averaged SSC (Figure 4b), but the total sediment flux down-estuary nonetheless increased and remained elevated for ∼40 days following peak river flow (Figures 4c). For the remainder of the study period the sediment flux was weakly but consistently down-estuary. Decomposition of the total sediment flux time series reveals that nontidal advection was the primary mechanism of down-estuary transport at NC1 (Figure 4d). Tidal pumping and advective sediment fluxes were comparable in magnitude, but whereas the advective flux was directed mostly down-estuary under the influence of river discharge, the pumping flux changed from down-estuary during the first mooring deployment to up-estuary later in the year. This change in pumping flux orientation was most likely related to an increase in SSC in the NC1 area associated with landward migration of the estuarine null zone and turbidity maximum. The cumulative sediment flux averaged over the 20 day river flood period was over an order of magnitude larger than that produced during the ensuing 150 days (Figure 4e). During the second mooring deployment, the down-estuary advective flux of sediment was only slightly larger than the up-estuary tidal pumping flux such that the cumulative flux increased only incrementally.
 Effects of the river flood on velocity, SSC, and sediment flux were equally apparent at Station BC2. Four days after river discharge peaked at the river gauging stations, the residual depth-averaged velocity increased to a maximum of ∼25 cm/s at BC2 (Figure 5a). Tidal and residual values of SSC were generally 2−3 times higher at BC2 than at the other channel stations and increased during the river flood (Figure 5b). The residual sediment flux at BC2 was negligible prior to the flood, but it rapidly increased and peaked at 0.16 kg/m/s, the largest flux measured among channel stations during the period of study (Figure 5c). As was observed at Station NC1, decomposition of the total sediment flux time series at BC2 revealed that the orientation of the tidal pumping flux changed from up-estuary to down-estuary in association with the river flood (Figure 5d) and that the magnitude of the pumping flux was comparable to that of the advective sediment flux. These data suggest that down-estuary advection of suspended load is nearly matched by up-estuary sediment pumping near the estuarine null point (near BC2) under ordinary conditions of river discharge. However, during times of peak river discharge, the orientation of sediment pumping temporarily reverses to down-estuary with increases in the advective sediment flux. This reversal implies a feedback between the gravitational circulation, sediment delivery to the ETM zone, and tidal pumping. The total cumulative sediment flux at BC2 was dominated by nontidal advection, which increased significantly during the river flood (Figure 5e). The tidal pumping component of the total sediment flux reversed from down-estuary to up-estuary as the estuarine null zone and ETM migrated landward with diminishing freshwater discharge.
 In the lower estuary at Station BH3, the response of mean velocity and SSC to the river flood was surprisingly weak as the sediment pulse observed at the landward stations was not observed. Depth-averaged tidal velocities were strong during the observational period, exceeding 1 m/s during spring tides (Figure 6a), but the residual velocity was weak (±10 cm/s). Values of SSC at BH3 were the lowest among the three channel stations, but the spring-neap variation in SSC was most conspicuous (Figure 6b). The dramatic increase in SSC observed during spring tides is presumably related to increased bed shear stress and sediment resuspension. The total sediment flux at Station BH3 was strongly up-estuary, and only during spring tides was the total flux oriented down-estuary (Figure 6c). Decomposition of the total flux time series indicates an advective origin for down-estuary fluxes, which appear as spikes in the sediment-flux record (Figure 6d). Tidal pumping varied in magnitude and direction over the observational period, but overall pumping contributed little to the cumulative sediment flux (Figure 6e). Rather, sediment advection by baroclinic flow appears to have been the principal mechanism of landward flux in the lower estuary (Figure 6e).
 Interplay between tidal currents, suspended sediment, and sediment flux in the relatively stratified lower estuary can be clarified from vertical distributions of properties during spring and neap tides. In terms of flux magnitude and direction, the main factors differentiating spring and neap tides include differences in tidal velocity (Figures 7a and 7e), tidal and residual SSC, and the vertical structure of residual velocity. During strong spring tides, sediment is mixed into high-velocity waters where it is available for streamwise transport down-estuary (Figure 7b). The water column is relatively well mixed on spring tides such that landward gravitational flow near the bottom is weak to nonexistent (Figure 7c). This combination of residual SSC and velocity produces to a seaward advective sediment flux throughout the water column (Figure 7d). At the same time, near-bottom suspended load is tidally pumped up-estuary because more sediment is available at the onset of the flood tide than before ebb due to asymmetries in tidal velocity. Because the advective flux in the upper water column exceeds the near-bottom tidal pumping flux, the total flux integrated over depth is oriented seaward. This scenario can explain the seaward advective pulses of sediment observed during strong spring tides (see Figure 6c). During weaker spring tides the total sediment flux reverses to landward, because seaward sediment transport in the upper water column is offset by a larger landward advective flux near the bottom driven by baroclinic flow.
 During neap tides the near-bottom velocity is relatively weak (Figure 7e) and less sediment is resuspended and available for streamwise transport (Figure 7f). Still, more sediment is available during flood than on ebb due to velocity asymmetry and settling lag (Figure 7f). Two-layer residual gravitational flow is vigorous on neap tides due to suppression of vertical mixing by density stratification (Figure 7g). The landward advective flux of sediment below the null level is larger than the seaward flux above, because more sediment is available near the bed (Figure 7h). In the present case the advective flux dominates the total flux of sediment in the water column because the tidal pumping flux is limited by sediment availability. In summary, neap tides reinforce the general tendency for suspended sediment to move landward from the lower estuary to the ETM zone, and only during large spring tides are conditions favorable for transport seaward toward Delaware Bay.
4.3.2. Subtidal Flats
 Sediment transport over subtidal flats in the lower estuary was considerably more influenced by tidal pumping than in the estuarine channel, where nontidal advection dominates the residual sediment flux. At Stations BH1 and BH5, the direction of dominant tidal flow was nearly parallel to the channel axis during the observational period. On spring tides, tidal current velocities at 1 mab reached 50 cm/s at both locations with no apparent flood-ebb asymmetry in tidal velocity (Figures 8a and 9a). Although the residual current over the flats reached 10 cm/s (seaward) during the period of river flooding, more typically it did not exceed 1−2 cm/s. At both stations this weak residual current reversed from seaward to landward from the first to second mooring deployments (spring season to summer).
 Sediment concentrations varied tidally on the order of 50−200 mg/L at Stations BH1 and BH5 and were maximal during spring tides (Figures 8b and 9b). Concentrations increased during early flood and ebb, after the critical velocity for resuspension was exceeded (∼25 cm/s locally), peaked around the time of peak flood and ebb velocity, and decreased during late flood and ebb as slackwater was approached. At both stations, SSC remained high during the time of peak flood and ebb velocity, suggesting that the flow did not become sediment-starved through streamwise transport of locally resuspended sediment.
 Time series of total sediment flux computed for Stations BH1 and BH5 are shown in Figures 8c and 9c, respectively. At both stations the total flux was oriented seaward during the spring season and strongly seaward during the April river event and the magnitude of the total flux varied on the spring-neap cycle with higher fluxes observed on spring tides. Tidal pumping and nontidal advection were of equal importance in controlling the magnitude and direction of the total flux (Figures 8d and 9d); in contrast to the estuarine channel, tidal pumping over the subtidal flats accounted for roughly half of the cumulative sediment flux measured during the observational period (Figures 8e and 9e). At both sites the contribution of tidal pumping to the total sediment flux increased somewhat from the first deployment to the second, whereas the advective flux was nearly constant. Interestingly, the orientation of tidal pumping reversed from down-estuary during the spring season to up-estuary in summer, a change that forced a reversal in the direction of total sediment flux (Figures 8e and 9e). These data suggest that tidal pumping and advection transport sediment down-estuary over the flats when tidal currents and SSC are elevated by high river discharge, counteracting the general tendency for sediment to move up-estuary under the influence of tidal velocity asymmetry.
4.4. Salt Intrusion, Null Point, and ETM Locus
 To examine relations between the residual circulation and suspended sediment along the axis of the estuary, time series of low-passed filtered bottom velocity (1 mab), salinity, and SSC for the three ADCP stations were interpolated and contoured (Figure 10). The interpolated isoline of zero velocity, taken here to represent the estuarine null point, was projected on the contoured salinity field to show its position relative to the salinity intrusion (1 psu isohaline). For context, records of river discharge and estimated sediment load as described in section 4.1 are shown in Figure 10 with the contoured salinity and SSC data.
 Predictably, the salinity intrusion and null point migrated on a seasonal basis with river discharge and fortnightly with the spring-neap cycle (Figure 10a). During the period of elevated freshwater inflow from yeardays 90 to 140, the salinity intrusion was pushed down-estuary to just seaward of Artificial Island, and after yearday 150 it migrated back up-estuary and remained in the vicinity of Station NC1. Similarly, the null point moved ∼50 km seaward in association with river discharge and remained in the lower estuary until river discharge decreased to ∼500 m3/s (Figure 10b). By yearday 175 the null point was reestablished at its usual location in the vicinity of C&D Canal, seaward of salinity intrusion, just landward of the 5 psu isohaline. Superimposed on this seasonal migration were 15−20 km excursions of the null point in association with spring and neap tides (Figure 10a). During spring tides the null point moved down-estuary to more saline waters, presumably as a consequence of increased turbulent mixing, decreased density stratification, and reduced landward baroclinic flow. When the estuary restratified during neap tides, the null point moved back up-estuary under the intensified baroclinic flow.
 The estuarine sedimentary response to the river discharge event in 2005 was rapid and widespread. As shown in Figures 10c and 10d, SSC peaked at the three measurement sites within 2−3 days of maximum sediment delivery from the watershed on yearday 94. Although the mooring array could not resolve spatial variations in SSC within the reach between moorings, the data suggest that the ETM did not migrate as far seaward as the null point. SSC at the seaward end of the array was high when the null point at its furthest point down-estuary, consistent with ETM migration, but at all times concentrations were highest in the middle part of the array (Figure 10c). Hence, the ETM zone appears to have broadened and increased in sediment inventory in response to the freshwater pulse and sediment influx. Spring-neap variability was superimposed on this seasonal trend as SSC throughout the ETM zone increased and decreased in accord with spring and neap tides, respectively (Figure 10c).
5.1. Nature of the Turbidity Maximum
 Despite the weakly stratified to well-mixed nature of the Delaware Estuary, the behavior of its turbidity maximum is broadly comparable to that observed in partially mixed estuaries such as upper Chesapeake Bay [Sanford et al., 2001], the York River estuary [Lin and Kuo, 2001], and the Tamar, Weser, and Humber estuaries in Europe [Grabemann et al., 1997; Uncles et al., 2006]. Our observations suggest that the residual gravitational circulation controls the along-estuary limits and sediment inventory of the Delaware Estuary ETM zone, whereas some combination of residual circulation and tidal pumping influences the locus and sediment concentration of the ETM core. If gravitational circulation were the only mechanism of turbidity maintenance in the estuary, the salt intrusion, null point, and the ETM core would fall in the same general vicinity [Festa and Hansen, 1978]. In reality, tidal pumping of salt causes the salt intrusion to fall up-estuary of the null point, and tidal pumping of suspended sediment produces high turbidities landward of the null point. In some estuaries the nose of the ETM has been observed to fall landward of the salt intrusion under certain conditions [Grabemann et al., 1997; Uncles et al., 2006], but more generally the high-concentration core falls 10 s of km seaward of the salt intrusion, as was observed in this study.
 The hydraulic geometry of the Delaware Estuary most likely plays a role in buffering the impact of river discharge on the magnitude and direction of residual sediment flux. During the river discharge event in 2005, the salt intrusion and null point were advected 10 s of kilometers down-estuary, yet not far to weaken gravitational circulation and landward sediment transport at the seawardmost mooring (see Figure 4b). Although barotropic flow induced by river discharge is important in the tidal Delaware River and upper estuary [Cook et al., 2007; Wong and Sommerfield, 2009], its magnitude decreases significantly as the basin widens seaward. For example, the peak freshwater inflow of 7250 m3/s measured in 2005 is an order of magnitude smaller than the 1.5 × 105 m3/s M2 tidal volume flux estimated for Delaware Bay [Münchow et al., 1992]. Hence, the size of the tidal basin seaward of Bombay Hook contributes to the entrapment of suspended sediment landward in the estuary, even under the most extreme conditions of freshwater inflow. In contrast, river estuaries such as the Seine [Avione, 1987], Gironde [Castaing and Allen, 1981], and Columbia [Fain et al., 2001] are capable of discharging freshwater volumes approaching or exceeding the tidal volume flux of the basin. In these estuaries, the ETM can be advected the estuary mouth, perhaps resulting in sediment export to coastal waters.
 It is instructive to compare measured sediment fluxes at the landward and seaward limits of the Delaware ETM zone for insight on sediment sources and sinks, bearing in mind that the unit-width fluxes reported in this paper are representative for the measurement sites only. The cumulative mass of sediment transported down-estuary past Station NC1 during the study period (yeardays 72–258) was 240 × 106 kg/m, whereas −38 × 106 kg/m was transported up-estuary past BH3 averaged over the same period (excluding the gap in coverage between yeardays 157 and 178). By comparison, the estimated sediment load delivered to the estuary during the same period was 3400 × 106 kg. Two important points can be made from these data. First, the difference in the magnitude and direction of unit-width sediment fluxes measured at the upper and lower stations of the mooring array provides evidence for a convergence of flux (deposition) within the intervening estuary. Second, apparently more sediment was transported from the tidal river to the estuary (past Station NC1) than was delivered past the river gauging stations, suggesting that much of the sediment delivered to the ETM zone must have originated from bed storage within the tidal freshwater river reach of the estuary. This is suggested by the near-instantaneous increase in SSC throughout the estuary following peak river discharge and sediment delivery from the watershed (see Figures 10c and 10d). Lacking an internal source of sediment, several days would have passed between the times of peak sediment load at the river gauging stations and maximum sediment flux at Station NC1.
 It is not unlikely that remobilized bed deposits contributed to the suspended sediment inventory of the estuary during the study period; indeed, prior work in the upper Delaware estuary [Cook et al., 2007] and Hudson River estuary [Wall et al., 2008] has shown that sediment reworked from storage can generate sediment fluxes that exceed the allochthonous supply from the watershed. The potential for sediment storage in the Delaware Estuary is high considering that the tidal freshwater segment is long, accounting for 20% of the total length of the system, and, it stands to reason that new sediment from the watershed will reside in the tidal river for some time before being dispersed to the ETM zone. Our general understanding of sediment routing from watershed sources to depositional sinks in tidal basins is poor, and research focusing on sediment storage within the tidal freshwater segment of river estuaries is sorely needed.
5.2. Mechanisms of Residual Sediment Flux and Trapping
 Residual gravitational circulation, tidal velocity asymmetry, and tidal mixing asymmetry are the most frequently cited mechanisms of turbidity maxima formation and maintenance in stratified estuaries [e.g., Burchard and Baumert, 1998]. Although tidal pumping is a contributing process of suspended sediment transport in the Delaware Estuary, the relative roles of velocity asymmetry related to tide-topography interaction, lags in particle resuspension and settling, and tidal mixing asymmetry are less certain. From the present data we cannot conclude that tidal mixing asymmetry is a central mechanism of landward sediment flux and turbidity maintenance in the estuary, as has been observed in other estuaries [Jay and Musiak, 1994; Sanford et al., 2001; Scully and Friedrichs, 2003, 2007]. This is illustrated by time series data obtained at Station BH3. Tidal straining of the density field causes the water column to stratify and destratify during ebb and flood tide, respectively, with maximal stratification occurring at slackwater after ebb (Figure 11a). Associated with straining is tidal asymmetry in velocity shear with increasing shear during the ebb tide (Figure 11b). Although the maximum surface velocity is strongly ebb-dominant, peak flood and ebb velocities are symmetrical near the bottom, where most of the suspended load resides, and the amount of sediment resuspended during flood and ebb is comparable (Figure 11c). However, the period of slackwater after low water is shorter than slackwater after high water; thus, sediment resuspended during ebb does not completely settle before being transported streamwise by the ensuing flood current (Figure 11c). During both flood and ebb tides, sediment concentrations generally decrease before the time of peak near-bottom flow, suggesting that ebb and flood tidal currents are similarly sediment starved. As shown in Figure 11d, the near-bottom sediment flux is symmetrical because the flood and ebb asymmetries in tidal velocity and SSC average-out. Consequently, sediment pumping associated with tidal straining and the advective component of the total residual sediment flux are both flood directed, but the advective flux driven by the residual baroclinic flow is larger (Figure 11e).
 Another possible explanation for the observed ebb−flood asymmetry is lateral variability, specifically, advection of a more turbid volume of water past the mooring during ebb but adjacent to the mooring during flood. This could explain the ebb-to-flood decrease in SSC and sediment flux at Station BH3. The cross-channel structure of the streamwise flow is no doubt more complex than can be resolved by a single mooring, particularly one located adjacent to the channel axis. Hence, tidal mixing asymmetry may be more important a mechanism of landward sediment transport than implied by the available data, but verifying this possibility will require additional work to characterize the tidal variability of stress and eddy viscosity in the estuary [e.g., Scully and Friedrichs, 2003]. Despite this uncertainty, our observations support modeling by Burchard and Baumert  who demonstrated that tidal mixing asymmetry is not requisite for turbidity maxima formation and maintenance in stratified estuaries.
 In closing, the findings of this study are broadly consistent with the notion originally put forth by Meade  that fine-grained river sediment trapped within larger estuaries of the U.S. Atlantic coastal plain is a consequence of gravitational circulation. At the same time, our results make clear that gravitational circulation is more specifically a process of landward sediment movement within the deeper areas of estuaries, seaward of the null point. This distinction is important to make given that shallow subtidal flats and shoals comprise a large fraction of the total area of coastal plain estuaries, and considering that sediment trapping in these shallow environments is tidally driven [Dronkers, 1986; Dyer, 1995], as was observed in this study. The strength of gravitational circulation is inversely proportional to the cube of the water depth [Hansen and Rattray, 1965]; thus, its significance as a process of landward sediment transport is a strong function of estuarine morphology. We contend that gravitational circulation in the axial channel of Delaware Estuary is a fundamental mechanism of sediment entrapment within the ETM zone, but also that tidal trapping over the shallow subtidal flats is involved in the permanent sequestration of sediment delivered to the estuary as whole. This implies lateral circulation and sediment transport between the estuarine channel and subtidal flats, the nature of which will require further research to characterize.
 We thank the crew of the RV Hugh Sharp, Art Sundberg, and David Huntley for field assistance, and Susanne Moskalski and Hua Yang for help with data processing. We also thank two anonymous reviewers for their insightful comments and suggestions to improve the paper. This research was supported by the Delaware Sea Grant Program at the University of Delaware and a grant from the Delaware River Basin Commission.