Contrasting Seasonal and Fortnightly Variations in the Circulation of a Seasonally Inverse Estuary, Elkhorn Slough, California

Abstract

Three and a half years of hydrographic, velocity, and meteorological observations are used to examine the dynamics of upper Elkhorn Slough, a seasonally inverse, shallow, mesotidal estuary in central California. The long-term observations revealed that residual circulation in Elkhorn Slough is seasonally variable, with classic estuarine circulation in the winter and inverse estuarine circulation in the summer. The strength of this exchange flow varied both within years and between years, driven by the annual cycle of dry summers and wet winters. Subtidal circulation is a combination of both tidal and density-driven mechanisms. The subtidal magnitude and reversal of the exchange flows is controlled primarily by the density gradient despite the significant tidal energy. As the density gradient weakens, the underlying tidal processes generate vertically sheared exchange flows with the same sign as that expected for an inverse density gradient. The inverse density gradient may then further strengthen this inverse circulation. These data were collected as part of the Land/Ocean Biogeochemical Observatory and demonstrate the utility of long-term in situ measurements in a coastal system, as consideration of such a wide range of forcing conditions would not have been possible with a less comprehensive data set.

Appendix: Long-Term Mooring Considerations

Several issues arose with the data quality of the long-term observations that are discussed here. The foremost problem with this extended observational effort was the gradual failure of the pressure sensors. Through comparison against the water level observations at the National Ocean Service Monterey, CA tide gauge (9413450), it was evident that both the ADCP pressure gauge and the L02 pressure gauge began to experience a degradation in response time that made these data unsuitable for tidal-time scale computations (such as Stokes drift). The ADCP pressure gauge, in particular, failed completely during the final, thirteenth deployment. This was particularly problematic because ADCP depth was essential for both removing bad bins above the surface and referencing the ADCP to small depth differences between deployments.

In order to ensure that the degradation of the pressure signal throughout the observational period did not affect the analyses, we utilized the following procedure. First, tidal harmonics were computed from pressure observations (converted to depth) from the first 293 days of the observations, from January 11 to November 1, 2005; this harmonic analysis included extraction of 25 compound tide constituents that were required to capture the nonlinear tidal distortion in addition to the standard 68 astronomical constituents (Pawlowicz et al. 2002). Ultimately, a synthetic, predicted tidal signal could account for 99.1 % of the true record; the most significant tidal errors occurred during the large solstitial spring tides and a small amount of uncertainty may be associated with diurnal wind effects. The predicted tidal signal (excluding the fortnightly and semiannual constituents) was added to a subtidal water level record derived from the pressure sensor; we assumed that the gradual degradation of the pressure sensor response did not affect the ability to resolve subtidal water levels. This synthetic tide plus observed subtidal setup was used as the water level record for all subsequent analyses, including removal of bad bins above the surface. For each ADCP deployment, the relative vertical datum of the ADCP was determined by first finding the surface return based on acoustic signal strength. The average depth determined by the surface return was compared to the average depth from the same period in the quasi-synthetic water level record in order to account for subtidal water level changes.

Another issue with the quality of the long-term data set was the issue of biofouling of the conductivity and temperature sensors. Hydrography data from the LOBO moorings were visually de-spiked and de-trended based on service logs kept by Monterey Bay Aquarium Research Institute personnel (i.e., spikes and baseline shifts attributable to mooring servicing were assumed to be bad, and were removed or corrected, respectively).

Finally, battery constraints and biofouling required periodic recovery and redeployment of ADCPs at 2- to 4-month intervals. Servicing involved recovery of the ADCP by divers, on-site cleaning and battery installation, followed by redeployment; this process took 1 h. The short data gaps due to servicing were interpolated using a velocity record synthesized from tidal harmonics to create continuous velocity records at each nondimensional vertical coordinate. This regular servicing procedure resulted in slight differences in the horizontal placement of the ADCP. The ADCP was connected to the base of the LOBO mooring with stainless-steel wire rope, and so was always a constant distance from the mooring base, but varied slightly in later placement as conditions permitted. The inter-deployment difference was on the order of meters in the lateral and order of tens of meters in the axial directions, as the entire LOBO mooring was redeployed on October 5, 2006. The similarity in axial parameters across all deployments (Table 1) indicates that discrepancies between locations were negligible with regard to along-channel-directed velocities. Error velocity variance changed, however, between deployments. The changes in error velocity were discrete and associated with individual deployments, and not any discernible fortnightly, seasonal, or hydrographic pattern. Further inspection revealed that the lateral velocity structure changed significantly between individual deployments; for most of the deployments, near-surface velocities during ebb were directed north–northeast, towards the outer bank of the broad bend around Kirby Park (Fig. 1b). During several deployments, however, near-surface velocities were directed south–southwestward. This reversal in near-surface direction was observed to occur between two successive large ebb tides, separated by a redeployment of the ADCP; we surmise that wake structures caused during the large spring ebb tide affected the measurements. These errors significantly impacted the lateral and vertical velocities. The magnitude of both tidal velocities and the reversing axial subtidal circulation, however, is sufficiently large to have been unaffected by this relocation. Nevertheless, only qualitative results can be derived from this data set. Situations where data are of questionable quality are noted in the text.

Table 1 ADCP deployments at L02 in Elkhorn Slough