OFR 93-011: Farallones

This is the text portion of U. S. Geological Survey Open-file Report 93-11. The appendices are not included.

The Farallones Moored Array Data Report

Kaye Kinoshita, Marlene Noble and Steven R. Ramp

INTRODUCTION

In February 1991, a year-long study was begun of the circulation and the related transport of suspended materials in the Gulf of the Farallones and over the adjacent continental slope. As part of this program, an array of current-meter moorings was deployed in the region. The data from the moored array provides the core of this data report.

The information gathered by the moored array and by allied programs in the study was designed to provide both a basic description of the current field and an understanding of the underlying physical processes in the region. It was necessary to gather this knowledge so that models could be developed that would allow the Environmental Protection Agency (EPA) to choose appropriate sites for the deposition of materials dredged from San Francisco Bay. The observed currents will also be used with simple advection/dispersion models to allow the EPA to predict the ultimate fate of materials deposited at those sites.

There are four primary sites of interest to the EPA, one on the shelf and three on the mid to outer slope (Figure 1). Because the historical data record is sparse, an extensive measurement program was developed in the region so that the EPA could evaluate these sites . While currents have been monitored for several years on the shelf near candidate disposal site 2 (Strub et al., 1987; Sherwood et al., 1990), the historical data base for currents over the slope is almost nonexistent. It consists of a few short records collected mainly by the U. S. Navy in a variety of locations. Hence, the existing data records are too limited, both in duration and in spatial coverage, to provide the extensive knowledge necessary for the development of reliable transport models in the regions of interest.

The extensive, year-long field program begun in February 1991 gathered the data necessary to address the above issues. The program was developed and carried out by a consortium of agencies, companies and research institutions. Personnel from the EPA, U. S. Navy, U. S. Geological Survey (USGS), National Marine Fisheries, Naval Postgraduate School (NPS), Monterey Bay Aquarium Research Institute (MBARI), and Science Applications International Corporation (SAIC) cooperated in the design and execution of the project. The comprehensive program consisted of three elements. An array of moorings that collect measurements of current, temperature, salinity, and pressure was deployed for a year over the shelf and slope. Five quasi-synoptic surveys of the spatial structure of the currents above 300 m and the physical characteristics of the entire water column were conducted during February, May, August, and November 1991 and February 1992. Satellite images of the near-surface thermal structure were collected and processed. In addition, records of the wind velocity over the water and sea level height at the coast were obtained from the National Data Buoy Office and the National Ocean Service.

In March, 1991, the moored array portion of the Farallones project was begun. SAIC and NPS deployed an extensive array of equipment at 6 sites, A-F, in the region of the Gulf of the Farallones (Figure 1, Table 1). The array was designed to monitor the current velocity, water temperature and several other physical characteristics of the water column for an entire year in order that seasonal changes in the measured parameters could be resolved. The selection of the exact locations for the measurement sites was based on several criteria. It was important to determine the spatial structure of the measured parameters and how that structure changed with time, depth and horizontal location. The array was designed to allow an evaluation of possible connections between the circulation and sediment transport patterns on the shelf and the slope. The locations were also chosen to provide information on characteristics of the circulation in candidate dredge disposal sites. The mooring sites are either adjacent to or within all areas of interest to the EPA.

The main line in the array, which contains sites A-D, monitors the changes in the physical oceanographic parameters with water depth. Changes in water depth typically cause the largest spatial gradients in the circulation and sediment transport pathways. Site A monitors currents and other properties of the water column in 92 m of water, a depth found on the outer shelf (Figure 1 and Figure 2). Site B measures these same quantities in 400 m of water, over the upper slope. Sites C and D, located in 800 m and 1400 m of water, respectively, provide information on water movements over the mid and lower slope.

A primary purpose of the secondary line in the array, sites E and F, is to provide information on how the characteristics of the circulation patterns change with distance along the isobaths. Mooring F is in 400 m of water, the same depth but displaced from mooring B (Figures 1 and 2). Mooring E is paired with mooring D. Mooring E is also located on the inshore edge of candidate disposal site 5, the most northerly of the possible disposal sites.

Each mooring in the array had between 3 and 6 instruments that measured current and temperature at specific locations in the water column (Figure 2). Some of the instruments also measured salinity (or conductivity) and pressure at selected sites (Table 1). At each site, the instrument location was designed to resolve the expected vertical structure of the current and temperature fields. The instruments were closely spaced in the upper portions of the water column, where the vertical gradients were expected to be strongest (Figure 2, Table 1). The distance between instruments more than doubled in the lower half of the moorings, where vertical gradients were expected to be smaller. Instruments on all the moorings were placed at similar water depths in order to resolve changes in the horizontal structure of the circulation patterns with depth (Figure 2, Table 1). Currents in the upper 75 m of the water column were monitored at all locations by 1 or 2 instruments. Other common instrument levels were 150 m, 250 m, 400 m and 800 m. In addition, all moorings had instruments within 12 m of the sea floor in order to monitor the strength of currents that could possible resuspend deposited dredged materials. Unfortunately, not all of the instruments in the moored arrays worked and a more limited data set than planned was collected (Figure 3).

To ensure a quality data product, specialized current meters, either EG&G vector-measuring current meters (VACMs), Interocean S4's or Neil Brown acoustic current meters (ACM2's), were deployed in the upper 150 m of the water column (Table 1). These current meters are designed to monitor current fluctuations even when oscillatory currents from large surface waves are present. The current meters below 150 m were Aanderra RCM4, RCM5 or RCM8, instruments that provide quality data at the deeper sites where currents from surface waves do not exist. The sampling intervals were chosen to be appropriate for the instrument depth and type (Table 1). The moored instruments were recovered and redeployed on 3 to 6 month intervals in order that an entire year of data could be collected at each site.

The direction and strength of the wind during the measurement program was obtained from 4 sites along the shelf. Winds were obtained off Point Reyes, on the shelf opposite the entrance to San Francisco Bay, at a location adjacent to site A and at a station within Monterey Bay (Figure 1). The height of sea level along the coast was collected near 3 of the 4 wind stations (Figure 1). There were some gaps in the wind and sea level data records, but they existed for the most part over the entire year of the program (Figure 4).

This data report is a statistical and pictorial description of the data collected in the moored array portion of the field program and the wind and sea level data provided by the other government agencies. The different data sets have been decoded and processed through the initial quality control steps by the individual laboratories that collected the data (Table 1). The data have then been passed through additional quality control procedures at the USGS. Some common conventions have been used to standardize the basic data sets. Greenwich mean time (GMT) is used as the common time base for the data. All vector quantities, such as current and wind stress, have been rotated into a coordinate system that is aligned with or perpendicular to the mean alongslope isobaths. Variables that were sampled at less than one hour were averaged into an hour sampling interval. Data sets that had longer initial sampling intervals have been averaged into 2 or 3 hour sampling intervals. If the initial sampling interval was incommensurate with a 1 to 3 hour sampling interval, the records were averaged, then subsampled into one of the common sampling intervals. The data was subsequently lowpass filtered to remove fluctuations with periods shorter than 33 hours. Both the hour-averaged and lowpass records from the current meters are included in this data report. Only the subtidal wind stress and sea level records are depicted, though tidal parameters are calculated for the sea level records. The data in the report is grouped by type. The first 4 appendices describe and depict the currents (Appendices A-D). The wind and sea level data are in appendix E.

Moored current observations

The moored current observations were collected over the shelf and slope off of San Francisco, California from March 1991 to March 1992. The entire data set consisted of 29 current meters deployed at 6 locations (Figures 1 and 2). To insure quality data, the current meters located within 150 m of the surface were designed to function in a combined wave and current environment (Table 1). The instruments deployed below 150 m were suitable for the deeper waters over the slope. The sampling intervals for each instrument were selected to be compatible with the expected current environment (Table 1). The sampling intervals generally lengthen as the water deepens. All data sets were averaged to hourly or longer sampling intervals in the post-processing procedures. The filter principally removes the tidal currents from the record; hence the predominate flow directions are more easily seen. The filtered currents will be called subtidal currents. Both the hourly and subtidal currents are included in this report.

The currents were rotated into a coordinate system that is aligned parallel and perpendicular to the average orientation of the shelf break. Positive alongslope currents flow toward 328°; positive cross-slope currents flow toward 58°. This coordinate orientation is within a few degrees of the mean isobath orientation at all sites except site D. The isobaths at this deep station are nearly north/south. However, the subtidal, mid-water currents at site D, and at most of the other sites, flow parallel to the shelf break. Hence, currents at all stations were rotated into the same coordinate orientation. The near bottom current meter at site E was located within a small submarine canyon. Hence, currents at this site flow mainly along the canyon axis which is oriented in the cross-slope direction.

Several statistical quantities of the hour-averaged and subtidal currents are included in this data report (Appendix A). The mean, standard deviation, minimum and maximum of each current component are calculated whenever the current record existed for the time periods from March 1991 through August 1991, from September 1991 through March 1992 and for the entire year (Tables A1-A6). The statistics were broken into half yearly and yearly sections for two reasons. First, some instruments only recorded data for the first half of the measurement program (site C at 150 and 250 m). Other instruments only recorded data during the last half of the program (site C at 790 m). These records cannot be directly compared because they were not taken at the same time. The second reason to compute half-yearly statistics is to examine possible seasonal structures in the data set. The standard summer period for the shelf circulation off central California is March through August. The circulation returns to a normal winter pattern in September and lasts until the following March. Hence, the time periods for the half-yearly statistics were chosen to correspond to the standard seasons.

The statistical quantities for each instrument location were computed whenever the records existed in the temporal window of interest. Because of instrument failures, the data records were often broken into two or more sections within one temporal window. Hence, the statistical quantities given in the tables had to be calculated over several pieces of record. The mean in the tables is an average of the mean of each piece, weighted by the number of points in that piece.

Here, is the mean for the entire time period of interest. N is the number of pieces in the time period of interest, is the mean and is the number of points in piece i. The standard deviation given in the tables is similarly weighted by record size.

where is the standard deviation for the period of interest and is the standard deviation for piece i. The means for the hour-averaged and subtidal records have the same value. The amplitude of the standard deviation for the subtidal records is less than for the hour-averaged records mainly because fluctuations due to tidal currents have been removed for the subtidal records.

The error bars around the mean values are calculated at the 95% confidence levels. If the absolute value of the mean is not larger than the error bars, then the mean direction has no statistical significance. The sites that did have significant mean directions are highlighted in the tables.

The formula for the error bar around the mean is

where is the student t statistic with m degrees of freedom at the 100(1- )% confidence level. is the standard deviation of the subtidal data set. The degrees of freedom of a record is the record length divided by the autocorrelation scale. The autocorrelation scale for all the cross-slope currents and the mid-depth currents at site F is 2 days. The autocorrelation scale for the rest of the alongslope currents varied between 5 and 7.5 days.

The average correlations and covariances among the subtidal currents at each mooring site over the year of record are given in Table A7. The correlations indicate how similar subtidal currents observed at one depth on a particular mooring are to those found at a deeper depth. A value of 1 denotes perfect similarity. Values below 0.25 suggest that the characteristics of the subtidal current field change considerably between the two different depths.

The average annual correlations and covariances were not computed over an entire year of data record because there were gaps in the data records during various portions of the year (Figure 3). The values reported in Table A7 are calculated for the times when data at the two sites of interest both existed. The average coherences are a weighted sum of the individual coherence estimates, similar to the weighted sum for the standard deviation (Equation 2). The average covariance amplitudes have been adjusted slightly to correct for the expected temporal changes in the variance of each instrument over time. If a particular current record existed for less than 6 months and was within 150 m of an adjacent record that existed for most of the year, the best estimate of the yearly variance is reported in Table A7. The best estimate is defined to be

where is the estimated yearly variance at site a, is the measured yearly variance at the neighbor site, is the measured variance for the time period site a existed and is the measured variance at the neighbor site for the same time period. Hence, the ratio of the variances for the different time periods is the same for the two sites. The adjusted covariance was then calculated using the estimated yearly variance and coherence values.

The potential to disperse material in suspension or to resuspend material from the sea floor is enhanced at locations where the tidal currents are relatively large. Hence, the characteristics of the major diurnal (O1 and K1) and semidiurnal (M2 and S2) tidal constituents are calculated for each measurement site (Tables B1 and B2). The tidal amplitudes are a weighted average of the data within the collection times listed in the tables because the records of currents were often broken into several pieces. The strong tidal currents near the bed at sites B, C and E suggest that material deposited on the sea floor in these regions may be more frequently resuspended.

The next several sections of this data report contain a series of figures that display characteristics of the measured currents. The first collection of figures shows the time series of the hour-averaged along and cross-slope currents an entire year (Appendix B). The time axes and the vertical scales are the same for all current sites so that comparisons can easily be made among the instruments. All hour-averaged files are grouped together. The next section of appendix B contains
time-series plots of the subtidal data. Again, all plots share the same axes.

An additional plot type, a vector plot is shown in appendix C. These figures display the direction and amplitude of the subtidal current vectors simultaneously.

The energy in the currents at each measurement site is depicted as a function of the frequency of the process in appendix D. A variance-conserving autospectra is calculated for both the along and cross-slope currents. We chose a variance-conserving spectra in order to highlight those frequencies that had the largest energies; the dominant processes occupy the biggest areas on the plots. The tidal frequencies generally have a large fraction of the energy; these large spectral peaks are found at frequencies above .00001 hertz (1.2 days). Not all spectral plots have the same vertical scales because the tidal amplitudes had significant variations over the array. Nor is there a unique vertical scale for each measurement location because it would become impossible to compare the different spectra. As a compromise, there are only 4 vertical scales for the entire data collection, 400, 2000, 4000 and 6000 centimeters squared per second squared.

All spectra for the moored array are calculated with similar Fourier transform parameters. Each record was divided into pieces that contained at most 1/3 of the available data (Table A1). The pieces were windowed with a Hanning window shape in order to reduce leakage of energy from the dominant frequencies into the less energetic frequencies. The pieces overlapped each other by 50%, so a Fourier transform for a minimum of 5 pieces was computed at each station. The spectra for all pieces at one location were averaged together. The length of each piece changed among the measurement sites, but each length was chosen so that both the diurnal and semidiurnal tidal periods were near a natural Fourier frequency (Table D1). Note that the spectra are calculated at different times for different instruments. Spectra for the longest continuous record were calculated at each site.

Wind and sea level observations

Records of wind velocity, atmospheric pressure and sea level were obtained from the National Oceanographic Data Center in Washington D. C and the National Ocean Service in Rockland, Maryland from March 1991 to March 1992. Winds were measured by offshore meteorological buoys at 4 locations between Point Reyes and Monterey, California (Figure 1 and Figure 4, Table 2). The sea level stations were located at the coast, but in the same general area as the meteorological stations.

The hourly observations of wind velocity were converted into estimates of wind stress because wind stress represents the actual force that causes water to move over the shelf and slope. Wind stress was calculated from the formula by Wu (1980).

where is the air density (0.0012 g), the drag coefficient and the wind velocity. The drag coefficient increases with wind strength

where is the wind speed 10 m above the surface in meters per second. The wind stress was low-pass filtered and rotated into an alongslope/cross-slope coordinated system. The positive alongslope wind stress direction is 330°, parallel to this alongslope topography. The positive cross-slope direction is 60°.

The wind stress record at the site nearest our moored array, Buoy 46012 (Half Moon Bay), had a large data gap from April 11 through July 30, 1991. It was possible to fill in this temporal gap with data from the other meteorological sites because the along and cross-slope components of subtidal wind stress were very similar from Point Reyes to Monterey. The correlations among the alongslope wind stresses were uniformly greater than 0.88. The cross-slope winds had similar high correlations. The best correlations were between Buoy 46012 (Half Moon Bay) and 46042 (Monterey). Correlations between the most energetic component of wind stress, the alongslope winds, were above 0.95. Hence data from the latter site were used to fill in the gap in the Half Moon Bay buoy. The wind stress amplitude during the temporal gap was adjusted slightly to correct for a minor difference in wind stress amplitude at the two locations.

The coastal sea level records were converted into records of synthetic subsurface pressure (SSP) by adding atmospheric pressure to the sea level records. SSP is a better representation of the pressure forces that cause currents to move along the coast because changes in sea level caused by high and low pressure systems are removed from the record. The mean was removed from each SSP record. The record was low-pass filtered to remove the tidal signal.

The wind stress and sea level information is presented in appendix E. The tidal amplitudes computed from the unfiltered SSP records are given in Table E2. The basic statistics for the subtidal wind stress records are given in Table E3. The subsequent section displays the temporal and spectral plots of winds stress and SSP data.

REFERENCES

Sherwood, C. R., D. A. Coats, D. W. Denbo and J. P. Downing, 1990. Physical oceanographic processes at candidate dredged-material disposal sites B1B and 1M offshore San Francisco. Pacific Northwest Laboratory, Battelle/Marine Sciences Laboratory, Sequim, Washington.

Strub, P. T., J. S. Allen, A. Huyer, R. L. Smith, and R. C. Beardsley, 1987. Seasonal cycles of currents, temperature, winds and sea level over the northern Pacific continental shelf: 35oN to 48oN. Journal of Geophysical Research, Vol. 92, No. C2, p. 1507 - 1526.

Wu, J., 1980. Wind-stress coefficients over the sea surface near neutral conditions. Journal of Physical Oceanography, 10, 727-740.