DELILAH Stationary Instrument Data
This appendix includes information about the mounting, positioning, and naming of the DELILAH array instruments. It also provides details on the basic
measurements taken at the Field Research Facility (FRF), and a methodology for processing the data. Day-by-day sensor status reports (changes in instrument
height, orientation, etc.) were included based on the experiment notes of Dr. Peter Howd and Dr. Edward Thornton. Other factors such as biological growth,
bent sensor mounts, or changing local depths have not been included. Data from the surf zone arrays were collected continuously except for the time required to
[an error occurred while processing this directive]
The 19 instruments of the DELILAH array are shown in Figure E1. Nine
instrument packages form the primary cross-shore array of the DELILAH
array (indicated by squares in Figure E1). These nine gauges were provided
by the Naval Postgraduate School. Ten additional current meters, provided
by Scripps Institute of Oceanography, were installed to form three subarrays
(indicated by circles in Figure E1). One longshore subarray located in the
trough consisted of six current meters. The second longshore subarray was
formed by five current meters seaward of the bar crest, labeled the crest
subarray. The third subarray was a three-gauge cross-shore array over the
bar, between the two longshore subarrays.
Each primary cross-shore subarray instrument package included a small ball
(Model 512) Marsh-McBirney electromagnetic current meter, a
Paro-scientific pressure gauge (except instrument package 80), and a Setra
strain-gauge pressure sensor. The Paroscientific gauge was connected to a
thin copper tube that was buried in the bottom in order to isolate the orifice
from the current flow. The Paroscientific gauge was designed to measure
wave- induced setup while the strain gauge pressure sensors measured wave
energy. After considerable analysis at the Naval Postgraduate School, it was
that because of the uncertainty in sensor height and other factors, the
Paroscientific sensors could not be used to measure setup(1). Data from all
sensors (four channels) were internally converted to a digital signal and
transmitted as a serial data stream back to shore. Each instrument package was connected to a data collection system using four-conductor, double armor,
lightweight cable. Each electronics package was hose-clamped to the support pipes and connected to the cable with four-pin Brantner underwater connectors.
Except for the two seawardmost instruments (80 and 90), the current meter probes for the primary cross-shore array were mounted on a separate line of pipes
placed 1.5 m northeast of the instrument package/pressure gauge pipe (Figure 6 in the main text). At the five innermost positions (10 to 50), the current meter
stinger was mounted downward, 30 cm away from the mounting pipe. All other stingers were mounted upright. Some of the downward-mounted stingers were
adjusted during the experiment to keep them either above the sand or in the water.
A second data acquisition system was used to sample the analog output from the 10 current meters in the three subarrays. Five Scripps open frame
electromagnetic current meters mounted upward were used in the Crest sub-array. Five Marsh-McBirney current meters with their sensors mounted downward
were used in the trough subarray. A single upward-mounted open frame (54) completed the secondary three-element cross-shore subarray. Each of these current
meters was wired back to the base of the duneline using seven-conductor double armor cable. The cables were wired into watertight "Hoffman" boxes. Power to
the Hoffman boxes was isolated through the use of DC-to-DC converters as a preventative measure to reduce potential ground loops.
Figure E2 shows the location of the cross-shore instruments, including the initial position of the current meters relative to the changing bottom and water
surface. Note the nearness of some sensors to the water's surface. This was taken into account when processing the data from these sensors and is discussed in
the data analysis section that follows.
Current meter orientations were determined with an underwater digital compass mounted on a long nonmagnetic pipe so that it was not affected by the steel in
the sensor mount. For Marsh-McBirney current meters (Nos. 2101-2352) the compass was read with the meters aligned to the direction of -y flow of the sensor
ball. For the open frame meters, the compass was read with the meter aligned in the direction of +x flow. At the time of the DELILAH experiment, it was
believed that the pier axis was 70 deg east of true north, and the current meters were aligned relative to the pier. More recent measures have computed the pier
orientation at 71.8 deg. Current meter orientations are listed in Table E1.
A Met One anemometer was mounted on the top of the pipe at
position 35. The anemometer was hit by high waves and failed on
DELILAH array gauge numbering
The following tables list the DELILAH instruments by gauge
number and name. The gauge number is a four-digit number
uniquely identifying the gauge, while the gauge name is a short
alphanumeric string that identifies both the type of instrument and
the location. Both were required to collect and process data at the
Included in both labels is the location numbering scheme (Figure E1)
which uses a two digit-number to locate every surf zone instrument.
The first digit (1-9) refers to the cross-shore locations starting at the
shoreline and moving offshore. The second digit numbers the
longshore location starting with 0 at the primary cross-shore array and increasing to 5 at the southernmost gauge location.
The gauge number is stored in the header of the FRF's time series data (along with other information such as the gauge depth, gain, and bias). It is composed of
four digits. The first digit, always a 2, identifies the gauge as a DELILAH instrument. The middle two digits refer to the location code described above. The last
digit identifies the sensor (or channel) at that particular location. The sensors used were:
Current speed in the X direction.
Current speed in the Y direction.
Paroscientific pressure gauge.
Strain gauge pressure sensor.
Thus, a gauge number of 2503 is a DELILAH instrument, located at the 50 position, the fifth offshore gauge on the primary cross shore subarray, and is the
Paroscientific pressure gauge (type 3).
The Gauge Name is used only to describe the particular gauge. It is not stored with either the raw time series or the FRF's summary statistics file. The following
conventions are used:
CM - Current meter (Either open-frame or Marsh-McBirney).
PD - Pressure gauge to measure water Depth. This refers to the buried Paroscientific pressure sensors designed to measure setup.
PW - Pressure gauge to measure Waves. Refers to the strain gauge pressure sensors to be placed on the cross-shore array.
Y - Current measurement, positive to the south (also known as U).
X - Current measurement, positive offshore (also known as V).
SLED - This word precedes the names of instruments mounted on the sled.
Current meters and pressure gauges used in DELILAH are listed in Tables E1 and E2, respectively. The tables include gauge name, gauge number, serial
number plots, gain, bias, and coordinates relative to the FRF coordinate system. The signs of the gains may differ from those listed in Calibrations section. The signs
have been adjusted to compensate for instruments being mounted downward. A negative gain flips the data of the channel so that they correspond to the
+Y - a southward-moving longshore current.
+X - an offshore-moving cross-shore current.
DELILAH array data collection
Data were sampled at 8 Hz except for the Paroscientific sensors, which were
sampled at 1 Hz. The primary cross-shore array had an unanticipated 4 to
8-sec gap in the data which occurred approximately every 20 min as a result
of the microprocessors used. The 10 current meters in the three subarrays
did not have these data gaps. These gaps in the time series were filled with
values of -9999 so they are easily identified. There was no way to eliminate
Gains and biases applied can be found in the data file headers of each time series.
Several of the gauges (CM73 at 0939 on 20 October, CM71 at 1030 on 20
October, and CM50 at 1358 on 4 October) were rotated 180during the
experiment, noted as a change in the sign of the gains, in order to conform
to other Marsh McBirney current meter orientations. Other orientation
changes that occurred and must be noted include: gauge CM31 rotated 20 clockwise (visually estimated) at 1330 on 15 October, gauge CM10 rotated 2
clockwise between the beginning of collection at 0834 and end of collection at 1455 on 11 October and also re-aligned 17.6 clockwise between 0624 and 1330
on 18 October, and a re-alignment of gauge CM20 rotated 28.1 counter-clockwise between 0624 and 1330 on 18 October. In addition, a 10 landward bend from
vertical in Gauge CM73 was corrected at 0939 on 20 October. These rotations were noted in the time series headers.
Throughout the experiment, two current meters, CM10 and CM20, were in a zone that experienced large bathymetric change. This necessitated moving the
gauges up and down in order to ensure that they remained submerged in the water but were sufficiently far above the sediment surface. Gauge CM10 was
moved up on 7 and 8 October and down on 9,11, and 18 October. Gauge CM20 was moved up on 7 and 8 October. Measurements of how far up or down these
gauges were moved were recorded and the time series headers were modified.
The electronics package at position 60 (Figure E1) was lost during the high waves of Hurricane Lili at approximately 0700 on October 11. The sensors at
position 30 stopped functioning a few days later.
DELILAH array data analysis
Although several of the current meters were vertically adjusted to maintain submergence and to remain above the bottom, several of the shallower gauges
would occasionally became exposed in the wave troughs, particulary during low tide. A technique for determining when the current meters were exposed is
presented, and was used to mark time series headers with a data quality parameter for exposed sensors. Several methods for estimating gauge exposure could be
envisioned, the one presented here was used since it was fairly simple to employ. This technique finds the lowest trough elevation in each pressure gauge record
(Setra gauges), then determines if the position of the corresponding current meter is above the lowest trough. This was a reasonable approach, particularly for
the primary cross-shore array, since the current meters and pressure sensors were only separated by 1.5 m. The 10 current meters in the sub-arrays did not have
co-located pressure gauges, so comparisons were made with primary cross-shore array pressure gages at the same cross-shore coordinate. Longshore
homogeneity of the wave field was assumed.
DELILAH array pressure gauges recorded offsets and daily variations which were corrected by comparison to the Paroscientific pressure gauge LA33 (gauge
number 231) in the FRF's permanent 8-m array. Gaps sometimes occurred in the data from gauge 231. If the gaps were less than 30 minutes in length,
interpolation was used to attain water levels for those times. On 20 October, 1990 gauge 231 stopped functioning, from that time forward, the tide gauge located
at the end of the pier was used for water level comparisons (Gauge 1, Figure E3). The tide gauge was not used for correcting water levels through the entire
experiment because the signal from this gauge is typically not as
accurate as the signal from the Paroscientific gauge. It samples a
single point every six minutes which results in a noisier signal. This
method does not take into account any potential wave setup between
the 8-m array and the nearshore DELILAH array. The changes in the
water levels for the current meters resulting from this analysis are
presented in Figure E4.
An additional data quality analysis was performed on each current
meter record to determine the effect of biofouling on signal
attenuation. This procedure, referred to as the PUV-test, Z(s), used a
ratio of surface wave Hmo computed from pressure gauges and
current meters, to estimate a gain correction for the current meters.
The method to calculate the PUV-test follows.
Pressure gauge data (in units of meters of sea water) were surface
normalized by the pressure response function
where is gauge depth from
(ensemble) mean sea level,
is total water depth, and
wavenumber (related to radian frequency
by the dispersion
) immediately after Fourier transformation,
by dividing the complex Fourier coefficients at frequencies
Current meter data were surface normalized in a similar
fashion except that the Fourier coefficients were divided by
. In linearized wave theory, the auto-spectrum from
surface normalized pressure is equal to the sum of the
auto-spectra from the two surface normalized velocity components
and for and , respectively. A routine check on
data quality for co-located pressure and current meter data , the
-test, is to compute the function, Z2(s)=/[+], which was expected to be unity in
the wind-wave pass band of frequencies in regions where linear
The PUV-test analysis used the same gauge pairs as the exposed
current meter analysis, current meters without colocated pressure
gauges were matched with primary array pressure gauges at the same
cross-shore coordinate. Again, longshore homogeneity was assumed.
An average PUV-test is calculated for each current meter record over
a select frequency range of the Z(s). This PUV-test is averaged over
the half-power bandwidth in the pressure gauge energy spectra, for
the spectral peak that lies in the wind-wave band (0.4 to 0.05 Hz).
These PUV-test values are used as a multiplier to adjust the mean
current amplitudes. Inherent in this treatment is the assumption of
uniform fouling between both axes of each current meter. Corrected
and uncorrected current velocities, and PUV-test values are recorded
in the DELILAH statistics database. Plots of the PUV-test values are presented in Figures E5 and E6. These data are noisy because averages are taken over
short time segments. PUV-test plots for current meters that are not colocated to pressure gauges are especially noisy. The noisiness increases with the distance
between current meter and pressure gauge. Plots CM71, CM72, CM73 and CM74 demonstrate this spatial inhomogeneity. The consistent PUV-test value of
less than one for gauge CM90 indicates the pressure gauge was either lower in the water than believed or the current meter was higher.
Statistical data from each instrument in the primary cross-shore subarray, the trough array, and crest array are plotted in Figures E7 through E12. The shoreward most gauge CM10 (2101 and 2102) was exposed more often than not consequently, unlike all other gauges in figures E7 and E8, it did not undergo the PUV-test. Data from 11 October through 15 October appears to be the only portion of the collection when this gauge is consistently submerged. Gauge 1 in these plots is
a primary National Oceanic and Atmospheric Administration tide gauge located at the end of the FRF pier. These plots have been processed to handle data
gaps, establish gauge depths relative to mean sea surface, and eliminate gauges that became exposed at low tide. Analysis has indicated the data set is of highest quality between 6 and 16 October. In the first few days of the experiment problems existed in the collection system and gauge elevations and
orientation were being adjusted. After the 16 October biofouling of current meters had significantly attenuated the sensor response, this was especially true for
the Scripps Open Frame gauges. For these Open Frame gauges, there appears to be three identifiable portions of the experiment where three separate gains can
appropriately be applied to each Open Frame current meter. The PUV-test values remained fairly constant during the first portion of the experiment, from 1
October through 8 October, then increased from 9 October through 13 October, and stabilized again from 14 October through 19 October. When these PUV-test
multipliers are applied to the data, differences in the current velocity as great as 0.24 m/sec (27.5 % change in velocity) can result. For these reasons DATA
COLLECTED USING OPEN FRAME SENSORS SHOULD BE USED WITH CAUTION.
FRF Permanent Instrumentation
A group of instruments that is permanently installed at the FRF allows for the continuous collection of oceanographic and meteorologic data. During large
cooperative experiments, such as DELILAH, these instruments not only provide researchers with information on conditions during the experiment, but also
with an invaluable archive of data from which to compare and interpret trends. The permanent instrumentation includes the 8-m array; Baylor wave gauges;
waverider buoy; and gauges for wind speed, wind direction, tide, atmospheric pressure, and temperature. Table E3 lists the FRF permanent instruments by name
and number and includes gain, bias, coordinates relative to the FRF coordinate system, and gauge depth. Figure E3 is a schematic depicting the location of these
An additional numbering convention for FRF permanent instrumentation is used to refer to combinations of gauges. This is a four digit gauge number that
associates a name with the results of a multi-sensor data analysis. Among these four-digit combination gauges are 3111, which is the pressure sensors in the
8-m array; 3519, which is the vector averaged currents of the PUV gauge in the 8-m array; and 3932, which is the vector averaged wind speed and direction.
Statistics from these and other gauges show the basic climatology during DELILAH in Figure E13.
The 8-m array is composed of a linear longshore array of ten pressure sensors (Nos. 101-191), a cross-shore array of five pressure sensors (Nos. 211-251), and a
combination pressure/current meter (PUV) (Nos. 511, 519, and 529) within the array. The fifteen pressure gauges are mounted approximately 0.5 m off the
bottom in the vicinity of the 8-m isobath. The array employs pressure sensors, manufactured by Senso-Metrics Inc., with a range of 0-25 psir (pounds per square
inch relative to 1 atm). Each sensor was statically calibrated prior to deployment and mounted on a 2-in.-diam pipe jetted into the ocean bottom. A complete
analysis of the data results in a three-dimensional directional spectra as illustrated in Figure E14. 8-m array representative
spectra and data
for all days of the experiment are available for viewing and the downloading. Characteristic wave heights from spectral
observation are most frequently given as Hmo, which is four times the standard deviation of sea-surface displacement. It can be determined from the volume
under the frequency-direction spectrum by the equation:
It can also be found from the integrated frequency spectrum by:
which is its more conventional definition, or from the integrated direction spectrum (Equation 3) by:
Peak wave period (Tp) and peak direction (qp), can be determined by integrating the data in both direction and frequency as shown by the graphs on the vertical
panels in Figure E14. Note that as defined, the peak direction does not necessarily correspond to the waves which occurred at the peak period (i.e. peak
energy). The PUV directional wave gauge consists of a Marsh-McBirney electromagnetic current meter and a Senso-Metrics Co., Inc. pressure sensor. It is
mounted on a tripod located within the linear array. Wave direction is measured in degrees relative to true North and indicates the direction the waves are
coming from. A detailed discussion of the data processing required to transform measured time series to estimates of frequency-direction spectra may be found
in Long and Atmadja (1994)(2).
Surface Wave Gauges
The Baylor gauges and the waverider buoy are two different types of gauges used to collect wave information. The Baylor gauges (nos. 625 and 645) are surface
piercing inductance staff gauges mounted on the pier at the locations indicated in Table E3 and Figure E3. The waverider (no. 630) is an accelerometer buoy
located four kilometer offshore. Data analysis is similar for the two gauge types. Data were collected in 34 min records, which consisted of 4,096 data values
(representing the voltage output of the sensor) sampled at 2 Hz. After the voltages were converted to engineering units using the sensor calibration factors, the
time series were edited to eliminate erroneous jumps and spikes.
Measurements of the wind speed (no. 933) and direction (no. 932) were made at the seaward end of the research pier using a Qualimetrics Corporation Skyvane
Model 2101 anemometer. Wind directions are reported relative to true North with onshore winds having a direction of 71.8o. Summary statistics of 34 minute
mean values of these data are presented as the wind vector plot in Figure E11.
Water level data were collected by a National Oceanic and Atmospheric Administration (NOAA) tide gage located at the seaward end of the research pier
(Gage 1). A 6 minute mean value from the tide gauge, computed once per hour, is plotted as water level from NGVD in Figures E7 through E13.
Barometer and Thermometer
Atmospheric pressure (gauge 616) and air temperature (gauge 624) were measured by Yellow Springs Instrument Co. sensors installed at the FRF building.
Summary statistics of 34 minute mean values of atmospheric pressure is included in Figure E13.
1. Personal Communication, 29 April 1997, Dr. Edward Thornton, Dept of Oceanography, Naval Postgraduate School,
2. Long, Charles E., and Atmadja, Juliana. (1994). "Index and Bulk Parameters for Frequency-Direction Spectra Measured at CERC Field Research Facility, September 1990 to August 1991," Miscellaneous Paper CERC-94-5, U.S. Army Engineer Waterways
Experiment Station, Vicksburg, MS.
Longshore Subarray Directional Wave Measurements
The following discussion describes one approach to processing some of the
data collected during DELILAH. It is not the only acceptable processing
method, but it is described here in detail to demonstrate one way of handling
the data complexities. A preliminary analysis of the wind-wave directional
spectra from the two longshore subarrays during DELILAH was performed by
Dr. Charles Long. The arrays were not ideal for wind-wave analysis,
consequently simplifying assumptions as well as some modifications to the data
were applied. Some of the data processing techniques described here may also
be useful for other than wind-wave analysis.
The following assumptions were made:
Linear theory is valid throughout.
The wave field is stationary for durations of 2 hr 16 min (the total lengths
of records analyzed).
The wave field is uniform over the longshore length of each array.
Mean currents have no effect on the observations.
The two longshore subarrays are used in the analysis. The crest subarray,
seaward of the nearshore bar crest, consisted of the pressure gauge 2704, and
five two-axis (x/y) current meters: Marsh-McBirney 2701/2702, and the
Scripps Open Frame current meters 2711/2712, 2721/2722, 2731/2732, and
2741/2742. The set comprises an 11-element array with each current meter
representing two elements.
The trough subarray was generally on the trough side of the nearshore bar
crest. It used pressure gauge 2304, and five Marsh-McBirney two-axis current
meters 2301/2302, 2311/2312, 2321/2322, 2331/2332, and 2341/2342. This
set also constitutes an 11-element array.
The intent was to compare nearshore results with spectra from the FRF 8-m
array. 8-m array frequency-direction spectra were computed from records of
2 hr 16 min duration (16,384 data points at 2 Hz, or 8,192 sec) for each gauge,
with collection times at 3-hr intervals, starting daily at 0100, 0400, 0700, 1000,
1300, 1600, 1900, and 2200 Eastern Standard Time. To compare data of the
same duration and collection start times required finding 16 contiguous records
from the crest and trough subarrays that began at about (± 4 minutes) the same
times as the FRF collections. Because of the logistics involved with changing
tapes, not all FRF collections had a matching crest or trough collection, so
some cases are missing from the analyzed set.
To deal with the gaps, and make maximum use of available data,
processing occurred as follows:
The pressure gauge from each array (2704 for crest, or 2304 for trough)
was processed first. The 8-Hz data were subsampled at 2 Hz by taking every
fourth point, starting with the first point in the first of the 16 contiguous
records to be processed, and storing the points in a single long array of 16,384
points. The long array was then broken into 31 half-lapped ensembles of 512-
sec duration (1,024 points), and each ensemble was searched for data gaps.
If no gaps were detected, an ensemble record was accepted as it was. If a
gap occurred in the second half of any ensemble record except the first, the
bounding indices of the ensemble record were shifted to earlier times so that
the last point in the ensemble record was the last good data point before the
gap. If a gap occurred in the second half of the first ensemble record, that
record was not used in analysis. The ensemble bounding indices from the long
array were stored for later use with current meter data.
If a gap occurred in the first half of any ensemble record except for the final
one, the bounding indices of the ensemble record were shifted to later times so
that the first point in the ensemble record was the first good data point after the
gap. If a gap occurred in the first half of the last ensemble record, that record
was not used in analysis. Again, the ensemble bounding indices from the long
array were stored for later use with current meter data.
This procedure resulted in occasional redundancy of the ensemble records,
and occasional loss of one ensemble record (or half an ensemble's worth of
data points), so that there were only 30 ensembles instead of 31. However, this
method retained most of the useful data and none of the gapped data.
After the pressure gauge data had been analyzed for gaps, the current meter
data were processed. Current meter data were also decimated to 2 Hz from
8 Hz and written into long arrays. To ensure no offsets in time, the ensemble
record bounding indices from the pressure gauge were used to define the
current meter ensemble records. Consequently, a given ensemble spanned the
same period of time from all gauges in an array, so that relative phasing was
preserved for all ensembles. In this way, a working database of ensemble
records for each collection was created.
Modifications to pressure data
To establish realistic gauge depths relative to the (collection) mean sea
surface, the barometric (gauge 616) mean value (in meters of water) was
subtracted from the pressure gauge mean value (in meters of water). When
the (positive) gauge depth relative to NGVD was added to the barometrically
corrected gauge mean, the record should have reflected the tide plus any setup
or setdown from wind or radiation stresses. However, this was not what was
observed. Mean pressures from both gauges 2704 and 2304 seemed low.
High tide didn't get very high above NGVD, and low tide was much lower.
This was further pursued in an effort to determine an accurate estimate of
gauge depths by comparing means from gauge 2704, from the 2703 pressure
gauge, and from predicted tides. Figure E15 shows some results from 3-21 October 1990. In the relatively quiescent days of 3-7 October, the 2703
gauge appears to read high relative to predicted tides by about 0.10 m, and
gauge 2304 reads considerably lower. The difference between the means is
also plotted in Figure E15b and appears to fluctuate around 0.40 m.
As there was not a clear pattern to explain the differences, a bias of 0.40 m
was added to the means of both gauges 2704 and 2304 for all runs. Where this
is incorrect, it will have a strong effect on results because all gauge depths are
keyed on the pressure gauges; pressure and current meter wave signals were
surface corrected using these depths, and cross spectra were normalized with
the surface-corrected pressure auto-spectra.
Modifications to current meter data
In data processing, current meter data records were read in pairs, and the
component velocities were rotated from the orientations given in the time series
data headers into the FRF coordinate system. Additionally, the signs of the
rotated signals were adjusted so that the cross-shore velocity (U) was positive in
the onshore direction, and the longshore velocity (V) was positive in the
southerly direction. Due to the change in the coordinate system mentioned
earlier, this is the reverse of the previous convention.
Because of the biofouling on the Scripps open frame current meters (2711/
2712, 2721/2722, 2731/2732, and 2741/2742), the current meter gains for all
current meters were adjusted to be consistent with sea surface displacement
variance frequency spectra derived from the pressure gauge (2704 for the crest
array, 2304 for the trough array). To do this, the PUV-test, (described in the
"DELILAH array data analysis" section earlier in this Stationary Instrument Data chapter ) was used.
To adjust the current meter gains, the mean value of the (s) over the
range of (cyclic) frequencies 0.006 to 0.338 Hz was computed, and then the
gains listed in the header records were amplified or reduced until this mean
value was unity. Identical gain amplifications were applied equally to both
channels of a two-axis current meter under the assumption that the biofouling
or other signal degradation was horizontally isotropic, and because a way to
determine gain adjustments on individual current meter channels is not known.
This assumption is important because overcorrecting one current meter channel
relative to another changes the apparent wave direction at that current meter,
and degrades the directional estimator.
To illustrate the temporal behavior of the gains, the time series of the gains
for each subarray have been plotted. Figure E16 shows the gains from the
trough subarray, and Figure E27 shows the gains from the crest subarray. The
results make qualitative sense in that the Marsh-McBirney current meters
(station 70 in the crest subarray, and all stations in the trough subarray) had
gains that generally remained within 20 percent of unity, while the Scripps
open frames (stations 71-74 in the crest subarray) had gains near unity early in
the experiment, but drifted high as time evolved, suggesting the influence of
fouling. Note that all of the gain adjustments were keyed to the spectra from
the pressure gauges, and, because the pressure gauge depths were modified,
some noise may have been introduced in the current meter gain adjustments.
For instance, there appears to be a diel variation in the gains from the trough
array that is reminiscent of the pressure gauge differences shown in Figure
E25b. There may be a relationship between these phenomena, but what that
relationship is, remains uncertain.
Total water depths
In an earlier pass through the data, the total water depth, as listed for each
gauge location, was assumed constant. However, the depths were not constant
throughout the experiment, and erratic changes in depth occurred after the
energetic events beginning on 10 October. Subsequently, total water depths for
each gauge site were established by interpolating the bathymetric minigrid
surveys in both time and space, to obtain a more correct total water depth for
each gauge location and for each collection. Water depths found in this manner
are used in analysis and are plotted as time series in Figure E28 (trough
subarray) and Figure E29(crest subarray). When the bar moved offshore
during the energetic events, shoaling occurred under the crest subarray, and
deepening occurred under the trough subarray. More consistent results overall
were obtained when changing depths were used in the analysis.
Despite efforts to position the gauges so that they remained within the water
at all tide stages, at some low tide stages, some of the gauges became exposed.
Because such events add unacceptable noise to the cross-spectral matrix, all
such cases were eliminated from analysis. These were identified using a
method similar to that described in the "DELILAH array data analysis" section
earlier in this appendix. The one difference in this analysis was the use of a
0.40-m bias in the gauge depth adjustment, and the other analysis determining a
bias from correlating water levels with a tide gauge (gauge 1). This analysis
was done by saving the minimum instantaneous water level from the raw
pressure signal and comparing it to the depth of the shallowest gauge in the
array being analyzed. If it appeared that the shallowest gauge was at or out of
the water surface in the trough of any wave passing the pressure gauge, the
analysis was aborted. This may have been a liberal treatment because it nominally allowed the Marsh-McBirney current meters to function within several
ball diameters of the free surface for a few seconds in some runs, which is
generally considered unacceptable. The time series of velocity and pressure
from some of the marginal runs were visually inspected and it was not always
obvious that the signal was degraded. A more conservative constraint may be
appropriate if further analysis is pursued.
A check on data quality was performed by overlaying the frequency spectra
plots from all gauges for each run that survived the exposure test. Figure E30
is an example of this. Some features which may not be obvious include:
- The main graph shows the frequency spectra (in log coordinates) from
all gauges plotted out to the Nyquist frequency (1 Hz for the 2-Hz
subsample frequency) so that background noise floors can be seen.
- The pressure spectrum is offset downward by three orders of
magnitude to isolate its shape from spectral curves from the other
gauges. Spectra from longshore currents are offset downward by one
order of magnitude to isolate that group of spectra. Spectra from cross-shore currents are plotted with no offset.
- Raw spectra are shown throughout except in the low-frequency (including wind-wave) pass band of 0.006 to 0.338 Hz. In the low-frequency
band, the spectra have been surface normalized as described above
under "Modifications to current meter data." This helps to determine how
much surface correction is done even for these shallow gauges, and
indicates if noise is amplified or reduced in surface normalization.
- Modified current meter gains have been applied to the current meter
spectra in these plots to see if modified spectra form tight groupings for
each type of gauge.
- Curves near the part of the main graph ordinate labeled 'Z' are the
function described above (actually is used) after its mean
value is forced to unity. These functions are plotted on a linear scale.
These curves are informative in that if they are reasonably flat, the
surface normalization and gain modification might be considered satisfactory. The example shown is reasonably clean. Visual inspection of
the plot shows some degradation of these curves in time. In some high-energy conditions, there is considerable chatter at the lower
frequencies, due possibly in some part to shear waves, which have a
stronger velocity signal than pressure signal.
- Also shown in the upper right corner of Figure E30 are numbers
representing a possible modified pressure gauge gain (which should be
1.0 for all cases in these runs).
Data quality plots were used to identify cases with data that were clearly
bad. Surface-corrected velocity spectra should cluster together, with one
grouping each for alongshore and cross-shore velocities. Barring other sources
of noise, these spectra should follow a chi-square error distribution. For the
nominal 160 degrees of freedom with which these spectra were computed, the
95-percent confidence interval extends from about 21-percent below to about 2-percent above the true value of each spectrum, or a range of about 4-percent.
Though the true values are not known, members of a group of spectra were
considered satisfactory if the range of spectral densities agrees to within about
44 percent at each of the discrete frequencies in the nominal wind wave pass
band of 0.04 to 0.32 Hz. Pressure data were considered satisfactory if Z(f) was
reasonably uniform through the wind-wave pass band of frequencies, indicating
consistent frequency dependence of velocity and pressure data. Excepting
high-energy, low-frequency conditions, where shear waves may be influential,
the nominal acceptable range of PUV-test results is 0.8 < Z(f) < 1.2. Data
acceptance criteria given here served as rough guidelines. Minor exceptions
were tolerated in this analysis in an effort to maximize the amount of data
considered to be useable.
The mean current data contained either severe, persistent nonuniformity or
some yet-to-be understood bias. It remains unclear what the source of the
problem was or how to circumvent it . Consequently, any effects from mean
currents were ignored, though the wind-wave parts of the current meter records
seemed acceptable. Data that passed the quality control constraints were used.
Final data sets
Processed data satisfied three constraints: (a) start times and durations had to
match FRF collections,(b) there were no exposed gauges, and (c) pass band
frequency spectra satisfactorily met the constraints of the data quality plots.
Table E4 lists the cases from the trough and crest subarrays that satisfied these
constraints. More exposed gauges occurred in the trough subarray, and the
current meter at station 30 was lost sometime on 15 October. Therefore, the
total set of cases for the trough subarray is small.
Satisfactory data runs were processed for frequency-direction spectra
following the basic premises for IMLE described by Davis and Regier(1), and
Pawka.(2) Figures E21 and E22 are examples of trough and crest subarray spectras. Other representative spectra from these subarrays can be viewed, and the data are available. For the image plots, when
possible, two representative spectra for each sampling day are presented. As
summarized in Table E5, all gauges were used for the lowest frequencies, but
only the pressure gauge and two current meters were used for the highest
frequencies. Figures E23 and E24 summarize the directional data from the
trough and crest subarrays. Through the twenty-one days of the experiment, movement of sediment changed the depth from instrument to bed. Table E6 displays these depth changes with respect to gauge location
1. Davis, R. E., and Regier, L. A. (1977). "Methods for estimating directional wave spectra from multi-element arrays," Journal of Marine Research 35, 453-77.
2. Pawka, S. S. (1983). "Island shadows in wave directional spectra," Journal of Geophysical Research 88,
[an error occurred while processing this directive]