Report on DART Field Deployment into San Francisco Bay
To
Dr. Andrea Neal, BOSS
Prepared by
Dr. Arthur Lonne Lane
&
Lloyd French and Kenneth Manatt
of
Land Sea Discovery Group R&D, LLC
24 December 2012
Executive Summary:
• DART was easily deployed from the Brownlee to depths of 40 feet in the San Francisco
Bay. Current depth capability is 90 ft, and with modest investments for cables, can be
extended to 250 ft.
• The DART instrument characterized the three sites selected both for differences in water
purity and for some physical aspects of the mixing and stratification at each site.
• Supporting water samples were collected from several depths at each site and were
examined with a laboratory spectrofluorimeter.
• There was no evidence of benzene, toluene or xylene contamination at these three sites
in the afternoon time of the measurements, at least to the level of instrumental sensitivity
which currently being tested to provide a base limit number.
DART Instrument and Measurement Method:
The principal question that needed examination was to record and advance the
understanding of the potential of water quality assessment by fluorescence emission
patterns in the mid- and near- UV that might be produced by soft-pulse UV laser
illumination at 248 nm. Also quite important was the demonstration of the DART
instrument in a field deployed configuration along with acquisition of useful
scientific/environmental data.
Figure 1. Deployment of
the DART into San
Francisco Bay from the
R/V Robert V. Brownlee on
28 November 2012. The
instrument is suspended
from an incremented
measured length rope and
two electrical cables
provide power and
command/real-time
control with active data
return.
The DART instrument contains a 248.6 nm hollow cathode discharge laser that generates
an output pulse of 40 to 60 µsec duration, at repetition rates of 1 to 20 Hz. The output
intensity is variable over a modest range; most of the measurements reported below were
performed with 2 to 3 µJoules / pulse [2 to 4 x 1012 photons], at 3 Hz. Thus the illumination
cycle is low and the power density directed into the water column is modest when
compared to most pulsed UV laser sources. The instrument contains six narrow band pass
limited photomultiplier tube detectors (PMTs), covering the wavelength region between
270 and 410 nm. The detectors have gate-synchronized on-time with the laser pulse and
also make a pre-sequential ambient, non-laser light measurement with the same
integration time as that selected for the laser fluorescence signal measurement. Most nonoutdoor measurements have background detection corrections of 0.2 to 1 photon per
pulse, when averaged over a 300 pulse measurement sequence. Daytime outdoor
measurements typically require background corrections for the longer (redder) portion of
the detection spectrum because the Earth’s atmospheric ozone does not fully attenuate
wavelengths longer than about 340 nm. In aqueous environments such as the ocean or
lakes, measurements at depths of 15-20 feet and deeper usually need background
corrections only in the 360 and 380 nm detection bands.
As presently configured, the instrument optics provide optimum performance in signal
detection at a nominal distance of 68 inches (1.72 m) in air and a shorter distance in water
but the range of detection of peak is not narrow; best response covers a linear distance
range of about 4 to 12 inches (10-30 cm) around best focus. For most of the aqueous
environments, the detection zone is in the 18 to 30 inch range.
The measurement sequence consisted of 301 pulses of the laser, each of which produced an
elliptical ring illumination pattern that was about 7 mm x 5 mm in the principal axes. The
ring illumination thickness was about 1 mm for each pulse. The axes of the ellipse rotate
slowly, taking about 7 to 10 pulses to make a full 360° rotation. Thus no two sequential
pulses fall exactly on the same location in the fluid field. The time-averaged pattern is
approximately a circle of around 7 mm diameter and 1 to 1.5 mm in ring thickness.
The measurement data are recorded for each laser pulse. Since the laser is a hollow
cathode discharge, no two sequential pulses are alike in photon number density. Over the
course of the 301 pulse measurement @ 3 Hz, most measurements have an average photon
flux of about 3 x 1012 photons and the 2 sigma bound pulse envelope is around +/- 5% of
the average. If any anomalies, such as a laser miss-fire with low output flux occur, those
data are removed from the data set. For most measurement sets all 301 points were used;
only for one or two sets was it necessary to edit a block of points – but in all these cases at
least 220 to 250 good discrete points were available to construct a meaningful and useful
average for both laser output and return fluorescent signal intensity. The control
electronics reside in a single laptop computer that provides instrument control, real-time
display of laser performance, real-time display of the fluorescence detection and recording
of all engineering and data germane to each laser pulse. The measurement data are
currently processed separately after each day’s field work but the near-term future plan is
to have a second laptop processing and displaying measurement data in terms of chemicals
and materials of interest in real-time.
Niskin Bottle Sampling of DART Measurement Zones
A commercial Niskin bottle was used to sample water from each site where the DART
instrument was deployed in the San Francisco Bay during the short field outing with the
Marine Science Institute. Table 1 below provides the measurement sets for both DART and
the Niskin sampler. Limited deployment time curtailed generation of a denser
measurement matrix that would have been desired for a full science-oriented field study.
The water samples were collected in pristine, certified I-Chem analytical sample bottles
and they were refrigerated within 24 hours of the collection time. The collected water
samples were examined in the laboratory for comparison with and enhancement of the
DART data.
Location of Sampling Areas for DART & Niskin Bottle Collections
Figures A & B below provide an overview of the locations in the southern San Francisco
Bay where the DART was deployed and the Niskin samples were acquired. In these figures
A1 – A2 indicate the beginning and end locations of the first set of measurements. The ship
was ‘stationary’ with respect to the surface water; the drift from A1 to A2 is the result of
current flow over the sampling time with movement within the stationary geography.
Similarly, B1 – B2 is the track for the second main channel measurements, start and finish.
In the Industrial Dock area C1 – C2 is the motion of the ship.
Figure A. Track of the ship in the southern San Francisco Bay region during the
measurement sequences – purple colored trace. A1 - A2 are the start and finish of the
Location One data acquisitions. B1 – B2 are locations for the second DART data and Niskin
sampling. The Bay was under outflow conditions, with the water moving northward
towards the Bay Bridge.
Figure B. Location Three. Track of the ship in the channel adjacent to the Industrial Dock
area just south of the Marine Science Institute (lower left corner of image). C1 - C2 are the
start and finish of the Location Three data acquisitions. The Bay was under outflow
conditions, with the water moving out of the inlet and northward towards the Bay Bridge.
Table of Measurement Acquisitions
Table 1. Location, designation and depths for the San Francisco Bay activity with the
DART and Niskin Bottle sampling on 28 Nov 2012.
Measurements Discussion
Hitachi Spectrofluorimeter Measurements (Niskin Bottle Samples):
A Hitachi Model 4500 Spectrofluorimeter was used to examine the collected Niskin Bottle
water samples that had been refrigerated once they arrived in Pasadena. The measurement
configuration utilized spectrally verified, clean UV quartz cuvettes with a right angle
geometry: incident scanned excitation wavelengths were orthogonal to the viewing
emission detection. The cuvette holds about 3.3 mL fluid, but needed to be only half full to
cover the entire illumination-detection apertures. Optical pathlengths are 10 mm per side.
The liquid samples for study were withdrawn from each bottle with a clean, noncontaminating stainless steel syringe needle and deposited into the cuvette. The fluids
were permitted to warm from 4° C (refrigerated temperature) to around 12-16°C to
prevent frost deposits on the cuvette optical walls during the optical measurements.
This laboratory instrument employs a dual monochromator system: one to select the
narrow band excitation illumination and one to scan the selected range for potential
emissions. The parameters used were 5 nm steps for excitation illumination going from
200 to 400 nm, then coupled with full emission scans at 2 nm bandwidth over the range
200 – 600 nm for each excitation band pass step. Figures 2 to 5 below provide an overview
of the acquired data, with Figure 2 (pure water) annotated to show instrumental features
found in the Hitachi data and display.
Figure 2. Hitachi Excitation-Emission (EEM) diagram for 18.2 MOhm, 0.22 µm filtered, UV
irradiated NanoPure water. Laboratory standard for clean water.
At least two fluorescent entities are in the vertical water column at this location; the
bluer one should have a weak emission in the DART longest wavelength band (380-410
nm). These data imply good vertical mixing between these two depths at this location.
Figure 3. Hitachi measurements of water samples from Location One (see location map,
figure A). Two weak emission features are detected near excitation of 248-250 nm, which is
the wavelength of the DART laser. The emission of the redder wavelength feature around
480+ nm lies outside the detection region of the DART. The bluer feature around 380-440
nm lies in the longest wavelength detector channel of DART, Channel 6 whose band pass
covers ~380-405 nm. This particular Hitachi spectrophotometer has some scattered light
issues at very short and very long wavelengths. Our regions of spectral interest are not
compromised by these difficulties.
Figure 4. Hitachi measurements of water samples from Location Two (see location map,
figure A). No new features with any significant presence that were not seen at Location
One can be detected. This implies good fluid mixing between Location One and Location
Two.
Figure 5. Hitachi measurements of water samples from Location Three, near the Industrial
Dock Area (see location map, figure Y). This location has a very noticeable pair of emission
features that appear to have differing emission strengths (concentrations?) with depth.
DART measurements:
As indicated in Table 1, there were a total of 14 DART measurement sets at the three
locations selected. For each measurement set the DART instrument was suspended by a
measured interval rope from the ship’s stern port davit. The instrument was suspended
vertical with the optical entrance facing downward (see figure 1). DART measurement ML10 was performed very near the surface, about 1 foot below the air interface, at Location
Two. The data were distorted by the changing solar illumination caused by shallow surface
waves. At night this would not be a concern, but with intermittent cloud cover during this
day’s mid-afternoon period, there was a rapidly changing solar illumination environment in
the first one or two feet of water depth. This singular data set was not processed because of
the strong amplitude variations of the measurement background associated with each laser
pulse data count.
The DART data are presented in several formats. Each specific measurement place
(location and depth) data collection is formed into a single ensemble vector of the six band
fluorescence photometry that is generated by the 248 nm laser pulses. These are plotted
first as simple band intensities, then processed in a Principal Component Analysis (PCA)
that teases out the subtleties inherent in the six color intensity data. Those plots are
examined for potential associations and matches with existing and future libraries of pure
material signatures collected with the same instrument. The processed data sets are
presented below, first in single location collections, then as aggregates of the three
locations, and finally with comparisons to a few other materials.
Location One:
Six data sets are presented in this graphic. PCA side, left: Only three points are readily
visible. The data for the 40 ft, 30 ft (two splits of this set to clean a few anomalous points)
and the 20 ft depth are all coincident within a single plotted point, the 20 ft value. Slightly
different is the 10 foot depth point – by not by much. It is only separated by 0.0003 in the zaxis PCA space. The % relatedness value is derived from the PCA assessment for comparing
a specific point to the others being examined in the plot. The table (upper right corner) for
the values of relatedness show that the 40, 30 and 20 ft measurements are all at least 99%
identical (hence, all those four points are coincident in one location). The 5 foot depth value
is well removed from the locations of all of the other deeper measurements. Band intensity
plot; right side: Only the 5 foot depth measurement shows any real difference to those in
this collection. Channel 1 (nominal 280nm) is slightly less than the other points in
amplitude (more obscuration in the water) and Channels 5 and 6 (nominal 360 and 380
nm) also show amplitude attenuations. Compare this figure to the Hitachi data for sampled
waters at 35 and 10 foot depths where a very slight emission feature was present in these
wavelengths. These measurements support the statement that the water around Location
One were well mixed vertically.
Location Two:
Four data sets are presented in this graphic. PCA side, left: Only two points are readily
visible. The data for the 40 ft, 20 ft and the 10 ft depth are all coincident within a single
plotted point, the 10 ft value. Slightly different is the 5 foot depth point . The % relatedness
value is derived from the PCA assessment for comparing a specific point to the others being
examined in the plot. The table (upper right corner) for the values of relatedness show that
the 40, 20 and 10 ft measurements are all at least 99% identical (hence, all those three
points are coincident in one location). The 5 foot depth value is removed from the locations
of all of the other deeper measurements. Band intensity plot; right side: Only the 5 foot
depth measurement shows any real difference to those in this collection. Channels 5 and 6
(nominal 360 and 380 nm) also show amplitude differences (but brighter in this case as
opposed to what was observed at Location One). Compare this figure to the Hitachi data for
sampled waters at 35, 20 and 10 foot depths where a very slight emission feature was
present in these wavelengths, and were barely visible in the 3-D intensity plots. These
measurements support the statement that the water around Location Two was well mixed
vertically, except near surface.
Location Three:
Three data sets are presented in this graphic. PCA side, left: All three points are readily
visible, representing 30, 20 and 10 ft depth measurements. Although the axis scales
indicate rather small numerical differences in the three PCA vectors, the differences are
discernable. Band intensity plot; right side: Channel 6 (nominal 380 nm) also shows
amplitude differences (but brighter in this case as opposed to what was observed at
Location One). Compare this figure to the Hitachi data (figure 5) for sampled waters at 25
and 10 foot depths where a quite pronounced emission feature was present in nominal 380
band region. These measurements support the statement that the water around Location
Three has more contamination present than was detected at Locations One and Two. The
unusual aspect of the 20 foot depth measurement not being close to either the 30 or the 10
foot data is confusing, for no tread can be ascribed to this behavior. For this location, a
more thorough vertical and areal sampling would be necessary to further understand the
contributions from the industrial area as well as the water flow patterns that might cause a
concentration stratification of the chemical burden.
Comparison of Locations One and Two:
Locations One and Two are both in the main ship channel (deepest portions of the Bay).
Although the Hitachi measurements (figures 2 and 3, above) did not show much difference
in the measurements at those two locations, the DART instrument data indicate that there
is a real difference between the waters examined in those two places. Each location has a
tightly clustered PCA zone but each cluster is well separated from each other. The
‘greenish’ colored set of measurements at Location Two (40, 20 and 10 ft depths) are so
similar to each other on this plotting scale that only a single point appears in the PCA space
diagram (Also note that in the Location Two data above, all but the shallowest depth point
are also coincident.). The band intensity plot on the right side of this figure shows a
normalized difference in all bands except where the data are normalized in band 2
(nominal 300 nm).
Comparison of Locations One, Two and Three:
This figure compares all three sampled sites in the Bay. Locations One and Two are both in
the main ship channel (deepest portions of the Bay), and Location Three is near the
Industrial Dock area. The plot orientation and scale are virtually identical to the figure just
above that compares Location One and Two so that the addition of the Location Three data
(in shades of pink and purple) can be compared to the earlier sets. The DART instrument
data indicate that there is a real difference between the waters examined in these three
places. The wide spread seen in the figure for Location Three data only is substantially
compressed when placed into this viewing frame. The data for each location are well
separated from each other, but Location Three has a significant spread in its cluster. The
band intensity plot on the right side of this figure shows a normalized difference in all
bands except where the data are normalized in band 2 (nominal 300 nm). The character of
the band intensity plot show a significant variations between locations.
These two small figures provide views along orthogonal axes to further elucidate the
spread of the PCA assessment of the three sampled sites. They are quite distinctly different
from one another.
Comparison of the three Locations with ultra pure water:
This figure compares all three sampled sites in the Bay with a sample of 18 M, 0.22 µmfiltered, UV-irradiated water; this is the laboratory standard for ultra pure water. The
DART instrument data show that there is a real difference between the waters examined in
the three places in the Bay and very pure water (not surprising!). The band intensity plot
on the right side of this figure shows the substantial difference in all bands when
comparing the Bay waters to pure water. One important aspect of the DART measurements
that has not been easy to discern in the Bay water data only is the nature of the 280nm
band intensity. Based upon many laboratory measurements of aqueous samples containing
a wide variety of chemical and biological materials, the ‘brightness’ of the 280 nm band is
an indication of the clarity of the water sample being examined. When a water sample
contains more chemical ions and sub-micron particulate material, the brightness of the 280
nm band diminishes with some proportionality to the total amount of impurity in the
sample. It is likely that this affect comes from the absorption of the pure water Raman
scattering band around 3500 – 3700 cm-1 that contributes to the bright 280 nm emission
detection. This absorption may be caused both by chemical species as well as the microparticulates in the water samples that optically interact more strongly with the shorter
wavelengths in the detection band pass region of 275 to 410 nm.
Specific Conclusions:
The DART UV Fluorimeter instrument was easily deployed from the Brownlee into three
areas of the southern San Francisco Bay making water measurements between 5 and 40 ft
depths. Those measurements showed that the water in the main channel differed slightly at
the two selected locations, but that good vertical mixing was implied by lack of strong
stratification layers. The water at the third site, near the Industrial Dock area, showed
enhanced contributions of some contaminants and these conclusions were supported by
laboratory spectrofluorimeter instrument examination of the collected water samples from
several depths at each site.
Prior to the field trip we had assumed that solvents such as benzene, toluene and xylene
would be among the most likely contaminants to be detected. At the level of the laboratory
spectrofluorimeter sensitivity, no benzene-ring structured materials were detected. Some
additional testing with DART would be necessary to see if its enhanced sensitivity to
polycyclic aromatics might detect a trace level of these types of chemical compounds.