US JGOFS Antarctic Environments Southern Ocean Process Study (AESOPS)


Beam Attenuation Coefficient, Fluorescence, Light Scattering,
PM filtration protocols

Wilford Gardner, Mary Jo Richardson
Texas A&M University


Data Reduction Scheme

Beam attenuation via transmissometers

The primary purpose for measuring beam attenuation in JGOFS programs
is to determine the concentration and distribution of particulate
organic carbon (POC) or particulate matter (PM) in the water with
continuous profiling rather than with limited discrete samples.
Towards this end, a 25-cm path-length Sea Tech transmissometer was
interfaced with the R/V Palmer's SeaBird CTD for all Ross Sea cruises.
During the Polar Front cruises, a new model of SeaTech transmissometer
(pathlength=20 cm) was interfaced with the R/V Revelle's SeaBird CTD.
In addition, during the final process cruise (Process II, RR Kiwi-09)
D. Stramski and G. Mitchell (SIO) provided two WetLabs transmissometers
(660 and 480 nm) that were also interfaced with the SeaBird CTD for
casts from station 9-30 where the cast did not exceed 1000 m. Beam
attenuation data posted here for stations 9-30 (<<1000 m) of RR
Kiwi-09 were from the 660 nm WetLabs transmissometer because it was
more stable in the surface waters. For the casts deeper than 1000 m,
we provided an earlier model SeaTech transmissometer with 5000 m
capability, and those SeaTech data are posted in the data base. 
Transmissometer data were analyzed for the seven Palmer cruises
(NBP9604, NBP9604A, NBP9605, NBP9701, NBP9703, NBP9708 and NBP9802),
and the four Revelle cruises (Kiwi 06-09).  Data from the raw CTD files
were binned at 2 db intervals through SeaBird's SEASOFT program, which
has a spike removal subroutine that we have tested and found to remove
transmissometer data spikes properly. Beam transmission was converted to
beam attenuation coefficients using c = -(1/r)*ln (%Tr/100) where
c = beam attenuation coefficient (m^-1), r = beam path length (m),
and Tr = % beam transmission.

In order to determine the attenuation specific to particulate matter
(beam cp), the attenuation due to water is subtracted from the beam c
values (cp = c - cw). In principal, cw should be 0.364 for the SeaTech
transmissometers since they are set at the factory to read 0.364 in
particle-free water.  Practically, cw is determined as the minimum
attenuation measured during each cruise. It must be noted that this
minimum attenuation value is the "cleanest" water observed and is not
particle free. The WetLabs transmissometers are designed to yield cp
directly with final calibrations provided by R. Reynolds (SIO). To
match the cp values of the SeaTech and WetLabs profiles on simultaneous
casts, an additional correction of 0.0151 per m was required.  


The extreme cold conditions of the Southern Ocean and Ross Sea
presented some challenges because it is not practical to do a proper
bench or air calibration prior to each CTD cast, thus requiring other
calibration methods. Relative calibrations between casts can be made by
comparing the beam attenuation at a depth where the particle
concentration is relatively invariant - usually deeper than 300 m. The
primary concern is ensuring that the optical windows are uniformly
clean. Linear offsets in beam cp can be made to the entire profile when
clear-water values differ significantly. Occasionally the CTD casts
extended only to 150 m or less, which was usually shallower than the
particle minimum. The stations during some cruises covered a wide
geographic area, especially on the Polar Front cruises, so it was
possible that the particle minimum at depth could vary.  Washing the
transmissometer windows just before launching the CTD apparently caused
ice to form in the subzero air of the Ross Sea, and the ice sometimes
took many meters to thaw, making some surface data unusable.


The primary method for correlating beam cp with the concentration of
particulate organic carbon (POC) or particulate matter (PM) was by
linear regression of the data.  

Bottle POC concentrations


The concentration of particulate organic carbon was determined by
Walker Smith's group after filtration of water samples through a GFF
filter (0.7 m nominal pore size).  (Please see Chapter 15 of the JGOFS
protocols (1994) for POC methods.) POC was filtered on the four process
cruises in the Ross Sea and on the four survey and process cruises in
the Polar Front. Bottle samples were obtained primarily in the upper
200-250 m, so extrapolation to deeper depths is less certain.  In the
Ross Sea there were often bottom nepheloid layers of resuspended
sediment and the POC equations cannot be used to estimate POC
concentrations in those areas.


Bottle PM concentrations


Water samples were vacuum filtered (0.5 atm) in-line from the
Niskin bottles through pre-weighed 0.4 um pore size Poretics filters
and washed with three aliquots of either distilled water or 1M ammonium
formate, dried and returned to the lab for reweighing. Extensive
experiments during NBP9701 showed no difference between washing with
distilled water versus ammonium formate. Therefore all PM regression
data are from the distilled water washes, as in all other JGOFS PM
measurements. PM was filtered on one cruise in the Ross Sea (NBP9701)
and on the two process cruises in the polar front.


Regressions against beam cp


The beam c data for those bottle depths (chosen as the beam c
value of the 2 db bin within which the sample depth fell) are then
regressed against POC or PM using a Model II regression to determine
the slope of the regression. (Model II regressions minimize the
deviation of data points from the regression line in both the x and y
direction; Model I regressions minimize deviations in only one
direction.)


POC Concentrations

A prediction of the POC concentration (g/l) can be obtained
from the resulting equations for each cruise:

Ross Sea

     NBP9604a  -> POC = 593 * cp + 7.56       (r^2 = 0.710)

     NBP9701   -> POC = 635 * cp + 63.1       (r^2 = 0.818)

     NBP9703   -> POC = 752 * cp + 0.782      (r^2 = 0.065)

(Use of the NBP9604A equation would probably be better for NBP9703)

     NBP9708  - > POC = 519 * cp + 24.7       (r^2 = 0.878)


Polar Front

     RR_KIWI06, Survey 1  -> POC = 645 * cp + 10.3      (r^2 = 0.822)

     RR_KIWI07, Process 1 -> POC = 412 * cp + 8.69      (r^2 = 0.899)

     RR_KIWI08, Survey 2  -> POC = 367 * cp + 36.5      (r^2 = 0.839)

     RR_KIWI09, Process 2 -> POC = 510 * cp + 40.7      (r^2 = 0.749)


	POC is in ug/liter, and cp is attenuation per meter.


	Note that these are Model II regressions .


PM Concentrations


 Ross Sea

A prediction of the PM concentration was determined from
filtration during only one cruise, and is the only estimate for
determinations of PM concentration during other Ross Sea cruises:


   NBP9701 -> PM = 890 * cp - 5.02   (r^2 = 0.926) Distilled water rinse


 Polar Front

   RR_KIWI07, Process 1       -> PM = 1395 * cp - 21.5     (r^2 = 0.912)

   RR_KIWI09, Process 2 So    -> PM = 1771 * cp - 23.2     (r^2 = 0.956)

   RR_KIWI09, Process 2 No    -> PM = 666 * cp + 5.93      (r^2 = 0.896)

   RR_KIWI09, Process 2 Sta 15-> PM = 867 * cp + 3.82      (r^2 = 0.980)


	PM is in ug/liter, and cp is attenuation per meter.


Note that data for Process Cruise 2 were separated into the Southern
transect (So) and Northern transect (No) because of very different
relationships. Station 15 (part of the northern transect) was given its
own correlation and was not used in the Northern regression. These
differences were not seen in the POC or Fluorescence data.


Chlorophyll


Chlorophyll-a fluorescence distribution in the Ross Sea was determined,
in-situ, with a SeaTech Fluorometer. The fluorometer was interfaced with
the Sea-Bird CTD, and the data were acquired in the same format as the
transmissometer data. The Fluorometer is a standard irradiation/emission
system. When chlorophyll a is excited by blue light (425  nm), it will
fluoresce at a peak wavelength of 685 nm (red light).  The emission
detector is filtered to a peak response in order to make the measurement
insensitive to the excitation source. The amount of fluoresced light
detected is converted to a voltage range of 0 to 5 volts. The fluorometer
is set to sample with a three-second-time constant to smooth the data.
A baffle has been placed in front of the emission detector in an attempt
to make it insensitive to ambient light (SeaTech Fluorometer Manual).
The SEASOFT software converts the measured voltage into a relative
chlorophyll-a value using the equation:


       [volts * signal gain/5] + offset = mg chl-a m^-3


These relative values were calibrated using discreet chlorophyll
samples (taken by scientists from the Smith/Marra/Barber primary
productivity team and analyzed onboard the ship using a Turner
Fluorometer). Linear regressions between fluorometer-determined
chlorophyll-a fluorescence, and the chlorophyll-a concentrations
determined using a Turner fluorometer were very good when
concentrations were high (r^2 = 0.88), but were sometimes very poor
during periods of low concentrations (r^2 = 0.06). Regressions were
made for each cruise individually. Regressions between chl a
concentrations and fluorescence (Fl) are:


 Ross Sea

     NBP9604a -> Chl a  = 2.10 *Fl + 0.0395    (r^2 = 0.873)

     NBP9701  -> Chl a  = 0.665*Fl + 0.363     (r^2 = 0.876)

     NBP9703  -> Chl a  = 2.27*Fl - 0.0125     (r^2 = 0.061)

(Use of equation for NBP9604A would probably be better for NBP9703)

     NBP9708 ->  Chl a  = 0.908*Fl + 0.397     (r^2 = 0.88)


 Polar Front

During the Polar Front cruises in-situ fluorescence was measured
with a Chelsea fluorometer interfaced with the CTD. The Chl/Fl slopes
are much greater with the Chelsea instrument. Comparison of the bottle
chl a determined with a Turner fluorometer and the HPLC bottle data
showed a trend of decreasing Turner fluorescence during the sequence of
cruises. No such trend is apparent in the Turner fluorescence versus
Chelsea in-situ fluorescence (i.e. no temporal trend in the slopes
listed below). The Model II regressions are as follows:

     RR_KIWI06, Survey 1  -> Chl a  = 11.7 *Fl - 0.160    (r^2 = 0.884)

     RR_KIWI07, Process 1 -> Chl a  = 12.9 *Fl - 0.180    (r^2 = 0.807)

     RR_KIWI08, Survey 2  -> Chl a  = 8.67 *Fl - 0.119    (r^2 = 0.849)

     RR_KIWI09, Process 2 -> Chl a  = 11.7 *Fl - 0.217    (r^2 = 0.707)


LSS - SeaTech Light Scattering Sensor


Light back-scattering due to particles was monitored using a SeaTech
Light Scattering Sensor (LSS). The LSS projects light from two 880 nm
(infrared) LEDs into a sampling volume that varies depending upon the
concentration of particulate matter, but the volume is roughly the shape
of a stretched balloon. Back-scattered light from the particulate matter
is measured by a detector next to the LED. The range on the LSS was set
to 0 - 33 mg/l. The amount of light detected is scaled to a 0-5 volt
output, but most values were less than 0.5 volts.  The LSS output depends
upon the nature of the particulate matter and will vary with changes in
particle size distribution, shape, index of refraction, organic/inorganic
content etc. Therefore the LSS requires site-specific calibration. The
LSS was interfaced with the SeaBird CTD and the data were handled in the
same format as the transmissometer and fluorometer data. The LSS
malfunctioned occasionally during the later Ross Sea cruises and those
profiles were eliminated. For the Revelle cruises in the Polar Front, an
LSS was mounted inside the rosette pointing upward at the lanyards used
to close the bottles. The lanyards were in the LSS sensing volume, so the
absolute values of the data are not directly comparable with the Ross
Sea values, but the values among Polar Front cruises are intercomparable.
These LSS data can be regressed against particle concentration as with
POC or PM to make comparisons between Ross Sea values and Polar Front
values and to obtain continuous estimates of POC or PM.