Bering Strait 2013 Mooring Cruise CTD Data Archive
==================================================
Contact:  Rebecca Woodgate, woodgate@apl.washington.edu
          Cecilia Peralta-Ferriz, ferriz@apl.washington.edu
Version: 13th Oct 2015.

1) Platform: research vessel Norseman 2

2) Project Title:  Bering Strait

3) Data collection dates:  4th-10th July 2013

4) PI: Rebecca Woodgate (woodgate@apl.washington.edu)

5) Data collection method: SBE19plus CTD unit with six 4-litre
bottle rosette on conducting cable
(see below for full package details)

6) Data calibration method: manufacturer's pre-cruise calibration.
Processing described below, along with data format and known
data issues.  

7) Other related data sets:
CTD data have been collected (mostly annually since 2000) in the 
Bering Strait region as part of the Bering Strait mooring 
project.  For overview,
see psc.apl.washington.edu/BeringStrait.html, or data archives
at NODC.  

8) Conditions for use or citation:
Data freely available.  Please, as a courtesy, contact the PIs
before using these data.

9) Data qualifications or warnings:
As discussed below, downcast oxygens and salinities, and upcast
salinity concurrent with sharp temperature gradients, are dubious
in quality. 

10) Grant numbers: NSF ARC 0855748

11) Sample repository locations:
        National Ocean Data Center (http://www.nodc.noaa.gov)
        University of Washington (http://psc.apl.washington.edu/HLD)

Summary
-------
This archive is of data from 150 CTD casts taken during the 2013 Norseman2 
cruise to the Bering Strait.
For positions, see file headers or the cruise report accompanying these
data.

Data were preliminary processed at sea, as per cruise report, and bottle
files were obtained from this processing.
Subsequently, significant issues were found with the data as regards to 
lack of reliability of salinity and oxygen on the downcasts.  A reprocessing
(described below) attempts to address these issues. (Bottle data
were not reprocessed.) 

This archive includes:
== the raw data files (.hdr, .hex. and .XMLCON) for each cast,
so the user may reprocess the data to their own standards.
== 2db average UPCAST data for each cast, as per Oct 2015 data
processing.  As will be discussed below, these data provide (we believe) 
the most reliable salinity and oxygen data (although note no oxygen 
or salinity samples were taken for validation).  
** The 2db resolution reflects uncertainty in the depth of the data 
as the upcast data are being used. See further discussion below. 
** Casts with particularly suspect salinity data at times of sharp
temperature gradients (Casts 83, 90,91,94,100,101,103) are placed in
a subdirectory. 
== the .ros and .btl bottle data as obtained from the processing at
sea.  We encourage those seeking parameters to match bottle samples to
consider also the 2db averaged cast data.

Summary of known problems
-------------------------
Salinity values at times of sharp temperature gradient.  
Salinity and oxygen values on the downcasts.
See notes below. 

Funding Source:
---------------
This work was funded by the National Science Foundation 
Office of Polar Programs, as part of the Arctic Observing 
Network, grant ARC 0855748, "An Ocean Observing System
for the Bering Strait, the Pacific Gateway to the Arctic
- an integral part of the Arctic Observing Network"
PIs:  Rebecca Woodgate (UW), Tom Weingartner (UAF),
Terry Whitledge (UAF) and Ron Lindsay (UW)
UW=University of Washington, Seattle.
UAF=University of Alaska, Fairbanks.


CTD setup
---------
Configuration for Bering Strait 2013 cruise:
This instrument package was kindly loaned to us by
Peter Winsor, University of Alaska, Fairbanks.  
(Instrument calibration dates given in parentheses.)
 SBE19plus V2 Seacat CTD SN 6849 with strain gauge pressure (Calibration 21st Jun 2011)
 SBE43 Oxygen SN2136 (Voltage 0) (Calibration 21st Jun 2011)
 Wetlabs WETstar Fluorometer SN:WSCHL-1404 (Voltage 1) (Calibration 1st Jun 2011)
 Teledyne Benthos Altimeter:7601 SN53448 (Voltage 6)
 SBE 5T Pump
 SBE 55 Ecowater sampler (SN81) with six 4-liter bottles with internal rubber bands (Part 801714)
 SBE 33 Carousel Deck Unit
 NMEA input from the Norseman II, Aft-A-frame GPS

For deployment details, please see cruise report. 

Data citation
------------- 
Please cite these data as:
Woodgate, R.A., and the Bering Strait 2013 Science Team, 2013,
Bering Strait Norseman II 2013 Mooring Cruise Report, 67pp,
available at psc.apl.washington.edu/BeringStrait.html

Data processing and data issues
-------------------------------
Data were processed using the SBE standard software SBE Data
Processing version 7.21f, as per the details below.

== ** Known problems:
1) Salinity data from the downcast were frequently found not to
match salinity data from the upcast near the surface (upper
10-20m), differences sometimes up to ~0.5psu.
2) Oxygen data on the downcast were always significantly 
(10-20% oxygen saturation) lower than on the upcast in
the upper 20-30m.
3) Unrealistic salinity values were evident at times of 
traversing steep temperature gradients.  
4) Fluorescence data have (as described below) been despiked,
but are not further quality controlled. 

-- Re issues 1 and 2
On one cast, cast8, the CTD was yo-yoed, and the agreement of
salinity and oxygen on the 2nd lowering and raising of the
CTD with the first upcast suggest that:
a) upcast data are reliable, 
b) the problem arises from either from air trapped in the
system (e.g., due to insufficient wait time and/or a clogged 
vent hole) or lowering the CTD before the pump came on
(the pump delay is known at time of writing).
 
Thus, for the final data product, we use the upcast data.

However, analysis of the depth of the maximum temperature
gradient in upcasts and downcasts suggest a frequent
bias with the upcast maximum gradient being often ~ 2m 
shallower than the downcast maximum gradient, consistent
with the sweeping of water upwards as the CTD is recovered.
However, it is unclear if this (or spatial/temporal variability)
is the source of the discrepancy since, the problem is 
not ubiquitous to the data set (~ 20 casts show no bias, 
~ 20 casts show 1m bias, and ~ 20 casts show 2m bias); 
the problem is not more common on casts without bottle
stops; and during bottle stops there is only some vague
evidence of temperatures relaxing to the downcast data.
We conclude thus that SOME upcasts maybe prone to this issue,
and to represent this, we provide bin averaged data only
at 2db resolution.  

-- Re issue 3
Attempts were made during processing (as per SBE notes) to
remove this salinity spiking, which is typically due to
a timing mismatch between the temperature and conductivity
sensors.  However, as discussed below, it was found that
the optimum timing delay varied over the course of the cruise,
and an entirely satisfactory solution was not found.
The 2db averaged upcast data (which have been filtered as
per details below) generally do not show this problem, but
care must be taken interpreting salinity data near sharp
steps in temperature. The 7 most troublesome casts (casts
83,90,91,94,100,101,103) are in a subdirectory of this archive
to alert users to this issue. 

Details of data processing
--------------------------
Data were processed using the SBE standard software SBE Data
Processing version 7.21f, as per the following steps and options.
Unless otherwise stated, options were standard SBE defaults.

Step1: Data conversion
- uses .hex data and instrument calibration to obtain
calibrated values for Pressure, Temperature, Conductivity,
Raw Oxygen, and Fluorescence.

Step2: Filter
- performs low-pass time filtering on the data.
As per recommendations for SBE19plus V2, 
pressure data are filtered at 1s;
temperature, conductivity, and oxygen are filtered at 0.5s.

Step3: Align
- aligns temperature, conductivity and oxygen data with pressure.
Temperature is advanced relative to pressure by 0.5s, as per
standard recommendation for SBE19plus V2.
Conductivity is retarded relative to pressure by 0.25.
(i.e., setting is -0.25). Manufacturer's recommendation 
is to experiment with different values.  The values used
for this data set are those that minimize the sum of density 
inversions within each cast for the entire cruise.  
Note that for some casts, zero advance, or a retarding 
of 0.5s or 0.75s gave better results. 
Oxygen is advanced by 3s. Manufacturer's recommendations
are between 3s and 7s.  Due to oxygen issues on the downcast,
only cast 8 could be used to test this value.  That test
showed 3s was the best choice between 0s and 7s.
Fluorescence was advanced by 4s as per manufacturer's
recommendation, without further testing.

Step4: Cell Thermal Mass
- corrects for Thermal Mass of the cell. 
Thermal anomaly amplitude (alpha) taken at 0.04 and
thermal anomaly time constant (1/beta) taken at 8,
both as per manufacturer's recommendations for 
SB19plus V2.

Step5: Loop Edit
- flags as bad, data collected when descent/ascent
rate is less than a certain value.  (Can also
remove surface soak, however soaking routine
for these casts was to wait for ~ 1min at 2m and
then lower, and this soak can only be removed via
ascent/descent rate.)
Manufacturer's recommendation for descent/ascent
rate cut off is 0.25m/s, 25% of typical lower speed (1m/s).
However this removed 80% of our data.  
Since our descent rate was more typically 0.3m/s, 
we use instead a cutoff of 0.1m/s which, inspection shows, 
removes surface, bottom, and bottle waits, and looping 
during up and down casts. 

Step6: Derive
- computes derived variables of salinity, density,
Oxygen (ml/l and % saturation)

Step7: Window Filter
- despikes using a median filter. We chose here to
use an 11pt filter, which (with a drop rate of
0.2-0.4m/s and 4Hz sampling) is equivalent to
a 0.5-1m smoothing.  This filtering was performed
on temperature, conductivity, salinity, density,
all oxygen values, fluorescence, and latitude
and longitude.  Note this filter only works with
points flagged good. 

Step8: Bin Averaging.
- are bin averaged into 2db bins.  Given surface
issues and that the surface soak was at 2m depth, 
a surface bin was not derived for these data.

Final data accuracy
-------------------
Although manufacturer's specifications for instrument 
accuracy (including initial accuracy after calibration and
likely drift over the 2 years since calibration) 
are ~ 0.01degC for temperature, 0.007S/m for conductivity, 
0.3% of full scale range for pressure, and 0.3mg/m3 
(not including drift) for fluorescence,
effectively the data are far less accurate than this,
due to issues discussed above and the natural variability 
of the system, i.e., how well the data from the cast 
represent the real ocean at that time and place.  
A rough estimate of accuracy can be made by considering 
the difference between the good portions of
the up and downcasts.  In general up/downcasts
agree to within:
  Pressure ~ 2db
  Temperature ~ 0.05degC
  Salinity ~ 0.01psu (larger discrepancy near sharp temperature steps)
  Oxygen ~ 1% oxygen saturation
  Fluoresence ~ 5mg/m3

For an overview of previous and on-going Bering Strait mooring work,
please see http://psc.apl.washington.edu/BeringStrait.html.

For queries, please contact:
   Rebecca Woodgate   woodgate@apl.washington.edu   (206) 221-3268
   Polar Science Center, Applied Physics Lab, University of Washington
   1013 NE 40th, Seattle, WA  98105-6698   USA      FAX (206) 543-6785

