eMOLT Probe Comparisons and Calibration Results
Preliminary Comparisons of TDR probes
Formal Comparisons of TDR probes
VEMCO Fast Temperature Probe Trials
SBE37sm Microcat Probe Trials
- Nov-Dec 2000
- 2-day multi-probe test
- multi-week 2-probe test
- The results were fairly
satisfying, except the apparent bias among the TidBits of nearly
1 degree F. The VEMCO minilog is apparently offset from the
TidBit probes by nearly 0.3 degrees. Hence, more test of this sort
needed to be conducted with controlled conditions.
March-June 2001 with VEMCO, ONSET, and YSI
Results: All probes performed as expected with VEMCO being slighly higher
than the others by a few tenths
the YSI probe declined as expected due to increased biofouling in may and
june. This test proves the necessity of regular cleaning of the conductivity
cells. Several bottle samples were taken during this deployment to document
the effect of biofouling (not yet analyzed).
June-July 2001 with YSI
was collected during the Summer of 2001 after washing the conductivity
cell, but it went bad only a few days after deployment. At this point
I decided to clean off the cell and stick the probe inside the aquarium.
Overnight Ice Baths
Jan 2003 we tested 13 TidBit probes together in a large salad
bowl with ice. The probes were submerged in the ice bath at 2030
hours and then the solution was stirred at 2200 and allowed to melt until
the units were read the following morning. There results flagged one probe
#981, in particular, so we will need to check that probe especially when
it returns. There is a possibility that the time offset of the hourly reading
Feb-1 Mar 2003 we tested another 9 TidBit probes
together in a wine chiller at the Samoset Resort Fish Forum. As in the
Jan 2003 batch, the probes were let to sit overnight and read in the morning.
The results were very similar in that one probe (450046) looked suspect
but the offset may have been due to an offset in time rather than temperature.
In future downloads, strict attention (ie corrections) will be made to
time offsets as a one hour descrepancy results in large descrepancies in
the overnight changes of temperature. The other lesson learned in
this experiment is the probes need to remain in the melting ice solution
for more than 12 hours in order to experience the full range of temperature
5-6 May 2003 Patrice Farrey conducted her first icebath calibration check.
Here the probes were put in a large zip-lock bag with ice & water and
then allowed to melt overnight. Results were not satisfactory with
some probes recording as much as a degree different than others.
This procedure and all of the above were conducted BEFORE we were told
to use crushed ice in the refrigerator inside a insulated vessel.
30-31 Jan 2004 According to ONSET's suggestion,
probes were subjected to finely crushed ice inside a insulated cooler in
a refrigerator overnight. While I did not fill the container with
"distilled" water up to the level of the ice as suggested, I did find statisfactory
results with all 20+ probes recording to within 0.5 degC of 0.0 (except
one probe#321) .
- 16-19 Mar 2008 Prior to using the WHOI facility, I put 96 probes in an ice bath on mid-March with marginal success. It did find:
- Lotek probes 7195 or 7205 2-degC colder than others.
- The plot calprobe_mar08_zoom.jpg shows minilogs being 0.5 degC warmer than ONSET but that may be due to there being more buoyant?
- I tried to compare the results by the type
of probe (the 3 bold dashed colored lines represent mean for 3
different probe types) and, while I found a difference, the difference
may have been due to one or two bad probes.
- It was this experiment that made me realize I needed a more controlled experiment (Tank 2, 3, etc above).
- 26 Apr 2015
- The first version of Ruben's probe had two layers of gray PVC and therefore a long response time
- Similar results for JiM's version of the probe in one layer of white PVC when subjected to formal calibration text on 9 April 2015 at WHOI
The first test of the TidBit probes in the field demonstrated similar
results where a VEMCO & ONSET probe were set together at 120 fathoms
2000 and January
A more substantial comparison was conducted in mid-March 2001
with probes from four different manufacturers (VEMCO & ONSET)
The results of the 1-day controlled study in the NMFS Aquarium tank is
The results of the weekend-long off-the-dock study is presented here.
In both of these studies there are typical biases of a few tenths of a
degree (The only suspect probe is serial number 120 which has been
in the water for 3-months. I will need to monitor of performance
of this probe to check for drift). As in all "preliminary dockside comparisons",
the VEMCO probes evidently recorded higher temperatures than the ONSET.
At this point it was obvious that more formal test were needed. I
also had a YSI probe deployment here but did not analyze it.
After completing the "preliminary" tests described above, a set of more
formal experiments were made. The most controlled experiments were made
using Woods Hole Oceanographic's "Hart Scientific Well-Stirred High
Precision Bath (Model 7015) as described below as "tank" experiments #2-#4
below. Acknowledgements to Jason Smith (WHOI) who actually performed these tank studies on a
number of occassions.
Prior to these controlled experiments, a few tests were made as described
below were made "dockside"
as well in "overnight ice baths".
Tank 1 (29 March - 30 March 2001 with ONSET,
YSI, and SEABIRDs) including both temperature
and salinity performed in the NOAA/NEFSC
Aquarium seawater tank.
||Lab temp start*
|SEABIRD Model 39
*these lab temps are expected to vary with probes storage
location in the room but offer at least one check at set-up time.
Temperature Results: all probes performed as expected except the
YSI559 probe which had a 0.2 degree negative offset
Salinity Results: probe YSI563 is definitely suspect and YSI561 is probably
off. Conclusion: I will definitely need both NISKIN bottle samples
as well as calibration checks prior to deployment in the field, in the
beginning, with each bi-monthly battery change. These YSI6000 conductivity
sensors were evidently not calibrated during the YSI refurbishment. I need
to buy "conductivity reagent" to do so.
- Tank 2 (27 March 2009)
- Tank 3 (10 March 2010)
- first time where I looked closely at just the period of controlled temps as documented by Jason Smith's email a few days later
- documented steps in emolt_routine.html
- found 1499 way off so I removed this from analysis (down to 64 probes tested)
- Probe Type# 1 mean error = -0.10475 degC, max = 0.1103, and std = 0.066974 TIDBIT
Probe Type# 2 mean error = 0.067937 degC, max = 0.45067, and std = 0.019795 VEMCO
- calprobe_mar10_fig1.jpg shows two ONSET probes being colder than one stddev (dashed black) so this may have cause the mean result in figure 2
- Overall, VEMCO probes had the greatest deviations from the mean while TIDBITs had the greater overall error
- Tank 4 (12 April 2011) WHOI
- Had the probes recording hourly and soaked at each temperature for 1:20
- Found old probe #1673 had wild values (>2std) at most temps and #4519 and #4523 had wild values at a few temps
and should be watched closely in the next bath
- Line plot shows probes aligning fairly well and marker plot shows the overall results
- Aside from the near-freezing temperature, probes were in general
within a few hundredths of a degree and the standard deviation was
less than or equal to 0.16 degC.
- Tank 5 (27 Mar 2013) WHOI
- email from Jason Smith on 29 Mar documents the bath temps
- VEMCO Minilogs-1 did perform well within the 0.2 degC spec as more clearly pictured here.
- Tank 6 (9 April 2015) WHOI
Temperature Probe Trials
The VEMCO Fast Temperature Probe was deployed off the Glenna Jacob in late
July 2001. Seventy two cast were conducted over several sections across
the southern side of Georges Bank as part of the first Fleet Link demonstration
project. The first (very preliminary) look
at the data indicates the probe functioned as expected. eMOLT
participants interested in using this new probe should be encouraged to
- Tom McVain (Long Island, ME) deployed this probe on several occassions with some success seen in both profile and transect form.
SBE37sm Microcat Probe Trials (reported to SEABIRD
The first problem I had was apparently due to a dirty conductivity cell.
Data from the very first overnight deployment of the Microcat looked
good but the subsequent week-long deployment was not good. It is apparently
important that the conductivity cell be cleaned well after each use.
When I redeployed the instrument in the NMFS Aquarium for nearly a week,
it took that much time to flush out the apparently contaminated electodes.
This may have been due to my sticking it a 5 gallon bucket
with seawater that may have previously contained some chemical substance.
Or it may have been due to my immersing it in some dirty/oily Woods Hole
Harbor seawater. The week long deployment shown here
has suspect data and not in agreement with the
YSI6000 record obtained along with it. After a call to Dave Armstrong
at SEABIRD and a 10 minute flushing of the cell with fresh water, I redeployed
the instrument in the aquarium again. Armstrong suggested that the "electrodes
need exercising" . Fortunately, the third deployment indicated reasonable
values. All three deployments are shown
The second problem had to due with an apparent SEABIRD software bug. The
seaterm.exe routine offers a "yearday" variable to be exported in the cnv
files. First, it looks to me that it may be erroneously treating
year 2001 as a leap year since it added an extra day. Second,
one must be careful in defining which "yearday" is reported whether it
is a yearday that starts at zero or one. I was assuming it was going
to report the former when it appears to report the later. So, in any case,
one should be careful to check the validity of the "yearday" column in
the cnv files.
The third problem was a bad year in the ds display of seaterm header.
In late Fall 2001, I deployed the unit in two different tanks within the
NMFS aquarium and found that the salinity records were fouled
soon after I put it in the "resovoir" tank. So, in Jan 2002,
after a thoughough rinse, I redeployed the unit in the original "lobster"
tank and then in May 2002 to the "crab" tank.
Having solved the above problems after nearly a full
year of continuous sample testing, the instrument was then
delivered to lobstermen David Johnson in June 2002.