I. In The Beginning
What do you do? (when you have been using a toolkit forever, then log on to the site one day to find the following):
"The SeaWater library of EOS-80 seawater properties is obsolete; it has been superseded by the Gibbs SeaWater (GSW) Oceanographic Toolbox of the International Thermodynamic Equation Of Seawater - 2010, (TEOS-10)"(EOS-80, emphasis added). My solution was to change to using TEOS-10 for my thermal expansion and related oceanography oriented calculations (The World According To Measurements - 7).
Progress happens (such as: The World According To Measurements, 2, 3, 4, 5, 6, 7, 8, 9).
Sometimes I wonder, though, if others noticed the improvement.
That is because peer-reviewed papers relevant to the subject of thermal expansion of the ocean don't seem to mention the issue or either one of those toolkits (On Thermal Expansion & Thermal Contraction, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27).
Anyway, it is a struggle to use a vast database, as these logs show:
a Dredd Blog processing log:That is the "~billion records" I speak of every now and then.
1000_vc1 = 134,806,808
1000_vc2 = 133,628,647
3000_vc1 = 82,398,521
3000_vc2 = 80,352,621
5000_vc1 = 83,301,108
5000_vc2 = 82,227,068
7000_vc1 = 134,492,236
7000_vc2 = 133,943,219
new_vc1 = 67,156,207
new_vc2 = 67,061,513
Regular readers will know that the 1000, 3000, 5000, 7000, and "new" are WOD indicators for hemispheric quadrants: NE (1000), NW (7000), SE (3000), SW (5000).
The "new" indicates subsequent additional measurements for those quadrants about each quarter (WOD selection map, WOD updates).
The "vc1" and "vc2" indicate temperature measurements (vc1) and salinity measurements (vc2).
II. A Problem
One of several problems that are associated with the change to TEOS-10 is that the functions in the toolkit, such as gsw_alpha(SA, CT, P), require a pair or set of coherent measurements.
By coherent I mean taken at the same location, depth, and time.
Which led to the problem of organizing those ~billion measurement that way.
As it turned out, there were many measurements that did not meet that requirement.
There are "orphan" temperature and salinity measurements at various depths that can't properly be used in toolkit functions that require a group of values.
III. A Solution
The solution is to weed out the outliers and make a coherent group of measurements that will allow the proper use of the TEOS-10 toolkit.
The numbers themselves show that the CTD and PFL dataset counts are not one temperature measurement for each salinity measurement.
So, my software separates them into coherent groups while processing the WOD format into a CSV file format.
I have been perfecting that software for awhile now, and am very close to "no more problems" with that part of it.
IV. Recent Graphs
Recently we noticed that the salinity graphs revealed some radical departures from the norm of abstract maximum and minimum boundaries.
That is not unusual in a mean average situation, because there will be "the lowest" and "the highest" which will be moderated by averaging.
But as I fine tuned the pairing, some of the aberrations (but not all) subsided.
V. Today's Graphs
Today's graphs show that the salinity pair matching has improved, and I really don't expect it to change from here on out.
I am not talking about new data each quarter, I am talking about the fixed CTD and PFL datasets (1000, 3000, 5000, and 7000 quadrants).
In other words, I think the software is "there."
I did not include the S. Hemisphere ... cause I want to listen to some Santana when he was a youngster (after Mark Hanson's hot Santana session addition, I was forced to add the S. Hemisphere graphs: Fig. 3a - Fig. 3c).
Join the jam if you like Carlos:
The previous post in this series is here.