Blogpost Header

A "Wales" of a Good Time

Img 0337 Scaled
Figure 1. View from a Welsh promontory

Sometimes we get bogged down by the fishbowl effect.  Looking around your fishbowl at the other fish as they look back at you.  You swim around in a circle, they do the same, you’re staring at them again a second later.  Sometimes it’s nice to be put in a little baggy and taken halfway across the world and dropped in somebody else’s fishbowl.  The Natural Resources Wales ( fishbowl is very fine, shown in Figure 1.  This is after Day 2 of the 2019 Wales International Salt Dilution Workshop organised by Paul Griffiths of NRW.  The event brought together ~55 hydrometric professionals from 10 countries (Wales, Scotland, England, Norway, Sweden, Iceland, New Zealand, Canada, Austria, and France) plus Nick Everard.  It was an amazing 3 days of well organised presentations, socialising, hiking, drinking, field demonstrations, workshopping, drinking, and drinking.   It featured the beautiful countryside around the Brecon Beacons, idyllic babbling brooks, idyllic babbling presentations, and traditional Welsh public house merrymaking.
This wee memo documents the presentations to the best of my ability, and the subsequent shake-down, slam-down flow measurement throw-down pitting the plucky QiQuac (Quick Instream Q and Uncertainty Analysis Calculator) against the brutish and mercenary Sommer Tracer Q system.  As Christoph Sommer said, I did cheat wildly and yet, still, the two answers were surprisingly close, although I cannot explain the small difference between the two results:~5% different with the TQ 71 lps and the QQ 75lps when the probes were placed side by side.

Day 1: An Unexpected Party

Figure 2. Daniel and Mikael working the Welsh Fishbowl

The event began on hospitable grounds, much like when Bilbo welcomed the first few dwarves into his hobbit hole, with a welcome to the land, grand gestures of hospitality and goodwill, and some cheerful case-studies. Possibly the most entertaining presentation was by Daniel Wennerberg and Mikael Lennermark (Swedish Meteorological and Hydrological Institute) as they described the trials and tribulations of gauging in cold conditions. They had perfect comedic timing with Daniel playing the straight man and Mikael making wild conclusions and proclamations and waiting a second after each to see if anyone objected, like this: “The salt we buy in 1kg boxes is +/-1%, therefore this is acceptable, ok?” One-one thousand… They started to move on to their next assertion as though this was the court of public opinion, until I disagreed.  I said boxes can lose salt, more than 3%, and I’ve seen up to +3% overweight. So while using the store bought weight is possible in a pinch, it’s adding 3% uncertainty to the final Q.

Figure 3. Relationship between CFT and Background EC.T. The average is 0.486 [mg/L]/[µS/cm] +/-2.8%. There is a small but statistically significant dependence on BG ECT of 0.003%/[µS/cm] or 0.3%/[100µS/cm]. (Richardson, Sentlinger, Zimmerman, and Moore 2017)

They also wondered how some CF.T values were so low and they figured that as their solution froze, it left more salty solution in the jar, which they injected.  A saltier solution would result in larger steps and a lower CFT.  When the ice thawed in the standard, a less salty standard remained.  This in turn would result in a higher CF.T (smaller steps).  This is partly why we published “Quantifying the Relation Between Electrical Conductivity and Salt Concentration for Dilution Gauging Via Dry Salt Injection” to understand what natural variability in the CF.T to expect.  Through the Salt Portal, we’re also building a global database of CF.T values.  Just like in Hitchhiker’s Guide to the Galaxy, the  answer is 0.486+/-2.8%, and shown in Figure 3.
For any given site, the uncertainty is +/-2% of the site average, and if a calibration is performed under best case conditions, it can be as low as +/-1%, or in less than ideal conditions as high as +/-4% uncertainty.  From this last statement you can see the value of not performing the CF.T calibration each time as best-case conditions are not possible everytime by definition—less than half will be below 50% percentile conditions.

Figure 4. Roger gave a slide show of him “working” followed by 9 mins of photos of him at various pubs throughout the UK. Still, it was one of the best presentations.

Roger Bailey put on a very entertaining show to describe the value of standards and announced that the ISO would be looking to revisit ISO9555-3: Measurement of liquid flow in open channels — Tracer dilution methods for the measurement of steady flow — Part 3: Chemical tracers.  This workshop was the first stop on this long train ride bound for the glory of the next version of the standard. Surprisingly many people wanted to sign up for this grueling ride, which seemed to surprise Roger.  “It’s not fun, you know,” he’d implore people, “It’s often like pulling teeth and herding cats, while watching a kettle boil!” he said in his exasperated yet extremely engaging manner. “I’m interested, go on,” the listener would invariably say.  “You will hate it, you will pull out your hair, you’ll age prematurely, I’m 38!”  “Hmm, fascinating, sounds rewarding,” at which point Roger would throw up his hands and walk away.

Figure 5. King Nick holds court

While Nick Everard only did a quick show and tell of their new radio controlled ADCP boat (a joint effort between the UK-EA and HR Wallingford), it was still the highlight of the workshop, and wasn’t salt dilution related in the least.  In classic King Nick fashion, he declared to the group “Welcome! And thank YOU for attending my workshop,” Paul sighed loudly here, “I realize it’s a long way to travel to see me and the new Royal Yacht but you won’t be disappointed.  Behold! (here he quickly pulled a black velvet cloth from the ADCP boat). You’ll appreciate the new quick release handle on the bow (he demonstrated several times with raised eyebrows each release), the smaller propellers completely protected from damage, (he  spun them) and best of all, my handsome visage engraved on the transom.  Ha ha!”  The crowd clapped somewhat politely, but not enthusiastically.  “Thank you!  I declare this workshop Initiated!”  Paul rolled his eyes.
Evan Baddock presented a fascinating study of the AutoSalt system (made by Fathom Scientific) installed at Rollergate.  With his signature wit and charisma, he casually strolled about the front of the room and the crowd (especially the ladies, but also several men) were lost in his Jimmy Stewart charm.  The very strange and very high EC signal during low flows at Rollergate emerged as a point of discussion for the group.  The observation that the natural EC variation is much larger than those typically associated with the AutoSalt injections was noted.
Sam Everitt also presented automated flow measurement results from the RWT based AutoQ from Norway and the AutoSalt system from Fathom. With a keen eye for efficiency and total lack of regard for aesthetics, Sam and Andy transplanted the brains and guts of the AutoSalt into their own Frankenstein version.  The nice thing about it is the portability — requiring just 1 person to carry it to site for a quick deployment—and the adaptability – allowing Andy to purchase off the shelf plumbing.  When Sam flashed our patent-pending sprinkler pre-mix system on the screen there was a outcry from the Norwegian contingent. “This is an abomination!” Knut cried, “You’re masking the poor mixing of the site with this contraption. Ridiculous!”
Sam waited a moment for the unrest to subside, and in his calmest of tones, explained that results improved instantly when deployed.
“Its an experiment,” I further offered. But the Norwegian contingent remained unconvinced, and the tension in the room,  palpable.

Figure 6. NVE Figure comparing all Gauging Results (>3000 of each) to their respective Rating Curve as a measure of repeatability. This is essentially saying Salt is as consistent as the Streampro, more so than the Current Meter (Unspecified) but less so than the RioGrande.  That may also indicate gauging conditions, but this is a bundled comparison of conditions, equipment, methods, etc.

Frode from Norway presented some amazing results of their dilution gauging program. Over 30% of the gaugings in Norway are from salt dilution. He did some very clever things comparing SD results to existing rating curves.  He compared >3000 SDIQs with relevant Rating Curves, and other concurrent measurements and found no significant bias, and a repeatability (consistency is correlated to accuracy) on par with the StreamPro.  Figure 6. demonstrates this comparison.  If only these analytics could be automated somehow. Hmmm….
Day 1 evening event was a fabulous dinner at  the pub with way too much food and drink.

Day 2: SDIQ Showdown: There are No Shadows at High Noon

Figure 7: Brecon Reckoning. The crowd that had assembled expected a spectacle.

On Day 2 the three suppliers at the event, Fathom Scientific, Sommer, and NVE all presented their products in the classroom. After my presenation, I went out to upgrade the AutoSalt that Sam and Andy brought; that’s what I told people, but I was really going out to cheat a bit,  get an upper hand in the coming showdown…

Figure 8: Christoph loves this part of the job.

I must explain that when the group came out, I had just made an injection of 484g with the AutoSalt system. We then injected 300g, but I forgot to  update the mass in the QQ. When I reported 0.120 cms to the first group, it was the wrong mass.  Using 300g, I get 0.077 cms.  The average Q declined to 0.076 cms then 0.075cms on the last measurement.  I don’t know the first two Sommer measurements, but it was 0.071 cms on the last, a difference of 5%.   I still can’t explain why they are 5% different, as they should not be.
You can examine any of these measurements by using Chrome, Internet Explorer, or MS Edge, and going here:
user: NASH.user
password: Garwnant
Feel free to play around, but don’t save either the discharge or the CFT.  And Christoph is not allowed to enter the site and take notes.

Table 1. Results from Wales SDIQ Workshop.  Click to enlarge.
Figure 9. QQ CH0 Traces showing the rising BG ECT Slope. The last four pulses are when the group arrived.

Table 1 shows the measurements, starting on the day prior to the workshop.  The recent rainfall resulted in larger Q (0.403 cms) and lower BG EC.T (66 uS/cm) on the 17th.  On the 19th, the BG EC.T had risen to 131 uS/cm at the start of the day and 134 uS/cm at the end.  The sloping background can be seen in Figure 9 for the QQ Ch0.

Figure 10. The result of Kristoffer splashing around upstream of the CH0 probe.  Not sure why it dropped, then climbed up again.

The second injection was 300g. The 3rd 100g. Before the 3rd came down, Kristoffer asked if he could stir up the sediment just upstream of the probe.  I said “No, please don’t,” but he went ahead and did it anyways, and the trace in Figure 10 resulted. It seems odd that the EC.T would drop, then recover.  This SDIQ was aborted before calculating a Q.

Figure 11: Sommer TracerQ traces showing rising BG EC.T when 100g is injected on the last measurement. If the slope is not considered, the resulting Q is 65 lps; if it’s considered the resulting Q is 71 lps.

At this point the QiQuac seemed to be consistently giving higher flows (“We give you more water!” I exclaimed proudly) but there’s no reason why this should be. So for the fourth injection, Christina and I were keen to place the probes side by side and we injected 100g. Initially the Sommer kit was giving 65 lps but we then realized that the BG EC.T was actually climbing due to the receding  rainfall runoff from the previous day.  Therefore using the pre BG EC.T, the default Sommer behaviour, was resulting in an overestimation of the area and underestimation of Q.  When the sloping BG EC.T was entered the Q was steady at 71 lps. The QQ gave 75 lps however and I don’t know why they were different.
The QQ probes were on the stream channel bottom, while the Tracer Q probes were suspended mid-channel.  Looking at the last injection again on the Salt Portal, for CH1, it seems the 3rd injection may have still be coming downstream resulting in a falling BG EC.T which is not present in the suspended probe.  This may explain the small difference, although I’m not sure why. The

Figure 13. Closeup on BG EC.T on CH1. It was moved from downstream to upstream and left on the streambed. The sharp drop at :20 is likely due to moving upstream where the 3rd injection had already passed along further, but clearly has not completely flushed out of the system, resulting in a locally falling BG EC.T.
Figure 12. Last injection showing the falling BG EC.T due to the remaining tail of the 3rd injection in the QQ CH1 record.

high resolution of the T-HRECS sensor is on display in Figure 13 with a stability (Coefficient of Variation) of only 0.02%.  At 134 uS/cm this is a standard deviation of 0.03 uS/cm. This allows us to see features otherwise obscured in the coarser resolution of conventional sensors.  For example the small dip in EC.T just prior to the salt wave is apparent and the slightly falling BGEC.T from the previous injection.  If we can start to understand what causes variations in BG EC.T we can make better measurements using less salt.
After these harrowing Day 2 exertions, Paul wisely thought it best to spend some time in the mountains.  There are a lot of great photos from this excursion and some of them are shown below.   The hike was a great way to get some fresh air, stretch the legs, and sample some of Paul’s Scotch which somehow he won for best picture?  I wasn’t clear on how he ended up with it, but it was nice touch to fantastic day of workshopping and a vigorous hike up the spine of the sleeping dragon.

Day 3: Further Discussions

On the last day, it was a slightly smaller crowd.  We brought up several remaining issues, such as permitting, salt type, and next steps.

Permitting: My Naive Q&A with MoE Wizard Robin Pike

On Day 3, I showed some results of Robin Pike’s (BC MoE) dilution study to the group from the 1m, 2m, 5m, and DS mixing study.  We hope to have something more conclusive by the end of the year.  I think I fairly characterized Robin Pike’s approach to the permitting issue: know the Water Quality Guidelines (WQG) and sensitive species within a habitat and keep dosing below Cl- limits.
As a group, I believe we concurred that the following 3 points would be a good approach:

  1. measure the naturally occurring variation of NaCl in the stream.

Compare the difference in concentration at high and low values. One thing I realize now was that we were talking about conductivity and not necessarily Cl- so it would be important to determine what is the natural cause of the EC.T variation and what is the natural Cl- variation.  Looking at Table 6 in Quantifying the Relation Between Electrical Conductivity and Salt Concentration for Dilution Gauging Via Dry Salt Injection (Richardson, Sentlinger, Zimmerman, and Moore), it would appear that Na+ and Cl- are not the dominant ions in these samples, but rather Ca and Mg.  The following italicised comments are from Robin Pike:

Figure 14. Lucky number 14, Robin Pike was there in spirit and is working diligently on the permitting question, as we speak.
  • We are examining both Cl and Na at the moment as part of the dosing project.  As you can imagine the variability in natural background is time, place and discharge dependent.   My initial feedback on expecting operators to know the seasonal characteristics of Cl or Na,  a priori  is that it’s not operationally feasible.
  • Hence,  EC.T analogues were chosen as they are easily measurable and it is possible to directly estimate Cl and Na loadings from the salt we introduce.  As well,  if we assume Cl/Na loading in the ECbg, this will result in a conservative approach, even if the ECbg is low in Na/Cl and not caused by these constituents.
  1. Put Salt Dilution IQ Cl-  increases in the EC.T in context compared to road salt spikes and other chronic increases of Cl-. While we are not condoning increased levels of Cl-, it’s important to make others aware of the pre-existing conditions.  Robin Pike said:
  • I won’t be comparing this to Road salt spikes to justify operational additions of NaCl to water courses.
  1. note that when natural variations in Cl- exceed the critical threshold of freshwater aquatic species, such as wavy-rayed lampmussel, they are likely not present to begin with.  Robin Pike said:
  • Not necessarily.  The mussel you speak of is only sensitive during its glochidia stage (essentially when its reproducing).   It is possible to have a ECbg vary beyond critical thresholds for species on a seasonal basis outside of their sensitive time periods.  Hence assumptions of lack of presence based on ECbg may be erroneous.  

We also discussed the fact that this is probably a Master’s student thesis worth of work.  Robin Pike said:

  • I have lots of data and am always looking for partnerships with Academia.  We have just concluded a project working on another dataset (candidate was successful recipient of a Ph.D.)!

Perhaps this is a Yammer post?  Perhaps after a bit of back and forth within this group I should post it to yammer?  I started it as an update to you, Robin, and I’m not sure if you are on the Yammer groupchat?

  • Yammer.  Not at this time.  I don’t see any benefit.

So that’s Robin’s input.  Basically we are still in a holding pattern but Robin is open to collaboration to rigorously answer some of these questions.

Salt Type

There was some debate on the influence of salt type on the CF.T.  Practitioners I’ve asked here in BC have said they’ve never noticed a significant difference and therefore no longer keep track of it.  Unfortunately we didn’t include it in our CF.T Study. There is anecdotal evidence, like that from the Vienna flow regatta that some table salts have different constituents and therefore different CF.T’s, but there is no conclusive evidence.  Perhaps a Master’s Student would like to take this and a few other tasks on.

The Pipettor-Barometric issue

This issue just came to my attention.  Apparently there is some error introduced when an auto-pipettor, like that used in the QiQuac and Tracer-Q kit, is used at an elevation significantly different from that in which it’s calibrated.  The typical case is a pipettor is calibrated near sea-level and used at higher elevations.  This issue is documented here. In this case 1.3% error is introduced for 0.2 mL.  We hope it’s less for 0.5 mL and 1.0 mL: something to consider.

Bag End and Back Again

The 1st Annual Wales International Salt Dilution Workshop and Whisky Walk was a huge success.  Not only did we meet other practitioners from around the world, but we learned what others are doing, and confirmed what we all have no idea about such as: Permitting, influence of Salt Type, hanging vs resting probes.  It was great to see how closely the SD results were to ADCP in the Norwegian hydrometric record.  Paul was a fantastic host and proved himself exceptional at mundane bureaucratic paper pushing;  he clearly loved the thrill of a well organized meeting.
Fair play, butt?
P.S. I asked my Little Buddy Nick Page if the Featured Image of Paul and I was too much, and he replied, “No, I don’t think it too much. But it’s not that interesting either: two middle-aged guys drinking from plastic cups on the top of the highest mountain in Wales.”  So, there’s a sobering perspective.
P.P.S. Paul says “The mountain we went up was the highest in Southern Britain (Pen y Fan). Our North Wales colleagues would be spitting feathers.” Shows what Nick knows.
[rev_slider Wales]


Contact Us

We offer custom solutions to hydrometric and hydrological problems. We are keen to hear about particular solutions you might require. For more information, or a quote, please contact us using the form below.