Puzzling over Waterloo Creek Data

Puzzling over Waterloo Creek Data

Everyone loves a puzzle. Witness the popularity of the online game Wordle. But we have a different sort of puzzle to talk about today. Regular readers of this column will know it’s an exciting puzzle, because it involves graphs! The magical tool that turns information into insights.

In August of last year, we installed a gauging station in Waterloo Creek. The station records the water level of the creek every fifteen minutes, so we get a ton of data to review. The dull and dreary days between the holidays and the gardening season seemed like a good time to do some of that review.

In Graph 1, covering November through January, we can see the water level going up and down like an elevator. It’s the same pattern we see in our Mud Bay data. And, thanks to graphs, we know that high water levels correlate with significant rainfall and low water levels coincide with dry spells. So far so good.

But wait! What’s going on here? There’s hardly any rainfall showing up until the end of January. So why is the water level in the creek going up and down before that?

After some mulling and stewing and staring at the screen, we decided to zoom in a bit and remove the latter part of January. This changes the rainfall scale from 0-5000mm to 0-60mm. Graph 2 shows the new results. Suddenly, the rainfall spikes jump into the picture and the graph makes sense. The rainfall data in the first graph had been dwarfed by the amount of rain that fell in a few days after the 10th of January.

But wait again! Could those rainfall events at the end of January really be 80 TIMES greater than the other rainfall events? That doesn’t sound right. We would have been building Arks. We enjoy the sound of rain on our metal roof but that much rain would have deafened us! So how do we explain this?

All of the data we collect goes through a rigorous validation process, so we had confidence in the accuracy of our water level data. But could the rainfall data be wrong? There are multiple places online where one can find local weather data. Comparing several alternate sources revealed that our usually reliable source had some crazy and inaccurate numbers for the end of January. Plugging in the correct rainfall data from the new sources makes the rainfall spikes line up nicely with the rise and fall of water levels, all the way to the end of the month. The finished product is Graph 3. Puzzle solved!

 This is a great lesson: you can’t ever take data at face-value, no matter the source. Raw data must be examined, tested, validated. It must be puzzled over. Now if you’ll excuse me, today’s Wordle is still waiting.

*BWS would like to thank Weaver Technical Corp. in Courtenay for their generous donation of two high quality pH field meters. We can now add pH to our list of parameters being regularly sampled.

Leave a Reply

Your email address will not be published. Required fields are marked *