Tuesday, September 7, 2021

This post is a mea culpa and an apology.

You might recall the last post, which I will leave up for couple of days so people can link back to it as they read this. I added a line to the beginning saying the data is in error. The post outlined a nitrate event at the Decorah fish hatchery spring and the Upper Iowa River over the past 10 days or so.

We reported nitrate levels of nearly 50 mg/L at the hatchery, and 35 mg/L in the river.

We were wrong.

Both of those sensors had been in the factory for service just prior to deployment. We had reason to believe then that they were working perfectly, and in fact they were.

The problem is how nitrate is reported. In the U.S., we measure it “as Nitrogen (N)”. Nitrate is an ion of nitrogen trioxide—NO3-. In most of the rest of the world, nitrate is measured “as NO3”. There is a factor of 4.4 difference in N and NO3. Thus, 10 mg/L nitrate “as N” is the same as 44 mg/L “as NO3."

The sensor can report the data in either the N mode, or the NO3 mode. Evidently when the sensors were returned to us from factory maintenance, they had been switched from the N to NO3 mode. This had never been done before in 8+ years of the sensor network, and thus we had no reason to suspect that it had been done this time. We should have suspected. Also unfortunately, this episode happened in the worst possible circumstance.

We did have reason to suspect that the hatchery nitrate could have been 20 mg/L “as N” as the sensor was reading initially. The spring at the Manchester fish hatchery was at that level for two years, and Big Spring at Elkader often is at 13 mg/L or above.

But I started to worry something was amiss yesterday, when the river nitrate started to go up again. A stream that size, above 30 mg/L for that long, just started to seem too incredible to me. I spoke with our field staff and we agreed they should go up there first thing this morning.

Then I got to my desk this morning where I had an email in my inbox from a Luther College professor, reporting that he was measuring 8 mg/L in the river and 7 mg/L at the hatchery. We then masked the sites on IWQIS until we can get the sensors set appropriately.

So I’ve learned some things these last few hours. This was an honest mistake, but nonetheless we failed in that we didn’t have procedures in place to prevent it. That was my fault.

As Carl Sagan said, in science, extraordinary claims require extraordinary evidence. We didn’t have it. I should have known better.

This episode will cause me to reflect on some things. My youngest daughter, who is about to finish her degree in chemistry here at UI, was talking to me yesterday about some water quality data she had been reviewing. After one day of looking at data, her remark to me was that it is all so depressing. The whole situation here in Iowa, with the controversy regarding water quality, can be exhausting, day after day. Nonetheless, that isn't a license to make errors. But the only thing worse than making mistakes is not owning up to them, and so here I am.

I’ve tried to make my work “public facing”, i.e., bring the data and the stories to the masses, rather than just the rest of the scientific community. I thought this was needed, and still do. But this episode also exposes the risk of trying to serve an audience. I really have to give this some thought.

I am sorry about this. We will try our hardest to not let a mistake like this happen again.