Last year around this time, I reported on this blog that Wooster had just completed its third wettest year on record. A year later, the “wettest year” record has been broken. With continuous record-keeping beginning in 1900 at the OARDC weather station, the 1901 water year (Oct 1900 through Sep 1901) is the first full year, and 2019 is the 119th year on record. Amazingly, this was the wettest year ever recorded for Wooster. Here is an updated graph of the annual precipitation in Wooster with “line of best fit” and a more detailed curve. The black dot at the end of the time series is water year 2019. At 56.3 inches, it beat out the previous record of 51.0 inches set in 2004 by 6% — a large margin! Note that although, there has been a long-term increase in annual precipitation at Wooster, this year was so far above the trend line that it’s likely we’ll drop back down to around 42 inches next year.
The reason for this record was primarily because of an exceptionally wet period from May through August, peaking with a July in which we experienced about twice as much precipitation as normal. Late spring to early summer is usually our wetter season, but this year the summer storms were dramatic. However, as shown in the plot below, every month except September yielded above-average precipitation. (The green bars are the total precipitation in 2019 for each month; the blue dots are the average, and the black whiskers are the standard deviation.) In fact, the record was broken in August!
Finally, it’s worth noting that the maximum daily precipitation was 4.22 inches recorded July 22. That ranks 5th highest all-time in Wooster for daily precipitation. Only two days have ever had over 5 inches — September 14, 1979 and the infamous flood of July 5, 1969. (Note, because of when precipitation is recorded, much of the precipitation really fell on the 21st, 13th, and 4th, respectively.)
Thank you Dr. Crawford – I will be using your plot of the water year in my presentation tomorrow. That is an impressive record breaker.
Glad to be helpful! And good luck with presentation. I should be there to see.
Hi Dr Crawford – Would we be able to use these charts in an article being written regarding precipitation rates in Ohio and the Midwest? Thanks for these regardless.
Yes, that would be fine! The source for the data is the OARDC weather station (NOAA’s ID number for it is GHCND:USC00339312).
A couple of notes, Alex. I would like to see that first graph, but with one and two standard deviations plotted alongside the mean and the raw data–I feel like trend lines in data sets like this one are misleading, from a regional point of view, and the standard deviation information allows us to see how terribly variable the data are. I’ve plotted trend lines through temperature data sets (that look like they were plotted with a shotgun) with tens of thousands of points, and found more utility in looking at the data without trend lines, but with standard deviation lines surrounding the mean. I find it much easier to see statistical outliers that way, and it’s pretty clear what’s happening if one starts to see a “flood” of outliers on one, or the other, side of the standard deviation plots.
Also, while it has been exceptionally wet in much of Ohio, we, one-half hour south of Wooster, are less wet than we’ve been over the past two years. We were on track to have our third 60″+ (rain only) year-in-a-row, but fell off that trend with an exceptionally dry Sept and Oct (so far). With that in mind, and realizing that the OARDC is the only official NOAA meteorological station nearby (is there still one at Mansfield Lahm airport?), I think it is unfortunate to use single-locality plots except as a, “Wow, hasn’t it been rainy this year, in Wooster,” commentary.
I feel like it would be better to accumulate all the regional precipitation data (a much more onerous job, of course). I say this, because, as I’ve mentioned before, spring and summer Midwest rainfall is so thunderstorm-controlled, that one location is unlikely to be representative of the entire state. We see this all summer long, driving the back roads, where nearby corn fields show different levels of moisture, not from soil, bedrock, or slope differences, nor from planting-time differences, but for the simple, fickle, reason that the thunderstorm hit one and missed its neighbor.
Anecdotally, thunderstorms, here, are much less severe than they were, even 1-2 decades past. This would indicate, to me, a weakening fight between tropical and polar air masses. Stronger storms imply stronger frontal conditions and more pervasive, even, rainfall. Weaker air mass collisions translate into more isolated, spotty, storm systems. Perhaps this is all linked to a warming Arctic and a much weaker polar vortex (from the progressive loss of sea ice) and a destabilization of the boundary between Polar and Ferrel atmospheric circulation cells.
And, once again, thank you for putting all the effort into the posts! It is always fun to read them!
Hi Bill,
First, to speak on your last paragraph, which is an interesting anecdote, you talk about thunderstorms and then relate them to frontal processes (and beyond). My understanding is that many summer thunderstorms in this area form at a mesoscale level independent of the polar stream (which is usually hanging out farther north). They can form from an unstable atmosphere as opposed to forced lifting. I’m not a mesoscale expert, but I suspect that’s a considering one would want to make if formally looking at that hypothesis.
While I appreciate the detail of your plotting comments, I’m going to stand my ground on two of your critiques. First, a standard deviation around the mean is not appropriate. A regression line has error — of course it does — but that error is on the slope and intercept being estimated. Therefore, the error is not the same for all values of x and the standard deviation around the mean would give the misleading impression that the error is equivalent. So while what I’ve shown is withholds some information that could be shown, what you suggest adds inappropriate information for assessing trends. Moreover, when trying to show something descriptive (not predictive) for a potentially general audience, over-complicating the graph can be a poor choice. I’m not going to show advanced statistics that most people have never heard of unless it’s absolutely necessary; I’m not going to talk about test statistics and p-values unless it seems necessary to get a point across, and I’m going be sparing with my graph annotations to avoid overwhelming amounts of information. Honestly, I even hesitate to show the smoothed curve except it shows how Wooster is responsive long-term natural variability (i.e., it’s not just a long-term trend), and I wanted to mention that first time I used a graph like this in a blog post.
Second, no, it is not better to aggregate regionally for a post like this. This is indeed a blog post focused on a single locality and how, wow, there was as lot of rain in Wooster this year; therefore, this is an appropriate graph to be using. In no way am I trying to say this represents all of Ohio or proves some wide scale climate change. It is, however, a good example of a local manifestation of a broader-scale climate change. Also, to be clear, what you are asking for is a level of analysis appropriate for the Ohio State Climatologist or a full-blown research project. That is not the purpose here. Otherwise, absolutely, I’d gather as many stations as possible and handle the data gaps and aggregation issues that come from having time series that don’t have perfect overlaps. And I don’t know exactly how Dan DeBaun might use the graph, but I will trust other people to use it as a point example as opposed to a definitive measure of a large area.
Thank you for the detailed response, Alex, and so quickly! And please, don’t take my comments as critiques, because they absolutely are not. I love everything you post and the level of detail you bring to your discussions. There are very few places of higher education that take the time to post with such care, let alone depth. I think, perhaps, you have misunderstood what I tried to say (my fault for not being clearer in my comments), and I suspect we’re talking about different phenomena, different graphs, and philosophical questions about the presentation of The COW geoblog, as it has evolved from Fossil of the Week, in Mark’s original conception of it, if I am remembering correctly.
First, I wasn’t trying to relate our local weather to the position of the interface between Polar and Ferrel Cells (which is, indeed, at 55-60N), but the weakening boundary between the two does allow for the polar jet to deflect, abnormally. Furthermore, the strength of thunderstorms definitely relates to the relative strength of cold, dry air masses as they interact with warm, moist ones. In a changing, warming, climate, might we not expect that to occur, particularly if polar air masses weaken in intensity?
Also, I wasn’t suggesting placing SD lines around your trend lines, if that’s what you thought. I was thinking of a separate plot, showing the data, the mean rainfall plotted as a line through the “center” of the data, and SD lines around the mean. The SD lines allow us to see outliers, distinct from potential trend line graphs. The SD simply tells us about the variability of the data set, around the mean. We have rainfall data over many years, an average of that data, and a SD, showing the variability around the average. This type of graphic representation is quite common, and gives an at-a-glance reminder of data variability, the potential of trends as indicated by outliers, and the long term averages. My suggestion was simply offering a slightly different view of the graph, as presented, not a specific critique of applying trend lines to historical data.
Finally, I fully support your “single locality” rationale, but I think many people might interpret that single locality as representative of more regional phenomena, and that a disclaimer might be warranted, just to prevent misunderstanding of what you have stated. The COW geoblog has long been a mixture of purely professional material, often with no commentary, juxtaposed with content that is more lighthearted, but very informative, if on a different level. I suspect that there might be confusion amongst many readers of the blog as a result this admixture, something that needs to be carefully trod upon, particularly in politically fraught times, and especially when we begin talking about politically sensitive topics like the components of climate change, which rainfall might represent. If you read Mr. DeBaun’s comment carefully, it sounds like he would like to make this single-station data set representative of the entire Midwest, which it most clearly is not.
And, of course, that “research project,” accumulating all that data, sounds like the perfect IS topic, to me. I think that’s what I was, opaquely, hinting at when I suggested the larger data set. I never meant to suggest that you should, laboriously and without compensation, retrieve the NOAA data from multiple stations, just to present in a blog.