Just in time for February, it’s the return of the great annual International Cherry Blossom Prediction Competition! (This post is by Lizzie.)
Help scientists like me better understand the impacts of climate change and win cash prizes by predicting when the cherry trees will bloom in four cities across the globe. The competition is open to all with prizes for closest prediction as well as other categories.
Interested? Check out the website with all the details, including data, rules and how to enter here.
Last year over 80 contestants from across four continents entered with a variety of prediction approaches. You can read more about last year’s competition here.
A big thanks to the American Statistical Association, Caucus for Women in Statistics, Columbia University’s Department of Statistics, George Mason University’s Department of Statistics and Posit (formerly RStudio) for their support, and partnerships with the International Society of Biometeorology, MeteoSwiss, USA National Phenology Network, and the Vancouver Cherry Blossom Festival—as well as Mason’s Institute for a Sustainable Earth, Institute for Digital InnovAtion, and the Department of Modern and Classical Languages. Sponsors and partners will be updated on the website.
Organizers: Jonathan Auerbach and David Kepplinger (George Mason University) and Elizabeth Wolkovich (University of British Columbia)
Japan’s in the midst of one of its worst cold spells ever. More snow that usual everywhere (except, to my chagrin, Tokyo, sigh). Tokyo’s winter usually isn’t winter enough to crack violins and guitars, but I’m running a humidifier in my study to not take chances with the acoustic and archtop guitars I keep out.
So, for you betting folks, the cherry trees will, most likely, be on the late side in Tokyo this year.
Full bloom in Guangzhou already. Very mild winter so far
Of course, there’s the cherry tree analog to the “does a tree falling in a forest make a sound if there’s no one to hear it” conundrum; “did the cherry tree actually bloom if the country’s government has taken a “let her rip” approach to Covid-19 and everyone’s either dead or too sick to go out.” Here in Japan, that seems likely. January 2023 was the worst month for Covid deaths (over 10,000 deaths in one month*) of the whole pandemic so far and that very same month, the government announced it was downgrading the official seveverity of the disease to the same ranking as seasonal flu. “We’re tired of trying to do anything about it, so go get sick and die.” They seem to be saying.
*: The first TWO years of the pandemic saw 20,000 deaths, but the third year by itself saw 40,000 deaths.
About the photo – nice photo! However, it reminds me how disappointing pedal paddle boats are. I remember seeing one of these for the first time as a little kid and thinking, “Wow! A boat that you pedal! My bike goes fast, so this boat must go fast too!” Two minutes into a grueling unwanted cardio session, the realization that the lumbering pedal paddle boat wouldn’t clock more than a lazy walking speed at a max sprint was only matched in disappointment by the fact that now I had to get it back to shore. It must be some cruel life lesson intended by some designer that these things always seem to be fiberglass behemoths weighing in like an elephant seal, with no gearing, and a paddle that barely chips at the water.
Clearly, you should have taken the (mid-1970s) MIT NROTC course “Air, Water, and Interface Vehicles”. There you would have learned that (1) short wide boats are bad news (maximum water resistance), and (2) The US Navy realy really really believes in the English system and deals with the density of seawater in slugs per cubic yard. Really. They do. It’s amazing our boats work at all…
Well I was only a kid in the 90s, but perhaps this suggests pedal paddle boat designers didn’t go to MIT…
Perhaps they just like the ring of slug per cubic yard of seawater. Slug and seawater. Seems more literary than g per cubic cm or something boring like that.
This page shows how the date has varied in Washington DC since 1921:
https://www.epa.gov/climate-indicators/cherry-blossoms
The chart shows recent values returning to those recorded back in the 1920s through 40s, which fits surprisingly well with temperature curves that are adjusted for urban heat island effects. This is odd, since you would think that the DC cherry trees would be affected by urbanization. In any case, I don’t see much rationale for claiming an AGW signal here.
Nevertheless, the EPA writes:
“Peak bloom date for the cherry trees is occurring earlier than it did in the past. Since 1921, peak bloom dates have shifted earlier by approximately seven days.”
The only way to get that number that I can see is to, um, cherry-pick a start date around 1957 and ignore the previous data.
This prognosticating is all good clean fun of course. But I did notice this on the page describing the winning entries from last year:
“Joshua and Taehoon combined a popular statistical model (the logistic-growth model) with a popular biological predictor (growing degree days or GDD). The judges appreciated the simplicity and interpretability of this approach, with one remarking that the team provided “a wonderful demonstration of the power of using the simplest possible model that captures the key biological aspects of the question.”
The quote seems to me a concise example of a statement that might trigger Andrew’s frustration with the misuse of the concept of parsimony.
Matt –
> The chart shows recent values returning to those recorded back in the 1920s through 40s, which fits surprisingly well with temperature curves that are adjusted for urban heat island effects. This is odd, since you would think that the DC cherry trees would be affected by urbanization. In any case, I don’t see much rationale for claiming an AGW signal here.
(Like Phil below) I am also confused here. If “this” fits well with temperature curves adjusted for UHI, then why would be surprising since “you would think that the DC cherry trees would be affected by urbanization?”
Typo? Should it be “wouldn’t be surprising?”
Matt,
You say “The chart shows recent values returning to those recorded back in the 1920s through 40s, which fits surprisingly well with temperature curves that are adjusted for urban heat island effects. This is odd, since you would think that the DC cherry trees would be affected by urbanization. In any case, I don’t see much rationale for claiming an AGW signal here.”
I’m not sure what you’re saying there. I’m not saying I disagree with it, just that I don’t understand it. According to the plot in on the page you linked, the cherry bloom in DC is happening earlier (and with less variation from year to year) than was the case previously. I didn’t try to extract the data but it’s easy to look at how many years were below or above a threshold so:
Full bloom occurred April 10 or later:
First 80 years: 15 times
Next 40 years: 1 time
Full bloom occurred March 21 or earlier:
First 80 years: 3 times
Next 40 years: 5 times
I can’t eyeball either the mean or the median but they are both clearly earlier in the past forty years than they were previously.
So…are you saying that this is all attributable to the DC heat island effect? I would think that’s what you’re saying except that you seem to think something is odd about this signal and suggest that it is not attributable to a heat island effect.
You suggest that you have to cherry-pick a year (good one!) in the mid-1950s to get a pre-post difference but that doesn’t appear to me to be the case. I looked at the first 80 years and the subsequent 40, but you could also split the data in half and you’d see that the second half generally has earlier blooms than the first half. Or you could leave out the middle and just compare the first 40 years to the most recent 40 years. Indeed I think you’d have to cherry-pick in order to _not_ see a trend towards earlier dates. But I’m not sure you’re disagreeing with this, I think you may be saying there’s a trend but it’s not attributable to AGW, which, well, of course you can’t determine causality this way.
Anyway I’d be interested in what you are saying. Can you re-word it?
Phil and Joshua,
First one piece of background that might be useful – the 1920s through 40s in the US were quite warm, with temperatures similar to today if you subtract out a reasonable estimate of Urban Heat Island (UHI) effect from the recent urban data. Back then the west coast states filled with smoke every summer just as they have been doing recently.
I concede that the cherry blossom bloom date was later in the earlier years. My eyeball had a tough time picking it up, but the data is there in csv format. Now that I am retired I lack the tools to do anything with data, so I crunched the numbers myself. I came up with an average of 94.3 days since the first of the year for 1921-1950, and an average of 90.8 for 1980-2022. So half of the seven days cited by the EPA. To get to seven, I think you would have to start in the 1950s and disregard the previous data, but I did not crunch those numbers.
My second point was that we should see an urban heat island effect in any data related to temperature in Washington DC, but I was not eyeballing much of a change at all. Since the data indicates that the date is indeed earlier now by 3.5 days, that would be an entirely reasonable number for a UHI-induced change.
As the saying goes, the corrections do not change the conclusion. If there is any signal from CO2-AGW, as opposed to UHI-AGW, it is down in the noise. Cherry blossom bloom time, along with ice melt in some town in Alaska, have been cited as evidence for CO2-AGW, but the actual data for the cherry blossoms provides little support. I do want to be careful about straw men though. The fact that advocates sometimes say silly things about AGW does not in any way impinge upon the legitimacy of the theory.
Matt –
Seems to me that identifying a UHI effect for the DC area more generally and estimating its impact on the cherry trees blooming could get a little tricky. I don’t know anything about the relevant features where the trees are located but I would guess that an averaged effect over a metropolitan area might might not be directly applicable to any specific area. Also. I would imagine the degree of effect would vary by season and that might also be relevant to the impact on trees blooming in the early spring (when the impact might be less than other seasons?)
What do you think?