CET warming rates by month

[WYorksWeather]

With thanks to @LetItSnow! for the inspiration, I thought I'd take a look at the temperature trends over time by month, to try to answer the question of which months are warming fastest. I thought this might be of general interest so have left it here, but mods feel free to move it to the Climate Change area if you feel it would be more suitable there.

To be clear, I don't want to use this thread to beat people over the head with any sort of climate change debate, my interest is in the data and its implications for weather expectations, hence the reasoning for putting it in here. I'd politely ask that people try to stick to discussing the trends I've picked out of the data.

My big concern about doing this was how to do this in a way that avoid the obvious suggestion of cherry picking - in other words, that any conclusions are too sensitive to the choice of start years. So, I came up with a methodology to try to account for this.

Methodology (feel free to skip!)

Here, in (somewhat) simplified terms is what I did:

1. Start with the daily CET data (1772 to present)
2. Calculate monthly averages for each month in each year
3. Calculate 30-year averages for each month and year, where the 30 year average for is over the 30 years up to and including that year (e.g. May 2020 corresponds to a 1991-2020 average for May, 1970 to a 1941-1970 average, etc.)
4. Calculate the change in that 30-year average compared to a period 30 years prior (e.g. for April 2000, this value would be the 1971-2000 average for April, minus the the 1941-1970 average)
5. Divide the figure in point 4 by three, to convert it to an average change per decade
6. Plot a graph of the final outcome, by month, which hopefully doesn't look a mess!

Advantages

The main advantage of this method is that there is no cherry picking - I'm able to show the trend for any arbitrary starting year since 1831. I also managed to get a workable graph out of it.

Disadvantage

 The main disadvantage is that all this smoothing and averaging reduces the numbers of years you can use - the data goes back to 1772 in this dataset, but I can't actually use all that data, because 30 years gets lost in the rolling mean (well, it's incorporated, but you can't get a 30-year average for the earliest 30 years), and then another 30 years gets lost before the first year you can show differences for.

The 30-year averaging also creates a bit of a lag effect - measures for recent years are reflective of average conditions over 30 years, and so don't reflect further changes since.

Results

In the end, I got this chart below.

It's not immediately easy to interpret, but once you understand it, it tells you a lot. The value on the y-axis is the rate of warming for each month, over the 30-years prior to that date. For example, the value for February 2023 is 0.45. What this tells us, is that the average February in the period 1994-2023 has warmed by 0.45C per decade compared to the average 30 years prior, i.e. February months between 1964-1993. The overall area above/below the line corresponds to the change in temperature since the start of the data, and the height of the lines above or below zero indicates rate of change. Think of it like inflation - above zero is still an increase, and a fall in the line means a slowing in the rate of increase, not a fall in temperature (which only happens if the line dips below zero)

Main findings:

• All months are currently warming, though at varying rates
• The warmer months have generally warmed more slowly, and June in particular shows little change compared to the early 19th century
• Particularly rapid recent warming trends are seen in late winter and early spring, with February being the standout month in recent years
• The mid-20th century cooling trend is seen in around half of months

Restricting the data to post 1950, to look at more recent trends, we get the following (y axis scale is largely the same, x axis scale is of course restricted to post 1950):

Here, you can see a bit more clearly the following:

• December stands out as an outlier - it has warmed only very slowly in recent years, though this may be driven by some relatively recent cold Decembers including 2009 and 2010
• All other months are currently warming at a rate of at least 0.2C / decade
• More than half of months show a recent acceleration in the warming rate
• There is a recent trend towards more rapid warming in the summer and autumn

Finally, for a very recent view, here is 1990 to present, presented without comment.

Conclusion

Hope this analysis proves useful to some. Don't want to tag too many people, but drawing this to the attention of @Roger J Smith as everyone's resident CET expert! Happy to discuss any of this / explain stuff, or even do some further analysis.

[reef]

Interesting analysis.

February, March and April accelerating does not surprise me at all. We've had some very warm Springs in recent years and February often feels more like a Spring month these days than a winter month. I have a feeling that in 20-30 years February will be what March used to be in the 1961-1990 period. The same could perhaps be said on September on the other end: more like a summer month than an Autumn one.

May continues to be a strange one. We're well overdue a very warm one. We've had a 11.9C April and a 17.0C June now, but nothing above 13.4C in May in the last 31 years. I'm sure at some point soon there should be a 15C May. Its the only month without a top 10 CET in recent times.