'City climate twins': similarity measure for monthly temperature data

Question

Cities are getting hotter due to climate change. I want to communicate this intuitively through a concept of 'city climate twins': ie. which other city will my city's climate resemble several decades from now?

Available data:

(1) monthly average temperatures for a set of global cities: present-day reference period (eg. 2000-2019).

(2) monthly average temperatures for a set of global cities: future climate projections (eg. 2040-2059, high emissions scenario)

See table and charts below for the data format.

Take the city of interest ("City A" - eg. London). I want to take City A's future monthly temperature profile, compare it to the current monthly temperature profiles of all other cities, and select the most similar ones.

Example: "London currently has a mild climate but it's getting hotter. By 2050, projections suggest that its climate will be similar to Rome or Madrid today."

Questions:

• Easiest option: How can I get a suitable measure (and ranking) of similarness for these temperature profiles? (preferably something I can implement easily in Python)
• Slightly more complex option: How could I approach this if I included more variables, eg. monthly precipitation as well as temperature?

Thank you!

Answer 1

You would have to define exactly what you mean by "similar", either for the simpler or more complex case. A really easy measure of similar would be mean annual temperature. But you might want mean temperature for a particular month, or you might want the mean high temperature. (You have average daily maximum).

Then you might want to do something about seasonal variability. Many cities (like Lagos in your example) have roughly the same temp year round. But cities outside the tropics (like London) don't.

You could look at mean difference in average high temperatures by month, but that would not capture the highest temps in cities like London.

There are all sorts of possibilities, but you have to decide what you want. Once you've done that, maybe someone here can help (although you might want to ask on a Python list). For help deciding on a measure of similarity, the Earth Sciences stack exchange might be better than CrossValidated.