Dear @Vitaliy Kaurov ,
great catch! Thank you and thank you for your kind words. The country borders come out much better the way you do it. Much of what I do is an experiment in getting LLMs to use the Wolfram Language as a tool, and I use it to "explain certain ideas to the LLM", but the actual programming is nearly exclusively done by the LLM. I have developed a special setup for that. The skeleton for such a calculation is done within minutes now for me. If you look at my Moving Sofa Problem post, or the ENSO one, it was really just discussing the project with the AI and the rest was automated. It seems that something happened at the end of last year which changed my workflow completely. Before I had trouble "vibe-coding" with LLMs in the Wolfram Language, but now that is not an issue any more. It has very much changed how I use the Wolfram Language, but I produce much (!) more code. Sometimes the code is not ideal and does not use some of the high level functions. For example in the post on textured 3D objects I use an FEM approach when simulating the mug being filled with and submerged in hot water. Oliver Ruebenkoenig kindly pointed out (among other things) that I had not used VonMisesStress which would have been more elegant. I guess that we need one or two more LLM iterations so that they know all of the high level functions that MMA already has.
Regarding this post, you are right D-Day is, in fact, an interesting event to look at in relation to the weather. I asked GPT for battles after 1940 where the weather played a crucial role and it gave me :

I have a couple of posts liked up, but if possible, I will look at some of these battles.
Thanks again.