Background
My mom does not care at all about science. I noted that global warming didn't seem to be something that concerned her or that she was aware of, and when I pressed her about it I realized that numbers, physics, etc., don't mean anything to her. However...the way she noted changes was based on how experiences changed over time. For example...the state fair where I grew up is in October, and she noted the years that we didn't have to wear a coat, so that's the sort of experience I'm referring to.
I started thinking about what sorts of impacts might fit into that category, and the first one that came to mind was that people like her might notice if their area stops getting snow at Christmas.
I started thinking about what sorts of impacts might fit into that category, and the first one that came to mind was that people like her might notice if their area stops getting snow at Christmas.
Method
The most obvious method I could think of was to take the estimated daily precipitation and estimated minimum temperature for December 25th every year, and track whether you had a rainy day when it went below freezing. Trying to predict the rain level for a specific day 50 years from now is complete guesswork though so that felt wrong.
I tried then to think of a probabilistic definition of 'white Christmas' that is temperature based and came up with one that loosely means 'Is it cold for there to be snow on the ground at Christmas' and it is:
- For a given year, take the minimum daily temperature projection for all days from December 22nd - December 28th for that year and the two years before and after it; e.g., 2025 means Christmas week of 2023 - 2027
- If 24 of the days in that bucket of 35 days (~2/3) are below freezing, assign a value of 1 to represent the possibility of a white Christmas; else, assign a value of zero
- After doing this with all models, average the 1's and 0's and multiply by 100 to get the percentage of models that predict that a white Christmas is possible
Thus, the score I came up with can be loosely interpreted as 'percentage of models that predict that a white Christmas is even possible'. Given the way that this is defined, this is an overestimation of the likelihood of actually having snow at Christmas.
It should be noted that I am using the RCP85 assumption for this. I will likely run this with other assumptions in the future (it takes a lot of data so I can't have too many that rely on daily data on my laptop at one time).
It should be noted that I am using the RCP85 assumption for this. I will likely run this with other assumptions in the future (it takes a lot of data so I can't have too many that rely on daily data on my laptop at one time).
Results
I then had to think about how to best present these results. I settled on picking reference points at the middle of each decade and making a gif. The results are below:
Noting that 'whiter = colder' and 'greener = warmer', you see a clear reduction in the areas that have the potential for snow on Christmas as we move further into the 21st century. Eyeballing it here, maybe a quarter of the US will see Christmases go from well below freezing to well above it.
An obvious weakness of this technique is that I'm checking if the min temperature for the day is below freezing, and that could easily mean that it's only below freezing for an hour or so. Re-running again with the check being 'less than 28 degrees (F)' yields the following gif:
Another obvious weakness is that I am not factoring in precipitation. I'll have to think of some way to do that and run these again in the future.
You might look at where you have Topeka located on this map. It looks like it is placed somewhere in SE Oklahoma, which is incorrect.
ReplyDeleteThanks and you're right...I had a typo on its latitude. I've updated it.
Delete