nartreb
Well-known member
Rank each candidate location in assending order in each catagory for record of:
High Temp
Low Temp
Annual Liquid precip
Seasonal Frozen precip
Maximum Wind
Interesting way around the scaling problem, and it's an easy calculation to do, if you allow it to be defined relative to the pool of candidates. If you want to know where Mt W ranks in, say, record annual liquid precipitation compared to all weather stations worldwide, that's more data than I'm willing to search through.
Using absolute rankings in this formula would also mean that Mt Washington's #1 spot in wind [I'll concede that for the sake of argument] would be totally overwhelmed by its middling rankings in other categories. Remember, there are thousands, probably tens of thousands, of weather stations in the world. You get 1 + 1300 + 3478 + ... or something like that, where the size of the huge numbers (the ones that matter to the score) have rather little to do with the size of the difference in weather being measured. You still have a scaling problem, except now it's an arbitrary result of the number and distribution of the comparison set. But if that's the formula you want to use, we can give it a try - maybe restricting the data set to weather stations in the US or something for a start.
If you use a restricted comparison set, there's a new problem in that no location has a fixed score - the score is entirely dependent on which other locations are in the set of candidates. So if somebody says, "but how come you didn't look at location X?", you have to recalculate every other candidate's score.
Computationally that's not such a huge problem. The real interesting part is that you can change how two locations compare to each other by inserting or deleting other locations in your comparison set. Add a bunch of rainy spots, and suddenly a poor ranking in rain really hurts. The scaling/sampling problem becomes very important.
To really face that problem squarely, one option is to only apply the formula to sets of two candidates.
So, for example, I presume Mt Washington beats Miami, 3 factors to 2 (lucky thing there's an odd number of factors!) or, by your lowest-score-wins formula, 7 points to 8. I don't think that's exactly what you had in mind, though.
I still think all-time-records are a stupid way to determine "worst weather" (I mean, we're not trying to decide "worst weather seventy-two years ago") but I promised to take on all comers using their own formula, and I will. Shall we pick, say, 50 weather stations scattered through the US? One per state, whichever is first alphabetically by station code?
Notice I'm restricting myself to weather stations in the US - that's like fighting handcuffed and blindfolded since most of the spots we're really interested in - high mountaintops - don't have weather stations and aren't in the US. I'm hoping that Mt Washington, despite having 90 years of recording to pick its extremes from, is so far from the "world's worst" that some randomly-selected station will beat it. (Of course, I'm also curious to see what happens when the list is deliberately enriched with "extreme" places... one way to find out...)