|
Perry Mason Jar posted:endless monotony is pretending he doesn't understand variables (and their relationship to probabilities, particularly) to make a point in the direction of SirPablo regarding the modeling crisis. Technically endless monotony is more correct than SirPablo but since they went about it so poorly they get large demerits Probability isn't real on these scales, and it doesn't even matter if reality's deterministic or not! It's just an artifact of insufficient data! Then I took a break, took a poo poo, wondered how I could make this point in a way where anyone cared, and realized even I don't care about longwinded rants about modeling, I want to see the weather do wild poo poo, so I didn't post more about it.
|
# ? Jul 16, 2022 19:17 |
|
|
# ? Jun 4, 2024 00:53 |
|
endlessmonotony posted:Probability isn't real on these scales, and it doesn't even matter if reality's deterministic or not! It's just an artifact of insufficient data! what lol
|
# ? Jul 16, 2022 19:21 |
|
Notorious R.I.M. posted:what lol its difficult the degree to which we can say something was a 100 year storm, when accurate measurement is at best a couple hundred years, when the overall climactic shifts of the last millennium are significant, etc this gets even stranger when talking about thousand-year-storms, is something like the ARK storm that flooded Sacramento 150 years ago really an event that happens every 1000 years? would it be a 100 year storm in the little ice age and a 10,000 year storm in the medieval warm period? or the reverse probabilities of that? endlessmonotony is just stating we do not have anywhere near the data required to meaningfully describe rare events, its hard enough to predict specific behavior of hurricanes and stuff that happens dozens of times per year, something we may have seen once, ever, cannot be given a useful probability in the way we want to receive one
|
# ? Jul 16, 2022 19:39 |
|
hekaton posted:its difficult the degree to which we can say something was a 100 year storm, when accurate measurement is at best a couple hundred years, when the overall climactic shifts of the last millennium are significant, etc As far as I can tell he's yelling at dynamical weather model outputs that aren't even normed against a climatology (i.e. the models are showing absolute temperatures, not deviations from some 20 year climatological average). The temperature outputs of something like the GFS or ECMWF plots posted in the past several pages don't care about the return-rate incidence of events beyond that background signal possibly being relevant in the dataset that the models are backtested against.
|
# ? Jul 16, 2022 20:17 |
|
Does a hurricane on a treadmill take off?
|
# ? Jul 16, 2022 20:29 |
|
Notorious R.I.M. posted:As far as I can tell he's yelling at dynamical weather model outputs that aren't even normed against a climatology (i.e. the models are showing absolute temperatures, not deviations from some 20 year climatological average). The temperature outputs of something like the GFS or ECMWF plots posted in the past several pages don't care about the return-rate incidence of events beyond that background signal possibly being relevant in the dataset that the models are backtested against. Exactly. This other person is ranting about the wrong thing, because they don't understand what they are talking about.
|
# ? Jul 16, 2022 20:30 |
|
SirPablo posted:Exactly. This other person is ranting about the wrong thing, because they don't understand what they are talking about. That's because there was an additional point "and by the time weather models show useful results there's no longer real world probability because the energy's already in, so model probability is merely a reflection of our confidence in our models and data, and those are incomplete by nature and selected by people influenced by their expectations, so using probabilities as an argument is just an appeal to authority without clearly designating the authority" but I couldn't make that a good post either. It's still not a good post, someone post wildfires, I heard Lytton got more.
|
# ? Jul 16, 2022 20:46 |
|
endlessmonotony posted:That's because there was an additional point "and by the time weather models show useful results there's no longer real world probability because the energy's already in, so model probability is merely a reflection of our confidence in our models and data, and those are incomplete by nature and selected by people influenced by their expectations, so using probabilities as an argument is just an appeal to authority without clearly designating the authority" but I couldn't make that a good post either. This bolded part is why you post GEFS or EPS ensembles of temperatures past t=120h on the model runs instead of the GFS or ECMWF operational (single) runs. Using model outputs in windows where they are a) knowingly highly uncertain and b) have better alternatives (like ensembles) is the entire problem SirPablo and I have with the cherrypicked models posted in this thread. Most of the people doing this are either cherrypicking models with results they want or have no idea how to use the products altogether and shouldn't be divining any sort of insight out of them.
|
# ? Jul 16, 2022 20:51 |
|
Notorious R.I.M. posted:This bolded part is why you post GEFS or EPS ensembles of temperatures past t=120h on the model runs instead of the GFS or ECMWF operational (single) runs. Using model outputs in windows where they are a) knowingly highly uncertain and b) have better alternatives (like ensembles) is the entire problem SirPablo and I have with the cherrypicked models posted in this thread. Most of the people doing this are either cherrypicking models with results they want or have no idea how to use the products altogether and shouldn't be divining any sort of insight out of them. That's a you problem. Nobody's divining any insight from this thread, they're here to look at wild poo poo. Which math chat isn't.
|
# ? Jul 16, 2022 20:57 |
|
gently caress, I want to give myself a wedgie by now.
|
# ? Jul 16, 2022 20:57 |
|
endless do you even understand the difference between climate models, global weather models, ensemble models, CAMs, etc.? Add another layer - direct model output and statistical post processing? And a whole different branch is statistical modeling.
|
# ? Jul 16, 2022 22:03 |
|
math isn't going to gently caress you
|
# ? Jul 16, 2022 22:22 |
|
Good thread https://twitter.com/burgwx/status/1547641607154872329?s=20&t=gOl7bpx6I2H1Z_1yH6QrpA
|
# ? Jul 16, 2022 22:30 |
|
“There’s no modelling crisis!” I exclaim, as my weather forecasts make Nate Silver look like a steely‐eyed missile man.
|
# ? Jul 16, 2022 22:43 |
|
I've never heard the term "model crisis" outside this dumb thread and I literally look at this poo poo for hours every day.
|
# ? Jul 16, 2022 23:02 |
|
You fools. You ignorant Morons. Are you suggesting a world where math doesn't exist? If you are modeling car crashes and are then teleported to an alien planet you just humansplain to them their risk of getting into a car accident based on data and physics. There is no need to account for a core fundamental shift to the concept of a model because that isn't how statistics works. They have a number chance of getting into an accident with whatever they call cars however they drive. That's how reality works. There would be no issues with your modeling. Underlying behaviors and historical assumptions being defeated as the environment being modeled enters a new period of behavior we have not seen and don't fully understand isn't relevant. Numbers exist. I read lots of graphs. Checkmate.
|
# ? Jul 16, 2022 23:11 |
|
SirPablo posted:I've never heard the term "model crisis" outside this dumb thread and I literally look at this poo poo for hours every day. https://m.youtube.com/watch?v=V9v3sAck-QQ
|
# ? Jul 16, 2022 23:15 |
|
Cup Runneth Over posted:those paintings are hung like poo poo uh it's holland and it's called the dutch angle
|
# ? Jul 16, 2022 23:33 |
|
The UK forecast calls for a Dutch Oven.
|
# ? Jul 16, 2022 23:46 |
|
"there's a model crisis! there's a model crisis!" i insist as forecast error slowly shrinks and i transform into a corn cob
|
# ? Jul 17, 2022 00:03 |
|
Rectal Death Adept posted:math isn't going to gently caress you My experience with real analysis back in undergrad says otherwise. The experience was not mutual.
|
# ? Jul 17, 2022 03:28 |
|
SirPablo posted:Let's watch this on - will Fargo hit 115 on July 31st? This would break their all time record of 114 (from the dust bowl years). quote this post if you're rooting for this lil' underdog model to succeed
|
# ? Jul 17, 2022 03:43 |
|
neutral milf hotel posted:quote this post if you're rooting for this lil' underdog model to succeed
|
# ? Jul 17, 2022 04:32 |
|
neutral milf hotel posted:quote this post if you're rooting for this lil' underdog model to succeed
|
# ? Jul 17, 2022 04:38 |
|
SirPablo posted:Let's watch this on - will Fargo hit 115 on July 31st? This would break their all time record of 114 (from the dust bowl years). Climate proof Duluth
|
# ? Jul 17, 2022 04:40 |
|
The next GFS run was 30 degrees cooler. Which is right!!?? Model is a mess!
|
# ? Jul 17, 2022 04:43 |
|
SirPablo posted:The next GFS run was 30 degrees cooler. Which is right!!?? Model is a mess! SirPablo posted:There is no "modeling crisis" wtf
|
# ? Jul 17, 2022 04:52 |
|
I went to Vegas and rolled an 11 at craps, then my very next roll I only got 2?! There must be some kind of die crisis to get such different numbers.
|
# ? Jul 17, 2022 04:59 |
|
climate models have played us for absolute fools
|
# ? Jul 17, 2022 05:01 |
|
Weather thread:. A model, idiot
|
# ? Jul 17, 2022 05:07 |
|
neutral milf hotel posted:quote this post if you're rooting for this lil' underdog model to succeed ༼ つ ◕_◕ ༽つ Take my energy. ༼ つ ◕_◕ ༽つ
|
# ? Jul 17, 2022 05:08 |
|
If it's 110 where I am on July 31 I'm chain probing all of you
|
# ? Jul 17, 2022 05:10 |
|
Platystemon posted:༼ つ ◕_◕ ༽つ Take my energy. ༼ つ ◕_◕ ༽つ Stop! That's how I got cancer!
|
# ? Jul 17, 2022 05:12 |
|
People just throwing energy around like it's no big deal... WELL IT IS! Energy is how we got here!
|
# ? Jul 17, 2022 05:12 |
|
RandomBlue posted:People just throwing energy around like it's no big deal... WELL IT IS!
|
# ? Jul 17, 2022 05:13 |
|
RandomBlue posted:Stop! That's how I got cancer! A heatwave can't get cancer! ... right?
|
# ? Jul 17, 2022 05:13 |
|
endlessmonotony posted:A heatwave can't get cancer! A heatwave could affect the tropic OF cancer
|
# ? Jul 17, 2022 05:14 |
|
I rely on weather models for leisure activities and GFS has always been kind of a mess, in some locations it can be almost flawless but in others you'll end up consistently wild inaccurate reports, takes some local experience to interpret it. Seaside is generally better than inland
|
# ? Jul 17, 2022 05:28 |
|
10k people may die next week, but at least we'll see which model was right.
|
# ? Jul 17, 2022 05:48 |
|
|
# ? Jun 4, 2024 00:53 |
|
Fell Mood posted:10k people may die next week, but at least we'll see which model was right. Covid thread is over there.
|
# ? Jul 17, 2022 06:24 |