[HN Gopher] Google says AI weather model masters 15-day forecast
       ___________________________________________________________________
        
       Google says AI weather model masters 15-day forecast
        
       Author : lemonberry
       Score  : 86 points
       Date   : 2024-12-06 19:41 UTC (4 days ago)
        
 (HTM) web link (phys.org)
 (TXT) w3m dump (phys.org)
        
       | BrawnyBadger53 wrote:
       | I thought I remembered this model being released many months ago?
       | Or is this a new update to the model?
        
         | xnx wrote:
         | New model. Much better than the last one.
        
           | dylan604 wrote:
           | It's so new that it has no long term track record. I want to
           | see how its record looks after say 1/3/5 years. Any 15 day
           | forecast is just too far into the future for me to take
           | seriously. Even the 3 day forecast is loose at best. By the
           | time it reaches 7-10 days, some of the forecast is completely
           | off when it reaches the 3 day window.
           | 
           | "But DeepMind said GenCast surpassed the precision of the
           | center's forecasts in more than 97 percent of the 1,320 real-
           | world scenarios from 2019 which they were both tested on."
           | 
           | Great, so it's using historical data to make a prediction
           | about what has already happened. I want to see it on current
           | data where it is truly a prediction with no data to make that
           | prediction.
           | 
           | I'm not convinced the results haven't been juiced/salted
        
             | giarc wrote:
             | >Great, so it's using historical data to make a prediction
             | about what has already happened.
             | 
             | That's ok though because it doesn't know it's acting on old
             | data. Researchers would give it data and ask for the 15 day
             | forecast, the researchers then compare against that real
             | world data. And as noted, "it surpassed the centres
             | forecast in more than 97 percent" of those tests. This is
             | all referred to as backtesting.
        
               | IshKebab wrote:
               | It's still not as good as _actually_ new data. The
               | individual model may not overfit the existing data, but
               | the whole system of researchers trying lots of models,
               | choosing the best ones, trying different hyperparameters
               | etc. easily can.
        
               | zactato wrote:
               | yeah, but weather patterns data isn't different now than
               | 10 years ago, right? right? ...
        
               | staunton wrote:
               | Overfitting is always bad by definition. The model learns
               | meaningless noise that happens to help getting good
               | results when applied to old data (be it due to trained
               | weights, hyperparameters, whatever) but doesn't help at
               | all on new data.
        
               | gpm wrote:
               | In principle they trained on data up to 2017, validated
               | (tried different hyper parameters) on data from 2018, and
               | published results on data from 2019...
        
               | dagw wrote:
               | I wish these sort of papers would focus more on the 3
               | percent that it got wrong. Is it wrong by saying that a
               | day would have a slight drizzle but it was actually sunny
               | all day, or was it wrong by missing a catastrophic rain
               | storm that devastated a region?
               | 
               | I've worked on several projects trying to use AI to model
               | various CFD calculations. We could trivially get 90+
               | percent accuracy on a bunch of metrics, the problem is
               | that its almost always the really important and critical
               | cases that end up being wrong in the AI model.
        
               | gyrovagueGeist wrote:
               | Yep, this is also the problem that self-driving cars have
               | with the "accidents per mile" metric.
        
             | zamadatix wrote:
             | 2019 would already be a post-training data prediction:
             | 
             | > GenCast is trained on 40 years of best-estimate analysis
             | from 1979 to 2018, taken from the publicly available ERA5
             | (fifth generation ECMWF reanalysis) reanalysis dataset
        
         | mmmrtl wrote:
         | Yeah the preprint was last year
         | https://arxiv.org/abs/2312.15796
        
       | xnx wrote:
       | 6 days ago: https://news.ycombinator.com/item?id=42319062
        
       | bad_haircut72 wrote:
       | Obviously this team knows way more about this donain than me but
       | I have to ask, wouldnt this only be able to predict weather which
       | is in line with past weather patterns/indicators? I can imagine a
       | weather analyst might be able to see "between the data" and
       | recognise when some anomaly might be brewing, but an AI model
       | would not
        
         | mjburgess wrote:
         | Yup. There's fundamentally no way to automatically verify
         | predictive accuracy, the assumptions in ML papers about such
         | accuracy is (the almost always false presumption) that there
         | are no distributional shifts in the data generating process.
         | 
         | Here, since weather certainly changes its fundamental patterns
         | over time, there is no way of reliably predicting out-sample
         | performance.
         | 
         | It's highly likely that in 1-2 years time we'd take all
         | physics-based predictions and all predictions of current AI
         | models, and find the AI model accuracy drops off a cliff when
         | faced with a real out-sample.
        
           | wcoenen wrote:
           | Weather models can't compute some of the important physics
           | that happens at sub-grid scale, and use various tricks to
           | deal with that. How sure are you that these traditional
           | models aren't also heavily tuned on past data to maximize
           | performance? Perhaps they will also perform badly "out of
           | distribution".
        
             | mjburgess wrote:
             | Causal physical models have no distributional requirements,
             | so they are not sensitive to distribution shifts. ie., A
             | causal model accounts for all possible distributions.
             | 
             | The kind of risk with a causal model is that the model
             | itself, of reality, is incorrect. This is a different kind
             | of risk than there being 'no model at all' as in the case
             | of curve-fitting over historical data, which is radically
             | more fragile to expected shifts.
             | 
             | In general, we're quite good at knowing the limitations of
             | causal models, ie., specifying model risk here is much
             | easier. You even, exactly, point out a known problem in the
             | modelling which is impossible to state for an AI model.
             | 
             | Since the AI model is just a weak induction, there are no
             | terms/structures/etc. within that model which can be
             | analysed to understand what parts of it are sensitive to
             | what aspects of the system from which the data was taken.
             | 
             | All we can say is, we know that in general, train/test
             | distributions have to be "nearly exactly the same" for any
             | of these methods to show anything like cross-val levels of
             | accuracy. So we can very confidently predict when we know
             | train/test wont be the same, that this is a mumbojumbo
             | metric.
             | 
             | Indeed, in the vast majority of common ML examples you can,
             | right now, just go and look at real out-sample data
             | collected later than the "common dataset" and you'll find
             | the val accuracy is random or worse-than-random despite
             | arbitarily high cross-val scores.
             | 
             | The dataset which drives me most up-the-wall on this is
             | house price datasets, or pricing datasets in general.
             | Prices, generally, follow geometric brownian motion and
             | nearly all ML models are extremelhy bad at modelling
             | prices. So it's basically pseudoscience whenver anyone uses
             | these datasets to demonstrate anything, esp. predictive
             | accuracy.
        
           | staunton wrote:
           | Ultimately, a hybrid approach might win out in the end.
           | 
           | Use equations for what we know, use machine learning to fit
           | parameters in the equations, as well as terms we don't know.
           | 
           | Prople nowadays can solve differential equations where some
           | terms are "neural networks" and train those networks on data
           | while numerically solving the equation. (some people call it
           | "neural differential equations", if you want a search query
           | to start)
        
         | rtsil wrote:
         | I know nothing about weather, but aren't changes happening
         | gradually instead of overnight? There's no major anomaly that
         | appears all of a sudden. In which case we can assume the every
         | new change will be incorporated in the AI model's training.
        
           | throwway120385 wrote:
           | They happen pretty quickly sometimes in the PNW. The bomb
           | cyclone we had in November developed within a day and wasn't
           | very well predicted in advance.
           | 
           | You can get a lot of traction with your assumptions but it's
           | that 2-3% of predictions where being accurate matters a lot.
        
         | gpm wrote:
         | > But DeepMind said GenCast surpassed the precision of the
         | center's forecasts in more than 97 percent of the 1,320 real-
         | world scenarios from 2019 which they were both tested on.
         | 
         | > The model was trained on four decades of temperature, wind
         | speed and air pressure data from 1979 to 2018
         | 
         | The test set is taken from after the training set. I agree it
         | seems likely that it will need to be continuously trained on
         | new data to maintain it's accuracy going forwards, but this
         | seems like a promising sign that it should work as long as you
         | keep updating the model.
         | 
         | Of course the "can compute weather in 8 minutes" thing doesn't
         | include the ongoing training cost... so the "this is much
         | cheaper to run" claim seems slightly suspect.
        
           | lispisok wrote:
           | It's much cheaper to run for Google as long as the ECMWF
           | keeps cranking reanalysis for training data and keep
           | generating new initial conditions for forecasts which are
           | generated by combining new observations and the previous
           | forecast generated by the ECMWF physical models which is not
           | cheap to do.
        
       | atonse wrote:
       | This is great from a practical standpoint (being able to predict
       | weather), but does it actually improve our understanding of the
       | weather, or WHY those predictions are better?
       | 
       | That is my issue with some of these AI advances. With these, we
       | won't have actually gotten better at understanding the weather
       | patterns, since it's all just a bunch of weights which nobody
       | really understands.
        
         | CabSauce wrote:
         | But that's the point of a weather forecast. We don't do them
         | for the science. (I'm not arguing against weather research.)
        
           | atonse wrote:
           | I would think we also use these models to run simulations. So
           | maybe the AI models can be used to run simulations with
           | different kinds of inputs to see if doing X in one place
           | (like planting a lot of trees in one area) will have an
           | outsized impact rather than doing Y in another place.
        
         | SpaceManNabs wrote:
         | do you want a causal weather model or one that can predict the
         | weather well?
        
           | jacobgkau wrote:
           | When climate change occurs (which it is), we're going to want
           | a causal one so we can actually make a forecast instead of
           | predicting only based on past (unchanged) data.
        
             | hotstickyballs wrote:
             | What does a causal weather model even mean?
        
               | IshKebab wrote:
               | He meant one that is physically based and interpretable,
               | rather than a black box based on pattern matching.
               | 
               | But I'll take the accurate black box any day, at least
               | for weather forecasting. Climate modelling is a totally
               | different thing.
        
               | SpaceManNabs wrote:
               | Doesn't exactly answer your question, but maybe this link
               | will give enough intuition so that i can be lazy and
               | hand-wave away an answer
               | 
               | https://en.wikipedia.org/wiki/The_Book_of_Why#Chapter_1:_
               | The...
        
             | SpaceManNabs wrote:
             | ahhhh i was hoping someone wouldn't mention this. i wanted
             | to add a caveat but it made my comment look ugly lol.
             | 
             | but like someone else says weather and climate models
             | forecast on different scales and for different purposes
             | usually.
        
         | criddell wrote:
         | Is that similar to being uncomfortable with quantum mechanics
         | and the shut-up-and-calculate advice?
        
         | sgt101 wrote:
         | whelp...
         | 
         | Ok, how would you articulate an understanding of how the
         | weather works?
         | 
         | Let's face it, it's not going to be easy to frame in natural
         | language...
        
           | yunwal wrote:
           | You can take a meteorology course that explains how the
           | weather works. That course is the articulation.
        
             | jncfhnb wrote:
             | That's great but it's not going to scale to the level of
             | detail that makes this better than traditional meterologist
             | models
        
         | bobthepanda wrote:
         | I think it certainly doesn't foreclose on knowing more in the
         | future, based on what we observe with this new model.
        
         | chaos_emergent wrote:
         | How many people can scrutinize the esoteric ensemble of climate
         | models that existing climatologists use to generate weather
         | predictions today?
        
           | jacobgkau wrote:
           | The climatologists can, at least. Can they scrutinize the
           | esoteric ensemble of weights making up this AI model? And
           | which type of model's going to be easier to update based on
           | changing climate parameters, a climate model or an AI model?
        
           | dartos wrote:
           | More than 0
        
           | ianburrell wrote:
           | Climatologists use climate models to predict the climate.
           | Meteorologists use weather models to predict the weather.
           | They are different time scales and disciplines.
        
         | zaptheimpaler wrote:
         | I think this will be the next generation of science to some
         | extent. The things we can understand and call
         | explanations/reasons might be something involving 5 or 50
         | variables with not too many interactions between them. They
         | have some unifying principles or equations we can fit in our
         | head. I think many things inherently just involve far too many
         | variables and complexity for us to understand in a neat theory
         | and we are hitting those limits in biology & physics. Even so
         | I'm sure we will develop better and better ways to interpret
         | these models and get some level of understanding. Maybe we can
         | understand them but not create them past a certain scale.
        
         | dartos wrote:
         | We can work backwards from correct predictions and develop an
         | understanding that way.
         | 
         | But most people just need to know if it's going to be storming,
         | hot or, going to rain on a given day and that is where this
         | shines.
        
           | adriancooney wrote:
           | We can always work backwards, regardless of AI.
        
             | dartos wrote:
             | Sure, but with this new predictive model we will have
             | better predictions to work backwards from.
             | 
             | OC was saying (I'm going to paraphrase) that this is the
             | death of understanding in meteorology, but it's not because
             | we can always work backwards from accurate predictions.
        
               | wongarsu wrote:
               | Or we could wait 15 days and work backwards from what the
               | weather actually turned out to be.
               | 
               | I guess there could be some value in analyzing what
               | inputs have the most and least influence on the AI
               | predictions.
        
         | kolinko wrote:
         | Define "our understanding". With complex / chaotic systems
         | there sometimes are no higher level laws that govern them - all
         | we have is just modeling and prediction.
        
         | raincole wrote:
         | A fisherman was relaxing on a sunny beach, enjoying the day
         | with his fishing line in the water. A businessman, stressed
         | from work, walked by and criticized him for not working harder.
         | 
         | "If you worked more, you'd catch more fish," the businessman
         | said.
         | 
         | "And what would my reward be?" asked the fisherman with a
         | smile.
         | 
         | "You could earn money, buy bigger nets, and catch even more
         | fish!" the businessman replied.
         | 
         | "And then what?" the fisherman asked again.
         | 
         | "Then, you could buy a boat and catch even larger hauls!" said
         | the businessman.
         | 
         | "And after that?"
         | 
         | "You could buy more boats, hire a crew, and eventually own a
         | fleet, freeing you to relax forever!"
         | 
         | The fisherman, still smiling, replied, "But isn't that what I'm
         | already doing?"
        
           | joe_the_user wrote:
           | But AI models in no way free you from work and they are
           | hardly relaxing.
           | 
           | If anything, they "free" people from understanding but that's
           | an activity that many if not most people value highly.
        
             | raincole wrote:
             | Prediction is the only way to test whether the
             | "understanding" has any value.
             | 
             | You can create an understandable physical model. It can be
             | mathematically aesthetic. But if it can't predict the
             | results of real world experiments, it's just "not even
             | wrong".
        
             | jacobgkau wrote:
             | I _think_ that might have been their point. People moving
             | work to AI so they can  "relax" by working on more
             | complicated technical matters (or more AI) are the
             | businessmen, and the meteorologists just chilling out
             | predicting the weather as best as they can with science are
             | the fishermen.
             | 
             | Edit: Just saw their reply to you, so maybe I was wrong
             | about the parable coming across wrong.
        
         | efxhoy wrote:
         | This has been the case for years now, way before the AI craze.
         | We just used to call it machine leaning. The best performing
         | predictive models are black boxes which can't practically be
         | interpreted by humans the same way you can say a linear
         | regression model that gives easily digestible parameters as
         | output. Boosted trees are a great example of very well
         | performing models that quickly become impossible for humans to
         | understand once they get big enough to be useful.
        
       | Always42 wrote:
       | Based on the Gemini release it's hard to take what Google claims
       | at face value.
        
         | amazingamazing wrote:
         | Totally different constraints. It's like saying Nobel prize
         | winning physicists can't be trusted because they get basic
         | arithmetic wrong
         | 
         | Gemini is fast food - it's not supposed to be the best.
        
           | itake wrote:
           | what about Google's flu prediction model release?
           | 
           | https://blogs.ucl.ac.uk/pcph-blog/2018/01/23/google-flu-
           | tren...
        
           | moralestapia wrote:
           | Your analogy doesn't make sense.
           | 
           | Nobel winning physicists can definitely perform arithmetic.
        
             | Maxatar wrote:
             | It's trendy for accomplished people to talk down about
             | themselves as a way to sound cute. It's similar to software
             | developers who like to say "I don't know what I'm doing, I
             | just Google and Stackoverflow all day." It has a certain
             | charm to it, and certainly there are some people for whom
             | it's true, but overall it's just a misguided attempt at
             | being modest but ultimately a horribly misleading
             | statement.
             | 
             | "Do not worry about your difficulties in mathematics, I
             | assure you that mine are greater" - Albert Einstein
        
           | rurp wrote:
           | A Nobel physicist that couldn't do basic arithmetic would
           | definitely raise my eyebrows, but even taking your analogy at
           | face value Gemini was _not_ marketed as fast food slop.
           | Google can 't be trusted to hype products in a reliable way,
           | regardless of their technical details.
        
         | petesergeant wrote:
         | Almost as bad as stuff branded IBM Watson for over promising
        
         | fldskfjdslkfj wrote:
         | what did Google claim about Gemini?
        
         | computerex wrote:
         | Google's work on Alphafold produced a Nobel prize.
        
         | crowcroft wrote:
         | Aside from the fact you're conflating two very different things
         | comparing this to Gemini, what exactly is the problem with the
         | Gemini?
         | 
         | Specifically just that the release was a bit awkward and had
         | some problems? I've found the latest model releases to be very
         | good compared to other frontier models.
        
         | kolinko wrote:
         | Their Alpha* work from DeepMind is actually quite good and has
         | a good track record. LLM/Gemini - yeah, what you said, I
         | wouldn't trust a word their team says.
        
       | wslh wrote:
       | I really wonder if, like the weather, we should see new financial
       | prediction models, understanding that there is randomness but
       | also patterns.
        
       | brcmthrowaway wrote:
       | Meteorologists are quaking in their boots!
        
       | dumbfounder wrote:
       | "More accurate forecasts of risks of extreme weather can help
       | officials safeguard more lives, avert damage, and save money,"
       | DeepMind said.
       | 
       | Did the DeepMind AI say this?
        
         | magicalist wrote:
         | ...why else do you think it's important to predict hurricane
         | paths and tornado spawning storms and flooding rains and heat
         | waves further ahead of time?
        
       | ericra wrote:
       | It does seem like this is one of those domains where new AI
       | models could thrive. From my understanding, the amount and
       | variety of data necessary to make these models work is huge. In
       | addition to historical data, you've got constant satellite data,
       | weather stations on the ground for data collection, weather
       | balloons going high into the atmosphere multiple times daily per
       | location, Doppler radars tracking precipitation, data from ships
       | and other devices in the ocean measuring temps and other info,
       | and who knows what else.
       | 
       | It's incredible that we are able to predict anything this far
       | into the future, but the complexity seems like it lends itself to
       | this kind of black box approach.
       | 
       | *This is all speculation, so I'd be grateful if someone more
       | knowledgeable could tell if if I'm mistaken about these
       | assumptions. It's an interesting topic.
        
       | BadHumans wrote:
       | I'm friends with a meteorologist and the 15+ day forecast is the
       | bane of their existence because you can't accurately forecast
       | beyond a week so I would love to know how they are measuring
       | accuracy. The article doesn't say and I know the paper is going
       | to go over my head.
        
         | n4r9 wrote:
         | I would guess that everyday they're comparing the current
         | weather against the forecast from 15 days ago. Not a lot of
         | data points to be sure, but perhaps enough to have confidence
         | of very high accuracy.
        
           | dgrin91 wrote:
           | Alternatively the can do back testing - using historical data
           | they feed a subset into the predictor, then compare it's
           | predictions to actual history
        
           | diego_sandoval wrote:
           | I don't think that's necessary.
           | 
           | You can backtest the weather for any point in the past, as
           | long as you only use the data that was available until 15
           | days before the day of the prediction.
        
       | choeger wrote:
       | Is that a prediction in the form of "There's a 30% chance of
       | light rain that day." or "Temperature will reach 22,5degC at
       | 14:00."?
        
       | fancyfredbot wrote:
       | It takes 8 minutes to produce a 15 day forecast. That's actually
       | quite a long time for an AI model. I should probably read the
       | paper to find out why but does anyone know? Is the model
       | predicting the weather in 10 minutes time and just run
       | iteratively 2000 times for a 14 day forecast?
        
       ___________________________________________________________________
       (page generated 2024-12-10 23:00 UTC)