300 meters resolution SF Bay Area Forecast

https://sf.atmo.ai

by johmathe • 4 years ago

169 115 4 years ago

sf.atmo.ai

Very cool.

Pedantic but suggest title rename to say “SF Bay Area” or “inner SF Bay Area”. SF is just a small portion of the coverage area.

1-6• 4 years ago

TIL Mt Diablo gets cold

latchkey• 4 years ago

It even snows!

https://www.sfgate.com/weather/article/Mount-Diablo-Bay-Area...

nathancahill• 4 years ago

Wow, this is the future.

pvarangot• 4 years ago

Literally

lostlogin• 4 years ago

I wish it was.

I see no sign that forecasting has improved at all in my part of the world. Auckland, New Zealand.

nikisweeting2• 4 years ago

This release is just for the SF bay area, their models for elsewhere are not released publicly yet.

2Pacalypse-• 4 years ago

This is pretty cool if I do say so myself.

What does the picnic data show?

johmathe12• 4 years ago

It’s a forecast of the best spots to have a picnic in the bay :) Basically a proxy for the best place and time to be outdoors.

cleandreams• 4 years ago

Yes, yes, yes. Great. Thanks. Bookmarked!

mistrial9• 4 years ago

fun - but I am reminded of the excoriation that Windy-dot-com got here on YNews when facts-oriented people started comparing the visualization to more quantitative sources.. snipes aside, no doubts here that they will sell this and be paid for it.

blululu• 4 years ago

That is impressive. Pretty accurate about the early morning cold spots. Not sure how much of this is just matching historical trends (SF has a pretty consistent climate), but the level of spatial resolution on the data is amazing. Not sure how they did it really.

misradeepika• 4 years ago

I’m loving the picnic mode!!!!

Waterluvian2• 4 years ago

A few thoughts from a geographer (be prepared to shoot me):

- Basically every comment is wowed by this, but nobody questions what the accuracy is. I, too, can Krig interpolated surfaces to any resolution.

- off-nadir view doesn't seem to offer much

- We've been dealing with janky tile loading for like 20 years now. I really hope we'll get a much smoother approach for viewing these tiles as they load. The dissolve transition hides it a bit, but makes the data uncomfortable to view when playing the timeseries.

- I'm deeply curious about the Picnic data layer. Can someone share the ArcGIS/QGIS model for that one? =D

wmeredith• 4 years ago

Thanks for chiming in. I can make an animation say whatever I want. I am very curious about the accuracy.

Unrelated: The site breaks my back button in Chrome, which is an unforgivable UI sin.

johmathe12• 4 years ago

Thanks for the feedback. We will have the back button fixed quickly :)

samvher• 4 years ago

This seems to be a thing with weather models more generally. Somewhat relatedly, I've spent quite a bit of time evaluating weather models for use in India and Africa, and while predictions are easy to find, validation results for the predictions are very hard to find. And when you do find them, the results are pretty poor, with many models performing worse than if you would say "predict temperature on date X to be the average observed temperature on the same date in the past 10 years". But people still sell (and buy) these predictions!

Weather predictions seem to be accepted quite uncritically. Perhaps people have a lot of confidence in the smart people that built these predictions (a bit like how AI predictions can sometimes be accepted uncritically).

Waterluvian2• 4 years ago

100% agree. Scientists and engineers all know that you must provide validation results, accuracy/uncertainty calculations, etc. or your data is just a pretty guess. I think weather forecast models are so commoditized and useful for laypersons that we've UX'd all of the complexity (scrutinizing the data) out of the product. The most scrutiny I ever see are people discussing what "Probability of Precipitation" values really mean.

My grad thesis advisor encouraged me to actually get the Environment Canada models and learn how to run them (they're in FORTRAN). I could never make them spit out data consistent with what EC publishes. That's probably on me, but it was a real eye-opener to this whole domain's complexity.

lbrindze3• 4 years ago

Given this is in the SF bay there a number of high quality observations that you can use to validate the forecast skill (unlike India and Africa). I have not bothered doing this here since… well that’s too much like my day job.

I’m always excited to see new forecast products, generally. If I were to guess (as an above comment did) it looks like they are applying some dynamic downscaling on top of either a custom WRF model (expensive and complicated) or more likely already available weather model data like the HRRR, which still would represent a 10x resolution increase.

I’m more curious what the refresh rate is. Anyone can get a super accurate forecast for the next 3 hours that takes 10 hours to run, but at that point it’s no longer a forecast by the time the data is available.

I still think that windy has set the standards as far as modern weather visualization goes. Not saying everything has to be particles but other things (like the inclusion of isobars) is really clean and not trivial to execute.

Either way this has definitely piqued my interests and I will be keeping an eye on it, their advisory board looks legit (at least in the meteorology end)

duanem2• 4 years ago

I've been working with weather models for 10 years and I often get asked "How accurate is X?" or "Which model is more accurate?" Many people think "accuracy" is a single number or a single thing - it is more complex than this and depends on your needs.

This chapter on Numerical Weather Predictions [0] is great, especially the section on "Forecast Quality and Verification" (p777). The eye-opener for me was "Binary/Categorical Event". An example of a binary event is rain, one model could predict rain correctly but a second model might not predict the rain at all. This doesn't mean the second model was completely wrong, it still predicted the rain but it predicted the rain passing further to the south.

[0] https://www.eoas.ubc.ca/books/Practical_Meteorology/mse3/Ch2...

I've also noticed some model are better than other at predicting one phenomena while other models might be better in certain regions. For example, many people report that Canada's GDPS is better at higher latitudes whereas NOAA's GFS is better at equatorial regions.

One final note, just because someone is solving an WRF model without verifying the results, doesn't mean it's wrong. Many numerical techniques and physical models within WRF have been validated analytical and experimental models. But it is also true that someone can naively setup a WRF model that gives bad results.

I use a 900m WRF model that predicts the wind shadow around an island and we use it to find the best beach for a picnic - and it works. But this same model predicts the general pattern of rain but it doesn't get the start and stop time of rain correct.

People get fixated on accuracy as a single thing and use it as a single basis for argument but to take a quote from the chapter [0] above "One of the least useful measures of quality is forecast accuracy" (ref. p777, Forecast Quality and Verification, third paragraph).

_Microft• 4 years ago

I had been wondering if that was actually something special because I thought I remembered that the German Meteorological Service ("Deutscher Wetterdienst") offered accurate forecasts on a sub-kilometer grid for years already. At least if you are ready to spend money on that because that service is not free, so maybe that's the innovation here.

namibj3• 4 years ago

Where did you get the existence of sub-km (horizontal) grid forecasts from? Thus page https://www.dwd.de/EN/research/weatherforecasting/num_modell... contains the following sentence: "The operational NWP models of DWD currently employ horizontal grid mesh widths between 2.8 and 13 km."

The convective cell tracking for nowcasting seems finer, IMO reasonable as it's about predicting watersheds down to <10km² area flash-flooding and causing the local creek to swell to actually dangerous levels/requiring partial evacuation of a valley.

hadlock3• 4 years ago

Few thoughts as a sailor in the SF Bay area.

- Accuracy seems at least somewhat correct.That wedge shape you see in the late afternoon sailors call "the wind engine". Local sailing magazine Lattitude 38 has a special PDF that talks about doing a sailing trip around the bay accounting for this local wind phenomenon.

Correct stuff:

- The SF waterfront, out to the edge of the piers is mostly calm which is correct

- Berkely, Oakland, Emeryville getting blasted late afternoon is correct

- Back side of treasure island, immediately to the east is much lower than the west side, particularly near clipper cove

- Vast majority of alameda estuary is dead calm, that's correct for this time of year

- There's a big blast of wind between Daly City and OAK international where there's a gap in the mountains

Weird stuff:

- Most noticable, is the wind is still strong up to and south of the bay bridge. The bay bridge has been described by many as "a wall" when it comes to the wind. There's a drop off but it's not in line with the bay bridge. At all. at least 45 degrees off from true wind speed.

- There's a very windy patch between golden gate coast guard station and belvedere, it's usually really patchy wind here but I guess if the wind direction is just right it'll blow there

- Pointe Bonita (lighthouse on the west side of gg bridge about 2 miles, north coast) they are modeling the gap in the rocks there and you can see it funnel through which is neat

It's a cool visualization though, gives you a great idea of where the wind is, and more importantly where it won't be. There are a bunch of races that start in the bay and head south towards Santa Cruz and Monterey so it's nice to better visualize where the wind just dies off on the coast as it skips over the mountains.

Anyone who wants to see what the wind is like in the bay I recommend reaching out to YRA.org they can put you in touch with a boat who needs crew most likely. There are races 4-5 days a week through november all around the bay. It is modeling a distinct drop off of wind speed on the south side of the bay

johmathe12• 4 years ago

Very valuable feedback - thank you!

UniverseHacker2• 4 years ago

Also, as a person with experience sailing on the Bay, this model immediately seems unusually accurate. It shows, for example, the Angel Island wind shadow moving around correctly as it does during the day, which I have never seen in another model.

Could anyone with more understanding of meteorology (or OP) please explain what is different about this model vs say the ECMWF model that you can see in apps like Windy, that are supposedly great, but just don't seem to get these features right? Those models are incredibly bad when dealing with the unusual local geographic features on the bay. What resolution are they operating at?

s1artibartfast4• 4 years ago

Basically every comment is wowed by this, but nobody questions what the accuracy is

I am very skeptical. Does the San Mateo bridge really block 10 knot winds for the entire south bay? Similarly, the land temperatures all seem the same close to sea level.

hadlock3• 4 years ago

Tall bridges do weird things to the wind. I can confirm the bay bridge at surface level, there is functionally no wind for about half a mile downwind from it. Just glassy smooth.

georgeburdell• 4 years ago

Not working on my iphone. Slashdotted?

bombela2• 4 years ago

How do you switch to normal (for 97% of humans) units for temperature and wind?

froidpink• 4 years ago

The heaviest users of wind forecasting use knots

pklausler• 4 years ago

I learned to interpret wind speeds in m/s on a visit to Iceland and found it to be an intuitive unit for hiking and other outdoor purposes. And 1m/s is pretty close to 2 knots so it's easy to convert.

johmathe12• 4 years ago

We will be adding the Celsius switch shortly. We started with F since we expected most users of the forecast to be in the bay area but it is great to see interest from non Fahrenheit users :)

bombela2• 4 years ago

The bay area is probably the place with the fewest native users of Fahrenheit in the whole USA.

All those US customary units should never be the default in 2022. It's time to move on people. The whole world and the majority of the industry in the USA share the same system of measure. Let's get the American population there too.

modzu• 4 years ago

shameless plug of my little rain map that has the opposite approach --

zero interpolation, zero forecasting. just the real data from the radar feed in 10 meter resolution at the current time (plus 3 past snapshots at 1 hour intervals):

https://truweather.link

and the app versions:

https://play.google.com/store/apps/details?id=net.conceptual...

https://apps.apple.com/ca/app/truweather/id1537614881

i find it useful because every other weather app is wrong in some way ;) instead, it enables you to form your own mental models

nfriedly• 4 years ago

What does "300 meters resolution" mean in this context?

johmathe12• 4 years ago

It means that the underlying weather data is computed and validated at a 300x300 meters resolution. Hope this helps :)

carabiner3• 4 years ago

How are you doing validation?

lispisok2• 4 years ago

Weather models divide the atmosphere into cells where each cell has it's own forecast. There are several ways to do it but if you imagine the cells as cubes, the north-south and east-west dimensions of the cube is 300m long. 300m is very high resolution for weather models with most running at a few kilometers to a few dozen kilometers.

swimfar• 4 years ago

Also note that it should be "300 meter resolution". You don't use a plural when using it as an adjective. I only point this out to help people because this is a common grammar mistake for non-native speakers.

lispisok2• 4 years ago

If I had to guess they are running the WRF model [1][2]. The AI part is post-processing the model output. With a fair amount of reading the manual anybody can run their own WRF. WRF scales from running on a laptop to supercomputers with 1000s of cores.

[1] https://www.mmm.ucar.edu/models/wrf [2] https://github.com/wrf-model/WRF

RosanaAnaDana2• 4 years ago

I mean up or downsampling is trivial. The question isn't if you can make a raster at any resolution, its if you can make a raster thats accurate and precise at that resolution.

Its not clear to me that this is either.

johmathe12• 4 years ago

One of the interesting things the model captures at this resolution is the dynamics of the wind going in the bay through the golden gate. See for instance: https://sf.atmo.ai/wind@37.80911,-122.44543,11.68,36,0,16669...

lbrindze3• 4 years ago

sure anyone can do it, but it takes a lot of computers to do it at a reasonable refresh rate. HRRR is 3km and available for free. no super computers required by the end user. then you can apply what ever AI/statistical downscaling you want without having to try and run a better/more reliable forecast than the NWS

milancurcic• 4 years ago

Having worked with WRF for 13 years now, contributed patches to many releases, and built a SaaS business centered on WRF (https://cloudrun.co), I still discover new things about it and run into interesting scientific and engineering challenges. It's not the kind of software that's easily picked up and run by a non-expert. It's a large framework that's more niche, more obscure, and not as well documented as something like, say, Tensorflow. There's still a ton of value to derive from making WRF and similar models more usable by non-experts and without access to supercomputers.

neurostimulant• 4 years ago

I remember tinkering with WRF model in university using a single server. Depending on the resolution, generating forecast for the next day can take more than one day! Imagine getting forecast for tomorrow two days later! You do need quite a bit of computing resources and GPU if you want to use its forecasting feature meaningfully. In my case, I was interested in downscaling some historical data, so my limited compute was enough.

nicolashahn• 4 years ago

I've been wanting this for a long time. Any way I could get it as a live updating widget for my Android home screen?

dopeboy• 4 years ago

Opportune day you picked to share this, given this nasty cold front. Very cool.

asdff• 4 years ago

This needs to be done for more cities in California, especially Los Angeles with its huge geographic area and all the diverse micro climates contained within. Sometimes the weather changes more than 15 degrees in half as many miles from the coast. It therefore doesn't make sense to e.g. check "weather in LA" when its going to be somewhat wrong most of the time depending on where in LA you happen to be, since the little widget that pops up on google search for "weather in LA" doesn't exactly tell you where in the 500sq miles of LA they are putting the temperature probe.

issa• 4 years ago

I feel like you could make a whole movie about the weather in LA. It would change your life. Twice.

PaulHoule• 4 years ago

That's true, but an upstate New Yorker would wonder what the fine resolution forecast looks like for a place that really has weather.

CalChris2• 4 years ago

Not weather, sun, sun, sun, sun, sun!

spullara• 4 years ago

Recognized the reference immediately. What a movie.

https://tvtropes.org/pmwiki/pmwiki.php/Film/LAStory

sosodev• 4 years ago

Same thing in San Diego. I feel like the popular weather services can't make accurate predictions or even tell what is happening most of the time. My iPhone will tell me it's raining when it's sunny outside.

syntaxing• 4 years ago

How much computational power does it take to make something like this?

RosanaAnaDana2• 4 years ago

so whats the RMSE and how do we know that?

orangepurple• 4 years ago

Looks amazing! However, this model (like all models by default) is not validated and produces garbage unless proven otherwise.

modo_• 4 years ago

Running over in Rodeo valley (Marin) there was a very noticeable (and unusual) inversion this morning around 7:30 am. It must have been 5-10 degrees colder on the valley floor compared to up higher -- after ~100ft of elevation gain up out of the valley it warmed up very rapidly.

I don't see that reflected in this map at all fwiw. https://sf.atmo.ai/temperature@37.83200,-122.51075,13.53,20,...

Workaccount2• 4 years ago

It's likely because services like these use models as data feeds and not live/recorded data.

zestyping• 4 years ago

Is there a Celsius mode?

anyfoo3• 4 years ago

Growing up in Germany, before I moved to the Bay Area, I was wondering why weather apps and widgets were so prolific. Sure, knowing the forecast for next weekend was nice, but for anything closer I'd just get out of bed and look out of the window. That would pretty much tell me what weather it is, and it would usually change just slowly over a few days or so.

Then I moved to the Bay Area, and weather does not only change quickly, it may also be vastly, vastly different just a short distance away. Temperature differentials of 10°C or more within just 40 miles are interesting enough that it's a frequent topic of conversation with friends in Germany.

Everything suddenly made sense. The weather widgets. The hoodies: Easy to put on or off.

anyfoo3• 4 years ago

Another anecdote: In the South of Germany at least, long stretches of sunny days are often followed by sudden thunderstorms with equally sudden bursts of rain. That "fact" had been so deeply ingrained in me that it was subconscious. You'd have a careful feeling if it was hot for too long, suddenly you might find yourself running for the next awning to escape the torrential rain.

Sunny weather was a bit like building up a sort of pressure, that must release violently.

Took a while to let go of that feeling in California. It basically rains in winter, and does not rain in summer. Like, at all.

(Note that I moved away about a decade ago, climate may have changed in the meantime.)

s0rce• 4 years ago

Midwestern USA has similar stretch of hot summer followed by thunderstorms.

dan-robertson• 4 years ago

I hadn’t even thought of that until I read this. I had thought thunderstorms after hot weather were just a fact of life. I guess in places near German latitudes that get thunderstorms the hot weather is caused by high pressure systems but maybe that isn’t really the cause in California.

wrycoder• 4 years ago

When I frequented Portola Valley, we called it six months of mud, six months of dust. Those golden hills are actually brown.

gonzo41• 4 years ago

Thats what we used to have in Australia, and now it just rains. It's killing my morale atm.

Fatnino• 4 years ago

The climate did change. Now it hardly rains in the winter either.

prottog• 4 years ago

It basically rains in winter, and does not rain in summer. Like, at all.

You needn't have moved so far from southern Germany to experience that! Lack of rain in the summer is a defining characteristic of the Mediterranean climate. You would have experienced the same a few hundred miles south. :-)

FullyFunctional2• 4 years ago

Yes. I’m from Denmark, but I check the weather every night before stepping out as I have experience 13 C nights where the previous night was 20 C. And this is not uncommon.

When I first got here I was stunned by how noticeable nicer the weather was when driving from Santa Clara to Palo Alto, and more than once have I forgotten to bring a sweater to SF.

anyfoo3• 4 years ago

Yeah. Living in SF and working in the South Bay, it's common in the evening to get into my car sweating, and coming out freezing. I always pack a hoodie.

On the bright side, a hoodie is often all I ever need, all year long.

tiagod2• 4 years ago

My hometown is close to Portugal's westernmost tip, about 30km from Lisbon. Its not uncommon to see differences over 10C between the two, and the car ride will many times include rain or fog only in a specific spot of the highway, half way.

Even if you stay right next to the ocean/river during the whole trip by avoiding the highway you'll notice a big difference most days.

londons_explore2• 4 years ago

Weather forecasts are so hard for a user to evaluate... Are you going to check it every day and remember how many days it was right or wrong?

Please can weather providers just publish a headline statistic of "Our rain/no rain one day ahead forecast is right 85% of the time. That is better than NOAA (80%), Met Office (72%) and weather.com (65%)."

carabiner3• 4 years ago

This is kind of the purpose of the "50% chance of rain" things. The process is called calibration and is usually done with linear regression, and it means that in historical forecasts, the actual outcome was rain 50% of the time. Surface precip is notoriously hard to predict, so this is what we've got right now.

londons_explore2• 4 years ago

But they should publish that... And then compare that figure to their competitors... To demonstrate to their users that their service is actually better, not just has a shinier UI...

tiagod2• 4 years ago

Would be nice to have some extra error information. Like a probability for a number of mm precipitation buckets.

namibj3• 4 years ago

On part that's what ensembles are about: they run like 20 or so simulations of the near future, randomizing the start values and simulation parameters within the range of uncertainty, and get one potential future per simulation out of the process.

A 30% rain chance every hour of the entire afternoon until sunset tends to be less actionable than "85% chance of a 50~80 minute rain shower during the afternoon, unclear when, but likely a bit longer if it'd start very late", as one can often adapt around such, for example by scheduling the homework to get done "wherever it's raining, at the latest so it'll be finished by dinner time", spending the dry time with outdoor physical activity that doesn't care about wet ground (but getting wet from above isn't nice).

nikisweeting2• 4 years ago

Atmo built their business on beating their competitors on a particular ranking metric that does exactly as you describe. I discussed it with the founder but am not sure how much of that convo was public info so I hope they post it here.

s1artibartfast4• 4 years ago

Very interesting data but infuriating UI.

With so much heat map display, why is there no key? For wind, red/purple are counterintuitively lower speed than orange and yellow?

Also, who do so many applications force oblique views?

lucasmullens• 4 years ago

Also, who do so many applications force oblique views?

I'm able to adjust it by dragging with two fingers on my trackpad, which I think is the standard behavior for that (albeit hard to discover). But I do agree, it's weird for that to be the default.

There's also probably no key because the colors are mostly transparent, so it would be hard to make a key easy to understand. Labelling the contour lines seems like a reasonable approach imo.

s1artibartfast4• 4 years ago

Interesting, I don't get Contour labels on mine, so I have to continually reposition to get a sense of the magnitude

FullyFunctional2• 4 years ago

Sorry, I’m from the present. How does one change it to Celcius?

rnk• 4 years ago

Interesting. Their website has a real "Delos" feel, as in the West World corporation. I'd love a narrow cast for my area too. You could sell access to these maybe.

geuis• 4 years ago

Pretty cool, but the map really needs to have a street label option.

bandyaboot• 4 years ago

The secret sauce to making accurate forecasts at 300 meter resolution has to be something more than “AI” right? I mean, there’s little reason to think AI should be meaningfully better at filling in details at 300m resolution than it would be at picking lottery numbers, no? Am I thinking about this wrong?

labster• 4 years ago

Nah, just run the WRF model at a 300m horizontal resolution and use a regional model or GCM for the boundary conditions. No AI necessary.

evilotto• 4 years ago

There's lots of weird little vortices where the wind seems to start or end in a spiral. I'm guessing it's an artifact of the modelling. Funny, but overall very cool.

keepquestioning• 4 years ago

Is there a site that predicts the movement of SF fog?

sailor1234• 4 years ago

Love the app. The details that the map is displaying is really important if you want to know the forecast for a specific location, which is really important if you're on the sea or in the air. Looking forward to have the forecast for even more locations around the world

_tom_• 4 years ago

Displays nothing but garbage on mac/firefox.

I did briefly see something that looked like a map, but then it smeared into an abstract art looking blob.

300 meters resolution SF Bay Area Forecast

Related Stories