Users of Windy: why do you want historical weather data?
-
Hey @StephenBeste, welcome to the world of Windy! Ivo and the team have really created something special here. So, as others have mentioned, the "historical" data you need doesn't go back especially far... And it sounds like you want to understand model performance for your specific location. There's a technique called Model Output Statistics (MOS) that deals with this. Although model verification is a whole science in itself, I've also been interested lately in the idea of creating something like that on a more basic level. Maybe I'll look into it again...
@Franzus-0, of course your post counts! Do bear in mind though, this is merely a discussion; I must reiterate that I don't work for Windy, and I'm not collecting "requests" as such! Congrats on the new vessel, and glad you found your way in the end :) As I understand it, that was one of Ivo's main motivations for Windy; that everyone could have access to the best data possible for free. I was blown away when I saw that they had the full 9km ECMWF fields on discovering Windy; I found myself asking around about Windy at the next ECMWF workshop I was at :) As to the historical data, you seem to be looking for something similar to @rfleming – essentially a type of analysis chart.
Some kind of MOS-based analysis charts that go back a few months, perhaps with some satellite and MADIS data mixed in.... this might have to be my next project :) Thank you so much all for your input. Please keep the discussion going if you have any other perspectives on historical data!
-
@johnckeal Hi I've just registered as a member with Windy and my interest in historical weather data is for fire weather. Cheers Mel
-
Hey @Coop4433, welcome to the community! Fire weather sounds badass; can you be more specific? I assume you'd be interested in winds at multiple model levels, and maybe some particle dispersion models. Model history of CO2, air quality, aerosol... that sort of thing? Observations?
-
@johnckealy said in Users of Windy: why do you want historical weather data?:
some
I would love to have access to historical weather data for aviation purposes.
I work for a charter airline that operates to destinations all over the world. We would love to have this data regarding averages for a given month. This is crucial when trying to figure out our Take-Off and Landing performance for future operations.
Whats the average high for a given month or week for a specific time during the year? What are the average winds during a specific time of day? Do we need to add a stop somewhere because recent annual trends show its going to be too warm? Or can we make a specific nonstop route because its trending cooler?
Some websites try to provide this data...most seem inconsistent. However, we would love to see this information come from Windy.
-
Hi @garlandz, thanks for the comments. It's a little unclear what you mean. Sounds like you're talking about a climatology, are you not? The "average high" (do you mean temperature?) over a week would be quite different to that of a month, or for a certain month across many years. What are "recent annual trends"? Is that a reference to climate change? Can you give an example of a website that tries to do it? Are you speaking to purely observational data (at airfields), or is there a model component (remember that Windy are primarily focused on model visualization)?
-
@johnckealy said in Users of Windy: why do you want historical weather data?:
Hello. Yes climatology would be the correct term I am looking for. I would love for Windy to have "average highs and lows" data built in for airfields, especially under ICAO codes. Would be extremely helpful! Right now we take data off of Meteoblue. But I would like to have more accurate data taken from the history of METAR reports if it is possible.
I guess what I am looking for overall is a history of purely observational data to reference at certain airfields built into windy.
Thanks for your time.
-
Ah well here's a coincidence @garlandz; looks like the guys at Windy just added this!
https://community.windy.com/topic/8710/windy-predicts-extreme-wind-temperature-and-rain -
The new future is great! Thanks for the quick responses. I look forward to windy becoming better and better everyday.
Best Regards,
-
@garlandz I feel I should mention again (third time this thread!) – I don't work for Windy! This thread is a discussion between users :)
-
@johnckealy hot air baloonist and GA flyer here - my reasons and use for historical data:
(1) assume I do a flight and some meteorological situation crops up which I did was not aware of/did not look at
(2) assume some flight accident happened. The press typically reports sensationalistic nonsense like so: https://www.austrianwings.info/2019/06/heissluftballon-in-kilb-abgestuerzt/ . And a proper accident investigation takes forever, und you might not hear of the results. The national ATC organisations arent much better - they typically delete forecasts right away - no going back.
In both cases I would like to go back in time, and see myself if I could have avoided the situation - basically for my learning, and to answer questions of concerned people.
Both use cases involve a time window of a few days max. Two days would already be great!
Personally I have no use for historical data going back further in time.
-
Thanks for the comments @mhaberler. You've hit on an important concept here. Is there such a thing as a "historical weather forecast"? You seems to be alluding to model data. The important question is this – could you have avoided a dangerous situation if you had looked at the forecast an hour before flying vs. 3 days before flying? Which model run is the one to reference?
We've talked about creating more detailed surface analyses before. But with an investigation, for example, you would need to know what forecast data you had available at the time (and what's its lead time was). Not only that, but which model. Weather forecast models are an extremely fluid thing, and are just a tool. This is why, when stakes are high, companies shell out for private forecasting companies who employ real forecasters. And with a proper accident investigation, it's usually time to do that. I can't speak to other modelling centres, but I know the UK Met Office archive all their UM forecasts (which cost an obscene amount in storage), so that they can be referenced later.
But, if we're just taking about your own learning, it sounds like a surface analysis would do the job nicely. Surface analyses are traditionally just synoptic charts (isobars and fronts), but I think we can do better. I may just start working on something like that myself (though with one year left, I should probably make a stab at starting to write my PhD thesis).
Let me ask you this @mhaberler, what level of detail would you need to "see if you could have avoided a situation"? Highs and lows? Surface winds? Gusts? Winds at all levels? Precipitation? Visibility? As gridded data or just at airfields? I have some aviation forecasting experience, but we were very focused on just our own airfields, and the WAFC/LOLAN guys used to take care of the rest.
-
@johnckealy said in Users of Windy: why do you want historical weather data?:
Thanks for the comments @mhaberler. You've hit on an important concept here. Is there such a thing as a "historical weather forecast"? You seems to be alluding to model data. The important question is this – could you have avoided a dangerous situation if you had looked at the forecast an hour before flying vs. 3 days before flying? Which model run is the one to reference?
that is the secondary goal
First, I would love our wonderful and super-expensive ATC corporations would provide their forecasts not just from today onwards, but a couple of days back. Just what they publish on their aviation briefing website. That would cover 90% of my learning needs. Could I have avoided that situation if I had read the aviation briefing more diligently? Note: no historical 'data', no fancy UI and yadayada - JUST their briefs. So wrong forum here.
the other 10% of my needs would be going back in time for a few days max and look for stuff which was NOT in the aviation briefs. For instance, the accident linked above was due to a cold are pressure wave within the inversion layer, and it was NOT in the aviation briefs.
Let me ask you this @mhaberler, what level of detail would you need to "see if you could have avoided a situation"? Highs and lows? Surface winds? Gusts? Winds at all levels? Precipitation? Visibility? As gridded data or just at airfields? I have some aviation forecasting experience, but we were very focused on just our own airfields, and the WAFC/LOLAN guys used to take care of the rest.
I know.. the aviation weather forecasts are of so-so for ballooning and of moderate value at best if you do long-distance/high altitude. Example: predicting northerly föhn a few days ahead - in reality you get hints for föhn-affected terminal areas like Innsbruck and not much beyond.
For a short-range check a few hours before a pedestrian flight I mostly do precipitation, visibility, clouds/fog, average wind, wind gusts, shears. An important value for us is winds/gusts in the planned landing area, and here forecasts outside of large airports scarce and unreliable - we replace that by webcam views (sometimes you can judge wind that way), private weather stations, and experience about typical valley wind systems.
for long-distance over mountains a lot more is involved - vertical profiles, soundings, wave prediction, sometimes even likelyhood of a hydraulic jump. For longer-term prediction Rossby waves in the FL180 Geopotential forecast help - black line cutting the Alps N/S is a good predictor a few days out. As the flight comes closer, you look at the more local phenomena - ground wind etc like for a pedestrian flight.
-
Nice one @mhaberler, sounds like you know your stuff! The Föhn effect is a really tricky beast to forecast; actually, pretty much all mountain-based phenomena in general are a nightmare to get right. As people rely more and more on the models, the forecasters who can actually understand and predict this stuff become a dying breed... But that's a general thing, and for the marjoity of these phenomena, it's only the pilot that will ever see it, not the weather models, and not the observations network. So historical data wouldn't be much help anyway right? But then again, the charts can certainly show you the right conditions for these things to form. If pilots had a medium to add in local scale phenomena that they saw on a flight to a historical chart, that would be pretty cool. That way, they could reference what they saw when that setup reoccurs. I'm getting more and more interested in creating this myself...
By the way, with the aviation briefs, what's the "aviation briefing website"? Could you give me a link? Also, I just spent some time on working on my skewT platform tephigrams.org over the weekend, some vertical profiles, if it helps at all.
-
@johnckealy yes, historical data for would not be relevant back more than a few days. But then I am just an amateur pilot.
aviation briefing: well Air traffic control organsations run weather briefing and flight planning services, like for instance Nav Canada - I guess in the UK it's https://www.nats.aero
-
@johnckealy I see this thread is a bit old, but thought I would respond anyway. As an aviation accident investigator most all of my work is forensic in nature, and the ability to efficiently pull historical weather information is extremely valuable. Always looking for more "tools" to put in my toolbox, I really like how intuitive and simple to use this application is, and I can load up a flight track from radar or ADS-B in a matter of seconds. With historical weather information available, I can get a "quick look" at the weather at a particular time of an accident flight track in seconds and make this information available to other investigators responding before ever even arriving on site. Absolutely invaluable. Just thought I'd chime in and let you know maybe something some don't think too much about!
-
As a backcountry skier, being able to review the last 24 – 72 hours of wind history would be extremely helpful to understand where I'm likely to encounter slopes that have been wind-loaded with snow. This is frequently a major route planning consideration. Would be happy to pay for this as a "premium" feature.
-
very simply to plan a long passage. eg .shall I do it in may or september?
-
@Mais78
With the purpose of long passage, you better use the Pilot Charts, which give a probability of wind and current month by month. -
If historical model data would be available, I’d use them for evaluating in retrospect which model would have given me the best forecast. In my case data for the last week would suffice for that purpose.
-
In this case you could use the Observation vs Forecast feature. The forecast was displayed at same time with observed wind and temperature during several days. Unfortunately at the moment the past forecast is shown only for few hours. But this feature should be back in future.
https://community.windy.com/topic/9425/how-to-use-the-observation-vs-forecast-feature