Weatherkit Forecast Model
-
@Wheats I really liked your idea at first but then I tried to do it. The color picker in the app is so hard to use I actually had to turn on the accessibility magnification feature in Android to be able to get close to the colors I wanted (green, yellow, red). Finally they were good enough but when I went back to the forecast panel only the colors on the map had changed. The chart still uses the default colors.
Did I miss a setting somewhere?
I will admit with this change for the first time I'm seeing the map as slightly better than completely useless. When I zoom in to view a 44 mile round trip bike route the wind is the same everywhere and the map becomes a gigantic go/no-go light. A little dot would convey the same information.
-
It would also help if the customized color scheme applied to the widget as well. Then I could make the chart readable with colors and not have to try to read the tiny text.
Maybe I'm a unique case because if the gusts are below about 10-12 mph I go biking and if it's over I stay home. So I want to see a huge color change between 10 and 12 mph and everything above and below can just be two colors.
An option for customizing chart colors is the only way it makes sense to do this because most people probably like the default color variation which does show the data in a usable way if you memorize the colors.
-
@tessellahedron I don't think the colour customizations are linked in that way - between the map and the widgets/forecast panels. I like that idea though. You should put this in a clean post in the Feature Request section - custom colour scales in the app affecting the forecast panels and widgets.
It's interesting - you lean towards maps being useless because you are focusing on a certain location. In my world - I use both, but use the maps way more often to gain a picture or perspective of what's happening on a larger scale and what may or may not being coming in the future. I'll look at a line chart when I need to analyze a specific trend for an area or analyze specific data from an instrument...but not for a forecast - unless it's an ensemble or some long range trend analysis.
Personally - I've done all my settings changes using the website...much easier to use. So I can't relate to the trouble you have with seeing the text.
The website is MUCH better in my opinion. But I get it, some may not have the option of using the website. -
@tessellahedron For the maps, it's a matter of preference but they are also useful as you get to know your area better.
For example, if I just look at a graph output that says it's going to be a south wind of 10mph, that's helpful but if I look at a map version of a forecast, I can see much more clearly how likely that is to happen given the current setup.
The reason is I can see how a particular storm is "shaped", what is the tilt of the pressure gradient, what are the winds above the surface doing (those can enhance ground level winds depending how they setup, or steer a storm away), will a rain shadow likely form, etc..
All this amounts to me being able to make a better guess at a hyper local forecast of wind and rain.
Ultimately, no single source is going to be right all the time. For the most accurate forecast, you need to combine multiple models and sources of information. It's why professionals also use ensemble forecasting, where they run a model multiple times with slight variations to average out the inherent shortcomings of a single model run.
-
@illili @Wheats I see how a zoomed out map that can show storm "shape" combined with analysing conditions in the upper atmosphere could be useful to a meteorologist. That's their job after all. It means absolutely nothing to me.
All I want to know is whether or not I can go biking.
I liked the idea of comparing forecasts to get a better idea of what might happen than from using just one forecast, and eventually learning which model(s) is accurate in my area. I tried to do exactly that yesterday afternoon and had terrible results tho.
I think gusts are being added to Windy's forecast compare feature, but for now I used Flowx to compare 8 forecast models that included gusts (for some reason ECMWF has no gust info in Flowx). All but one of the models showed wind gusts below 11 mph. Basically it looked like a great opportunity to bike, so I went for a ride.
Before leaving I checked the route forecast using Windy, leaving it set to the default model (ECMWF I think). That forecast showed gusts of 16 and 17 mph, a definite no-go. I went anyway hoping it would be more like what the other 8 models predicted.
It wasn't. I have no way to validate wind forecasts with actual readings while I'm biking (that's another issue I've wanted to solve but haven't been able to find an anemometer or other method that would work on the bike), but the wind was whistling around my helmet and I've heard that doesn't happen until at least 18 mph.
If I had to estimate the gust speeds 15-20 mph would sound about right, but that's a wild guess just based on how it felt and past rides/forecasts. With turns in the route and lulls in the gusts I would say only 1/3 to 1/2 of the ride was really unpleasant. For a 3 hour ride (my normal route) that means being battered by gusts for over an hour. Plus, a bad cross-wind gust right when a vehicle is passing could literally get me killed.
I don't know what to do. Idiots in other forums have suggested an exercise bike in the past and while that's the stupidest, most irritating suggestion imaginable, it's also the only one that's guaranteed to solve the problem completely, other than maybe moving which I can't do right now. I doubt words can convey how much that pisses me off.
-
@tessellahedron I downloaded that Flowx app, it's pretty nice. What's worth noting is you're looking at the same data as on Windy for comparable models. If you prefer that interface, then it might be a good choice for you to add to the mix.
Something about Windy that I'm not sure if you've used yet; you can use the wind gust map for any of the available models in the map view we've mentioned.
I also just wanted to highlight something we've touched on a few times but it bears repeating:
You seem very interested in getting more precise than what your average app (which is built for average folks who don't care about the minutiae of weather) can show.
Unfortunately, no model/forecast on any app is 100% accurate. Even high-resolution models like the HRRR, NAM, HRDPS, etc. all have weaknesses and will not provide an exact forecast. If you want to get more precise than what you are finding in graphical readouts/standard apps, you have little choice but to become a bit more of a weather nerd to learn your local area better and utilize more tools.
I certainly don't think you need to quit biking, but I do think you'll need to take more of an interest in forecasting weather if it's impacting your rides that much.
-
@tessellahedron @illili said it right. It's the reason people in my profession get yelled at for getting the forecast wrong haha. They put to much faith into guidance models that WILL have varying issues that you as a weather enthusiast need to learn and account for. The only way to get a more precise and accurate forecast for your specific area is to keep doing exactly what you're doing, checking all the models, comparing, checking the actual conditions and asking yourself - why was that wrong on all these models? Even then, you'll probably get it wrong from time to time because the atmosphere is unbelievably complicated.
Also, a pretty solid weather instrument that you could probably mount on a bike - my suggestion is Kestrel Instruments.
-
@illili I didn't think it was unreasonable to expect the wind not to whistle around my helmet (ie, it's >18 mph) when the forecasted gusts are 10 mph or less.
Since I've consistently had problems like that for over two years now I'm starting to conclude that forecasts are just inaccurate to the point of being useless around here and all I can do is move.
If nine out of ten forecast models are wrong for this area, and those were created by people who dedicate their lives to studying weather, what are the odds that I'm going to be able to sort out some kind of accurate forecast based on applying correction factors to the models that are wrong. I used to think that might be possible but now that I've watched long enough to see that all the forecast models are wrong by varying amounts it doesn't seem even worth trying.
-
@Wheats Kestrel Instruments look decent. One of the main requirements tho was for the meter to have a threaded hole so I can screw it onto a little photography mounting arm that clamps on my handlebars.
Just over a year ago I actually bought one of those handheld anemometers and mounted it on a bike with that arm. The propeller spun and it seemed to be working but the LCD display was totally illegible in sunlight. I had to crouch and get my chin basically on the handlebars to read it. The display seemed to be about average size.
I could use one that records but then I'd have to somehow merge that data with my speed and direction throughout the ride to see accurate wind readings. I had planned to estimate wind direction and do the vector arithmatic in my head to subtract the bike's speed as displayed on the bike computer. That's why I couldn't use the anemometer with the smartphone app acting as the display, I need to run a bike computer app on the smartphone to know my speed.
The only successful project with an anemometer on a bike that I'm aware of is https://cdacrr.blogspot.com/
It uses a Weatherflow device, either Windmeter or Weathermeter. The app isn't made for wind readings, it's meant to help racers perfect their aerodynamic tuck position. In that position the cyclist's nose is basically touching the handlebars, so tiny text in the app is probably ok. I don't bike like that tho and since CdaCrr is an app I'd be once again stuck with no bike computer or having to use two phones. It doesn't seem likely to work for me so I've been hesitant to buy the hardware I'd need to try it.
I don't know what to do other than give up on validating forecasts. It's really frustrating to ask myself if the wind I'm observing is above, equal to, or below the forecast and not be able to come up with anything better than a wild guess.
I could practice guessing and then measuring wind speeds on foot, but that wouldn't translate to how it feels on a bike at all due to the fact that aerodynamic drag varies with the square of the wind speed (doubling the wind quadruples the drag). I'm often amazed at how I can be getting slammed with horrendous wind on the bike that disappears almost completely when I come to a full stop.
-
@tessellahedron A few other notes and ideas:
Is a new helmet an option? I cycle and mountain bike and neither helmet I have whistles in the way you describe at apparent head wind speeds up to ~40MPH.
As for forecast challenges, I think it's worth noting that mountainous environments are notoriously difficult for a pure numerical output to forecast. The rapid changes in elevation create many gaps that create additional variables (venturis, thermal effects, etc.) that are particularly hard to pin down. That's where local knowledge starts to come in.
To your question of how to be more accurate; there is always a limit to forecast accuracy, that's ultimately why it's a forecast and not a guarantee. Numerical outputs from any source are better thought of as averages, ranges, and percentages rather than an absolute fact.
To share a short story: I'm a kiter. I rely entirely on the wind for my sport. The difference between 5 MPH worth of wind can make or break a session. Locally, I need to be in tune with some of the things we've discussed: Temperature, pressure gradients (and their associated angles), clouds, etc.. The numerical models frequently miss the mark in an absolute sense, but usually I'm able to read between the lines and score a session based on what I know is useful in the models and where it might under/over emphasize certain things. Without them it would be nearly impossible for me to predict.
You seem like an observant person who cares about details. I'd hazard a guess that with some time you can learn to read between the lines of your forecasts for your area too. You won't get it perfect, but you can probably improve your odds of a better ride!
-
@illili The whistling around the helmet doesn't bother me. I actually like the fact that it whistles because that's the only evidence I have that proves the wind isn't below 11 mph as forecasted, and that I really am seeing wind resistance four (or more) times higher than normal.
I'm not in the mountains. This is the edge of the great plains. The valley I live in is about 350 feet deep on the south (highest) side and about 5 or 6 miles wide. The "plattaeus" on either side are basically completely flat with some gentle rolling. I suspect that's why we have wind problems. The state highways and county roads are all oriented in the cardinal directions and are completely straight, sometimes for the entire length of the county.
If I was in mountains with winding highways and peaks all around there would be isolated spots were the wind was terrible, but just a few miles later things would change and it could be dead calm. That would be nice. Out here there are times that I have to ride straight into highly variable gusts for over an hour to get home. The headwind is more demoralising but the cross wind gusts are very frustrating and dangerous.
Once again the forecast for today is split. 6 of 8 models in Flowx say low gusts. Hyperlocal says low gusts. Windy's route forecast shows gusts of 16 to 20 mph, and I think my limit is 12 mph. I'll probably go out on the highway just to see which forecast is right. I guess I'd have to buy an anemometer (I returned the one in the picture) and stand out there taking wind readings to really know so I'm limited to guesstimating the speed based on how it feels.
This situation happened on Wednesday and Windy was right. I don't understand how Windy's route forecast using the global model ECMWF is more accurate than the consensus of 6 or 7 other models, multiple of which were designed specifically for the North American continent. Maybe learning which models to trust is what you meant by reading between the lines but I'm just hoping I get to move away from this place long before I get good enough at reading bad forecasts to make accurate go/no-go decisions.
Like I'm stuck using multiple forecast models, I'm also stuck using multiple weather apps to see them. To do it all in Windy I would need gusts in the forecast compare pane, preferably graphed, and an easier way to see the general 10-day weather forecast (temps, rain, sunrise/sunset times, etc).
Also, the widgets in Windy and Flowx could be better. Windy's widget looks more polished but Flowx's are more customizeable (tho still not completely). I was able to adjust the line in the middle of the graph to be my wind threshold, about 12 mph, which was the only way I found to make those graphs legible.
-
@tessellahedron said in Weatherkit Forecast Model:
I guess I'd have to buy an anemometer (I returned the one in the picture) and stand out there taking wind readings to really know so I'm limited to guesstimating the speed based on how it feels.
There is a lot of weather stations in your are that report many parameters (including gust). This one right on the route 76 reports data every 10 minutes. That might be easier than getting an anemometer and taking readings on your own
-
@Filip_K I tried using weather stations for a while. There are a lot out here, which is good since it takes a lot more stations to accurately indicate wind than for things like temperature and precipitation.
The problem was that many of them aren't online and there isn't one site/way to see them all easily. I was having to manually load three different websites and click on each weather station to see current gusts, and it still didn't seem very accurate, even a couple miles from the station. Of course that's all based on my wild guesses since I don't have a way to actually measure the wind.
I spent some time looking for a website/app that could display a custom list of weather stations with their gust data in an easy to read format but I never found one so I just gave up on the idea of weather stations.
If it's gonna be this much hassle I might as well do it with forecasts. I'm currently having to check Hyperlocal, Flowx and Windy, plus I've got windfinder.com and windalert.com open to remind me to check if I can make them useful somehow.
It looks like it's gonna be another crappy week out here, ruined by high gusts every single day.
-
There are weather stations right on windy.com. Just click "Reported wind" in the lower menu
-
@Filip_K I just tried that using the app and it does seem to work. I can see the weather stations around here. They're named incorrectly, which I can get over, but Windy also doesn't display any gust information on the map or in the details for the station. I don't see gusts anywhere, even for weather stations I know provide gust readings like the one the CO Department of Transportation has at the i76-Hwy 385 overpass (CO0034 I think).
Google's weather forecast omits gusts and it sent me on dozens of terrible bike rides in high gusty wind before I realized that a wind forecast without gust information is basically useless around here.
I didn't notice any weather stations that I know exist and are online missing from Windy's map, so it might actually be the most complete map I've seen. If gust information was added, especially if it was added to the map in a way that doesn't force the user to click or mouse-over each station to see the gust info, Windy would be the most complete source of weather station data in NE Colorado that I'm aware of.
-
@tessellahedron the information about gusts is included (see my screenshot below where wind is 7m/s, gust 10m/s). It is unfortunately not displayed in the graph and cannot be displayed on the map either
-
@Filip_K I think it would be beneficial to have gusts included in the Compare forecasts panel. Is this something Windy is considering?
-
Hi @Wheats. @ondřej-šutera might know more as he manages requests from users
-
@Filip_K I just clicked on "nearest weather stations" by accident. If gusts were listed there it would be exactly what I was looking for when I was trying to use weather stations: a way to view the most current gust readings from all nearby weather stations on one screen.
I find it hard to believe that there's currently no way to do that but I certainly haven't been able to find one. Again, Windy has a chance to be the first/only with this feature.
-
@Wheats At this moment, it is not planned to implement this into the Compare forecast feature. However, I will keep this suggestion.
-
F Filip_K referenced this topic on
