Hello there!
We are building a weather app specifically for climbers as part of a University assignment and would be extremely grateful if you could fill out this form. Thanks in advance!
https://docs.google.com/forms/d/e/1FAIpQLScNM5E-ri5jkNU_sr2Jwws1klGGRYfOj24...
Here's what I would value:
A 10 to 14 day winter conditions forecast that is really easy to interpret, but site specific. It would tell me, for example, that the weekend after next Lochnagar will be experiencing a thaw, but gullies will still be well banked out with consolidated snow.
> Here's what I would value:
> A 10 to 14 day winter conditions forecast that is really easy to interpret, but site specific. It would tell me, for example, that the weekend after next Lochnagar will be experiencing a thaw, but gullies will still be well banked out with consolidated snow.
Have you seen this? https://www.winterclimbingforecasts.co.uk/
A couple of observations on the questions- you might want to clarify winter climbing or summer climbing, as my temperature preferences are quite different. Also 'time I usually climb" can be all day from a pre-dawn start. Weather often determines what I'll climb rather than whether I'll climb, so asking for 'preferred conditions' is a little bit moot.
You list “wind speed/direction” but that’s very generic. I can get that from almost anything with alerts for fixed locations etc.
Give me a map of crags in an area visually scored to indicate which crags are expected to have less wind based on the magnitude of the wind speed and how anti-parallel their aspect is to the wind direction. Now you’ve made something nobody else has.
You could add in the midge forecast!
You could add in the midge forecast!
That is the only thing I need that I can't get from a standard weather forecast. Good shout.
"Glen Coe, Hope Valley. Northeast 3 or 4. Occasional rain. Moderate or poor."
Include the previous few.days weather so people can know if it rained much rhe day before for example and then can guess if the rock or land will be dry/boggy.etc.
> Have you seen this? https://www.winterclimbingforecasts.co.uk/
It's almost useless.
This. It's always maddening that the forecast doesn't include previous days, both to know how much rainfall there's been, but also because I might not immediately know what 11 degrees feel like tbh, but 'notably colder than yesterday' is quite helpful!
I'd also add that 'would you consider climbing after rainfall' entirely depends on the aspect, the wind, the sun, the type of rock. Too many variables to answer that.
> It's almost useless.
Interesting. I have to admit I haven't used it, but got quite excited when I first heard about it - winter conditions prediction seemed ripe for ML. Is it not at all accurate in your experience?
> Interesting. I have to admit I haven't used it, but got quite excited when I first heard about it - winter conditions prediction seemed ripe for ML. Is it not at all accurate in your experience?
I was very interested in it too, to the extent that I paid for a subscription. It's not wholly inaccurate, it gets the easy calls right most of the time, but then you don't need it for the easy calls. When it comes to the harder calls, I found it to be frequently wrong at both ends of the scale. Pondering about why it's so wrong I think it's partly down to the accuracy of the input data. One of the inputs is the weather forecast which can of course be wrong. Another is the route details which I think have been incorrectly entered in a lot of places. There are clearly a whole lot of other inputs whose accuracy is unclear or questionable. Put crap in, get crap out.
I think a much simpler system could be more useful. Obviously climbers want to know about specific routes but maybe a forecast by corrie/route type would work better. The very precise nature of the output also fills me with suspicion. I couldn't sort routes in to 5 condition categories when standing beneath them so I've no idea how an app could do that in any meaningful way. It doesn't have the accuracy to support the level of precision it offers.
Hi all,
Adam creator of Winter Climbing Forecasts here. Just to respond to a couple of the points raised on the thread.
"I think a much simpler system could be more useful. Obviously climbers want to know about specific routes but maybe a forecast by corrie/route type would work better."
So this has been a feature on the app for a couple of seasons now. Summary charts of areas with the type of routes predicted in condition. Had some great feedback on this.
"I couldn't sort routes in to 5 condition categories when standing beneath them so I've no idea how an app could do that in any meaningful way."
This was a conscious design decision. The prelaunch app had just three categories, but it was decided five worked better for tracking the progress of water-ice routes slowly forming, but knowing that level of granularity was a trade off on other route types.
You could just view the original three categories as terrible, poor/marginal and reasonable/good, if you wish. The app effectively does this when you look at the summary or filter what's in condition, as it does them on reasonable/good.
"Another is the route details which I think have been incorrectly entered in a lot of places. There are clearly a whole lot of other inputs whose accuracy is unclear or questionable"
Totally baseless.
"It's not wholly inaccurate, it gets the easy calls right most of the time, but then you don't need it for the easy calls. When it comes to the harder calls, I found it to be frequently wrong at both ends of the scale."
Behind the scenes the models have constantly improved and are totally different to four seasons ago. The key difference now is the amount of user feedback confirming where the predictions were right and where they were not. Thanks to the community heavily using this feature, the total now numbers into the tens of thousands and the forecasts are only going to get better and better as time goes on.
New to this season just gone was fine-tuning the model every night on this data, so any feedback say on the Friday would be reflected in Saturday's forecasts. Lots of positive comments from the community on this too.
"It's almost useless."
Anyone who follows WCF closely will be aware Dave has been a extremely vocal critic since the start. However, the vast majority of the community has been very supportive and regularly feedback to me how useful they find the tool. So thank you all for that.
See you all at the start of next season...
Adam
Rain, scale of 1-5.... is 1 important or 5 important etc.. I gave up.
PS. Skill level: "I'm amazing" is that helpful? Or perhaps you need some more specific choices?
Pps. Did you test run this on anyone first?
Ok, you and I have definitely got off on the wrong foot.
I'm sorry I said the app was mostly useless, that was unfair. The rest of my criticisms come from having used it and found it to be unreliable. I'm not seeking out opportunities to rubbish it or being critical for the sake of it, it's just my honest appraisal having used it for a season. I've also been a bit concerned by the level of faith some users have immediately placed in what is new and in my experience, not particularly reliable technology and I think it's important to share concerns like that.
> Behind the scenes the models have constantly improved and are totally different to four seasons ago. The key difference now is the amount of user feedback confirming where the predictions were right and where they were not. Thanks to the community heavily using this feature, the total now numbers into the tens of thousands and the forecasts are only going to get better and better as time goes on.
Thanks for taking the time to respond. Out of interest what is the accuracy, false positive/negative rate of your forecasts?
This would be excellent addition, especially if updated to actual weather for the preceding days as best as possible. Really useful for winter too for route and avalanche conditions. Would be hit or miss depending on where local weather stations were though!
Meteoexploration used to have a "hindcast" option on their website to pull archived conditions.
MWUK app had links to lots of weatherstations but not sure if it's still active.
In reply to OP.
An option to switch between models (like windy) or if its working by accessing metoffice, mwis forecasts etc being able to compare side by side would be good.
Also showing probability banding for rain. Wind etc would be useful.
Finally keep it focused on weather / conditions rather than becoming a mountain safety broadcast.
Cheers,
Stu
> This. It's always maddening that the forecast doesn't include previous days...
That would be a 'past-cast' you're after! The clue's in the name
> This. It's always maddening that the forecast doesn't include previous days, both to know how much rainfall there's been, but also because I might not immediately know what 11 degrees feel like tbh, but 'notably colder than yesterday' is quite helpful!
The Meteoblue app does this. There's an option to include the previous day's forecast.
'aft‑cast'
I've always thought an ideal weather app for climbers would take account of the current and recent weather, then give confidence ratings for how pleasant it would be climbing at different crags / venues. Lots of things to consider that won't come up in a normal forecast, e.g.
- Is it quick drying?
- Is it exposed / will catch the wind?
- Which direction does it face?
- Are midges an issue?
- Does it only come into condition at certain times of year?
Once you get to know an area well you can make judgement on which crag will make a good choice on a given day. But there's no easy way to do this if you don't know an area, without cross checking back and forth with weather forecasts / guidebooks / UKC etc.