Accuracy of J2Ski's Powder Alarms

It's at the bottom of the forecast page. The summary of which is at the top. It has changed to 51 cm today. I have seen it as high as 497cm which is clearly impossible.

But let's not get hung up on this. Snow overnight and a lovely skiing day in prospect!

I thought it was common knowledge that all these snowfall forecasts are generated by computer based weather forecasting models, such as the Global Forecast System (GFS), European Centre for Medium-Range Weather Forecasts (ECMWF) etc. I presume that J2Ski uses the precipitation data generated by the GFS as they are not copyrighted and are available for free, whereas the data from the other models has to be paid for. The price for a maximum charge commercial ECMWF licence is 140,000 EUR/Year (+ delivery charges).

Unfortunately the modelling of precipitation is still not very reliable. If you compare the modelled results with observed values you will see that the model is nothing better than a computer generated guess. The GFS model is run in two parts: the first part has a higher resolution and goes out to 180 hours (7 days) in the future, the second part runs from 180 to 384 hours (16 days) at a lower resolution. The resolution of the model varies in each part of the model: horizontally, it divides the surface of the earth into 35 or 70 kilometre grid squares; vertically, it divides the atmosphere into 64 layers and temporally, it produces a forecast for every 3rd hour for the first 180 hours, after that they are produced for every 12th hour. Because of the low resolution the model can't see the complex terrain of the Alps, Pyrenees. and other mountainous areas very well. The precipitation data has to be multiplied by a factor to get the predicted amount of snowfall which is another source of error.

J2Ski's forecast for Zermatt last Thursday



Reality on Monday



Zermatt claims that there is 20cm of snow at village level! After a personal inspection yesterday I reckon about 10-15cm fell on the slopes over the weekend.

Swing Beep you nicely demonstrate that J2Ski forecasted 49cm whereas only 10-15cms fell. This is exactly the type of over forecasting snowfall I have noticed in recent years and hence my original post.

Just imagine they under-estimated, you cancelled and went to the beach - disaster. Even though I know the 14 day forecast is effectively a nonsense there is a certain addictive element to looking and seeing an apocalyptic amount of snow might be heading your way.

If I ran my life on a computer model it would probably conclude a high probability of waking up tired, body won't seem very willing = best stay in bed. Skiing is about the small probability that some trips, some days, some runs everything comes together and you get that bottomless bounce feeling and shot blasts. I would conclude this happens less than 5% of days I ski. By that measure as long as the 14 days forecast is no worse that 95% inaccurate I am in.

A few days have now passed since my original post. J2Ski initially forecast c.32cms I seem to recall, then a day or so later forecast 23cms still to fall, then a day or so later forecast 10cms for the following Wed. The reality was that only 10cm fell over this period and on the 10cm forecasted day none fell at all. It should be noted that several other weather forecasts got it right for this period whereas J2Ski had it wrong. So it wasn't just a one off, these were 3 or 4 forecasts over a number of days, all of which were wrong.

As I said originally, I have noticed this trend over a number of years i.e. J2Ski not concurring with other forecasting sites.

Furthy34 wrote:you nicely demonstrate that J2Ski forecasted 49cm whereas only 10-15cms fell.

I think, in fact, this demonstrates the issue you have. The amount of snow we forecast over 4 days is not the same as the "amount that fell over the weekend" and that amount is not the same as the snow on the ground on Tuesday. Those numbers are not measuring the same thing.

As soon as snow falls, it begins to transform (settles and sinters) and compacts under its own weight and the layers of snow that follow. Anyone who's been out skiing on a powder day will know that light powder can quickly "go off" - it gets denser and compacts, and surprisingly quickly. Sunshine, warm air, temperature gradients in the snow pack and - especially - wind all have a significant effect.

So if you get 2 falls of 10cm of fresh on successive days, you won't - ever - see the base depth increase by 20cm.

Given the complexity, you simply can't "hindcast" and say that "well, there seems to be 10cm fresh on the ground, so that means 10cm fell 2 days ago". It could easily have been double that.

If you'd like to learn more about the subject, then get hold of the book "Snow" by Robert Bolognesi (handbook size and less than £10).

Furthy34 wrote:hence my original post.

You titled the post to refer to our powder alarms. I fully accept it's a very small sample but from the ones I checked - that we issued on Thursday/Friday last week - every one appears to have under-stated the snow that followed. I'm sure there were a few "false alarms" but I couldn't find any.

Furthy34 wrote:A few days have now passed since my original post. J2Ski initially forecast c.32cms I seem to recall, then a day or so later forecast 23cms still to fall, then a day or so later forecast 10cms for the following Wed. The reality was that only 10cm fell over this period and on the 10cm forecasted day none fell at all. It should be noted that several other weather forecasts got it right for this period whereas J2Ski had it wrong.

This is just wrong.

When you made your post, Obergurgl's own site (linked in my initial reply) reported depths of 20cm (low) and 100cm (high). Their own forecast, that morning, stated that no significant snow was expected for the next 4 days.

Whilst the snow we forecast for yesterday didn't materialise (it was cloudy all day but never snowed), the reported depths are now at 30cm and 120cm respectively - with the upper depth incrementing by 10cm on Monday and Tuesday mornings IIRC.

Allowing for settling and compaction, they must therefore have had more than 20cm in the two falls that we did forecast correctly.

So that's 2 out of 3 falls happened that we forecast, none of which appeared in the forecast on Obergurgl's website.

Admin Man

I am well aware of snow compaction and hence I was viewing how much fresh snow had fallen in Obergurgl every day.

The fact is that nowhere near 32cms fell. Then after the initial fall of about 10cms you forecast 23cms but nowhere near that amount fell, even at the top.

You then forecast 10cms for Wed and none fell. On Tuesday you changed to 5cms on Wed and none fell.

Whilst all of this was going on I was checking other forecasters including the one that Obergurgl and many other resorts use and they were nowhere near as optimistic as you and as it turned out they were right.

As I have already stated, my OP was made having noticed the accuracy of J2Ski's forecasts over several years.

If you are keen to improve the accuracy of your reports I would politely suggest that you consider using other forecasters or weather stations as your source. What do you think of this idea?

You will have noted the current results from the poll on this and it seems I am not alone in my opinion (Btw, I voted "not very much" rather than "none").

Forecasts are just a guide so after the 48 hour rule (I personally prefer 24 hours) you simply have to use your common sense it's not rocket science (sometimes not even science at all!!) And you should always look across other sites to get a general picture