Mountain Madness

In July 2023, Tech Evangelist Tom Eldridge headed off to test his mettle and the EURO1k weather model in Italy’s best climbing destination, the Dolomites.

Mountains are a large part of why I moved to Switzerland. I’ve been a climber for the greater half of my life, and eventually every British mountaineer knows Snowdonia and the Lake District like the back of their hand, and seeks out newer, higher challenges. But even before this, mountains exerted their influence on where I would eventually end up. My decision to study meteorology was driven in large part by my love of spending time in the mountains, and mountains themselves exert a large influence on the weather, making forecasting in mountainous areas very challenging.

Meteomatics: High-Resolution Weather Forecasting

This is a challenge that Meteomatics knows well. As a weather company situated in St Gallen, we’re intimately familiar with the Alpine landscape, and well aware of the reliability issues faced by weather models with a coarse spatial resolution. Your typical global weather model divides the Earth’s surface into grid cells of the order of tens of kilometers, and correspondingly has to make approximations.

Mountains are defined by their rapid elevation change, often varying in height by a couple of hundred meters above local base height over the course of a few kilometers. This may not sound like much, but has a significant effect on the Earth’s atmosphere, largely because the temperature of the atmosphere, and hence how much moisture it can hold, varies hugely over a few thousand meters of elevation. Now, a weather model, in order to run in a reasonable amount of time, must divide the Earth up into regular cells of horizontal distance, and the properties of the Earth system within these cells must be averaged, just like the light signals contributing to a camera pixel. Below I show the effect of modeling the Dolomite mountains in northern Italy at two different grid cell sizes: one 10 km long, the other 1 km. The effects of averaging elevation over this spatial scale are quite dramatic.

This is one of the reasons that Meteomatics was inspired to develop the EURO1k weather model. With a modeling resolution of one kilometer, EURO1k is designed to take account of local changes in orography and produce much more realistic weather forecasts in regions with rapidly varying surface elevation. Our own verifications show that, statistically, EURO1k performs very well when compared against global models, but perhaps the clearest demonstration of its ability for me came on a climbing trip in July.

The Dolomites are, to my mind, some of the world's most dramatic mountains. Whilst the highest stands almost 1,500 m lower than the peak of Mont Blanc, the rate at which the peaks of the Dolomites burst up through the surrounding landscape can make a Swiss Alp look almost like a gentle slope in comparison. Consequently, the Dolomites are a famously difficult area to forecast: convective thunderstorms form from air forced upwards against the steep, rocky faces, before rapidly dissipating in a rush of lightning and hail. This kind of weather can be hazardous for mountaineers, who are often fixed in position on mountainsides by safety ropes, unable to take shelter and exposed to anything the sky might choose to throw at them.

The forecast the week before my trip to the Dolomites this summer made my heart sink. According to medium-range weather models, the entire region would be underwater for much of the period of our trip, with 10-20 mm of precipitation expected on average every day. Nevertheless, as many of our party had arranged travel from England and Finland, we crossed our fingers and pushed on regardless. Perhaps this is the way of the British mountaineer — if you let a bit of inclement weather get in your way in the UK, you’ll never get anything done!

A kind of black humor hung over us like a stormcloud on our first evening in Rocca Pietore. We’d all seen the forecast; we all knew we’d achieve nothing with our week. Still we sat down to make some optimistic plans for the next day. To help us with this, I grabbed my laptop and opened up MetX — the browser based weather visualization tool from Meteomatics. MetX allows users to divide their screen into parts and compare the output from different models.

I started looking at precipitation data for the next day in ECMWF’s IFS, NCEP’s GFS and our own EURO1k. The difference was apparent at once. Whilst ECMWF and NCEP both predicted widespread downpours, the rainfall expected by EURO1k showed much more local variation. Whilst some peaks would certainly experience powerful storms, others were completely free of clouds. Although this seemed too good to be true, we made several plans for the next day and agreed to check the forecasts again the next morning and decide on one.

Because of the chaotic nature of the Earth’s weather and climate system, forecasts become more reliable with smaller lead times. Still, when we woke up on the Sunday, ECMWF and NCEP were still predicting that we’d have a pretty wet time wherever we went. The only break in the clouds was provided by EURO1k. This shred of optimism was enough for us however, and we set out to climb the Sella Towers.

This pattern was to continue: every evening we’d compare the forecasts using MetX; in contrast to the global models, EURO1k would always provide at least a few dry spots; we would prepare a few alternative plans, wake up in the morning and confirm one of them; then we went to our destination, had a successful day climbing, and returned home mostly dry.

The sharp rocky features of the Dolomites force air to rise and thunderclouds to form with sometimes alarming rapidity, and from our vantage points atop the peaks we could see these systems forming all round us, just as EURO1k predicted. Our adventures though were always conducted in the safest of conditions.

Of course, no weather forecast is perfect. There are a large number of reasons for this, amongst them imperfect model physics, the necessity of statistical representations of bulk properties of fluids, the non-analytical nature of solutions to the Navier-Stokes equations, and ultimately also finite measurement precision, chaos theory, and the quantum nature of the universe. During our stay though, we never had cause to complain about the accuracy of the forecasts we experienced from EURO1k.

By contrast, if we’d relied on the predictions made by the coarser models like IFS, we’d have stayed at home all week due to its blanket predictions of constant rain. Whilst this is accurate on average, it misses the finer details: this is what EURO1k excels at.

EURO1k’s lead time of two days meant that forecasting our entire trip at the beginning of the week was impossible. However, given the dynamic nature of weather in the Dolomites, it’s not surprising that the estimate should update regularly, and only be valid for short lead times.

Ultimately, EURO1k is almost perfect for excursions such as this one, and I would certainly trust it to deliver similar results again.

Now, if you’ll excuse me, there are more mountains to climb, and there’s no rest for a brand ambassador!