Weather forecasting in mountainous regions

Gilles Brunot, director of Météo-France’s Chamonix office was at CREA Mont-Blanc last week for our third Science Sandwich of the year.  His presentation helped us understand a little more about what goes into the weather bulletins that we all read in the morning before picking out what to wear.  Read on for the answers to all of our questions!

The Aiguille du Gouter playing hide and seek in the clouds  ©  Alex Hernandez


How are weather forecasts made?

Measurement tools situated on land, in the ocean, the atmosphere and space collect a variety of different kinds of data:  temperature, humidty, atmospheric pressure, precipitations, snowpack, wind, etc.  By combing all of this data, forecasters are able to, several times a day, get a good idea of the “initial state” or a base analysis of what is going on with the weather on the scale of the whole planet.  Next, analyses of the initial state by several different forecasting models followed and interpretations by forecasters allow for the creation of a weather forecast.

Why are weather forecasts more reliable today than in the past?

For multiple reasons.  First, developments in computing capacity have had huge impacts.  The computing power of today’s supercomputers allows for a less simplified calculation of some complex weather phenomenons, as well as improvements in the resolution of the of predictions, accounting for increasingly fine spatial analyses.  For example, the size of grid cells (i.e. initial pixels) has improved from around 200 or 300 kilometers squared around thirty years ago, to between 1.2 to 10 kilometers today, depending on the meteorological model used.  This level of precision is especially important in mountainous areas where altitude can change enormously in the space of 200 km.  The addition of new satellites in the 1980s also allowed for more precision in the analyses of the initial base states, as well as for the creation of more base states, more models and thus more predictions to compare and interpret.



Evolution of the computing power of the different super computers used by Météo-France since the beginning of the 1990s.  Source: Météo-France



Will we see totally automatic forecasting soon?

Entirely automatic weather forecasts exist already.  However, the expertise of forecasters in interpreting the outputs from models continues to be essential for adjusting forecasts in mountainous areas (for example, adjusting the snow-rain line or the difference between precipitation quantities at different altitudes).  Local weather forecasters contribute their knowledge of a region and their understanding of weather phenomena that are specific to mountains (like the “foehn effect” and the formation of cumulus and cumulonimbus)–elements that models continue to have difficulty accounting for.

Weather forecasting: a good example of international cooperation?

In order forecast the weather at the scale of a territory like France, weather forecasting models need to receive daily inputs of data from all around the world.  This requires data-sharing between countries, which is coordinated by the World Meteorological Organization and is a great example of international cooperation.  In addition, an open-source ethic is common in the world of weather forecasting and you can often directly find weather forecasting models, like ones developed from the American GFS model, online.

Does climate change make weather forecasts less reliable?

While the frequency and severity of extreme weather events increasing with climate change, weather forecasts aren’t based on statistical analyses of past weather phenomenon, but rather on physics (more specifically, on thermodynamics).  Therefore, forecasting methods are independent of the changes in climate ongoing over the last 100 years.


Written by Colin Van Reeth and Hillary Gerardi

Pour en savoir plus

The 2019 Science Sandwich Program
Are you a scientist interested in sharing your subject and your love of the mountains with us? Contact us !

Leave a Reply

Your email address will not be published. Required fields are marked *