Weather forecasting models: which one to choose?

Weather forecasting models: which one to choose?

Every weather forecast is based on a forecasting model. The names of these models may not mean anything to you, but you're probably familiar with the platforms that use them: BBC Weather, The Met Office, Weather Spark, AccuWeather, UK Weather... But how do you know which weather model is the most reliable? What forecasting models does Sencrop offer?

What if the best solution involved using several methods? Discover all the details in this article!

What is a weather forecast model?

Weather forecasting models are powerful tools that observe and transform data collected by various sources - weather stations, satellites, radar, sounding balloons, aircraft and ships - into weather forecasts.

These algorithms simulate and anticipate the complex movements of the atmosphere by combining physics, mathematics and computer science. Everything is simulated: temperatures, wind strength, cloud density, rainfall intensity, etc. Behind each forecast are thousands of calculations and equations designed to produce the most reliable predictions possible.

Sencrop integrates the most reliable, proven weather models into its application.

The resolution of weather models

Each model divides the atmosphere into a three-dimensional grid, with each grid cell representing a small volume of air. These are known as "meshes", and their size varies from one weather model to another. This is why each model has a different resolution.

The smaller the mesh size, the greater the accuracy, but the greater the computing power required.

Source : Météo France

For each grid cell, the model simulates the evolution of the atmosphere over time, making it possible to predict the weather conditions over the next few hours or even days.

How reliable are weather models?

There are many weather models available today, each with its own advantages and limitations. Their reliability can vary depending on the region, the measurement or the time horizon.

For example, a model may be very accurate in predicting the average temperature in a region, but less effective in anticipating night frosts.

It's a bit like comparing a sprinter and a marathon runner: some models are effective over short distances or short durations, while others are better suited to longer-term forecasts. But both are good at what they do!

No model is perfect for all situations, and the choice of the appropriate model depends on the specific needs of the weather forecast to be carried out.

The best weather models: comparison


AROME (Application of Research to Operations at Mesoscale) was created in 2008 to improve short-range forecasts.

  • Created by : Météo-France
  • Forecast duration: 36 hours (just under 2 days)
  • Resolution: 1.3 km
  • Territories covered: France and Overseas

🟢Advantages: Highly detailed grids provide accurate forecasts for local phenomena, such as thunderstorms and heavy rainfall.
🟠Drawbacks: Coverage limited to the French territory and to 36 hours.


ARPEGE (Action de Recherche Petite Echelle Grande Echelle) is Météo-France's main numerical model. Its resolution varies depending on the geographical area. 

  • Created by: Météo-France
  • Forecast duration: 4 days
  • Resolution: 5.1 km over mainland France to 24 km at the antipodes of France.
  • Territories covered: World

🟢Advantages: Allows multi-day global forecasts.
🟠Drawbacks: Larger grids, sometimes insufficient to forecast ultra-local phenomena.


GFS (Global Forecast System) is one of the most widely used models in the world (notably by Météociel and Meto60). 

  • Created by: National Centers for Environmental Prediction (NCEP)
  • Forecast duration: 7 days
  • Resolution: 22 km
  • Territories covered: World

🟢Advantages: Long-term forecasts available free of charge.
🟠Drawbacks: Lack of precision for local forecasts.


ICON (Icosahedral Nonhydrostatic) is a German model from the meteorological service of the Federal Republic of Germany. 

  • Created by: Deutscher Wetterdienst (DWD)
  • Forecast duration: 5 days
  • Resolution: 7 km (2 km in Germany)
  • Territories covered: Europe

🟢Advantages: Quality weather forecasts for Europe.
🟠Drawbacks: Resolution sometimes insufficient for ultra-local phenomena.

There are, of course, many other models such as UKMO, COSMO and NEMS, which are sometimes more accurate in mountainous regions or for specific weather data, for example.

All the weather models in the Sencrop application

As you've likely realised, relying on a single model isn't always the most reliable approach. That's why it's best to use multiple models. Fortunately, the Sencrop ag-weather platform offers all the weather models in one place!

Here are three key features to ensure maximum weather forecast reliability:

Screens showing the new Sencrop fezatures : Forecast comparator - Model ranking - Sencrop forecast (from left to right)
  • The forecast comparator: Compare the rainfall, temperature, humidity and wind speed forecast by each model at a glance. A probability indicator shows you how many models are predicting the same thing. Analyse trends and identify extremes.
  • Model ranking: Evaluate each model on the basis of the reliability between what it predicted and what was actually recorded by your own local weather stations. Stay informed and in control of your decisions.
  • Tailor-made Sencrop forecasts: Benefit automatically and continuously from the most reliable weather model possible for your locality and for the data observed, based on the actual readings from your station. No more juggling between different forecasts!


Defining the most reliable weather forecasting model can be a complex task! Numerous parameters come into play, including: the grid, the time resolution, the weather data observed (temperature, wind, etc.) and the topography. There is no single weather model that is always the most reliable, for all of them. And a weather forecast will never be 100% reliable (unfortunately, it remains a forecast...). A multi-model analysis is therefore an interesting way of checking probabilities, but it remains complex.

There are ways of simplifying this analysis, including the Sencrop solution. The application brings together all the most reliable weather models, allowing you to compare forecasts with each other, and also to compare the weather forecasts of each model with your actual readings. The application automatically and continuously adapts the most reliable weather model for your location and observed data using Sencrop forecasts.

Discover the Sencrop solution

Create your profile, connect to a station close to your plots and benefit from a free trial, without obligation.

Try the application for 14 days