The weather forecast is developed in three basic steps: observation, simulation of the evolution of the atmosphere using numerical models and analysis of results by forecasters.

Observations Collection and Data Assimilation

To predict the weather tomorrow, we must already know the current weather. Observations are the first step of a forecast. The 80% of the observational data used in the forecasting models comes from meteorological satellites. The remaining 20% ​​is provided by ground or ship stations, drifting buoys, radiosonde and aircraft's sensors.

All these observations are then processed to extract the "useful" information in the forecasting model: this is called data assimilation. Data from the observations are combined with other information, such as very recent forecasts, to establish an initial state of the atmosphere that the model will use. The observations are also used by forecasters to monitor the weather situation to detect and correct errors of prediction.

Review of the Predictions by the Forecasters

The simulation results from the models are not yet weather forecasts. These are scenarios for the main meteorological parameters at all points of the grid that represents the atmosphere. At the end of the chain, the experience of the forecaster, his thorough knowledge of his own country weather, can analyse the observations to validate the results of model simulations, to turn it as appraised forecasts. These predictions can be provided to users or serve as a basis for decision-making.

What is MESSIR-NWP ?

MESSIR-NWP is a turnkey solution for Numerical Weather Prediction based on HPCC (High-Performance Computing Cluster) hardware platform and modelling software (Atmospheric, Marine, Climate, Dispersion)

MESSIR-NWP supports a large range of numerical models:

  • WRF model
  • WAVEWATCH III AND SWAN models
  • COSMO model
  • COSMO-ART model
  • COSMO-CLM model
  • ...
COSMO Model simple output (comulated precipitations)
WaveWatch III sample output (Significant wave height and direction)