Atmospherical modelling by the RMI

An atmospherical model simulates the time evolution of the atmosphere by solving the mathematical equations for the meteorological variables (temperature, precipitation, wind speeds, etc.). We first create an atmospherical initial state based on the available meteorological observations. The evolution of the situation via time steps is then calculated further in time until we reach the desired time in the future. Models, due to the limitations in the used observations and the limitations in the mathematical representation of reality, make errors in the calculations. These errors grow as time goes on.

Atmospheric models divide the atmosphere into cubes and do the calculations in grid points in the center of those blocks. The smaller the distances between the grid points, the more detailed the model results. We speak of resolution, indicated by the distance between the grid points. By carrying out the calculations on a limited area, we can limit the calculation time or increase the resolution. This is what we call models on a limited area or regional models.

Atmospherical models are used for weather forecasts in the short and medium term and for climate simulations. They are also used in probabilistic weather forecasting systems. An ensemble of model runs is performed with a dozen model versions that all differ slightly from each other. The differences are generated by disturbances in the initial state or by using different configurations of the model. On the basis of the distribution of the different members of the ensemble, probabilities can be estimated for the predicted variables (e.g. risk of precipitation, or the probability that a given meteorological phenomenon occurs).