The Integrated Forecasting System (IFS) is the production weather forecasting application used by the European Centre for Medium Range Weather Forecasts (ECMWF). It supports various models of production weather forecasting, including high resolution deterministic forecast, 4-dimensional variational analysis (4DVAR), ensemble prediction (EPS) and ensemble data assimilation (ENDA).
The code itself is capable of multiple resolutions covering a range appropriate for climate simulation, ensemble calculations and high-resolution forecasting. For instance, a high-resolution forecast (T2047L137) has a horizontal grid point spacing of 10~km and 137 vertical levels. Future requirements will increase the resolutions significantly resulting in an increased need of computational resources. For instance, the consequence of halving the grid spacing is a 16-fold increase in the computational cost (a doubling for each spatial dimension and the time step).
IFS uses a spectral, semi-implicit, semi-Lagrangian numerical scheme. Data exists in three different spaces, namely, grid-point, Fourier and spectral space. In a single time step (see Figure below), data is transposed between these spaces so that the respective grid-point, Fourier and spectral computations are independent over two of the three co-ordinate directions in each space. Fourier transforms are performed between grid-point and Fourier spaces, and Legendre transforms are performed between Fourier and spectral spaces.