Python codes for ensemble data assimilation using a simple two-level primitive equation spectral model on a sphere.

• twolevel.py: model. To generate an animation of the solution, run python twolevel.py (matplotlib must be installed).
• enkf_utils.py: EnKF code (for serial filter and LETKF).
• pyspharm.py: Spherical harmonic routines (uses shtns).
• run_twolevel.py: generatre nature run (requires netcdf4-python).
• enkf_twolevel.py: run EnKF experiment.
• enkf_twolevel.py_iau.py: run EnKF experiment using 4D IAU.

1. To generate a nature run, execute python run_twolevel.py.

2. To run an assimilation experiment, execute python enkf_twolevel.py <localization length scale> <inflation factor>. The localization length scale should be specified in meters (i.e. 4500.e3). If the inflation factor give is between 0 and 1, relaxation to prior spread (RTPS) inflation is used. If the inflation factor is exactly 1.0, then Hodyss and Campbell's proposed inflation is used, with a=b=1. Global mean error and spread statistics for each analysis cycle are written to stdout.