pyPDAF.PDAFomi.set_use_global_obs

pyPDAF.PDAFomi.set_use_global_obs()

Switch for only use process-local observations for i-th observation type.

The function is typically used in user-supplied function py__init_dim_obs_pdaf.

The filters can be performed in parallel based on the filtering communicator, comm_filter. This is typically the case for the domain-localised filters, e.g., LESTK, LETKF, LSEIK, LNETF. In this case, observation vectors are stored in process-local vectors, obs_p. Each local process (obs_p) only stores a section of the full observation vector. This typically corresponds to the local domain corresponding to the filtering process, based on model domain decomposition.

By default, use_global_obs=1. This means that PDAF-OMI gathers the entire observation vector for all processes. One can choose to only use process-local observations for global filters, or within localisation radius for by setting use_global_obs=0. This can save computational cost used for observation distance calculations.

However, it needs additional preparations to make PDAF-OMI aware of the limiting coordinates of a process sub-domain using pyPDAF.PDAFomi.set_domain_limits() or pyPDAF.PDAFomi.set_domain_limits_unstruc().

See also

https

//pdaf.awi.de/trac/wiki/OMI_use_global_obs

Parameters:

  • i_obs (int) – index of observations

  • use_global_obs (int) –

    Swith to use global observations or not

    1. Using process-local observations;

    2. using cross-process observations (default)