Skip to content

cov_regularization⚓︎

Scripts used for localization in the fwd_sim step of Bayes.

Changelog
  • 28/6-16: Initialise major reconstruction of the covariance regularization script.

Main outline is:

  • make this a collection of support functions, not a class
  • initialization will be performed at the initialization of the ensemble class, not at the analysis step. This will return a dictionary of dictionaries, with a triple as key (data_type, assim_time, parameter). From this key the info for a unique localization function can be found as a new dictionary with keys: taper_func, position, anisotropi, range. This is, potentially, a substantial amount of data which should be imported as a npz file. For small cases, it can be defined in the init file in csv form:

    LOCALIZATION FIELD 10 10 fb 2 2 1 5 1 0 WBHP PRO-1 10 PERMX,fb 7 7 1 5 1 0 WBHP PRO-2 10 PERMX,fb 5 5 1 5 1 0 WBHP INJ-1 10 PERMX (taper_func pos(x) pos(y) pos(z) range range(z) anisotropi(ratio) anisotropi(angel) data well assim_time parameter)

  • Generate functions that return the correct localization function.

localization ⚓︎

__init__(parsed_info, assimIndex, data_typ, free_parameter, ne) ⚓︎

Format the parsed info from the input file, and generate the unique localization masks