Standard pyCIF control vector

The control vector plugin includes sub-routines to compute operation related to uncertainty matrices, and stores the control vector meta-data and data themselves. Only a standard control vector including most commonly used control vector shapes is implemented yet.

Configuration

The control vector is defined with the following parameters:

• components:

  the different types of components in the control vector; accepts: fluxes, prescrconcs, prodloss3d and inicond

  • fluxes: fluxes to be optimized

  • species: each species is defined through a paragraph with the following parameters:

    • hresol: the horizontal resolution; should be one of:

    • hpixels: for individual pixels

    • bands: for zonal and meridional bands

    • regions: for pre-defined regions

    • type: (optional) the type of increments to deal with; should be one of:

      • scalar: (default) increments are applied to the scaling factor of the prior

      • physical: valid only with component`=`fluxes and hresol`=`hpixels

    • errtype: (optional) the type of error; if not specified, a scalar is applied to the prior value; if max is given, computes the max of the neighboring cells (spatially and temporally)

    • err: error as a proportion of prior fluxes

    • dir: directory where data files are stored

    • file: file format of the data

    • hcorrelations: (optional) information on horizontal

      correlations if any

    • filelsm: path to the land-sea mask file

    • dump_hcorr: (optional) dump correlation matrix if True;

      default is True

    • dircorrel: directory where to save the correlation matrix

    • sigma_land: correlation distance over land in km

    • sigma_sea: correlation distance over sea in km

    • tcorrelations(optional) information on temporal

      correlations if any

    • sigma_t: correlation period in hours

    • dump_tcorr: (optional) dump correlation matrix if True;

      default is True

    • dircorrel: directory where to save the correlation matrix