recentre

Perturbations refer to the small variations centred around a nominal value of a parameter. When dealing with ensemble forecasting, the perturbations are related to the difference between ensemble members and their given centre.

The recentre function computes a set of new ensemble members centred on a different centre from previous ensemble members using the following formula:

\[members_{new} = centre + ( members - \overline{members} )\]

Additionally, some variables must be non-negative to have a physical meaning (e.g. accumulated variables or specific humidity). To ensure this, positive clipping is performed using the alternative formula:

\[members_{new} = max(0, centre + ( members - \overline{members} ))\]

The current implementation enforces that the following variables are positive when using the recentre function:

Variable	Description
q	Specific humidity
cp	Convective precipitation
lsp	Large-scale precipitation
tp	Total precipitation

It uses the following arguments:

members

A reference to the ensemble members.

centre

A reference to the new centre requested.

Examples

data_sources:
  members_source:
    mars:
      class: ea
      expver: "0001"
      grid: 20.0/20.0
      levtype: sfc
      param: [ 10u, 10v, 2t ]
      type: an
      stream: enda
      number: [ 1, 2, 3, 4, 5, 6, 7, 8, 9 ]

  center_source:
    mars:
      class: ea
      expver: "0001"
      grid: 20.0/20.0
      levtype: sfc
      param: [ 10u, 10v, 2t ]
      type: an
      stream: oper

input:
  recentre:
    centre: ${data_sources.center_source}
    members: ${data_sources.members_source}