Stabilized Rpresenter Matrix Array Modes and Clipped Analysis

[arango]

• A new driver array_modes_w4dvar.h was added to compute either the requested array mode eigenvector (ARRAY_MODES) or clipped spectrum analysis (CLIPPING) of the stabilized representer matrix during weak constraint, variational data assimilation. The array mode analysis can be used to assess the observing system, prior covariance hypothesis, and the stability of the generalized inverse.

• The following new CPP options are added:

  ARRAY_MODES         use if computing W4DVAR representer matrix array modes
  ARRAY_MODES_SPLIT   use to separate analysis due to IC, forcing, and OBC
  CLIPPING            use if computing W4DVAR representer matrix clipping analysis
  CLIPPING_SPLIT      use to separate analysis due to IC, forcing, and OBC
  RECOMPUTE_4DVAR     use if recomputing W4DVAR in analysis algorithms
  

• A new parameter, Nvct, is added to input script s4dvar.in. If ARRAY_MODES is activated, Nvct is the eigenvector the stabilized representer matrix to process (Nvct=1 is the less important, and Nvct=Ninner is the most important). If CLIPPING is activated, only eigenvectors Nvct:Ninner will be processed and the rest will be disgarded.

Many thanks to Andy Moore for his help coding and testing this new driver.