[Dart-users] Nonlinear and Non-Gaussian Data Assimilation Capabilities in DART v11

[Data Assimilation Research Testbed users]

Version v11 of DART is now available and provides a Quantile-Conserving
Ensemble Filtering Framework (QCEFF
<https://docs.dart.ucar.edu/en/latest/#nonlinear-and-non-gaussian-data-assimilation-capabilities-in-dart>).
The QCEFF gives better results for non-gaussian and bounded quantities.

Please note, although the default assimilation algorithm remains the same
as v10, there are several namelist (input.nml) changes that will require
changing your existing input.nml files when you move from v10 v11.

The DART development team (dart at ucar.edu) would be happy to hear about your
experiences and is excited to build scientific collaborations using these
new capabilities.

Release notes <https://github.com/NCAR/DART/releases/tag/v11.0.0>:

   -

   Adds a Quantile-Conserving Ensemble Filtering Framework (QCEFF) to DART.
   Publications: QCEFF part1 <http://n2t.net/ark:/85065/d7mk6hm4>, QCEFF
   part 2 <http://n2t.net/ark:/85065/d7nv9pbt>, QCEFF part 3
   <https://docs.dart.ucar.edu/en/latest/_static/papers/QCEFF_3_submitted.pdf>
   .
   -

   The default QCEFF options are EAKF, normal distribution (no bounds).
   -

   User interface changes:
   -

      filter_kind is now a per-qty option through QCEFF table not a
      namelist option
      -

      Two new required namelists (add to input.nml files):
      -

         probit_transform_nml
         -

         algorithm_info_nml
         -

      assim_tools_mod namelist:
      -

         sort_obs_inc namelist option applied to ENKF only, so default is
         now .true.
         -

         spread_restoration is not supported in this version
         -

      algorithm_info_mod QCEFF options read at runtime from .csv or .txt
      file
      -

   New probability distribution modules:
   -

      beta_distribution_mod contributed by Chris Riedel
      -

      bnrh_distribution_mod (bounded normal rank histogram)
      -

      gamma_distribution_mod
      -

      normal_distribution_mod
      -

      probit_transform_mod
      -

      distribution_params_mod
      -

   Update to lorenz_96_tracer_advection:
   -

      positive_tracer
      -

      more tracer namelist options available and changed defaults
      -

      updated perturbation routine
      -

      bug-fix: real(r8) rather than real(i8)
      -

   Fix: obs_def_1d_state_mod (oned forward operators):
   -

      For non-integer powers, fix up values for negative bases
      -

   Documentation:
   -

      main page section on Nonlinear and Non-Gaussian Data Assimilation
      Capabilities in DART
      -

      QCEFF instructions: Quantile-Conserving Ensemble Filter Framework
      -

      Example to work through: QCEFF: Examples with the Lorenz 96 Tracer
      Model
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.ucar.edu/pipermail/dart-users/attachments/20240116/7c11da3a/attachment-0001.htm>