SpEnOI is an ensemble harmonic data assimilation software suite for spectral tidal modeling. Ensemble is built by perturbing essential tidal modeling parameters:

  • bottom friction coefficient
  • bathymetry
  • boundary conditions
  • internal (tide) wave drag

It is a C/C++ suite, with OpenMP optimisation. It includes data assimilation code and companion tools for ensemble generation. It needs the COMODO tools package to be compiled.

Main features

The assimilated data are harmonic constants (from tide gauge or altimeter time series harmonic analysis).

  • multi-discretisation (on unstructured triangle elements)
  • mono/poly-chromatic

The present version is dedicated to the assimilation of elevation harmonic constants analysed from tide gauge and altimetry-derived sea level time series. This not a restrictive list, any other sea level analysis might be used, such as GPS-derived data. Ensemble are used to build-up the model error covariances needed by the data assimilation. Mono-chromatic data assimilation means that only one tidal  constituent is solved at a time (a quite standard procedure, used in previous FES or TPXO atlases production). Poly-chromatic data assimilation involves two or more tidal constituents solved together. It has been designed to take profit of the strong correlation between a given constituent and its non-linear harmonics. Those harmonics are barely observable from altimetry, as their amplitude are usually close or smaller than the background noise (more precisely non-tidal ocean signal at tidal aliased frequencies). Using the generating constituent(GC)/non-linear generated constituents (NLC) correlations allows for controlling the NLC with GC data. In other words, M4 constituent will be improved by assimilating M2 data in a poly-chromatic M2/M4 configuration, even in absence of M4 data.

Data assimilation is based on a quadratic cost function reduction, and formalism is based on representer approach. As time is eliminated thanks to the spectral (harmonic) approach, data assimilation reduces to an optimal interpolation problem, where data and model error covariances are carefully estimated. Limitations are the usual OI limitations, i.e. error space is assumed to have gaussian distribution properties. It can provide some basic diagnostics (penalty function for the prior and optimal solution, representers on demand,…) and some more synthetic ones (SVD eigenvalues and vectors for the error covariances matrices).

Future plans include multi-parameters (such as tidal currents) and 3D data assimilation, plus quadrangular element extension (therefore compatible with structured grids).

SpEnOI installation is based on auto-tools. A detailed installation manual is in preparation.

Recent applications

  • Persian Gulf pilot implementation
  • COMAPI project (regional tides)
  • FES2012 project (global ocean tides)

Main contributors

Laurent Robloulaurent.roblou@aero.obs-mip.fr
Florent Lyardflorent.lyard@legos.obs-mip.fr