Skip to content

Latest commit

 

History

History
46 lines (38 loc) · 2.49 KB

forecast.md

File metadata and controls

46 lines (38 loc) · 2.49 KB
Raw
layout title permalink prevtext prevlink nextext nextlink
pagemd
Quantifying chaos and uncertainties in the ocean evolution
/forecast/
Exploring ocean fine scales motions
unravel
Understanding the ocean role in the climate machinery
climate

Ocean dynamics is described by non-linear equations, which connect the evolution of oceanic state variables (velocity, temperature, salinity, etc). Some theories predict that because of nonlinearity, the ocean evolution is not totally determined by external drivers such as the atmospheric variability: MEOM's high-resolution simulations confirm that the ocean evolution also depends substantially on small uncertainties in initial conditions, even over long (decadal and longer) periods. The ocean evolution is thus not fully deterministic in the turbulent regime, but partly chaotic over a wide range of timescales: it should be simulated from ensembles of multiple realizations, described and studied from probability distribution functions that represent this inherent uncertainty.

*Figure 1 : Probability distribution of sea surface elevation as described by a 96-member ensemble: free ensemble simulation, with stochastic parameterization of uncertainties (cyan), constrained ensemble simulation, with assimilation of satellite altimetry (green), and 10-day ensemble forecast (blue). Source : Candille et al., 2015*

The ocean models are also uncertain by construction due to e.g. unresolved (e.g. subgrid-scale) processes, or to the unperfect representation of the real dynamics. The MEOM team is working on methods to explicitly include these modelling uncertainties into ocean simulations by adding stochastic parameterizations : ensembles of such simulations represent the range of possible oceanic evolutions given these uncertainties. They may then be consistently corrected through data assimilation techniques to better match the observations and provide fully probabilistic forecasts.

MEOM research aims to design ensemble and stochastically-perturbed simulations, for characterizing, studying and reducing these uncertainties. Current projects include in particular:

  • the [OCCIPUT](add link here) project, which investigates the climate-related fingerprints and implications of the ocean's chaotic behavior

  • our studies of modelling uncertainties, in order to improve operational oceanic forecasting systems (eg. at Mercator-Océan).