Search

Oceanic convection, parameterized through vertical mixing schemes, is still not well captured by ocean general circulation models. A preliminary step necessary to improve these schemes is to evaluate and compare how the models behave for different forcing regimes. Literature often proposes single-case comparison (either on a specific location or a specific time or with a specific metrics). The goal of our work is to propose a more systematic framework allowing evaluations and comparisons over a larger range of forcing regimes. For doing so, we define a parameter space which has been derived thanks to a theoretical 1D model of the mixed-layer depth (MLD) evolution. This parameter space is formed by two dimensionless numbers : λs which describes the relative contribution of the buoyancy flux and the wind in the surface layer, and the Richardson number Rh which characterizes the stability of the water column at the mixed layer base. In this presentation, I will highlight the key features of this parameter space and I will illustrate its physical robustness with an ensemble of 1D simulations. These simulations were conducted by applying a 10 years JRA55-do 1.4.0 atmospheric forcing within a 1D standalone code making use of the NEMO Turbulent Kinetic Energy + Enhanced Vertical Diffusivity (TKE + EVD) scheme. Then, I will present a test case to study the impact of the horizontal resolution on the convection regimes for a TKE + EVD scheme in 1D, 1°, 1/12° and 1/60° realistic NEMO simulations. I will define convective regimes by sorting the values according to the normalized evolution of the mixed layer depth dt MLD / MLD and I will show that these regimes are almost kept in the parameter space between 1D and 1° but become generally less convective / more restratifiying when increasing the resolution, highlighting the restratification processes by lateral fluxes. Moreover, I will show that the dynamics in the Mediterranean is much more affected by the increase of resolution than the Labrador sea, suggesting that it involves more intense lateral restratification processes.

International audience; Periglacial rock walls are affected by an increase in rockfall activity attributed to permafrost degradation. While recent laboratory tests have asserted the role of permafrost in bedrock stability, linking experimental findings to field applications is hindered by the difficulty to assess bedrock temperature at observed rockfall locations and time. In this study, we simulated bedrock temperature for 209 rockfalls inventoried in the Mont Blanc massif between 2007 and 2015 and 209 000 random events artificially created at observed rockfall locations. Real and random events are then compared in a statistical analysis to determine the results significance. Permafrost conditions (or very close to 0 °C) were consistently found for all events with failure depth > 6 m, and for some events affecting depths from 4 to 6 m. Shallower events were likely not related to permafrost processes. Surface temperatures were significantly high up to at least 2 months prior to failure with the highest peaks in significance 1.5 to 2 months and 1 to 5 days before rockfalls. Similarly, temperature significances at scar depths were significantly high, but steadily decreasing, 1 day to 3 weeks before failure. The study confirms that warm permafrost areas (> -2 °C) are particularly prone to rockfalls, and that failures are a direct response to extraordinary high bedrock temperature in both frozen and unfrozen conditions. The results are promising for the development of a rockfall susceptibility index but uncertainty analysis encourages to use a greater rockfall sample and a different sample of random events.