To estimate environmental uncertainties, various scenarios were computed daily as a function of different initial condition estimates, assimilation procedures, modeling domains, numerical/physical model parameters and time of day. Different initial conditions were created based on different: sea surface temperature estimates, (re)-starts of simulations and initial pertubations. The sound section fields for different initial conditions sometimes differed quite a bit because of the limited real-time data in the wider Pianosa-Elba region.
Real-time high-resolution atmospheric forcing is in use, and there appears to have been some wind mixing on the 18th and 19th. The thermocline differed slightly from the fields used yesterday (which didn't have the full real-time SST and forcing impact).
The mean sound-speed background estimate has been improved as well as the estimate of its error std.
With updates to the error estimates and field estimates, the new yoyo code was able to capture the afternoon warming (daily thermocline) on the return path in the afternoon. These real-time computations of yoyo parameters on Jul 20 (for Jul 21), using the ocean realization corresponding to Run 2 above, are available here.
Note that due to the predicted higher wind conditions, sound speeds are smaller in surface and changes have also occurred in the thermocline depths.