ESSE Uncertainty Forecasts

ESSE Uncertainty Forecasts, Data Assimilation and Adaptive Sampling

Using the error subspace forecast for August 28, 0000GMT T and S data collected during August 27-28 are assimilated using ESSE. During this process, the T and S data correct the T and S fields, but also the velocity fields since ESSE error covariances are multivariate. Uncertainties are reduced and the output is a corrected ocean field estimate (an analysis) and an updated error estimate (the posterior error subspace). Starting from these posterior uncertainties, a ESSE ensemble error forecast is computed for August 31, GMT00.

Data were assimilated on August 28 PST, updating the forecast ocean fields and reducing the corresponding forecast errors. The size of the error subspace used to do so was 500 (retaining all 500 ensemble members). The error standard deviation estimates after data assimilation are illustrated by the table below.

It is interesting to compare these posterior error fields with the prior ones The impacts of the WHOI gliders, SIO gliders and different ship data are clearly visible. In particular, note that the prior error covariances are non-homogeneous, non-isotropic and multivariate. Error reductions thus occur with scales that vary with regions and directions, and across all velocity variables.

ESSE uncertainty initialization procedure and results

The dominant 400 eigenvectors of the posterior error covariance estimate for August 28 PST were utilized to perturb the ocean fields. A white noise of an amplitude proportional to the estimated absolute and relative errors in the observations is added to this random combination, in part to represent the errors truncated by the error subspace. An ensemble of forecast simulations, each forced by forecast COAMPS atmospheric fluxes issued for Aug 29, was then carried out.

A total of 500 perturbed forecasts were carried-out. The spread of the resulting 500 ensemble members is illustrated by the standard deviations shown in the table below.

Analysis of ESSE Results

Dynamics and uncertainties. In Monterey Bay, forecast uncertainties remain larger at the mouth of the Bay, especially near the Monterey Bay Peninsula (e.g. see T and S field at 30m).

Offshore, in the full domain, the effect of the lack of data along the southern boundary is clearly visible. In the interior of the domain, the meandering of the coastal current (see psi field and T, S, u, v fields at 0, 10 and 30m) continues to impact the uncertainty forecast. At 200m, the impact of the California Under-Current (near the coast south of PS, but meandering in the middle of the modeling domain elsewhere, is also visible.

Adaptive sampling recommendations

The gliders should focus on the mouth of the Bay and on the corresponding fronts, especially just offshore from the Monterey Bay Peninsula.