1 |
heimbach |
1.1 |
Surface fluxes and Ocean state estimates in the eastern |
2 |
|
|
Subtropical North Atlantic. |
3 |
|
|
|
4 |
|
|
Olwijn Leeuwenburgh, Patrick Heimbach and Detlef Stammer |
5 |
|
|
|
6 |
|
|
A coarse resolution regional ocean model of the eastern |
7 |
|
|
subtropical North Atlantic is used in a 5-year run to |
8 |
|
|
investigate the effects of two different surface forcing |
9 |
|
|
formulations and of several strategies to minimize a cost |
10 |
|
|
function in the context of ocean state estimation. |
11 |
|
|
The model is a 2 degree resolution application of the |
12 |
|
|
MITgcm which is forced at the open boundaries by |
13 |
|
|
temperature, salinity and current velocities obtained |
14 |
|
|
from a global implementation of the same OGCM. The assimilated |
15 |
|
|
data consist of T/P and ERS altimetry, a mean sea surface |
16 |
|
|
estimate, and Reynolds SST, and the results are compared |
17 |
|
|
with independent buoy data obtained from the Subduction |
18 |
|
|
Experiment. Different strategies to bring the model |
19 |
|
|
into consistency with ocean observations are being discussed. |
20 |
|
|
All of them use the models adjoint to adjust control variables, |
21 |
|
|
such as the initial state, lateral boundary conditions and surface |
22 |
|
|
forcing. Control terms include either the surface fluxes of momentum, |
23 |
|
|
net heat and freshwater fluxes, or the atmospheric state as control |
24 |
|
|
variables. Early results indicate that there are distinct differences |
25 |
|
|
in magnitude and behavior of the cost function depending on |
26 |
|
|
the type of forcing and the manner in which control |
27 |
|
|
variables are included and adjusted. While a bulk-formulae |
28 |
|
|
type forcing leads to better initial agreement between |
29 |
|
|
model and data, the optimization is more effective (faster |
30 |
|
|
decay) with the traditional type of flux forcing. |
31 |
|
|
|