ACPI Elements 1 and 2: Initializing the Coupled Model from Observed Ocean Conditions and the Ensemble Runs
Abstract
The first two ACPI-demo elements were to initialize the parallel coupled model (PCM) with observed ocean conditions, and then perform an ensemble of integrations. The challenges of these elements were: 1) Lack of comprehensive observations of the three-dimensional temperature (T) and salinity (S) structure of the ocean. 2) The likelihood that mismatches between real and model-implied atmospheric heat and freshwater fluxes would cause systematic model drift, obscuring the signals of interest. 3) Potential contamination of the initialization field by interannual or decadal variability, which was not of interest in this application. To address the lack of comprehensive observations, we used assimilated data from a project that uses a full, three-dimensional adjoint model to incorporate observed surface T, S, and altimetric data into a physically self-consistent density field. Oceanic heat content anomalies in the assimilated data set agreed well with the available observations, lending confidence to the quality of the assimilated data. To address the problem of systematic drift, we added the anomalies of the assimilated data set onto the coupled model's climatological T and S fields. This reduces the flux mismatch problem by at least an order of magnitude, since the anomalous heat transports associated with the anomalous heat contents are no more than O(0.1). Later analysis of the PCM's own heat content anomaly field in the 1990s showed it to be in remarkable accord with the observations, which suggests that the model's anomalous surface heat fluxes are close to the actual ones, and that the drift using this initialization method is negligible. To address the problem of decadal and interannual variability, full density anomalies were only imposed below 200 meters, and linearly ramped to zero at the ocean surface. This preserves virtually all the heat content signal (which extends to 3000m depth), but eliminates higher-frequency variability, which is strongest near the surface. The ensemble of integrations includes three cases with the assimilated initial conditions, starting at year 1990 and ending at 2099, plus one control case (forcing fixed at 1990 values). In addition to monthly fields, some values were stored at 6-hr intervals for later use by the hydrological modeling effort (ACPI-demo element 3).
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2001
- Bibcode:
- 2001AGUFM.H21G..01P
- Keywords:
-
- 1600 GLOBAL CHANGE;
- 3337 Numerical modeling and data assimilation