The subsurface density of the North Atlantic displays considerable variability on decadal timescales. In this thesis the variability is examined from two directions. First I show direct observations of oceanic Rossby waves, using historical subsurface temperature data in the North Atlantic. Previously, such waves have only been directly observed using satellite data. Subsurface temperature data are used to investigate westward propagating Rossby waves across the North Atlantic at 30.5°N to 34.5°N, during 1970-74 and 1993-97. The latter period is when high resolution satellite altimetry data are available for comparison. The comparison was very favourable, with observed phase speeds east of the mid-Atlantic ridge, 3.6 cm s<sup>-1</sup> in the altimeter data and 3.4 cm s<sup>-1</sup> for the corresponding period in the hydrographic data. The phase speed of Rossby waves (west of the mid-Atlantic ridge) was observed to be 6.1 cm s<sup>-1</sup> in 1970-74, compared with 4.3 cm s<sup>-1</sup> in 1993-00. Interannual variability of the phase speed was also observed in the altimeter data, this variability was greater in the west than the east. Data from a coupled climate model (HadCM3) are used to assess the ability of an assimilation scheme, based on temperature-salinity preservation, to represent salinity given temperature data. The assimilation scheme was amended to account for meridional frontal movements. The adjustment improves the pre-assimilated salinity in comparison to the true-salinity in all areas south of 55°N, in the North Atlantic. Finally initial condition experiments are performed to determine the predictability of sea surface temperature (SST) and heat flux. The temperature and salinity fields resulting from the assimilation scheme are used as initial conditions in a twin experiment of HadCM3. When the temperature and salinity of the initial ocean state are perfectly known, the SST anomaly forecast skill is up to 5 years. When the assimilation scheme is used to provide these initial conditions for a forecast, the skill of predicting the SST anomaly is almost 2 years. There was no skill in predicting the fluxes in HadCM3 on a seasonal or annual timescale. However the average winter flux anomaly, when averaged over 5 years was found to be predictable.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:642087 |
| Date | January 2002 |
| Creators | Brown, S. C. |
| Publisher | University of Edinburgh |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.0026 seconds