Wind-stress variability over the Benguela upwelling system

Bibliography: leaves 119-133.Regional wind-stress variability over the Benguela Upwelling System is described using 16 months (01 August 1999 29 November 2000) of satellite derived QuikSCA T wind data. The QuikSCA T data are compared to the climatologies presented by Kamstra (1985) and Bakun and Nelson (1991), as well as the long-term climatology (1968-1996) of the surface vector wind speed field off the coast of southern Africa, as derived from the 2.5° resolution NCEPINCAR reanalysis dataset. Broad scale similaritie"s are found between the QuikSCA T and the long-term NCEPINCAR climatology (1968-1996) data sets. This allows one to have confidence in using this scatterometer data to investigate details of spatial and temporal variability over the Benguela System. During summer, wind-stress maxima are found at approximately 17, 29 and 34°S. These maxima strengthen in late summer. The seasonal northward migration of the South Atlantic Anticyclone becomes apparent in late autumn, when the strongest wind-stress occurs north of 28°S. A significant wind-stress minimum is observed to develop slightly north of Cape Columbine (33°S) during autumn. To the north (10-23.5°S) the Benguela is characterised by relatively strong south-easterly wind-stress during winter. To the south (24-35°S) the Benguela is characterised by relatively weak westerly to south-westerly wind-stress during winter. A southward migration of southeasterly wind-stress is observed during early spring. By November the entire Benguela Upwelling System is once again characterised by southerly to south-easterly wind stress. Wind-stress variability is investigated using both a type of artificial neural network, known as the Kohonen Self Organising Map (SOM), as well as a wavelet analysis. Two independent SOM studies are conducted. The first study produced a 6x4 SOM output array, which is used to examine seasonal variability as well as the temporal evolution of two synoptic-scale wind events. For the second study both a SOM and a wavelet analysis are applied to an extracted data set to find that the system can be divided into six discrete wind regimes, 10-15°S; 15.5-18.5°S; 19-23.5°S; 24-28.5°S; 29-32.5°S; and 33-35°S. The wavelet power spectra for these wind cells span a range of frequencies from 4 to 64 days, with each region appearing to contain distinct periodicities. To the north, 10-23.5°S, the majority of the power occurs during winter, with a 6-16 day periodicity. Further south, 24-35°S, the majority of the power occurs in the summer. Here a bi-modal distribution occurs, with peaks of 6-16 and 35-40 days. Lastly a case study sequence of the spatial distribution of wind-stress, windstress curl and SST, at a location off the west coast of southern Africa (25-300S and 12-17°E), is discussed in relation to an intense, upwelling favourable, wind event that occurred from 11-20 February 2000