The sea surface temperature field in the Eastern Mediterranean from advanced very high resolution radiometer (AVHRR) data - Part II. Interannual variability

A ten-year dataset of AVHRR-SST (Sea Surface Temperature) with 18 km space resolution and weekly frequency has been analyzed in the Eastern Mediterranean. In Part I of the present study, we examined the seasonal variability of the basin and we defined the time and space scales of the monthly climatologies of the sea surface temperature distributions. A detailed and quantitative comparison was also carried out between the SST distribution of September 1987 and the surface temperature map of the corresponding hydrological survey of the POEM (Physical Oceanography of the Eastern Mediterranean) Programme. In this second part, we extend the analysis to examine and quantify the interannual variabilities. We summarize our results as follows. The analysis reveals the presence of an interannual cycle in the Ionian basin in the decade 1983-1992. The climatological pattern of the isotherms oscillates between two main states corresponding to the winter (zonal) and the summer (meridional) distributions. The apparent interannual cycle appears in the SST distribution only in winter and only in the Ionian basin, while the Levantine basin shows only fluctuations around the climatology. Thus, the interannual signal in our dataset is present but is much less important than the seasonal one. To confirm and quantify the above conclusions, we evaluate the Empirical Orthogonal Function of the SST time series. Gradient EOF's modes show the presence of a very strong yearly signal explaining 60% of the variance and a less intense seasonal signal that account for 16% of the variance. No interannual variability is represented by EOF gradient modes even though it can be observed in the increasing trend of the time varying amplitude of the first EOF mode, To observe the interannual variability signal, a seasonal temperature cycle was subtracted from each image instead of the ensemble (spatial) mean. A new EOF mode that accounts for 20% of the variance emerges between the first and second previous gradient modes. The time behaviour of this mode clearly indicates that it represents the interannual signal of the overall SST field. (C) 1999 Elsevier Science B.V. All rights reserved.