A re-analysis of Black Sea surface temperature

Two decades of AVHRR Pathfinder data have been used to obtain a daily series of optimally interpolated sea surface temperature (SST) maps over the Black Sea at 1/16 degrees spatial resolution. The interpolated data and associated errors have been first compared with in situ measurements, and successively used to characterize SST variability. The validation evidenced that interpolated data present a mean bias error (MBE) of about -0.05 degrees C and a root mean square error (RMSE) of about 0.62 degrees C, while no regional differences or significant temporal drifts were found. The main patterns of variability at different temporal scales have been identified through several Empirical Orthogonal Function (EOF) decompositions. The analysis of the weekly SST data and associated error (assumed as a proxy of cloud cover) evidenced strong interactions with the overlaying atmosphere at short timescales, while examining the EOF's computed from a low-pass filtered SST time series, a significant correlation (up to -0.65) to the North Atlantic Oscillation was found. The SST variations at longer timescales followed the NAO forcing with a relatively large time delay (between 2.4 and 3.5 years), indicating that the processes relating the Black Sea SST to the atmospheric conditions on the longer timescales are more probably linked to the thermal and dynamical balance of the basin as a whole, and not limited to the immediate response of its surface layers. (C) 2009 Elsevier B.V. All rights reserved.