Search

Abstract: Interannual variability of wind speed at 10 m above the sea surface is investigated over the Mediterranean Sea, with a particular focus near land–sea boundaries. For this purpose, monthly mean winds are formed at a resolution of 0.25° using twice-daily, rain-free, wind measurements calibrated to equivalent neutral winds from the QuikSCAT satellite during 2000–2006. A stability correction applied to these satellite-based winds reveals that the equivalent neutral winds are typically stronger by 0.2 ms−1, and can even be stronger by >0.5 ms−1 in some regions, including the Adriatic and Aegean Seas. Thus, the impact of air–sea stratification on the winds cannot be neglected, even on monthly time scales. Winds from a numerical weather prediction (NWP) model, the 0.5°-resolution Navy Operational Global Atmospheric Prediction System (NOGAPS), are found to have close agreement with the satellite-based winds. However, major differences arise near coastal boundaries where the winds from NOGAPS over the sea are contaminated by wind values over the land. Land–sea mask values from NOGAPS are introduced to examine the extent of the contamination, which can be severe (e.g., 2 ms−1 weaker) in comparison with QuikSCAT winds, especially near the northern boundaries. A creeping sea-fill methodology applied to the NOGAPS winds typically results in better agreement with the satellite winds. The accuracy of the NOGAPS winds is further improved by correcting them based on the satellite winds using a linear regression analysis. Finally, a 3.5-km-resolution HYbrid Coordinate Ocean Model (HYCOM) is forced with these original, sea-filled and regression-corrected NOGAPS winds. Sea-surface temperatures (SSTs) simulated by HYCOM demonstrate that the land-contaminated original winds from NOGAPS result in a relatively warm SST bias of >2 °C in comparison to a satellite-SST analysis near the coastal boundaries. Sea-filled and regression-corrected winds significantly improve the accuracy of the SSTs from the model. The results presented in this paper clearly reveal that (1) stability dependence in the satellite winds cannot be ignored and (2) winds from a NWP product (NOGAPS here) over the sea may not be accurate near land–sea boundaries due to contamination by land values and can be improved either locally or via a regression against QuikSCAT winds. [Copyright &y& Elsevier]