A wind and wave atlas for the Mediterranean Sea Luigi Cavaleri Institute of Marine Sciences National Council of Research

1. A wind and wave atlas for the Mediterranean Sea Luigi Cavaleri Institute of Marine Sciences National Council of Research Venice - Italy

2. The atlas presented has been commissioned by the French, Greek and Italian Navies via the Western European Armaments Organisation (W.E.A.O) Research Cell. It is the result of the Medatlas project led between 1999 and 2004 by a consortium of six companies located in France, Italy and Greece. NTUA (GR) - Statistical Analysis of wind and wave data - Production of charts and tables - Design and compilation of the Printed Atlas CS (F) - Co-ordination of the Project and validation of the Electronic Atlas ISMAR - CNR (I) - Supply of buoy data - Analysis of wind and wave model data - Calibration of model data vs. satellite data THETIS S.p.A. (I) - Conceptual design and compilation of the Electronic Atlas SEMANTIC TS (F) - Contribution to validation and calibration of altimeter data vs. buoy data METEO - FRANCE (F) - Supply of model data and satellite data - Contribution to validation and calibration of altimeter data vs. buoy data

3. The aim of the atlas is to provide reliable long term wind and waves statistics at specified points of the Mediterranean Sea at practically every offshore location (at about 50 km intervals). The information contained in the Atlas can be used in various ways and for different applications in Ship Design and Operational Planning, as well as in Offshore and Coastal Engineering, e.g.: • Operational Planning and Optimization of existing ships or fleetcalculate operability indices, optimize the overall efficiency of ships • Comparation of different designs by means of their operability indices • Operational planning of offshore activities (naval or civil) • Ship Design • Assessment of offshore wave energy resources.

4. Data Calibration Results

5. We have at disposal two main sources of information: • Satellites – measured data, but sparse in space and time • Model data – continuous in space and time, ok on the oceans, but often with problems (underestimation) in enclosed seas. • Solution: combine the two sources, using their overall information The two basic sources of data have been the altimeters on board of the ERS-1/2 and Topex/Poseidon satellites and the operational wind and wave models results of the European Centre for Medium-Range Weather Forecasts. Wind and wave data are provided by altimeters (wind speed U and significant wave height Hs) and scatterometer (wind speed and direction). We have made use of the data from ERS-1/2 and Topex/Poseidon, available respectively since 1991 and 1992.

6. The ERS altimeter data, made available by Meteo-France, provide point estimates, at about seven kilometer intervals, along the ground track of the satellite. Ground tracks at different passes differs by about 80 km. Having been used with data assimilation, hence considered in the ECMWF analysis, the scatterometer data could not be considered for this project.

7. Topex has been following an orbit with a 10 day period (instead of 30). This triples the number of data per point with respect ERS 1-2, at the expenses of tripling the distance between adjacent tracks, now 240 km. Many regions without enough data for a proper analysis.

8. The analysis results of the European Centre for the Medium-Range Weather Forecasts (ECMWF, Reading, U.K.) are available at 6-hour intervals (10 years = about 14,600 fields).

9. Since July 1992 the European Centre for Medium-Range Weather Forecasts run, parallel to their meteorological model, a wave model. The wave model used at ECMWF is WAM, an advanced third generation model developed with the co-operative effort of most of the experts available at the time. WAM considers explicitly all the fundamental processes that affect the evolution of the wave fields. As for the wind, wave fields are available at 6-hour intervals at all the grid points.

10. The wind and wave integrated wave parameters available from the ECMWF archive have been retrieved from July, 1st 1992 to June, 30th 2002. The data have been retrieved with 0.5 degree resolution between 6° West and 36° East for longitude, and 30° and 46° North for latitude. This corresponds to a 85x33 point grid, out of which about 950 are sea points.The electronic version of the atlas includes 239 points. Most of these are at one degree interval both in latitude and longitude, complemented by some sparse points in the most critical areas. For the printed version the number has been reduced to 129.

11. Data Calibration Results

12. For each data-set and for each single satellite measurement, e.g. the ECMWF and the Topex measured wind speeds, the model data have been interpolated in space and time at the satellite pass time and position. Then, each couple of data (model value and satellite measurement) has been assigned to the closest grid sea point. This has provided for each grid point a sequence of couples of data, sparse in time, suitable for a local analysis. For each point the model-sat distribution has been approximated with a best-fit straight line, passing through the origin. The slope of the best-fit line is the estimate of the average ratio between model and sat data.

13. Substantial underestimates by the model, particularly in the northern parts of the basin. Strong underestimates, for both wind and waves, in the areas with the more complicated geometry, and possibly with a complicated orography in the nearby land. It is easy to identify specific problems in the northern part of the Tyrrhenian Sea, in the Adriatic Sea and in the Aegean Sea.

14. Following the procedure outlined above, the single calibration coefficients for wind speed and wave height have been derived at the various grid points. Because the variability of the geographical distributions of both coefficients, the overall information has been summarised in the calibration coefficients only at the selected points, whose values has been derived with a careful analysis of the values at the surrounding not-considered-for-atlas grid points.

15. The wave period has been corrected using the square root of the calibration factor used for the wave height. No correction has been introduced for wave direction. The corresponding scatter indices (SI= rms error / mean measured value) have been evaluated point by point, and their distribution provides a clear indication of the reliability of the results in different areas.

16. Once we had the best-fit slopes throughout the basin, both for wind speed and wave heights, the next steps have been: • correct the model data according to the estimated best-fit slopes, • derive the overall and point statistics from the calibrated model data. The procedure has been applied separately to the data before and after 20 November 2000, i.e. when the substantial change of resolution in the operational ECMWF meteorological model implied a change of quality of the wind, hence wave, fields. Therefore for the first period we have at disposal 101 months of data, and only 19 for the later one.

17. The calibrated results were cross-checked by: • Direct comparison with measured data, not used during the calibration phase. • For waves, we have several buoys scattered in the Mediterranean Sea. Slope = 0.98 Slope = 1.05

18. TOPEX ALTIMETER DATA WIND WAVES • For wind we do not have many open sea measurement points. However, • we can cross-check the consistency of the wind and wave fields • The direct relationship existing between wind and waves implies a consistency between their statistics. This is obviously present in the original data that are model derived, and it should be expected also in the calibrated data. A further analysis of the calibrated data suggests the “calibrated winds” to be too low with respect to the “calibrated waves”.

19. WAM WAM This has been verified by hindcasting one year of wave data (same model as in the original data) using as input the calibrated wind fields. The resulting wave heights are too low by about 10% with respect to the calibrated values. As the latter ones have also been verified versus accurate buoy data, this suggests that the the algorithm used to derive the altimeter wind speed provides values lower than the actual truth in the Mediterranean Sea by a few percent. WIND WAVES 10% calibration calibration WAM WAVES ECMWF WIND

20. Data Calibration Results

21. The atlas provides synoptic charts and point by point data tables. The charts show the geographical distribution of the mean values of the wind and main wave parameters, plus the probability contours for different U and Hs values.

24. The tables provide bivariate frequency distributions of wind speed and direction, wave height and peak period, wave height and direction, and wave height and wind speed. All these results are provided on yearly and seasonal basis.

25. Conclusions: • we have presented what we believe is presently the best atlas of the wind and wave conditions in the Mediterranean Sea, • the wave results have been verified versus independent measured data, • the wind values in the atlas are probably too low by about 5%, • the reason is the underestimate of the wind speed by the altimeters in the Mediterranean Sea.

26. The atlas is available upon request luigi.cavaleri@ismar.cnr.it Thank you for your attention