The word ``Metocean'' was constructed from the contraction of meteorology and oceanology.

Metocean activities act as a support for design and operation of the offshore structures of the Industry: platforms and other oil & gas production systems, ships, barges, and of coastal planning and management: dykes and breakwaters, sewage outlets, hydrodynamics of sediment and pollutant transport. Metocean studies rely mainly on statistics of oceanographic measurements, on hindcast using numerical models, and on the identification and estimation of characteristic quantities relevant to the phenomena of interest, i.e. essentially waves, wind and currents.

Real-time forecasting plays only a very small part in this activity, in comparison to the analysis and study of datasets; the values that are searched for, usually design values, pertain to the entire life of a structure or to an operation of long duration, extending thus far beyond the few days for which it would be possible, at a given date, to prepare a forecast.

Created by decree of 5 June 1984, modified on 18 February 1998, the French Research Institute for Exploitation of the Sea 'Ifremer' is a public institute of industrial and commercial nature. From January 2002, it will be placed under the joint supervision of four ministries: Research, Agriculture and Fisheries; Transport and Housing; Environment.

The Metocean group maintains daily, as internal tool, reference lists on the subjects related to its activities.

[1] Athanassoulis, G.A., Soukissian, T.H., Stefanakos, C.N., 1995, "Long-term variability and its impact to the extreme-value prediction from the series of significant wave height", 4th International Workshop on Wave Hindcasting and Forecasting, Banff, Alberta, pp. 343-358. Abstract ->

[2] Debernard, J., Saetra, O., Roed, L.P., "Future wind, wave and storm surge climate in the northern North Atlantic", Climate Research, vol. 23, no. 1, pp. 39-49.

In this paper we consider a possible change in future wind, wave, and storm surge climate for the regional seas northeast of the Atlantic: the northern North Atlantic. Conclusions are based on a statistical analysis of the results derived with state-of-the-art wave and storm surge models run for two 20 yr time-slice periods, one for the period 1980 to 2000 (control climate) and one for the period 2030 to 2050 (future climate). Forcing was derived by extracting atmospheric wind and sea level pressure from a state-of-the-art regional atmospheric climate model for the same two 20 yr periods. These regional atmospheric simulations constitute dynamical downscales of the Max Planck Institute's global scenario which includes greenhouse gases, sulphate aerosols (direct and indirect effects) and tropospheric ozone. Generally the changes we found are small. However, there are some important exceptions, such as a significant increase in all variables in the Barents Sea and a significant reduction in wind and waves north and west of Iceland. Also, there is a significant increase in wind speed in the northern North Sea and westwards in the Atlantic Ocean, and a comparable reduction southwest of the British Isles in the autumn. The same change is suggested for wave height, but this is not statistically significant. There is a significant increase in the seasonal 99 percentile of the sea level in autumn in the southwest part of the North Sea. These results are consistent with earlier studies predicting a rougher maritime climate in the northern North Sea in autumn.

[4] Gulev, S.K., 1999, "Climate Changes of Wind Waves in the North Atlantic Over the Last Several Decades", Proc. ISOPE Conf., vol. III, pp. 164-167.

On the basis of visual wave observations available from the Comprehensive Ocean-Atmosphere Data Set (COADS) we consider secular tendencies during 1963-1993 in the height of wind sea and swell of different directions in the North Atlantic mid latitudes. Analysis shows that growing swell in the North East Atlantic does not result from the swell propagating from the Central Atlantic where it is associated with increasing wind sea. Possible alternative mechanisms of growing swell are discussed.

There is evidence of changes in sea level and wave height on various time-scales, globally and regionally. For example, the North Atlantic Oscillation appears to be correlated with increasing wave height in the North Atlantic over recent decades. These changes may or may not be anthropogenic but must be planned for, in an integrated coastal management plan. The impact of any increase in wave height in the North Atlantic at the coastline in the North Sea, the Irish Sea, the Malin/Hebrides Shelf and the English Channel will be quite different. The effect of increasing sea levels, due to global warming, and any changes in tides and surge height and frequency, is combined with changes in offshore wave height. Effects of change in wave period and direction may also be significant. Shallow water wave modelling, using the WAM and SWAN wave models, provides a useful tool for examining the possible impacts of climate change at the coast. This is part of a project studying the vulnerability of the UK coast to changing wave climate and sea level. Initial results validating the wave models and testing simplified NAO scenarios are presented.

[6] Dong, P., Chen, H., 1999, "A probability method for predicting time-dependent long-term shoreline erosion", Coastal Eng., vol. 36, no. 3, pp. 243-261.

This paper describes a practical procedure for predicting the probability distributions of long-term shoreline erosion. It treats the dynamical response of the shoreline over time as that of a time-dependent stochastic system. The input to the system are the long-term wave climate and shoreline properties while the output of the system are the probability distributions of the maximum shoreline recession within any prescribed time period. The procedure enables the combined effect of the longshore and cross-shore sediment transport processes on the shoreline erosion to be taken into account as both processes are generally responsible for the maximum shoreline recession. A series of simulations using idealised long-term wave distributions which have relatively narrow directional spread was carried out to evaluate the capability of the method. Based on these simulations it was found that both the distributions and variability of wave climates have significant influence on the predicted shoreline erosion probability distributions while the chronology effect is less significant, especially for long simulation time.

[8] Kriebel, D.L., Dean, R.G., 1993, "Convolution method for time-dependent beach-profile response", J. Waterway Port Coast. Ocean Eng., vol. 119, no. 2, pp. 204-227.

A simple analytical solution is presented for approximating thetime-dependent beach-profile response to severe storms. This solution isin the form of a convolution integral involving a time-varyingerosion-forcing function and an exponential erosion-response function.The erosion-forcing function reflects changes in the nearshore waterlevel and breaking wave height. In this paper, an idealized storm-surgehydrograph is considered from which an analytic solution is obtained forbeach and dune erosion associated with severe storms such as hurricanesor northeasters. It is shown that for a given initial beach geometry andsediment size, the peak water level and the incipient breaking waveheight determine the maximum erosion potential that would be achieved ifthe beach were allowed to respond to equilibrium. Because of the assumedexponential erosion rate, beach response obtained from the convolutionmethod is found to lag the erosion forcing in time, and is dampedrelative to the maximum erosion potential such that only a fraction ofthe equilibrium response actually occurs.

[9] Francescutto, A., 2002, "Sea Waves and Ship Safety - State of Art in Current Regulations", Proc. ISOPE Conf., vol. III, pp. 150-155.

Sea waves play a major role in ship safety, yet stability criteria both for intact and damaged ship generally contain only indirect reference to a sea state or a rough description of sea action introduced as interim long ago and never updated. In this paper this matter is reviewed with reference to international rules and recommendations and drawbacks are evidenced especially for modern ships with large rolling period.

[11] Guedes Soares, C., Krogstad, H.E., Prevosto, M., 1994, "WAVEMOD Project: Probabilistic Models for Coastal Site Investigations", Proc. OCEANS 94 OSATES Conf., vol. I, pp. 493-497.

[12] Kawabe, H., 2002, "Contribution of supposed wave condition on the long-term distribution of a wave-induced load", Journal of Marine Science and Technology -Tokyo-, vol. 6, no. 3, pp. 135-147.

[13]

[14] Lebas, G., Van Dyck, J., Veneziano, D., 1991, "Extrapolation of design waves in the gulf of Guinea", Proc. 10th OMAE Conf., vol II, pp. 33-40.
copy not available at Ifremer "Metocean team"

This paper summarises a method for the estimation of design wave heights in the gulf of Guinea at locations other than those where wave heights have been observed. The proposed method combines the data observed at a limited number of sites with the results from a mathematical ocean wave model (hindcast data). The latter results are used to correct for systematic differences in sea-state severity between the site of interest and the recording sites. The method accounts for various sources of uncertainty, including: 1. the limited record at the measurements site(s), 2. the limited accuracy of the ocean wave model, 3. the limited amount of results generated by the ocean wave model. The paper describes the different proposed models, indicates how to estimate its parameters and discusses how the method can be used to determine spatial maps of design wave height. Finally, some results are given.

[16] Smith, D., Birkinshaw, M., 1996, "The extreme weather hazard - airgap determination", Proc. ERA Conf., pp. 7.1.1-7.1.10.

[17] Tromans, P.S., Vanderschuren, L., 1995, "Response based design conditions in the North Sea: Application of a new method", Offshore Tech. Conf., no. OTC 7683, pp. 387-397.

[18] Soukissian, T.H., Chronis, G.T., Nittis, K., Diamanti, C., 2002, "Advancement of Operational Oceanography in Greece: The Case of the Poseidon System", The Global Atmosphere and Ocean System, vol. 8, no. 2-3, pp. 93-107.

Operational Oceanography in Greece has become today a necessity, since it can provide the means and the solutions on societal, economic, environmental and scientific problems related with the coastal environment. A rational approach to these problems can be based on integrated coastal zone management, which, in turn, is supported by the systematic and continuous operation of marine monitoring systems. A real-time monitoring and forecasting system (called the "POSEIDON" system) is currently in operation in the Aegean Sea. The three main components of the POSEIDON system are: (a) an integrated marine monitoring network consisting of oceanographic and wave buoys (POSEIDON network), (b) a telecommunication system for real-time data transmission and (c) the Aegean Operational Forecasting System (AOFOS) including a meteorological model, an offshore wave model, a general circulation ocean model, a surface pollutant dispersion model and a shallow water wave model. Three years of experience has shown that the POSEIDON system is an essential and necessary tool for developing Operational Oceanography in Greece, being also the only integrated and multipurpose system of this nature in the whole Mediterranean Sea.

A preliminary analysis of atmospheric and surface oceanic observations from the POSEIDON buoy network is presented. The data set consists of concurrent observations of physical properties at the atmosphere-ocean boundary layers as well as chlorophyll-f (chl-f) and dissolved oxygen content of the subsurface layer. The quality of data is evaluated using standard procedures and comparison to reference measurements with special focus on bio-fouling effects. The extended temporal and spatial coverage of the observations provides, for the first time in the Aegean Sea, the opportunity to study synoptic and mesoscale variability in the upper oceanic layer and its response to atmospheric forcing. Furthermore, examples of high frequency variability of physical and biochemical processes in the mixed layer and the seasonal pycnocline are given.

[20] Gaspar, P., Labroue, S., Ogor, F., Lafitte, G., Marchal, L., Rafanel, M., 2002, "Improving nonparametric estimates of the sea state bias in radar altimeter measurements of sea level", Jour. Atmospheric and Oceanic Techn., vol. 19, no. 10, pp. 1690-1707.

A fully nonparametric (NP) version of the sea state bias (SSB) estimation problem in radar altimetry was first presented and solved by Gaspar and Florens (GF) using the statistical technique of kernel smoothing. This solution requires solving a large linear system and thus comes with a significant computational burden. In addition, examination of GF SSB estimates reveals a marked bias close to the boundaries of the estimation domain. This paper presents efforts to improve both the skill and the computational efficiency of the GF SSB estimation method. Computational efficiency is rather easily improved by an appropriate kernel choice that transforms the linear system to be solved into a very sparse system for which fast solution algorithms exist. The estimation bias proves to be due to the GF choice of a rudimentary NP estimator for conditional expectations. Use of a more elaborate estimator appears to be possible after a slight adaptation of the method. This solves the bias problem. Further improvement of the estimation skill is obtained by a local tuning of the kernel bandwidth. The refined estimation method is finally used to obtain a new NP estimate of the TOPEX SSB. This estimate yields larger SSB values than most previous estimates, in better agreement with recent in situ observations.

Extreme sea levels from 18 ports in the Aegean and Ionian Seas were analysed in the present study. The joint probability of ther-largest annual sea-level observations and the revised joint probability method have been used. Return periods and associated errors are produced from the two methodologies. The estimated return values are shown to be spatially coherent with enhanced values at the tide gauges located in the North Aegean and in enclosed gulfs. In spite of the low tidal and surge signal in the area, the estimated return values depend on the model chosen to represent the asymptotic distribution of the maxima of sea level.

[22] Cooper, C.K., 1997, "A comparison of tow criteria derived from satellite- and ship-based observations", Proc. ASME ASIA '97.

[23] Dyck, J., Van Veneziano, D., Lebas, G., 1990, "The risk of towing operations", Proc 9th OMAE Conf., vol. 2A, pp. 151-159.

Sizing the sea-fastening system for offshore structures to be towed from the manufacturing site to the operational site requires a choice of design sea state along the most critical part of the route. This procedure is unsatisfactory because insufficient account is taken of the non-homogeneity of the sea state on the route, of spatial and temporal persistence of sea conditions and of trip duration. Nor does it consider the relatively accurate sea state predictions that can be made now, using weather forecasting and ocean wave modelling. Using these the risk can be reduced by postponing the departure or by rerouting to a safe shelter. These factors can be accounted for by building a simulation model that simultaneously generates seastates and forecasts for different times and places along the route. The convoy progression is also simulated and sea state severity record to produce a probability distribution of maximum wave parameters or barge responses. The paper presents the formulation, estimation and application of the model for an Atlantic route along the French and Portugese coastsSea-state parameters.

[24] Lee, B.-C., Fan, Y.-M., 2002, "Analysis of Wave Characteristics over HsinChu Water", Proc. ISOPE Conf., vol. III, pp. 143-149.

This paper adopts statistical methods to understand the wind and wave characteristics over the Hsinchu Water, where there are three observing pots: the coastal wind station, the Hsinchu data buoy station, the Kuokwang oil platform station. Based on the results of analysis, the following features over this water have been found. The wind fields over this water are dominated by monsoon. The wind speeds near coasts are less than those near offshore. However, the trends of observed wind speed among stations are similar. The differences of wind direction show that the deviations are smaller as the wind speed increased among three stations. The wave periods are normal distributions observed at buoy and platform stations, but the wave heights are not. There are, sometimes, two wave systems appeared in the join wave-period distribution. High correlations and lag time at +00-hr, or 2-hr are found between the wind and the wave observing item in each offshore station. The correlated formula for wave heights itself are power regression type for the summer, winter, and typhoon seasons. The typhoon cases are the highest correlation. Furthermore, the wave heights in Kuokwang oil platform area can be estimated from calculating the product of the Markov transfer matrix and wave height measured by the Hsinchu data buoy station. The error decreases when the wave heights increase, and it has the highest error in the summer season, smallest value in the typhoon duration. For the interesting high wave height, the error is less than 20%.

In this paper, a statistical analysis on wind and wave buoy measurements and wind and wave model forecasts obtained during a two-year period (1999-2001) is presented with reference to four characteristic near-shore sites of the Aegean Sea. The measurements are a main product of the "POSEIDON" system aiming at the monitoring and forecasting of the state of the Greek seas, operated by the National Centre for Marine Research (NCMR). Although the two-year period is rather short for a thorough analysis of the local wind and wave climate, yet the obtained results, presented herein for the first time, reveal some interesting features of the corresponding wave and wind characteristics. Comparisons between the measurements and the forecast results are also performed at the locations under consideration. It is found that (i) wind speeds obtained from the POSEIDON weather forecasting system are, in general, in agreement with the measurements, except for high wind speeds which are systematically underestimated, (ii) the WAM model can successfully follow the monthly and over year trend of the evolution of wind and wave characteristics, but face significant problems for efficient sea-state forecasting. Finally, the overall pattern of the wind/wave climate for the entire Aegean Sea as obtained from the models is presented by means of the spatial distribution of the mean annual wind and sea-state intensity.

[26] Athanassoulis, G.A., Skarsoulis, E.K., Belibassakis, K.A., 1994, "Bivariate distributions with given marginals with an application to wave climate description", Applied Ocean Research, Vol. 16, pp. 1-17.

[27] Labeyrie, J., 1988, "Joint probability distributions of Met-Ocean parameters", LEMPAB Rep., no. 88/LEMPAB/RCH/01, p. 15.

[28] Atanga, J.N., Wyatt, L.R., 1997, "Comparison of inversion algorithms for HF radar wave measurements", IEEE Journ. Ocean Eng., vol. 22, no. 4, pp. 593-603.

[29] Bentamy, A., Grima, N., Quilfen, Y., 1998, "Validation of the gridded weekly and monthly wind fields calculated from ERS-1 scatterometer wind observations", The Global Atmosphere and Ocean System, vol. 6, pp. 373-396.

[30] Bern, T.-I., Barstow, S., "ERS-1 Cal/Val in-situ measurement, wind and wave measurements during extreme situations".

[31] Bitner-Gregersen, E.M., Bonicel, D., Hajji, H., Olagnon, M., Parmentier, G., 1996, "World-wide characteristics of Hs and Tz for long-term load responses of ships and offshore structures", Proc. 6th Intl Offshore & Polar Engng Conf, 26-31 May 1996, Los Angeles, USA, vol. III, pp. 95-102.

The paper compares the satellite data provided by the CLIOSat database with the Global Wave Statistics as well as WRB data, and assesses the variation in the long-term load responses and fatigue damage arising from use of the different data types. The North Atlantic ship route is considered. In the first part of the paper, methods for derivation of Tz estimates from SAR image spectra are discussed. Further, the Tz estimates are computed with the Global Wave Statistics periods and WRB data. Finally, case studies illustrating the uncertainty in response characteristics arising from use of the Global Wave Statistics and satellite wave data are given.

[33] Hasselmann, K., Hasselmann, S., 1991, "On the nonlinear mapping of an ocean wave spectrum into a synthetic aperture radar image spectrum and its inversion", Journ. Geophys. Res., vol. 96, no. C6, pp. 10,713-10,729.

[34] Krogstad, H.E., Barstow, S.F., 1999, "Satellite wave measurements for coastal engineering applications", Coast. Eng., vol. 37, no. 3-4, pp. 283-307.

Measurements from the GEOSAT, ERS-1 and 2 and Topex/Poseidon satellites have now accumulated to over 15 years of global ocean wave and wind data. Extraction of wave height, wind speed and wave period from the satellite altimeters and directional wave spectra from the synthetic aperture radars are reviewed along with recent validation and calibration efforts. Applications of the data to a variety of problems illustrate the potential of satellite wave measurements.

[36] Queffeulou P., Chapron, B., Bentamy, A., 1999, "Comparing Ku-band NSCAT scatterometer and ERS-2 altimeter winds", IEEE Trans. Geos. Rem. Sens., vol. 137, no 3, pp. 1662-1670.

[37] Quilfen, Y., Chapron, B., Vandemark, D., 2001, "The ERS Scatterometer Wind Measurement Accuracy: Evidence of Seasonal and Regional Biases", J. Atmospheric and Oceanic Technology, vol. 18, no. 10, pp. 1684-1697.

A validation of European Space Agency (ESA) remote sensing satellite (ERS) scatterometer ocean wind measurements is performed using a formalism recently proposed for and applied to NASA scatterometer (NSCAT) and Special Sensor Microwave Imager (SSM/I) measurements. This simple analytical model relates scatterometer measurements to true winds, taking into account errors in the satellite winds as well as errors in the data used for reference. In this study, National Data Buoy Center (NDBC) buoy winds are the chosen reference. In addition, ECMWF analysis winds are used as a third data source to completely determine the errors via a triple collocation analysis. According to this development, the resulting wind speed error analysis indicates that ERS scatterometer estimates are negatively biased at light winds. This result differs from recent results determined using standard regression analysis. It is also shown that ERS and NSCAT measurement accuracies are comparable in an overall sense.

This error model provides a more certain measure of both random and systematic terms and the authors use this tool to look at possible systematic scatterometer wind speed biases in two separate long-term (1992-98) ERS datasets. The chosen approach examines temporal and spatial variation between ocean buoy and ERS-derived winds to identify both seasonal and regional ERS wind error signatures. First, data indicate a time-dependent bias between NDBC and ERS winds that is strongly correlated with the seasonal cycle. Buoy-derived long-wave and atmospheric stability parameter averages exhibit similar cycles and are the likely geophysical links to this scatterometer error. An illustration of regional or spatially varying error sources is further provided using ERS data collocated with Tropical Atmosphere and Ocean (TAO) buoy array measurements. In this case the long-term average wind speed bias between TAO and ERS exhibits well-defined spatial structures within the equatorial belt (10°N, 10°S). Bias variations show qualitative agreement with a near-surface current climatology map for this Pacific region and also with the limited available buoy current measurements. Overall results indicate small but systematic nonwind sea surface effects on scatterometer products. It is concluded that there cannot be one set of values for ERS scatterometer wind validation parameters. Accounting for surface effects on scatterometer measurements may need consideration to ensure proper assimilation of scatterometer data into weather forecasting and climate prediction models.

[39] Wave age

[40] Fontaine, E., 2001, "On the Evolution of High Energy Wind Induced Ocean Waves", OMAE Conf., no. OMAE2001/S&R-2172.

The evolution of high energy wind induced ocean waves is studied theoretically. The modeling assumes that wave groups evolve independently of each others in a state of local equilibrium between wind pumping and breaking induced dissipation. The well-known fetch and duration laws appear as natural solutions of the model in the case of a constant wind speed. The link between these two experimental laws is emphasized and the results are successfully validated against various experimental data.

The paper considers the effects of sea roughness and atmospheric stability on the wind wave growth by using the logarithmic boundary layer profile including a stability function, as well as adopting Toba et al.'s [J. Phys. Ocean. 34 (1990) 705] significant wave height formula combined with some commonly used sea surface roughness formulations. The wind wave growth is represented by the non-dimensional total wave energy relative to that for neutral stability used by Young [Coast. Engng 34 (1998) 23]. For a given velocity at the 10 m elevation, spectral peak period and stability parameter, the wind wave growth is determined.

[42] Hughes, S.A., 1995, "Nearshore waveheight during storms, by M.J. Tucker: comments", Coastal Eng., vol. 26, pp. 105-107.

[43] Thornton, E.B., Guza, R.T., 1983, "Transformation of wave height distribution", J. Geophys. Res., vol. 88, no. C10, pp. 5925-5938.

The transformation of random wave heights during shoaling, including waves breaking in the surf zone, was measured with an extensive array of instruments in the field. The initially Rayleigh height distributions in 10-m depth were observed to be modified by shoaling and breaking into new distributions which are again nearly Rayleight but with some energy loss. Using locally measured H sub(rms), the Rayleigh distribution describes the measured central moments of H1/3 and H1/10 with average errors of -0.2% and -1.8%, respectively. The Rayleigh distribution is used to describe the random nature of wave heights in a single-parameter

transformation model based on energy flux balance.

[45] Athanassoulis, G.A., Soukissian, T.H., Stefanakos, C.N., 1995, "Long-term variability and its impact to the extreme-value prediction from the series of significant wave height", 4th International Workshop on Wave Hindcasting and Forecasting, Banff, Alberta, pp. 343-358.

[46] Athanassoulis, G.A., Stefanakos, C.N., Soukissian, T.H., 1993, "Non-stationary modeling of long-term time series of significant wave height with application to extreme-value prediction", WAVEMOD Project Report, no. WOR-3.3-02, p. 13. Abstract ->

[47] Athanassoulis, G.A., Soukissian, T.H., 1995, "Return periods of extreme sea states from long-term time series of Hs", WAVEMOD Project Report, no. TEC-3.3-01, p. 26.

[48] Goda, Y., Hawkes, P., Mansard, E., Martin, M.J., et al., 1993, "Intercomparison of extremal wave analysis methods using numerically simulated data", Proc. 2nd Int. Symposium Ocean Wave Measurement and Analysis, pp. 963-977.

[49] Labeyrie, J., 1991, "Time scales and statistical uncertainties in the prediction of extreme environmental conditions", Reliability Eng. and System Safety, vol. 32, pp. 243-266.

[50] Labeyrie, J., 1987, "Prévision des valeurs extrêmes des conditions d'environnement", LEMPAB Rep., no. 87/LEMPAB/RCH/03, p. 18.

[51] Naess, A., Clausen, P.H., 2001, "The Impact of Data Accuracy on the POT Estimates of Long Return Period Design Values", OMAE Conf., no. OMAE2001/S&R-2128.

The main focus of this paper is an investigation of the ef- fect which the accuracy of data representation may have on the results obtained by using standard peaks-over-threshold (POT) methods. It is shown that this effect may lead to a substantial shift in the resulting predictions of long return period extreme values for speci_c types of estimators combined with the POT method. A brief discussion of the implication for the estimation of confidence intervals on the various estimates will also be given.

[53] Vledder, G. van, Goda, Y., Hawkes, P., Mansard, E., et al., 1993, "Case studies of extreme wave analysis: A comparative analysis", Proc. 2nd Int. Symposium Ocean Wave Measurement and Analysis, pp. 978-992.

[54] Zachary, S., Feld., G., Ward, G., Wolfram, J., 1998, "Multivariate extrapolation in the offshore environment", Applied Ocean Research, Vol. 20, no. 5, pp. 273-295.

We consider the estimation of the extremes of the metocean climate, in particular those of the univariate and joint distributions of wave height, wave period and wind speed. This is of importance in the design of oil rigs and other marine structures which must be able to withstand extreme environmental loadings. Such loadings are often functions of two or more metocean variables and the problem is to estimate the extremes of their joint distribution, typically beyond the range of the observed data. The statistical methodology involves both univariate and multivariate extreme value theory. Multivariate theory which avoids (often very inappropriate) prior assumptions about the nature of the statistical association between the variables is a fairly recent development. We review and adapt this theory, presenting simpler descriptions and proofs of the key results. We study in detail an application to data collected over a nine-year period at the Alwyn North platform in the northern North Sea. We consider the many problems arising in the analysis of such data, including those of seasonality and short-term dependence, and we show that multivariate extreme value theory may indeed be used to estimate probabilities and return periods associated with extreme events. We consider also the confidence intervals associated with such estimates and the implications for future data collection and analysis. Finally we review further both the statistical and engineering issues raised by our analysis.

[55] Olagnon, M., Nerzic, R., Prevosto, M.,1999, "Extreme Water Level from Joint Distributions of Tide, Surge and Crests: a case study", Proc. 9th ISOPE Conf., vol. III, pp. 95-100.

When designing the air-gap of an offshore platform, current draft ISO standards recommend that the extreme "green" water level be set by either reliability considerations or experience and judgement. Account may be taken of the joint probability of tide, surge height and crest heights if the metocean database allows and a reliable model for crest statistics exists. This paper reports an investigation of green water highest levels, using the joint probability of tidal elevation and crest height, based on a case study where observed joint distribution of storm and significant wave height was provided. In that case, use of the actual joint distributions reduces the 100-year level by 0.6m, and the 10000 year level is similar to combination of 10000-year crest, MHWS and 100-year storm surge. These differences are much smaller however than those related to the choice of the crest model, when crest measurements are not available for the location of interest. Caution and judgement should thus indeed be exercised when combining the components of the extreme "green" water level for an offshore structure, but further research is needed for the modelling and description of extreme sea states before one can reliably set water levels corresponding to the now commonly requested safety probabilities.

[57] Defu, L., Shuqin, W., Liping, W., 2002, "Compound Bivariate Extreme Distribution of Typhoon Induced Sea Environments and Its Application", Proc. ISOPE Conf., vol. III, pp. 130-134.

This paper proposes a new bivariate probabilistic model - Compound Bivariate Extreme Distribution (CBED) for the estimation of extreme sea environments. The new model is obtained by compounding a discrete distribution of Typhoon frequency with a continuous bivariate joint distribution of the maximum wave height and wind velocity in each Typhoon process. Poisson- Gumbel Mixed Distribution (PGMD), one particular form of CBED, is used to estimate the joint distribution of significant wave height and wind velocity in the East China Sea, and shows its primary advantage in greater accuracy and simple use.

[59] Panchang, V., Zhao, L., Demirbilek, Z., 1999, "Estimation of extreme wave heights using GEOSAT measurements", Ocean Engineering, vol. 26, no. 3, pp. 205-225.

Satellite technology has yielded a large database of global ocean wave heights which may be used for engineering applications. However, the sampling protocol used by the satellite leads to some difficulties in making use of these data for practical applications. These difficulties and techniques to estimate extreme wave heights using satellite measurements are discussed. Significant wave heights for a 50-year return period are estimated using GEOSAT measurements for several regions around North America. Techniques described here may be used for estimation of wave heights associated with any specified return interval in regions where buoy data are not readily available.

[61] Donelan, M., Skafel, M., Graber, H., Liu, P., Schwab, D., Venkatesh, S., 1992, "On the growth rate of wind-generated waves", Atmosphere-Ocean, vol. 30, no. 3, pp. 457-478.

A new approach to fetch-limited wave studies is taken in this paper. Using data from five towers arranged along a line from the eastern shore of Lake St. Clair, the differential growth between towers is explored as a function of local wave age. It is argued that this method avoids the usual fetch-limited pitfall of inhomogeneity over long fetches and, in particular, the changes in wind speed downfetch of an abrupt roughness change. It is found that the growth rate decreases uniformly downfetch as the waves approach full development. This differential method leads to a smooth transition from rapidly growing short fetch waves to the asymptotic invariant state of full development. When the variation in wind speed after an abrupt (land to water) roughness change is taken into account, the idea of a universal fetch-limited growth curve is called into question.

[63] Carniel, S., Umgiesser, G., Sclavo, M., Kantha, L.H., Monti, S., 2002, "Tracking the drift of a human body in the coastal ocean using numerical prediction models of the oceanic, atmospheric and wave conditions", Science and Justice, vol. 42, no. 3, pp. 143-151.

This paper describes the use of numerical models to infer the path of a floating human body in the Ligurian Sea (north-west Mediterranean) during the month of January 2001. The prevailing oceanic currents were obtained from a state-of-the-art real-time nowcast/forecast ocean circulation model, while the sea state was inferred from a numerical model of the surface gravity waves, both driven by regional atmospheric models. The surface currents (from the ocean model) and the drift ones at the ocean surface, as inferred from the wave model, were used to drive a Lagrangian model of the drifting body to deduce its plausible trajectory along the Ligurian coast. The inferred path is reasonably consistent with location and time of the discovery on the French coast. This note illustrates the utility of numerical prediction models at the disposal of modern forensic science in the fields of ocean sciences.

[64] Mansour, A.E., Preston, D.B., 1995, "Return periods and encounter probabilities", Applied Ocean Res., vol. 17, no. 2, pp. 127-136.

[65] Athanassoulis, G.A., Stefanakos, C.N., "A non-stationary stochastic model for long-term time series of significant wave height"

[66] Athanassoulis, G.A., Stefanakos, C.N., Soukissian, T.H., 1993, "Non-stationary modeling of long-term time series of significant wave height with application to extreme-value prediction", WAVEMOD Project Report, no. WOR-3.3-02, p. 13.

[67] Athanassoulis, G.A., Soukissian, T.H., 1993, "A new model for long-term stochastic prediction of cumulative quantities", Proc. 12th OMAE Conf., vol II, pp. 417-424.

A new long-term stochastic model is applied to the calculation of the probability structure of random quantities accumulated over long-term periods (long-term cumulative quantities). The underlying process (e.g., the sea-surface elevation, or a structural response of a ship) is modelled, in the long time, as a two-level (doubly) stochastic process, by distinguishing between the fast-time scale, in which the corresponding spectral characteristics are slowly evolving. As a first approximation, the time series of spectral parameters is given the structure of a renewal process, whose inter arrival times are the durations of successive sea states. Then, long-term cumulative quantities can be considered as the "accumulated cost" of a renewal-reward process, and their probability distribution is obtained in terms of the joint statistics of sea-state duration and spectral parameters, using a central limit theorem for renewal-reward processes. The limiting distribution is Gaussian with a mean value equal to that predicted by Battjes (1970). A new formula for predicting the variance is given in this paper. The long-term number of waves (cycles) having amplitude greater than a threshold value U, denoted by Mu, is treated as an example. Numerical results are presented for a site in the central North Atlantic, based on 20-year hindcast data. The statics of sea-state duration and spectral parameters is first obtained and discussed. Then, using these results, the probability distribution of Mu is estimated. It is found that the variation coefficient of Mu may be as great as 0.60, a fact indicating that Mu might not be adequately represented by its mean value. More elaborated models incorporating the dependence between successive sea states are expected to improve the prediction of the variance, without affecting the asymptotic normality of Mu.

A new approach for calculating the long-term statistics of sea waves is proposed. A rational long-term stochastic model is introduced which recognizes that the wave climate at a given site in the ocean consists of a random succession of individual sea states, each sea state possessing its own duration and intensity. This model treats the sea-surface elevation as a random function of a "fast" time variable, and the time history of the spectral characteristics of the successive sea states as a random function of a "slow" time variable. By developing an appropriate conceptual framework, it becomes possible to express various probabilistic characteristics of the sea-surface elevation, which are sensible only in the fast-time scale, in terms of the statistics of sea-states duration and intensity, which is meaningful only in the slow-time scale. The proposed model can be extended twofold: either by replacing some of the simplifying assumptions by more realistic ones, or by extending the model for treating the corresponding problems for ship and structures responses.

Forecasting of ocean wave heights, with warning time of a few hours or days, is necessary in planning many operation-related activities in the ocean. Such information is currently derived by numerically solving the differential equation representing wave energy balance. The solution procedure involved is extremely complex and calls for very large amounts of meteorological and oceanographic data. This paper presents a complementary and simple method to make a point forecast of waves in real time sense based on the current observation of waves at a site. It incorporates the technique of neural networks. The network involved is first trained by different algorithms and then used to forecast waves with lead times varying from 3 to 24 h. The results of different training algorithms are compared with each other. The neural output is further compared with the statistical AR models.

Linear models of time series are used in this work to describe the sequence of the significant wave heights in two locations of the Portuguese coast. The time series of the monthly averages and standard deviations are studied and the seasonal component of those series are identified. The models are then applied to the data of Sines and Faro and predictions based on these models are presented. The models of the means and standard deviations are used to deseasonalise the series of three hourly observations which is then modelled by autoregressive models. It was found that models with orders up to 20, but without all terms, described well the data. Simulated data was shown to describe well the autocorrelation function of the observed data up to lags of 20.

[72] Labeyrie, J., 1990, "Stationary and transiant states of random seas", Marine Structures, vol. 3, pp. 43-58.

[73] see [80]

[74] Stefanakos, C.N., Athanassoulis, G.A., Barstow, S.F., 2002, "Multiscale Time Series Modelling of Significant Wave Height", Proc. ISOPE Conf., vol. III, pp. 66-73.

In the present paper, a compound stochastic model is formulated and validated, resolving the state-by-state, seasonal and interannual variabilities of Hs. The model is a combination of two cyclostationary random processes modelling the variability of mean monthly values and mean monthly standard deviations, respectively, and of a stationary random process modelling the residual, state-by-state, variability. In this way, the time series of significant wave height is given the structure of a multiple-scale compound stochastic process. The present model is a generalization of the nonstationary stochastic model introduced by Athanassoulis and Stefanakos (1995).

A new methodology for the analysis, missing-value completion and simulation of an incomplete nonstationary time series of wave data is presented and validated. The method is based on the nonstationary modelling of long-term time series developed by the authors [J. Geophys. Res. 100 (1995) 16,149]. The missing-value completion is performed at the level of the series of the uncorrelated residuals, obtained after having removed both any systematic trend (e.g. monotone and periodic components) and the correlation structure of the remaining stationary part. The incomplete time series of uncorrelated residuals is then completed by means of simulated data from a population with the same probability law. Combining then all estimated components, a new nonstationary time series without missing values is constructed. Any number of time series with the same statistical structure can be obtained by using different populations of uncorrelated residuals. The missing-value completion procedure is applied to an incomplete time series of significant wave height, and validated using two synthetic time series having the typical structure of many-year long time series of significant wave height. The missing-value patterns used for validation have been obtained from existing (measured) wave datasets with 16.5 and 33% missing values, respectively.

[76] Athanassoulis, G.A., Stefanakos, C.N., 1998, "Missing-value completion of nonstationary time series of wave data", Proc. OMAE Conf.

A new methodology for the missing-value completion of an incomplete nonstationary time series of a certain structure is presented and applied to measured data. The method is based on the modelling of long-term time series of wave data as a nonstationary stochastic process with yearly-long periodic mean value and standard deviation (periodically correlated stochastic process), introduced by the authors (Athanassoulis and Stefanakos, 1995). After a detrending and seasonal standardization, a low-order ARMA model is fitted to the (incomplete) residual stationary series using appropriate estimation techniques. The raw spectrum, calculated as the Fourier transform of a consistent estimate of the corresponding autocovariance function, is used for the estimation of the ARMA coefficients and the variance of the residuals. The incomplete time series of uncorrelated residuals is then completed by means of simulated data with the same first-order probability structure, and used, along with the ARMA model and the estimated deterministic components, to construct a new time series of the same structure without missing values. The above procedure is applied to two measured time series with different percentage of missing values. Comparisons of various statistical characteristics of the initial (incomplete) and reconstructed (completed) time series are satisfactory.

[78] Athanassoulis, G.A., Stefanakos, C.N., 1995, "A nonstationary stochastic model for long-term time series of significant wave height", Journal of Geophysical Research, vol. 100, no. C8, pp. 16,149-16,162.
copy not available at Ifremer "Metocean team"

First, an analysis of annual mean values is performed in order to identify overyear trends. It is suggested that an increasing trend is present in the examined hindcast data. The detrended time series Y(t) is then decomposed, to a periodic mean value m(t) and a residual time series W(t) multiplied by a periodic standard deviation s(t) of Y(t). It is shown that W(t) can be considered stationary, and thus Y(t) can be considered periodically correlated. This analysis has been applied to hindcast wave data from five locations in the North Atlantic Ocean. It turns out that the spectrum of W(t) is very weakly dependent on the site.

Missing values are frequently encountered in time series of wave measurements and this presents difficulties in the continuous application of autoregressive models to the series. In this paper, the use of autoregressive models to describe the time series of significant wave height is described and different methods are presented to fill missing values in the series. For small gaps it is possible to forecast from models adjusted to partial series, but for large gaps a method is proposed to account for the spatial correlation of the data. Application examples are provided for the time series of Figueira da Foz in Portugal and its correlation with data from Sines in Portugal and La Coruna in Spain.

A new methodology for the analysis, missing-value completion and simulation of an incomplete nonstationary time series of wave data is presented and validated. The method is based on the nonstationary modelling of long-term time series developed by the authors [J. Geophys. Res. 100 (1995) 16,149]. The missing-value completion is performed at the level of the series of the uncorrelated residuals, obtained after having removed both any systematic trend (e.g. monotone and periodic components) and the correlation structure of the remaining stationary part. The incomplete time series of uncorrelated residuals is then completed by means of simulated data from a population with the same probability law. Combining then all estimated components, a new nonstationary time series without missing values is constructed. Any number of time series with the same statistical structure can be obtained by using different populations of uncorrelated residuals. The missing-value completion procedure is applied to an incomplete time series of significant wave height, and validated using two synthetic time series having the typical structure of many-year long time series of significant wave height. The missing-value patterns used for validation have been obtained from existing (measured) wave datasets with 16.5 and 33% missing values, respectively.

[81] DelBalzo, D.R., Schultz, J.R., Earle, M.D., 2003, "Stochastic time-series simulation of wave parameters using ship observations", Ocean Engineering, vol. 30, no. 11, pp.1417-1432.

A stochastic simulation technique was used with ship wave observations, which form the largest world-wide data base of wave information. Twenty years of wave parameter (height, period, and direction) observations from the Comprehensive Ocean-Atmosphere Data Set (COADS) were used as the input data. Simulations were compared to four years of wave parameters from a National Data Buoy Center (NDBC) data buoy near Monterey Bay, CA. The comparisons are satisfactory with differences mainly caused by biases between ship observations and buoy data. The stochastic simulation technique is attractive because it is computationally efficient and few decisions are required for its application. The applied techniques can be employed with global COADS data to simulate wave conditions at many world-wide locations where measurements and hindcasts by computer models do not exist.

[83] Monbet, V., Prevosto, M., 2000, "Bivariate simulation of non-stationary and non-Gaussian observed processes: Application to sea state parameters", Proc. 10th ISOPE Conf., vol. III, pp. 91-94.

[84] Scheffner, N.W., Borgman, L.E., Mark, D.J., 1996, "Empirical simulation technique based storm surge frequency analyses", J. Waterway Port Coast. Ocean Eng, vol. 122, no. 2, pp. 93-102.

This paper describes theory and application of the empirical simulation technique (EST), a statistical procedure for simulating time series and frequency-of-occurrence relationships for nondeterministic multiparameter systems. Procedures described are applied to a storm surge analysis for tropical events along the coast of Delaware. The approach involves the numerical simulation of historical events, the analysis of the parameters associated with each event, the application of the EST to that data, and the generation of frequency-of-occurrence relationships for 40 locations within the study area. The development of the input data set of descriptive storm parameters and their respective responses require the use of planetary boundary layer and very large domain hydrodynamic modeling techniques. Development of this input database is described herein. Implementation of the EST is based on N repetitions of T-year simulations; therefore, mean-value frequency relationships are computed and assigned confidence limits such that probability of occurrence is defined with error band estimates. Accuracy of the approach is presented and advantages of the technique over the traditional joint probability method are demonstrated.

This paper describes a procedure for generating simulated time sequences of wave height, period, and direction data at specific locations. The technique uses a finite length wave record to compute a matrix of coefficient multipliers, which are used to generate arbitrarily long time sequences of simulated wave data, preserving the primary statistical properties of the finite data set. The procedure was developed for simulating time series from the Wave Information Study (WIS) data base, a 20-year hindcast of wave height, period, and direction provided at three-hour intervals for locations along United States coasts and the Great Lakes. Application of the methodology is demonstrated in this paper through comparisons of simulated data with hindcast data corresponding to a Gulf of Mexico WIS station near the entrance to Mobile Bay, Alabama. Analysis of the results indicate that the simulated time series does exhibit the primary statistical properties of the WIS data, including winter and summer seasonal patterns and wave sequencing.

A probabilistic description of the ocean environment is developed for applications in areas such as the assessment of long term degradation in soil and structural properties of offshore structure-foundation systems. The sea is characterised by a series of storms considered to occur when a characteristic wave height (e.g. significant wave height), which is a function of time, exceeds a predefined threshold. The storms themselves are described by a duration, an average intensity, and a non-stationary random process tracing the evolution of the characteristic wave heightwith time. This last feature allows modelling of the initial build up and the terminal decay of storm intensity. The distribution of inter-arrival times between storms and that of the occurrence of storms are derived in addition to the distribution of duration of storms, calm periods, and the joint probability distribution of storm duration and average intensity. The non-stationary storm trace is represented by a Fourier transformation, the parameters of which are probabilistically defined for storms of a given duration and average intensity. Fourteen and a half years of wave records from the North Atlantic, recorded at hourly intervals, were used to estimate the parameters of the various distributions. Storms were then simulated on the computer by a random generation scheme. The 'synthetic' storms when compared with observed data appear to be realistic and in reasonable agreement with the data.

This paper presents a new long-term random sea-state model that traces the changes in energy and frequency content of the sea surface elevation in time. The model assumes that is suffices one of the short-term sea-state descriptors defining the wave spectrum. The parametric form of the wave spectrum is considered to be time-invariant. The modelling strategy involves transforming the Weibull or lognormally distributed characteristic wave height trajectory into a zero-variance, Gaussian random process. A spectral density function is then estimated for the derived random process using a classical Fourier transform based spectral estimator.

[89] Anastasiou, K., Tsekos, C., 1996, "Persistence statistics of marine environmental parameters from Markov theory, part 1: analysis in discrete time", Appl. Ocean Res., vol. 18, pp. 187-199.

[90] Athanassoulis, G.A., Stefanakos, C.N., Soukissian, T.H., 1995, "Missing value completion of nonstationary time series of wave data with application to sea-state duration calculations", WAVEMOD Project Report, no. TEC-3.1-03, p. 31. Abstract ->

[91] Graham, C., 1982, "The parameterisation and prediction of wave height and wind speed persistence statistics for oil industry operational planning purposes", Coastal Eng., vol. 6, pp. 303-329.

[92] Hogben, N., Standing, R.G., 1987, "A method for synthesising time history data from persistance statistics and its use in operational modelling", Journal of SUT, vol. 13, no. 4, pp 11-18.

A method for converting weather severity persistence data, derived from conventional probability distributions, into time history format is presented and illustrated with a simple numerical example and with wave height simulations prepared with data from the Sevenstones Light Vessel and the NMIMET climate prediction program.

The statistics of sea state duration (persistence) are dependent upon the recording interval Dt, since the graph of a time series of an environmental parameter such as the significant wave height has an irregular, "fractal" geometry. The mean duration can have a powerlaw dependence on Dt as Dt -> 0, with an exponent equal to the fractal dimension of the level sets of the time series graph. A more practical quantity for operational purposes is the "useful mean duration", where each interval satisfying the appropriate criterion is weighted by its duration ti. The above results are illustrated using wave data from the Frigg gas field in the North Sea.

[95] Mathiesen, M., 1994, "Estimation of wave height duration statistics", Coast. Eng., vol. 23, no. 1-2, pp. 167-181.

A theoretically founded parametric model for the estimation of duration statistics for significant wave height is established. Use of the model requires information both on the distribution and the average absolute rate of change of significant wave height. Results from analyses of data indicate that the model provides fairly accurate estimates of mean duration of exceedance and non-exceedance of a threshold level. The durations of exceedance and non-exceedance are found to vary considerably about the mean.

[97] Sanchez-Arcilla, A., 1984, "Long-term analysis of wave climate using short-term techniques", J. Energy Resour. Technol., vol. 106, no. 2, pp. 228-233.

This paper proposes a number of theoretical probability distributions for variables of interest when planning operations at sea. Examples of such variables are; duration of storms, duration of calms, intensity of storm peaks, etc. The theoretical distributions are obtained by means of random process theory applied to the curve of evolution of H sub(s) (significant wave height). The model is checked with six years of wave data at the north coast of Spain.

[99] Tesson, C., 1988, "Analyse de la houle naturelle - Rapport no.3: Etude des durées de calme à l'aide d'un modèle markovien", rapport EDF/LNH no. HE-42/88.07, p. 27.

[100] Tesson, C., 1986, "Analyse de la houle naturelle - Rapport no.2: Développements du modèle statistique d'étude du couple hauteur-durée de tempêtes", rapport EDF/LNH no. HE/42/86.14, p. 117.

[101] Tesson, C., 1984, "Etude statistique du couple hauteur-durée de tempêtes à partir de l'analyse de la houle naturelle", rapport EDF/LNH no. HE/42/84.41, p. 81.

[102] Tsekos, C., Anastasiou, K., 1996, "Persistence statistics of marine environmental parameters from Markov theory, part 2: analysis in continuous time", Appl. Ocean Res., vol. 18, pp. 243-255.

[103] Cieslikiewicz, W., Graff, J., 1996, "Sea state parameterisation using empirical orthogonal functions", Proc. 25th Coastal Engng Conf., vol. 1, pp. 703-716.

[104] Clément, C., McCullen, P., Falcão, A., Fiorentino, A., Gardner, F., Hammarlund, K., Lemonis, G., Lewis, T., Nielsen, K., Petroncini, S. et al., 2002, "Wave energy in Europe: current status and perspectives", Renewable and Sustainable Energy Reviews, vol. 6, no. 5, pp. 405-431.

The progress in wave energy conversion in Europe during the past ten years is reviewed and current activities and initiatives in the wave energy sector at National and Union level are described. Other important activities worldwide are summarized. The technical and economical status in wave energy conversion is outlined and important wave energy developments are presented.

A stochastic method has been developed to evaluate the average performance of an oscillating water column wave energy device equipped with an (assumedly linear) Wells turbine. The wave climate is represented by a set of sea states, characterized by their power spectra, the free-surface elevation being a Gaussian random variable in each sea state. The variance and the probability distribution of the air pressure inside the chamber are computed for each sea state, it being assumed that the chamber hydrodynamic coefficients and the turbine curves are known. This allows the average performance of the turbine and of the plant to be obtained for each sea state and for the annual wave climate. Numerical examples are worked out for given chamber geometry and turbine shape, showing how the turbine size and rotational speed may be optimized for maximum energy production. Controllable rotational speed and the use of a valve system for turbine flow control are considered.

The distribution of wave energy and power as functions of longitude and latitude are presented for the Northern Hemisphere at 12 noon GMT, October 2, 1975. Both the large peak of the distribution in the Atlantic Ocean and the smaller peak in the Pacific Ocean are found to be at longitudes towards the eastern end of the ocean basins. This "eastern accumulation" of wave energy and power offers interesting contrast to the western intensification of currents. Distribution of wave power with latitude shows peaks of wave power in the mid-latitudes. The total surface wave energy in the seas of the world for the same time is estimated to be 1600x1015 J. The corresponding total wave power estimate is 90x1015 W. The rate of renewal of wave power is estimated to bs 1012-1013 W, about the present level of world power consumption.

The aim of this paper is to provide a general view of wave energy resource assessment. First, a review of the origin of waves and the transformation they undergo as they propagate towards the coast through waters of decreasing depth is presented. Following this, the wave and wave-energy parameters and the statistics required for resource characterization are described. The various types of wave data and their usefulness for the present purposes are summarised. A common methodology for assessment of the wave energy resource is developed. Finally, a general description of the global open ocean resource is presented.

[108] Kestens, E., Teugels, J.L., 2002, "Challenges in modelling stochasticity in wind", Environmetrics, vol. 13, no. 8, pp. 821-830.

This article is an attempt to summarize some of the existing problems in stochastic aspects of wind. Different types of wind are listed with their specific properties. For most of them no statistical model or stochastic process has been constructed as yet. At the same time, existing problems with data are very diverse and possible improvements are proposed. For example, the quality of wind speed and wind direction data might be upgraded by a careful inclusion of measurable covariates while developing models. Other problems are dealing with extreme winds that can hardly be measured accurately. In this connection, interesting and important questions for insurance companies and for construction engineers can be tackled by applying extreme value theory. This incomplete overview hopes to encourage stochasticians to put more interest in wind problems. At the same time it is hoped that meteorologists will help them with high quality data, needed in the verification of the offered models.

In the peaks-over-threshold (POT) method of extreme quantile estimation, the selection of a suitable threshold is critical to estimation accuracy. In practical applications, however, the threshold selection is not so obvious due to erratic variation of quantile estimates with minor changes in threshold. To address this issue, the article investigates the variation of quantile uncertainty (bias and variance) as a function of threshold using a semi-parametric bootstrap algorithm. Furthermore, the article compares the performance of L-moment and de Haan methods that are used for fitting the Pareto distribution to peak data.The analysis of simulated and actual U.S. wind speed data illustrates that the L-moment method can lead to almost unbiased quantile estimates for certain thresholds. A threshold corresponding to minimum standard error appears to provide reasonable estimates of wind speed extremes. It is concluded that the quantification of uncertainty associated with a quantile estimate is necessary for selecting a suitable threshold and estimating the design wind speed. For this purpose, semi-parametric bootstrap method has proved to be a simple, practical and effective tool.

Copyright © 2004, IFREMER. All rights reserved.