The storm climate over northwestern Europe has been studied using geostrophic winds derived from triangles of stations with three air pressure observations per day 1881-1995. We have used 95- and 99-percentiles of the distributions of geostrophic winds as main storm indices. Generally, there is a weakening of the strong gradients and strong winds from the late 19th and the beginning of the 20th century up to around 1950-1970, but then the frequency of storms has increased and is now again on levels or almost on levels typical of the first decades. Especially the more maritime areas, including the British Isles, the North Sea and the Norwegian Sea have experienced a strong increase from a minimum around 1965, while in large parts of Scandinavia and Finland this recovery is much less pronounced. The main results are summarized as averages for the western (British Isles, North Sea, Norwegian Sea) and eastern (Scandinavia, Finland, Baltic Sea) parts of the area in this investigation. As a rule of thumb we find that extreme (stormy) years in the western part are very likely to be above average in the eastern part (but not very often extreme years). And the same going from east to west. The correlation coefficient between the averaged 95-percentiles in these two regions is 0.74. We have also investigated the relation between our storm indices and the NAO index showing that there is a positive coupling, although it is not so strong. Also the generally high storm frequency during the first decades does not coincide with a maximum in the NAO-index, in contrast to the recent increase, where the storm curves and the NAO-index show a more similar behaviour.
The use of the wavenumber-direction spectrum in wind wave models results in an effective loss of model resolution for waves traveling from deep to shallow water and in additional numerical disadvantages, when a conventional invariant spectral grid is used In this paper we present a theoretical studs of how the effects of variable depths and currents may be incorporated in a spatially varying wavenumber grid. It is shown that effects of currents cannot be efficiently incorporated in the grid. Effects of variable depths are incorporated in a wavenumber grid which is equivalent to a spatially invariant frequency grid. The resulting equations are nearly identical to the conventional equations for the frequency-direction spectrum, but include a more elegant way to address effects of temporal variations of the water depth. Furthermore, the technique employed to derive the equations for the variable grid approach closely resembles that used in the conversion between different spectral descriptions. The formulations presented in this paper may therefore serve as a basis for discussion of the selection of spectral descriptions.
Three sets of hydrographic data from October 1957, August 1981, and July-August 1992 are used to examine the variations in meridional circulation and ocean heat transport at 24.5°N in the subtropical North Atlantic. From these sections general warming has been found at intermediate levels between 800 and 2700 dbar in the middle of the North Atlantic subtropical gyre, over the entire 35-year period in a zonally uniform manner. On isopycnal surfaces, the waters have become warmer and saltier nearly everywhere above 2000 db. On level surfaces, maximum warming of 0.1°C per decade occurs at 1100 dbar ; on isopycnal surfaces the salinity in the main thermocline is increasing at a rate of 0.018 per decade. By requiring that the ocean interior transport balances the Gulf Stream transport through the Straits of Florida and the Ekman transport over the Atlantic (Hall and Bryden, 1982), we have performed a heat flux calculation using similar methods on each dataset. Even though the temperature and salinity have changed, the three cruises exhibit similar features in the large-scale velocity fields and similar zonally averaged meridional transport. Ocean heat transport performed by similar calculations on the three cruises are indistinguishable, which is the same conclusion reached by Roemmich and Wunsch (1985) in their comparison between the 1957 and 1981 sections based on a number of inverse models.
A 40-year hindcast simulation (1955-1994) forced by meteorological data provided by the Norwegian Meteorological Institute (DNMI) has been run with a two-dimensional storm surge model covering the north-west European continental shelf. A comprehensive data set of surge elevations derived from long tide gauge records was assembled to validate the model results. Extensive comparisons with observed surges show that the accuracy of the model results is good (of the order 10cm). Estimates of extreme storm surge elevations were then derived independently from the model results and from the observations using the Generalised Extreme Value method. This method was applied to the data set comprising the 7 largest surge maxima in each year. Estimates of extreme (50-year return period) surges from model and observations were found to agree in the North Sea i.e., the model and observations generally agreed within the statistical uncertainty given by the standard error. Some discrepancies, in estuaries and in the Irish Sea and Bristol Channel, are probably due to the coarse (35 km) model resolution. The study has demonstrated that realistic estimates of surge extremes can be derived directly from model simulations forced by good atmospheric data sets and hence such estimates can be made in areas (e.g., offshore) or situations (e.g., of changing climate) where no observational data exist.
A geostatistic method, known as the kriging method is employed to produce a gridded wind field using ERS-1 scatterometer-retrieved wind vectors. The method, which differs from the classical interpolation methods, makes use of spatial and temporal empirical structure functions for wind speed, eastward wind component u and northward wind component v, and the Gauss-Markov theorem to objectively analyze this vector field and yields an expression for the variance error expected in this estimate. The stationarity assumptions needed to apply the kriging method are tested statistically. Resultant three-, seven- and ten-day average wind fields computed over the tropical region indicate that the method reproduces large and small-scale features. The accuracy of the gridded wind fields is determined by comparisons with three-day averages calculated from European Centre for Medium-range Weather Forecasts (ECMWF) wind vector analysis, with ten-day averages derived from ship measurements over the tropical Atlantic, and with seven-day averages computed from moored-buoy wind measurements, which were recorded by Pacific Marine Environmental Laboratory (PMEL) in the Pacific ocean. In all cases, there is good agreement between the data sets. The rms values of the difference between satellite and numerical analysis, ship and buoy-derived wind fields are 1.67m/s, 1.17m/s and 1.05m/s, respectively.
Wind vector retrieval from synthetic aperture radar (SAR) images over the ocean is shown to be feasible. A set of 16 radiometrically calibrated ERS-1 SAR images, including analog to digital converter saturation power loss correction, are compared with two ERS-1 scatterometer wind retrieval models driven by accurate in situ wind vector mesurements. It is shown that, if the wind direction is known and by using the CMOD4 or CMOD5 wind retrieval models, the agreement is to within ± 1 dB, representing a wind speed extraction error of ± 15 m/s, for the 3 to 12 m/s wind speed conditions encountered. This result validates the feasibility of extracting wind speed from a suitably-calibrated ERS-1 SAR image using a scatterometer wind retrieval model We demonstrate that the wind direction may be deduced from the low-wavenumber portion of the SAR image spectrum for the conditions represented in our data set. It is shown that agreement in wind direction is to within ±24°, thus presenting the opportunity for extracting the full wind vector from SAR images on a routine basis.
The SOFIA experiment which took place in the Azores region during ASTEX, in June 1992 was devoted to surface flux and marine atmospheric boundary layer investigations at the mesoscale. The different instrumented platforms used are described: shipborne instruments, drifting buoys, airborne instruments, satellites. After a description of the main characteristics of the whole experiment which corresponded to perturbed cloudy meteorological situations with a latent heat flux close to 100 W/m 2 , a sensible heat flux smaller than 30 W/m 2 and moderate wind speeds <10m/s, an analysis of two observing periods of June 8 and 17 is undertaken. For these two days we illustrate the use of several kinds of equipment such as the airborne Lidar LEANDRE, the radiometer POLDER, the acoustic system OCARINA and the microwave radiometer DRAKKAR. In particular, the wake effect due to Santa-Maria island is documented with the constant level balloons and the instrumented aircraft ARAT.
The aim of this work was to test the accuracy of simulated sea surface temperature (SST) in the tropical Atlantic Ocean for the period 1982 to 1990 using a three-dimensional primitive-equation oceanic general circulation model. The SST reference was provided by monthly analysis of observed data obtained from selected ships. The annual mean difkrence between simulated and observed SST did not exceed 1.5 C, but monthly mean differences of several degrees were found for some arcas. There were two main problem areas, one along the western equatorial zone and the other within the northem pan of the Gulf of Guinea. The first, characterized by excessively cold SST, corresponded to a westward early shifting of seasonal equatorial surface cooling; and the second, characterized by excessively wann SST. concerned a regon of the Gulf of Guinea where winds were light and the sea surface warm. A series of test experiments was performed during one year (1989) using direct measurements of the meteorological variables which enter into the computation of turbulent heat flux forcing. with all other conditions remaining unchanged. The generally disappointing results may relate to the heat flux sensitivity induced by weak variations among basic meteorological vanables.
