Objective
A.BACKGROUND
Global scale measurements of sea state has long been pointed out as a pressing need in several domains. Firstly, systematic measurements of surface wave fields would be very useful to further improve the prediction of sea state and ocean storms from wave forecast models. Secondly, for navigation and maritime industry both the prediction of wave fields and the climatology of sea states have to be known. Thirdly, management of the coastal areas requires a better knowledge of the characteristics and climatology of water waves generated in the open ocean. Finally, in the perspective of our understanding of the coupled ocean-atmosphere system as a whole, and of the evolution of our climate, the influence of surface waves must be considered since they play a role in the energy and mass transfers at the interface between ocean and atmosphere.
It is recognized that improvements in all these domains will come from wind and wave measurements combined with numerical wave prediction models.
Modern remote-sensing techniques have already been developed with this aim. However they suffer from different drawbacks which are recalled hereafter.
Satellite-borne radar altimeters now provide information on a nearly operational basis on significant wave height (quantity related to the energy of the waves) and surface wind speed.
Recent work shows the usefulness of such measurements in assimilation schemes for numerical wave forecast models. However, as suggested by these studies, the main weakness of this approach lies in the lack of information concerning the wavelength and direction of the waves. Information solely about significant wave height limits the impact of the assimilation. These shortcomings clearly indicate that directional spectral information have to be taken into account in order to improve the quality of the analysed wave field and its forecast.
Today, Synthetic Aperture Radar (SAR) provides the only existing opportunity to measure the directional ocean wave properties from a space borne sensor. Unfortunately, it is well known that the wave-like patterns visible in an SAR image of the ocean surface may be considerably different from the actual ocean wave field: distortions frequently occur on the mapping due to the displacement of the surface scatter during the formation of the image. As a result, the extraction from a SAR scene of meaningful properties of the two-dimensional wave spectra is not straightforward and relies on assumptions, which have not all been verified. To overcome this problem, new methods have been proposed to combine SAR data and numerical wave model spectra to retrieve from the SAR images, directional wave spectra constrained by the model information. These methods have been validated, but several remaining problems must be solved or clarified to demonstrate their potentialities from an operational point of view.
Although SARs are not ideal to measure directional spectra, deployment in space of these radar systems will continue because they have other applications over the continents. Therefore, it is important to further progress in the inversion and use of SAR observations over the ocean.
In parallel, alternative systems have to be studied. Airborne radar-systems based on a technique which does not present the limitation of the SARs have demonstrated their capabilities to obtain the directional spectra of ocean waves. The extension of these types of radar to space borne systems has already been proposed. Preliminary studies have shown its feasibility. Further studies are necessary to simulate the expected performances of such systems, and to prepare the use of this type of observation into assimilation schemes of wave prediction models.
For all these topics there is a need for coordinated analysis between remote-sensing observations, sea-state modelling, and in situ measurements of wind and waves. Coherent data sets from a dedicated field campaign would bring valuable information to carry out these studies.
Considering that we will be able in the future to obtain directional wave spectra over the ocean basins on an operational basis, methods have to be developed to assimilate these data into numerical wave models. Assimilation of observations into wave models is much more recent than assimilation into Global Circulation Atmospheric Models (GCAM). For the assimilation of directional wave spectra, only limited studies have been performed until now. Therefore, there is an important need to develop assimilation techniques for the specific case of wave forecast.
Finally, in the domain of engineering applications (maritime industry and navigation), there is a need to use wave spectral information. However, the links between the engineering and scientific communities must be developed; the potential advantages of using new measurement techniques are not well known, by the first one, and the exact requirements for the observations are not well known by the last one.
B.OBJECTIVES AND BENEFITS
The main objective of the Action is to improve the measurement and use of the directional information of sea-states. To achieve this, six activities have to be carried out:
1)to further improve the retrieval of wind and wave information from SAR data alone
2)to improve the inversion schemes which provide directional wave height spectra from SAR image spectra using external data (numerical model, in situ observations)
3)to define and propose new missions as alternatives to SAR observations, for the measurement of the directional spectra of ocean waves at a global scale
4)to define new, or use existing, field experiments
5)to develop further strategies for the analysis, validation and assimilation of waves observations into numerical models for sea-state prediction
6)to develop the relations between the scientific and the engineering communities for application purposes.
The expected benefits of this Action are manifold. This Action will be a significant contribution to the improvement of weather and surface wave conditions prediction, based on the use of satellite data into numerical models. Benefits are also expected in a better understanding of the radar signals backscattered from the sea surface. Moreover, the Action should contribute to the definition of new instrument concepts for the measurement of the ocean waves spectral parameters. It should also serve as guidelines for future field experiments. Finally, it would contribute to tighten the links between the world of scientists and the world of engineers in marine technology, by proposing exchanges between the two communities.
C.SCIENTIFIC PROGRAMME
The Action is divided into six activities (six research tasks) detailed below.
1.Improvement of the retrieval of wind and wave information from SAR signals alone
Space borne SAR has the capability of high resolution measurements (~10 m to 1 km). This advantage is not yet widely exploited to obtain information on the scattering mechanisms, and to improve SAR image spectra inversion into wave spectra. Recent studies have shown that the statistics of the backscattered signal amplitude are sensitive to the physical conditions on the surface (wind and waves), as well as the imaging conditions (incidence angle, frequency, polarization). Moreover, these statistics provide one with information on the non-linear effects of wave imaging. It must also be noted that SAR instrument potentially captures sub-resolution scale information on the frequency of breaking waves during the along-track temporal integration of the backscattered signal. For this Action, cooperative studies will be undertaken to develop statistical analyses of the SAR images in order to gain knowledge in both scattering process and modulation mechanisms which both affect geophysical parameters derived from SAR image spectra, such as the significant wave height, and the peak wavelength and direction.
The account of speckle noise is also quite important for the retrieval of meaningful wave spectra from SAR images. Recent developments have shown that the use of cross-spectral analysis between successive looks can decrease the effect of speckle noise on the wave spectra retrieval. Moreover, this technique seems promising to resolve the 180? ambiguity in the wave propagation direction. These new methods will be more extensively tested in order to develop their use in the future.
The current theory to transform SAR image spectra into wave spectra is based upon a transfer function which takes into account three modulation effects: the velocity bunching (effect due to the motion of scatters), the hydrodynamic modulation and the tilt modulation. Theoretical expressions for these functions have been proposed in the literature. They are described in terms of the radar parameters and surface characteristics (distribution of motion scatters, coherence time, hydrodynamic interaction between long and short waves, wave spectral distribution for short waves). Direct experimental estimates of the modulation transfer function are few, but most of them show disagreements between theory and observations. From these experimental studies, new transfer functions have been proposed for various radar and surface conditions. They will be introduced in the SAR transformation and the sensitivity of the results to variation of the modulation transfer function will be extensively tested. Further experimental studies will also be proposed to assess the modulation transfer function (see task 3).
2.Improvement of the inversion methods, which make use of external data (numerical model, in situ observations)
Several aspects of the inversion algorithm for SAR images are still under discussion. The first one deals with the transfer function for the hydrodynamic modulation which is not quite well known (see point 1 here above). A second point deals with the estimation of the azimuth cut-off wavelength due to the motion of scatters. Third, questions on the dependence of the inverted wave spectrum upon the first guess remain. Improved inversion methods have already been proposed. Further studies will be developed to compare this algorithm to a complementary approach based on an adjoint modelling. Both suffer the same limitation due to the necessity of a first guess information. The combination of existing data from overlapping space and time regions (like the combination of ERS-1 SAR, SIRC SAR, aircraft measurements and buoy data) would be beneficial to settle the questions addressed above.
One of the main applications of directional wave spectra is in a combined wind and wave data assimilation for which an operational implementation of the inversion algorithm is presently under way. As one of the requirements is low computing time, any possible speed up of the inversion technique is desirable. An alternative approach might be the use of neural networks for operational purposes. However their potential use and advantage in terms of computing time has yet to be demonstrated. The data set for their training could again be provided by wave spectra inverted from SAR image spectra.
3.Definition and proposal of alternative missions for the measurement of the directional spectra of ocean waves on a global scale.
Because the SAR imagery for the measurement of directional ocean wave spectra has important limitations, there is a need to envisage alternative space-borne systems better adapted to this measurement. Following the call for ideas from the French Space Agency, a proposal called VAGSAT has been made to develop a radar based upon a real-aperture technique. The idea is to use a radar pointing at small incidence angles (~10? from nadir), using the real aperture of the antenna, and scanning in order to cover the horizontal plane over 360? in azimuth. Obtaining the directional spectra of the waves relies on an analysis of the modulation of the radar signal backscattered within the antenna footprint.
In order to prepare this possible mission, several studies will be undertaken. In particular, an analysis of the global performance of the measurements following the principle of VAGSAT as compared to SAR measurements will be carried out. Studies about the statistics of the ocean wave field are also necessary to be undertaken as a preliminary step to the development of assimilation techniques. Finally, constraints on the sampling strategy will be defined with the help of sea-state forecast model outputs.
4.Definition and use of field experiments
Field experiments are essential to progress in the study and use of ocean surface waves observations. For most of the topics described here, the combined use of in situ observations to measure surface and atmospheric parameters (e.g. wind, wind stress, waves, atmospheric stability) and remote sensing techniques from ground, platforms, aircraft and satellites would be of great interest.
In the context of this Action, we propose to make an inventory of the existing data bases, and to facilitate access to them. In parallel, we will work on the definition of new experiments to study aspects not covered by previous field experiments. We will take into account in the definition of these experiments, the existing instrumental possibilities of the different countries participating to the Action.
The main experimental objectives are listed hereafter.
Among the different unresolved problems, the behaviour of the hydrodynamic modulation is an important one, because the uncertainty in this term affects both the determination of wave peak directions and of significant wave height from SAR images. Moreover, it probably affects
other microwave measurements such the upwind/downwind difference in radar cross-section measured from wind scatterometers. Further experimental studies are necessary to better know this hydrodynamic modulation in various situations of wind, waves, current, atmospheric stability, and for various radar conditions (polarization, incidence angle, radar frequency). This will also be useful to prepare the use of future SARs which will enable observations over a large diversity of incidence angles, radar frequencies and polarization.
Other effects on the radar signal, like the breaking of large waves or the effect of rain are poorly documented from in situ measurements. Further studies are necessary for example to investigate the relationship between breaking wave events, Doppler radar signal, speckle noise level. Concerning the rain effects, most of the existing studies concern either laboratory measurements or statistical analyses of satellite data. In situ measurements are necessary to examine the effect of an actual rain on the surface and on the radar signal.
Field experiments are also essential to better investigate the spatial evolution of wave fields. First, the directional behaviour of the waves and of the associated radar signal is not well known in fetch-limited situations. The JONSWAP experiment performed in the 70s provided very little information on this aspect. It would be very useful to propose a new experiment aimed at this topic. Closed sea conditions with frequent fetch-limited situations, like in the Baltic Sea or in the Mediterranean Sea
would be the most adequate to carry out such an experiment. A second goal for the study of the spatial variability of wave fields is related to the development of the combined use of sea-state models and space-borne measurements. Indeed, SARs provide information on the wave spectra from a scale (~ 10 km) much smaller than the models (100 to 200 km), but with samplings depending on the satellite possibilities. With the recently proposed VAGSAT system, the obtained wave spectra would be representative of a scale similar to that of models. In order to use models and observations in a complementary way, it is necessary to study the variability of the wave field at scales ranging from about 10 km to about 200 km. This variability may be important in the vicinity of large storms and atmospheric fronts. Experimental studies devoted to this problem should help to define sampling strategies for future satellites, and to progress in the combined use of remotely sensed and modelled wave spectra.
5.Strategies for the analysis, validation and assimilation of satellite observations into numerical models for sea-state forecast
Numerical models of ocean surface gravity waves have been used for many years to produce analyses and forecasts of wave conditions. Until recently, because observations of waves were sparse, the wave field analyses were produced without reference to observations of the waves themselves, but were diagnosed from the wind field. Following the launch of ERS-1, many meteorological centres have introduced schemes to improve the starting states used by the wave models. These schemes combine the satellite observations of wind and wave height with the fields from the wave model,
allowing correction of errors caused by weaknesses in the physics represented by the wave models and the winds used to drive them.
Although the altimeter observations have been very useful for operational wave forecasting, their impact will remain limited because many assumptions have to be made to relate the observed wave height to the full wave spectrum that is needed by the wave model. Direct assimilation of wave spectra, from SAR or other sources, would not need these arbitrary assumptions, but the techniques to achieve this on an operational basis are not yet available. Therefore, there is a need for international cooperation to develop and assess methods for assimilating directional wave spectra. This development will benefit from the techniques already existing for GCAM, but specific studies have to be carried out to take into account the specificity of the wave prediction models. In particular one difficulty is that the assimilation process has to modify in a consistent way both the initial state and the forcing of the model (wind or friction velocity). Various configurations of assimilation can be envisaged (assimilation of wind and waves together, assimilation of waves alone, assimilation of observations provided by different kinds of sensors (altimeter, SAR, VAGSAT), and the impact of the assimilation must be evaluated in each case in order to propose an optimal strategy.
Although the variational technique seems presently too difficult to implement assimilation of observations into wave models on an operational basis, studies on the respective performance of the variational technique and of the more classical optimal interpolation method must be analysed.
One of the difficulties is to develop a method which will modify in a consistent way both the initial state of the model and the forcing parameters (wind or friction velocity). Methods to reduce the number of variables in the assimilation scheme have also to be further developed: the "spectral partitioning technique" must be adapted to several kinds of observations (buoys, SARs, VAGSAT) and its use in various assimilation schemes has to be validated. Besides, a method allowing a combined processing for the inversion of SAR spectra and for their assimilation has to be explored. Finally, an assimilation scheme taking into account several complementary remote-sensing systems (radar-altimeters, wind scatterometers, SARs or other wave spectrometers) has to be envisaged.
An intermediate step to these studies will be to agree on methods for comparing ocean wave spectra from different sources (buoy observations, radar observations, model outputs).
The outcome of this component of the Action is expected to be a definition of statistical methods that participants will use to compare wave spectra to validate the models and new observing methods, and collaboration between centres developing data assimilation techniques for wave spectra resulting in a consensus on the suitability for various applications of the methods used for data assimilation.
6.Development of the relations between the scientific and the engineering communities
An important task of this COST Action is also to develop the relations between the scientific community which defines the sea-state measurements, analysis, and modelling works, and the users community which needs sea-state information to design ships, offshore and coastal constructions, to improve sea-keeping and optimize ship routes. The relevance of wave spectral information both to model assimilation, as well as to end user applications, will be addressed by means of a dialogue between data providers/producers and representatives of the marine engineering and industry. A cross disciplinary workshop will be held one year into the run time of the Action. Scientific designers from ship, offshore and coastal construction will be invited to voice their views on the relevance of wave spectral information. Further, key representatives of the operational side, i.e. ship masters and harbour captains will be invited to do the same. These sectors of industry have international coordination committees which will be approached in order to find the individual experts. Their views will be significant for both science and data production in the scope of wave spectral information.
D.ORGANIZATION AND TIME-TABLE
The Action will last for four years. Since there are important interactions between the six activities described in Section C they will be investigated in parallel, within working groups. Results will be presented, where appropriate, during a workshop and will be published in the form of reports, according to the schedule given in Table 2.
The Action will carry out these activities through sharing the information on the state-of-the-art methods and facilities, and through exchange of scientists, data and software where appropriate.
The Management Committee (MC) will prepare a detailed workplan based on Section C, for approval by the Technical Committee (TC). The MC will produce an annual in-depth progress report for the TC. For each of the TC's meetings, other than the one where the in-depth report will be examined, it will prepare a short written progress report.
The Action will insure that European organizations dealing with meteorology, oceanography, space and marine engineering, will be involved.
At the end of the Action the MC will publish a final report.
E.ECONOMIC DIMENSION
Experts from six countries have participated in the establishment of the Technical Annex: Finland, France, Germany, the Netherlands, Norway and United Kingdom. When the TC on Meteorology approved the Action, at its 11th meeting, six delegates made known that they will advise their respective governments to sign the MoU.
On the basis of six countries joining the Action, and considering information given by countries having participated in the writing of the Technical Annex, it is expected that, as a mean value, each country will contribute with the equivalent of three scientists working full time for the duration of the Action. Therefore, for each of the four years of the duration of the Action, the equivalent of eighteen full-time scientists will participate, leading to an estimate of the scientific personnel cost of ECU 4 million. Annual overhead costs being estimated at ECU 0,2 million, the minimum total cost of the Action will be of ECU million 4,8.
Current status
The Action has been divided into six activities (six projects) :
Project 1 : Improvement of the retrieval of wind and wave information from SAR signals alone
Project 2 : Improvement of the inversion methods, which make use of external data (numerical model, in situ observations)
Project 3 : Definition and proposal of alternative missions for the measurement of the directional spectra of ocean waves at a global scale
Project 4 : Definition and use of field experiments
Project 5 : Strategies for the analysis, validation, and assimilation of satellite observations into numerical models for sea-state forecast
Project 6 : Development of the relations between the scientific and the engineering communities
With a view to establishing its workplan, the MC has started an inventory of activities carried out in the different countries.
It has decided to select IFREMER (F) as the organisation in charge of administrating Short Term Scientific Missions.
Programme(s)
Topic(s)
Call for proposal
Data not availableFunding Scheme
Data not availableCoordinator
1049 Brussels
Belgium