Unlike atmospheric data assimilation, which started in the 1960's, wind-wave data assimilation emerged only in the 1980's. Satellite wave data are assimilated to produce the analysis and to improve the forecast of the wave model. This has proven to be of great value to compensate for the possible model errors due to the modelling limitations (e.g. numerics and parametrisations) and the errors in the driving forces (mainly wind speed). The more severe those errors are, the more impact altimeter wave data assimilation would have. The more data available for assimilation in NRT, the more impact would be resulted.

At the European Centre for Medium-Range Weather Forecasts (ECMWF) we have gained over the past 20 years a considerable experience with the use of altimeter (onboard ERS-1, ERS-2, ENVISAT, Jason-1 and Jason-2) wind and wave data products. Since the launch of ERS-1 in 1991, the availability of the altimeter significant wave height data in near-real time (NRT) has enabled the operational assimilation of the wave height data in the ocean wave model that runs operationally at ECMWF from 15 August 1993 onwards. This had a significant impact on the wave model analysis and forecast. Assimilation of altimeter wave height data at ECMWF continued with data from ERS-2 (1995-2003), ENVISAT (since 2003), Jason-1 (since 2006) and Jason-2 (since 2009). It is worthwhile mentioning that the ECMWF wave model assimilates Synthetic Aperture Radar (SAR) wave spectra from ENVISAT as well.

The assimilation of ocean wave products was found to have positive impact not only on wave model products but also on the atmospheric products as well. This is only only possible through the two-way interaction between the atmospheric and the wave components of the model.

The ECMWF forecasting system will be briefly introduced and the benefits of ocean wave data assimilation will be presented.

We aim to constrain a coastal OGCM with sea surface height (SSH) and temperature (SST) satellite data. The objective is to provide a more realistic ocean state estimation at monthly time scales, with specific focus on the surface layers and heat content variability. Modelling and assimilation in coastal areas present specific challenges because of the numerous physical processes that need to be taken into account as well as the wide range of their associated spatial and temporal scales. In particular, high frequency atmospheric forcing on the shelf, tides, coastal waves, mesoscale eddies and instabilities of slope currents are critical mechanisms for the dynamics and hydrology on the shelf as well as on the slope. In such a context, data assimilation can effectively constrain the model if the method is able to take into account the complexity of the model error space that is due to the richness of the processes at work and to the specificity of the studied region. For this reason, we are working on an Ensemble Kalman Filter (EnKF) method where the full multivariate forecast error covariances are used. The SYMPHONIE OGCM is used in a realistic configuration of the Bay of Biscay (North East Atlantic). The model has 43 generalized sigma levels and 3 km horizontal resolution. It is forced with the Meteo-France ALADIN 3 hourly atmospheric fields and with MERCATOR products at the open boundaries. The ensembles are generated by randomly perturbing the wind stress forcing. We use the SEQUOIA data assimilation software that includes a code for the EnKF. SSH data are processed through a tool dedicated to coastal altimetry (XTRACK). SST fields are high resolution products, derived from satellite measurements. Before attempting to assimilate the altimetric and SST datasets we need to answer the following questions: which processes are the altimetric and SST signals representing? Is the model able to simulate these processes? To what extent does the signature of such processes compare in the simulations and in the observations? These issues are addressed through a study of a slope current in the southern part of the basin: we investigate the observability of the Iberian Poleward Current in the satellite datasets and compare the observed signals with the SYMPHONIE simulations. The model is also used to interpret the observations. Then, as a first step towards the assimilation with real data, we set up twin experiments that allow a straightforward evaluation of the assimilation impact. The influence of both kind of observations (SSH and SST) is studied through the analysis of the assimilation results and of the space-time structure of the representers. Our objective is to estimate and characterize the impact of the data on the Iberian Poleward Current, as well as on the mixed-layer properties and on vertical mixing.

This work is part of a group effort at POC (Pole d'Oceanographie Cotiere, Toulouse, France) to develop tools for coastal areas (XTRACK, SYMPHONIE, SEQUOIA, TUGO) and use them within research projects that combine observational and modelling studies.

The impact of the number of space altimeter satellites and the data number SLA observations available for assimilations is assessed using one eddy resolving experimental data assimilation system PSY2V2. And twin experiments performed in delayed-time conditions when 4 datasets are available. The assimilation system is based on the Reduced-Order Optimal Interpolation algorithm and uses 1D vertical multivariate EOFs to extract statistically-coherent information from the observations. In first step, we analyse here their respective impact on the analysis. We led several simulations in parallel that we compare with a simulation that we shall call after reference simulation Sref. In the reference simulation Sref, we assimilate the altimeter data (J1, En and Gfo). For all the simulations presented, we leave the same restart. We made these simulations over six months. First of all, we shall show that it is important to use the altimetry data stemming from various satellites, in particular when their spatial resolutions are different (J1, En, Gfo and Tp). In a second step, we compare the performance of fast delivery products with respect to delayed time data. The validation with independent in-situ data (tide gauge and drifter data) demonstrates a clear degradation of real time in relation to delayed time. To obtain the same quality we need: 1 altimeter for hindcast, 2 altimeters for nowcast and 4 altimeters for forecast. This is essentially due to the fact that to compute the real time only observations of the past are accessible.

The operational oceanography European project MyOcean is part of the Global Monitoring for Environment and Security GMES program. During the next 3 years, 61 European partners in 29 different countries will work to build a pan European ocean monitoring and forecasting capacity. The "marine core service" will be produced by ocean forecast centers and data centers working together. MyOcean is particularly attentive with the setting of quality control, including the scientific validation of the products.

The computation of various forecast scores and the inter-comparison of these scores between the various systems is done with an ensemble of metrics defined in the context of MERSEA and GODAE. Based on these metrics and on various data comparisons, this contribution will give an overview of the quality of the product of the state-of-the-art analysis and forecast system.

We will look at the global Ocean system which is run at Mercator-Ocean and is based on the ocean and sea ice modelling system NEMO and on an assimilation system based on Kalman filter/SEEK. It is declined in eddy permitting and eddy resolving configurations: The current version of the global system has a 1/4° horizontal resolution, with a North Atlantic (including the tropics) and Mediterranean zoom at 1/12° , and a global 1/12° system is under development which will be the reference global system at the end of MyOcean.

One of the aims of this quality report (which will probably be updated on a quarterly basis) is to interact with the scientific community and other users so that one can derive the level of confidence (or the correction one can make) for the use of the products in one’s own application. We will show that measuring the quality of the systems points out the importance of the real time observation network. In order to monitor the ocean we need a perpetual relatively high resolution spatial and temporal coverage, as an input for ocean analysis and forecast systems as well as for a validation purpose. We also need reliable references like long ocean reanalyses in order to validate these systems but also to provide useful information such as interannual or decadal anomalies (for instance for users who whish to initialize seasonal forecast, decadal forecast).

Some activities to evaluate impacts of observing systems are conducted in JMA/MRI using an ocean data assimilation and prediction system, MOVE/MRI.COM. From the activities, we here introduce two major results, the singular vector analysis of the Kuroshio large meander and the evaluation of the impacts of TAO/TRITON array and Argo floats on ENSO forecasting.

Singular Vector (SV) analysis is a way to identify the most unstable perturbations that grow up rapidly in a certain period and affect following phenomena effectively. We applied SV analysis to the formation process of the Kuroshio large meander reproduced in the western North Pacific version of MOVE/MRI.COM. The analysis result shows that an anticyclonic perturbation contacting the Kuroshio path in the southeast of Kyushu grows rapidly and affects the large meander path two month later. This implies that observations in that region are likely to benefit the forecast of the variability of the Kuroshio Current south of Japan.

Effects of assimilating TAO/TRITON array and Argo float data in the global version of MOVE/MRI.COM and its impact on the JMA seasonal forecasting system has been evaluated by an Ocean System Experiment (OSE). The impact of TAO/TRITON array on 1-7 month SST forecasts is remarkable in a most part of the central and eastern tropical Pacific, showing the importance of TAO/TRITON array for ENSO forecasting. In contrast, the homogeneity of data from Argo floats causes an impact on SST forecasts in a broader area.

The atmospheric conditions prevailing since September 2007 to put the global space geoclimatic of the Western Mediterranean to cool conditions, characterized by moderate air temperatures, abnormal negative pressure, positive temperature anomalies of surface water "SST" of the Atlantic Ocean and the Western Mediterranean, and the return of precipitation in North Africa and in South-West Europe. These rains have already caused flooding and damage in fall 2007 in Spain, Algeria and Tunisia. Morocco was spared compared to its neighbours, despite the spring rains have left some minor damage, and dispersed. The same conditions will strengthen and continue to the present, and should continue until August 2009, they are manifested by early rains and violent storms across Morocco, and should continue during the autumn season. The first flood have already left huge damage. In the year 2006, an event "El Niño" appears to the Pacific and was expected to drought in Morocco for winter 2007. Indeed, there has been no rain in winter 2007, although the "SST" off the country were surplus to the normal, and precipitation were not made until spring 2007. Events 2006-2007 that was not seen for at least 20 years, the period during which the ocean and the atmosphere are systematically monitored by satellites, such as Seasat, Topex-Poseidon, Jason and Envisat. It turned out that the phenomenon "El Niño 2006-2007: an event upset" as was stated in a press release from the IRD in March 2007, has not performed as usual scenarios. Indeed, instead of "El Niño" reaches its peak in December 2006, a turnaround was achieved by the return to normal. In the year 2007, the phenomenon has continued its progress towards the other extreme negative "La Niña", which continues to persist until now. This was manifested by storms and floods in our region in autumn 2007, first in Iberia and North Africa (Algeria and Tunisia). Morocco and was protected by a dorsal air did not allow the fall of precipitation occur during this period. It was not until the spring for the rain returned to Morocco, and was brutal in some places. This year 2008, we noticed a lack of heat waves during the summer, or even a drink, but the average temperature in September is still high, During this time of year, and under these conditions, there is the installation of a system of atmospheric circulation called "transition" between summer and winter. This system is characterized by the appearance of depression centers in the Western Mediterranean and Northeast Atlantic, and the rocking of the circulation from West to the South in the form of a meridional circulation. This swing takes the form which creates wave of South-West North-East and North West South East, which swept North Africa and South-West Europe, thus leaving achieve a energy conversion often in the form of rain, which could cause dangerous floods, such events of the fall 2007 in Algeria and Tunisia, and end in September 2008 in Morocco. These conditions should persist throughout the autumn season and continue in winter, which should give us a wet year at the national level. Oceanic conditions of the regional climate system is characterized by the appearance of surface ocean "SST" with a warm anomaly, which facilitates the exchange positive vertical heat between the ocean and atmosphere, and allows it to convey a tremendous amount of moisture, which turns into a torrential rain in the event of rain, which become more common in these conditions. These weather events are now known to scientists, and can be tracked and predicted using space technology and know-how and should serve as an aid to decision makers for our country, with a view to planning effective against the risks of environmental and social security for sustainable development.

We have produced high-resolution reanalysis data in Japanese coastal ocean by using the JCOPE2 ocean forecast system as a part of a cooperative study between FRA and JAMSTEC. We found that incorporation of the in-situ temperature and salinity data obtained by Japanese local fishery research agencies into JCOPE2 reanalysis data significantly improved biases for temperature south of the Japanese coast.

The forecast skill of decadal climate predictions is investigated using two different initialization strategies. First we apply an assimilation of ocean synthesis data provided by the GECCO project (Koehl and Stammer 2008) as initial conditions for the coupled model ECHAM5/MPI-OM. The results show promising skill up to decadal time scales particularly over the North Atlantic (Pohlmann et al. 2009). However, mismatches between the ocean climates of GECCO and the MPI-OM model may lead to inconsistencies in the representation of water masses. Therefore, we pursue an alternative approach to the representation of the observed North Atlantic climate for the period 1948-2007. Using the same MPI-OM ocean model as in the coupled system, we perform an ensemble of four NCEP integrations. The ensemble mean temperature and salinity anomalies are then nudged into the coupled model, followed by hindcast/forecast experiments. The model gives dynamically consistent three-dimensional temperature and salinity fields, thereby avoiding the problems of model drift that were encountered when the assimilation experiment was only driven by reconstructed SSTs (Keenlyside et al. 2008, Pohlmann et al. 2009). Differences between the two assimilation approaches are discussed by comparing them with the observational data in key regions and processes, such as North Atlantic and Tropical Pacific climate, MOC variability, Subpolar Gyre variability.

In recent years, efforts have tried to derive wind vectors from SAR images. The wind direction can be estimated by measuring the orientation of the wind-induced streaks visible in most SAR images. Vachon and Dobson (1996) used the absolute radiometric calibration of the radar images, in conjunction with a wind retrieval model function that relates the ocean wind speed to the normalized radar cross section, the relative wind direction, and the local incidence angle. This model function was developed for C band VV polarization ocean wind scatterometry. For C band VV polarization radars such as the ERS SARs, there are several well-developed Geophysical Model Functions (GMFs), for example, CMOD-4 and CMOD-IFR2. But for the C band HH polarization RADARSAT-1 SAR, similarly well developed and validated wind retrieval models do not exist. Horstmann et al. (2000) derived a polarization ratio relationship, deduced from a comparison between the NRCS obtained from C-band HH polarization ScanSAR images of RADARSAT-1, and observations of the C- band ERS-2 scatterometer in VV polarization, collocated in space and time. Vachon and Dobson et al. (2000) compared observed values of NRSC in HH polarization from RADARSAT-1 with values of NRCS in VV polarization estimated from in situ wind measurements and the empirical CMOD2-IFR3 model from IFREMER. They concluded that the parameter in the empirical polarization ratio mode proposed by Thompson et al. (1998) should be 1.0 rather than 0.6, and they showed that leads to an overestimate in wind speeds (e.g. for high winds). Additional efforts have been made in recent years to estimate by Elfouhaily et al. (1996) and Mouch et al. (2005). In this presentation we present an empirical parameterization based on RADARSAT-2 data.

An operational ocean forecasting system has been developed and demonstrated for the entire Mediterranean Sea and its coastal areas. From observations to modeling, the system operationally demonstrates the quality and feasibility of short term ocean forecasts together with end-user applications. The Mediterranean Operational Oceanography Network (MOON) designed and implemented the basic operational oceanographic service for the Mediterranean area. In this talk we overview the status of development of the MOON products and services after ten years of development. The basin scale system, both for the observing and modeling components, were implemented in three sequential EU-funded projects. The MOON components are now:

a) the Real Time-RT satellite and in situ Observing system;
b) the forecasting system at basin scale and with downscaling in sub-regional and shelf areas, connected to Numerical Weather Prediction (NWP) forecasting and ecosystem numerical models;
c) an information management system for observations/analyses/forecasts production/dissemination/exploitation, also called the Core Service;
d) end-users applications or Downstream Services.

MOON coordinates the RT system and the numerical modeling and data assimilation components and it foster the improvements of most of its parts. The regional approach and the sharing of responsibilities makes the effort sustainable and effective. Several applications of the generic forecasting products will be shown.

COSYNA: Improving regional forecasting capabilities for the German Bight

The presented studies are part of the COSYNA (Coastal Observation System for Northern and Arctic Seas) project. The objective of COSYNA is to enable a long-term observational network for the southern North Sea and Arctic coastal waters, which will be linked to pre-operational models for scientific and management purposes.

The presented poster gives an overview of COSYNA related activities at the GKSS Research Centre with a focus on data analysis and numerical modeling. Investigations are carried out concerning surface waves, suspended sediment, sea surface temperature (SST), sea surface salinity (SSS), and water level. Numerical circulation models and ocean wave models are used in combination with optical satellite data to estimate suspended particulate matter (SPM) concentrations in the North Sea. A detailed study on the impact of currents and waves on sediment transport processes is carried out for the East-Frisian Wadden Sea using a nested model approach. A new method to re-construct SST and SSS fields from data acquired by ferry ships (FerryBox) is described. The approach is based on EOF decompositions of both the 2-D parameter fields and the corresponding 1-D measurements provided by ships. A special technique is applied to interpolate discontinuous FerryBox observations.

Some key results obtained in a study on the assessment of observational networks are presented in the context of water level measurements with tide gauges and satellite altimeters in the German Bight. The method takes into account measurement errors as well as the background covariance structure and can also be applied for the optimisation of observing networks.

Furthermore, first steps towards the assimilation of water level data into the circulation model GETM using a Singular Evolutive Interpolated Kalman filter (SEIK) are presented. A twin experiment is set up to assess the performance of the method based on simulated observations.

The next steps of the COSYNA project are summarised. Of particular importance is the systematic merging of numerical models and observations using assimilation techniques, which are suitable for operational use.

The Mediterranean Forecasting System (MFS) is operationally working since year 2000 and it is continuously improved in the frame of international projects. The system is part of the Mediterranean Operational Oceanography Network-MOON and MFS is coordinated and operated by the Italian Group of Operational Oceanography (GNOO) at the National Institute of Geophysics and Vulcanology (INGV).

The latest upgrades and integration to MFS has been undertaken in the EU-MERSEA, BOSS4GMES and MyOcean Projects. Since October 2005 ten days forecasts are produced daily as well as 15 days of analyses once a week. The daily forecast and weekly analysis data are available in real time to the users through a dedicated ftp service and every day a web bulletin is published on the web site (http://gnoo.bo.ingv.it/mfs). A continuous evaluation in near real time of the forecasts and analyses produced by MFS has been developed in order to continuously verify the system and to provide useful information to the users. The MFS forecast system production is done using an OGCM implemented on the Mediterranean Sea and an assimilation scheme able to assimilate all the available in situ and satellite data. At present two different systems, SYS3a2 and SYS4 are running in parallel every day. SYS3a2 is the official one while SYS4 is under evaluation. SYS3a2 is composed by the numerical code of OPA8.2 implemented on the Mediterranean sea (Tonani et al., 2008) and 3DVAR assimilation scheme (Dobricic et al. 2008). SYS4 uses NEMO as numerical model and 3DVAR as well for the assimilation. The major differences between the two systems are the boundary in the Atlantic ocean which are closed in SYS3a2 while are nested into GLOBAL - MERCATOR in SYS4.

The Adriatic Forecasting System (AFS) is nested into the Mediterranean Forecasting System - MFS (Pinardi et al, 2003; Tonani et al., 2008) - as well managed in Bologna by the Operational Oceanogaphy Group. AFS has been implemented within the framework of the ADRICOSM Partnership (ADRIatic sea integrated Costal areas and river basin Management system). This system provides the forecast of the main physical fields of the sea, such as temperature, salinity, currents, air-sea fluxes, sea surface elevation, and disseminates the data for reaserch and commercial purposes via ftp and via web, and publishes a daily bullettin on the web (http://gnoo.bo.ingv.it/afs/) in image format.

The numerical forecasting model used (AREG, Adriatic REGional model) is based on the Princeton Ocean Model. Its implementation covers the entire Adriatic Sea and extends into the Ionian Sea and a detailed description of the model implementation is described in Oddo et al., 2005. The tides have been introduced into the model, since December 2008, following the formulation of Flather (1976) on the barothropic velocities. The tidal signal has then been introduced in the model through the lateral open boundary condition, where the Adriatic Forecasting System nests into the Mediterranean Forecasting System, at 39° North. The open boundary conditions are taken from the daily simulations and forecasts of the Mediterranean Forecasting System, while the atmospheric forcings come from the ECMWF data at 0.25° degrees of resolution, with a frequency of 6 hours, provided to INGV by the italian Air Force.The precipitations used in both forecasts and simulations come from the climatological dataset by Legates and Willmott (1990), while all the rivers flows, except for the Po river, come from the climatological dataset by Raicich (1994), to which some corrections have been applied, especially along the eastern coast. For what concerns the Po River, daily means at the section of Pontelagoscuro are considered for the simulations, while the last available datum is persisted for all the daily forecasts.

The new global ocean forecasting system developed at Mercator Ocean will be the reference at the end of the European MyOcean project. A global ocean and sea ice high resolution model with a horizontal resolution of 1/12° and 50 vertical levels based on the NEMO OGCM and a data assimilation scheme named SAM2v1 (based on the SEEK filter) are the two main components of the global ocean forecasting system. The third important component is the observation data set routinely used both by the data assimilation scheme and by the validation procedures. This multivariate data assimilation system is able to assimilate in real time both in situ and remotely sensed data (SLA, SST) in order to provide the initial conditions required for numerical ocean prediction. Thus, this new global ocean forecasting system offers a new perspective on the global ocean mesoscale monitoring. First results of different simulations will be shown.

The Surface Marine observation programme (E-SURFMAR) of the EUMETNET (Conference of European National Meteorological Services) Composite Observing System (EUCOS) started in April 2003. It is an optional programme supported by 17 countries and is managed by Météo-France. The main aim of EUCOS is to improve the quality of numerical weather prediction (NWP) and general forecasts over Europe, for which the most important parameter is surface air pressure over the North Atlantic, the adjacent Arctic Ocean and the European regional seas, which cannot be measured from space. The programme delivers marine observations from Voluntary Observing Ships (VOS) operated by EUMETNET members, drifting and moored buoys. As the priority for E-SURFMAR is to increase the density of in situ air pressure observations, the focus of E-SURFMAR has been on the introduction of Automatic Weather Stations (AWS) on ships alongside a significant increase in the number of drifting buoys deployed. Key issues for the programme are to improve the quality and timeliness of the observations, whilst reducing operating costs, e.g. through the use of Iridium for communications

This poster describes the development of an operational analysis and dynamical/statistical forecast system for mesoscale and sub-mesoscale variability in a coastal transition zone: the Mid Atlantic Bight. The analysis system uses adjoint-based data assimilation techniques to integrate a high-resolution 3-dimensional coastal model (ROMS; the Regional Ocean Modeling System) with data from a coastal observing system comprised of surface current radar (CODAR) installations, autonomous gliders, satellite imagery, moorings, and XBT/CTD acquired during the ONRs Shallow Water Acoustics 2006 (SW06) field program. Comparison with not-assimilated temperature and salinity observations suggest that the regional model has an skill superior to other data-assimilative global models. We attribute the added skill to a bias reduction due to assimilation of climatological information and to the correct projection of the satellite information by the adjoint model.

The Netherlands has started a number of activities to work towards a new generation of web-based information and data systems connected to integrated forecasting systems. The MATROOS system compiles, stores and distributes real-time data for forecasting of various parameters, such as wind, waves and storm surges. It also stores all the model forecasts and provides tools for analysis of previously made forecasts. Within the NOOS network, MATROOS is used for model comparison tasks. The MATROOS architecture is flexible to data providers and the data formats confirm to international standards. On the user side, all data are directly addressable via web-based interfaces and protocols. In the future, MATROOS will form part of the backbone of Dutch forecasting systems, also including water quality and ecological forecasting. As a data achieve system, it will become very valuable for research on ocean properties.