workshops 06-Aug-2014     
Organising Committee
Scientific Committee
Programme Overview
Seed Questions
Session Summaries and Recommendations

Session Summaries and Recommendations

Wave mode Algorithms, Validation and Assimilation


(by Vincent Kerbaol  &  Susanne Lehner)



        Improvement of the ENVISAT ASAR level-2 wave mode product as part of the reprocessing campaign:

         Detrending, MTF at low wave-numbers, ambiguity removal, cut-off estimation

         Improvement in the geophysical quality

        Empirical approach tuned against NWP

         Global statistics of conventional sea-state parameters

         Requires additional “ training data” for tuning the algorithm.

        Measurement of global map of individual wave parameters from 3 weeks of ERS-2 wave mode data performed at DLR and validation campaign initiated.

        Cross-spectra algorithm is particularly well suited for WSS products and first results have been presented.

        Wave algorithms exhibit limitations for high to extreme wind conditions.

        Very limited in-situ data sets on the behaviour of the normalised RCS at high winds (suitability of CMOD models questionable and would require further tuning to high wind speeds sea states).



        Use of historic ERS wave mode archive for climatologic analysis has been suggested and reprocessing of 2-years archive to imagettes has been completed at DLR. Global maps of wind speed and conventional sea state parameters generated



         Improvement in the geophysical data quality after the wave mode Level-2 algorithm improvement (reprocessing of ASAR WV mode data since December 2002 started at Ifremer):

        Decreasing of the rejection rate for assimilation

         Comparisons w.r.t. models/buoys show:

        Overall agreement in the SWH

        Larger variations on the shape of the directional spectrum



        Use of Wave mode products at different wave forecast centres

         ENVISAT ASAR Level-2 products ASA_WVW_2P at Meteo-France

         ERS-2 SAR spectrum at ECMWF (until 1st February 2006)

         ENVISAT ASAR Level 1 ASA_WVS_1P (from 1st February 2006 at ECMWF)

        Qualitative indicators are crucial and shall be provided in the products:

         SNR, Normalized Variance Wind Speed (in-homogeneity test)

        Impact of the level-2 wave products in the assimilation

         10% correction after reprocessing with the upgraded algorithm (compared with independent data)



        Increase sampling along the satellite track (to be considered for future missions)

        Increase size of Wave mode imagettes to improve the wind analysis (to be considered for future missions)

        Plan acquisitions at higher incidence angles in VV and HH polarizations (during limited test periods to gather suitable test data for incidence angles other than IS2 angles)




(by Johnny A. Johannessen & Fabrice Collard)




Existing applications


        Operational use of SAR data by offshore industry (GM and Wide swath mode) for current front and eddies tracking have been demonstrated based on purely visual interpretation of sea surface roughness (combined with other source (IR/visible/altimetry), in particular in area where cloud are frequents and ocean model not accurate enough.


        Internal wave are monitored in various ocean region maybe and there is an opportunity to built a global database (time of observation, IW crest geocoded poly-lines)




        Forward modeling is the base of any future inversion algorithm: promising development have been recently made to model effect of radar cross section and Doppler velocities modulation by variable surface currents.


        Some image analysis techniques based on wavelet techniques have been investigated for feature extraction and tracking. This analysis benefits from use of sequential SAR acquisitions a few hours apart from different SAR platforms.


        Radial components of surface current estimation from Doppler centroid anomaly can be routinely estimated but requires a posteriori knowledge of satellite attitude at the precision of a few milli-degrees. A validation campaign would be required to estimate the errors associated to this measurement.





        Develop feature extraction algorithms based on dual polarisation (even in wide swath mode).

        Have a background mission monitoring (GM and wide swath mode) in strong currents/eddies area (from shore to continental shelf break).

        Increase the temporal repeat time (at least one image every day). May be achieved by constellations like Cosmo-Skymed. Continuity on long term (mission not finished before algorithms are finally developed and validated).

        Provide restituted attitude information to users as a regular product (such as restituted orbit product).



(by J. Horstmann & E. Attema


Answers to Seed Questions


1. What are the major methodologies and which of these can be used on a fully operational basis?


Wind directions from wind streaks via:


FFT Method                                

Local Gradient Method           

Projection Method                     

Major differences besides the techniques are the utilized scales


Wind Speeds from the normalized radar cross section via:


C-Band models, for HH the polarization ratio has to be taken into account:

Thompson et al., 1998 (q) tuning via a (between 0.4 and 1.2)

Elfouhaily, 1998 (q)

Mouche et al., 2004 (q, F)

Engen et al., 2004 (q, F,u10)



2. Are there new promising methodologies?


Bayesian approach       

Spectral method           

Wind directions using incidence angle dependence of the NRCS

Combination of different approaches


3. What are the resolutions, accuracies and limitations of the available methodologies?


Wind direction:

Resolution of approximately 10 km with an error of typically 20 deg decreasing with wind speed. Dependent on the presence of streaks


Wind speed:

Resolution of ~ 500 m (depending on SAR resolution and number of looks). Wind speed error is typically 1.6 m/s for wind speeds below 20 m/s.



4. Where are the most useful applications of SAR wind field retrieval and what are the requirements?


Stand alone:

Tuning of high resolution wind models

Energy assessment in offshore wind farming

Investigation of turbulence (scientific)

Hurricanes (forecast, scientific) data are required near real time


Winds play an important role in most other SAR ocean applications, e.g. slick detection and ship detection.


5. What are the needs for accurate SAR wind retrieval (calibration, modes, polarization etc.)


Accurate radiometric calibration < 0.5 dB

Incidence angles preferable at incidence angles > 25deg

Better knowledge of the polarization ratio





·        Perform a detailed investigation of the polarization ratio

·        Analyze what is the accuracy and where are the limitations of the CMOD5 model at winds above 20 m/s

·        Investigation of fetch dependencies as well as the presence of swell on wind speed retrieval.

·        Investigation of the error introduced by using the C-band models for very high resolution.

·        Acquisition of ASAR wave mode data at different polarizations and incidence angles.

·        Regular acquisitions of ASAR data in Hurricane and Typhoon conditions and their availability in near real time (special BRM is already implemented in the Golf of Mexico & Caribbean Sea in ScanSAR mode)







Wave Applications

(by J.C.B. da Silva & F. Ocampo-Torres)





1.      Accidents in severe weather events using satellite radar data


       Wind and wave fields around offshore oil platforms (Ekofisk, Draupner and Fino) were retrieved from SAR, analysed and compared with environmental data collection, laser arrays and WAMOS directional wave radar

       SAR yields measurements over a 100km x 100km area and thus shows the spatial structure of the wave field (e.g. wave refraction around islands)

       Direction of the wind measurements and the high frequency wave components of the laser array fit very well


2. Surface wave spectrum in the Gulf of Tehuantepec


       Detailed analysis of SAR images provides information of spatial variability of wave spectra, also observed from direct measurements with wave staff array revealing the variable nature of waves

       Influence of swell on the momentum flux between ocean and atmosphere is determined directly and an enhancement of the drag coefficient is also observed


3. A global quality-controlled spectral wave database


       Integration of observations and models are essential to get quality controlled consistent basic databases to build applications


4. SAR retrieval of 2-D ocean wave spectra (PARSA)


        A parametric inversion scheme for SAR retrieval of 2-D ocean wave spectra from look cross spectra was presented

        The PArtition Rescaling and Shift Algorithm (PARSA) provides estimates of complete 2-D ocean wave spectra using prior information, making use of the complex information provided by cross spectra

        Validations based on triple collocations with buoys (WAM, NDBC buoy and PARSA), indicates PARSA improves spectrum when compared to WAM


5. SAR Wave Mode Imagette under extreme wind and wave


        Good coverage during hurricanes and typhoons offering the opportunity of a  statistical  study of storms dynamics, in the four quadrants. The main objective of project is to make a synergistic use of satellite data

        Spectral information provided by SAR contains valuable information about the ocean wave dynamics in storms


6. Fast Internal Waves from Luzon Strait


        Non-linear IWs in the South China Sea propagate westward of Luzon Strait, with typical phase velocities of Cp=3m/s (fastest reported so far). In situ measurements support velocities retrieved from sensor synergy

        Envisat Wide Swath mode is necessary to study the generation and propagation characteristics of such large-scale internal waves



Answers to Seed Questions AND RECOMMENDATIONS


What special SAR signatures characterize very fast propagating internal waves in deep ocean? What are the most useful SAR imaging modes to identify internal wave generation sites, and internal wave generation mechanisms? Could future satellite missions with Along Track Interferometric capabilities be used to measure surface currents associated to internal waves and tides? Have we improved our capabilities to distinguish atmospheric and oceanic internal waves observed by SAR?

Wide Swath mode is essential to study generation and propagation of internal waves, although narrow swath is also very important for finer scale observations. Along-track Interferometry should be used in future to retrieve current fields associated with internal waves and tides. Oceanic and atmospheric origin of internal waves is still a possible cause of misinterpretation. Methods to distinguish those signatures need further development. The ability to get further environment information (other physical variables) through remote sensing, as well as including multi polarization modes, would lead us to a better position to make the distinction


What new SAR products can be developed for coastal engineering studies? Can we develop current algorithms to be applied to the near-shore and the surf zone (include wave breaking processes)? Should this be addressed in future spaceborne SAR missions?

SAR images with better than 5m resolution are required. Special attention to depth dependence of inversion schemes should be given, where wave dispersion relation and SAR imaging theory with general validity should be taken into account.


Is it possible to use methods developed to detect extreme wave events for an early warning system, based on SAR observations (faster delivery of real time data)?

Extreme internal wave events are still a main concern for offshore operations. It is important to further study the generation mechanisms in order to be able to predict time and space variability


 Is there any theoretical work to support scatterometer signal response as a function of wind speed for ocean surface in the presence of swell?

Presence of swell and also fetch dependency on scatterometry. Polarimetric measurements would provide further information on sea state characteristics and roughness influencing backscatter


How can we make use of  SAR to improve our ability to predict the transformation of offshore spectra (from wave forecasting models) into inshore regions with sensitive engineering and navigational needs?

Main source of error is probably the bathymetry information. Finer resolution for SAR images are required to determine the wave spectrum nearshore. In order to provide useful information for operational purposes (engineering, navigation, rescue, etc) real time delivery of SAR images is required


Can we identify special test-sites for validation of wave products where characteristic conditions exist and in situ measurements are being conducted?

For calibration and validation purposes a multinational experiment should be conducted. Even if one of the main driving force could be target detection, most environmental variables should be retrieved, wind, waves, currents, etc. Since there are no new planed missions with scatterometer, SAR images should be considered for wind field retrieval. Participation of major space agencies is recommended


Discuss the use of the historic ERS wave mode data set

Wave mode re-analysis to integrate the various missions is required


What is the climate requirement for 2-d sea-state measurements from SAR?  (e.g.  directional wave height variance spectrum over what height/wave-number range?

Long term statistics of wave conditions has been reported with possible association to global change. Uniform processing providing cross spectra from imagettes would assist in building a long term data base. Applications (insurance for instance) require wave statistics over longer periods than ten years




(by J.A. Lorenzzetti, J. Schultz-Stellenfleth, W. Alpers)


The session was held on January 25, 2006 from 09:00 to 16:00. A total of 16 papers were orally presented. The paper’s authors included 9 European countries, United States, Mexico and Brazil.

In a broad sense, the presented papers can be categorized into three groups: a) mathematical models of data analysis (5); b) operational experiences of oil spill and ship detection (7); c) commercial use of SAR data for the monitoring of oil seeps (1), and d) science oriented applications of SAR data (2).

At the end of the oral presentations, a discussion plenary session was held, which was guided by a set of pre-defined list of seed questions.

The following main points can be highlighted from the oral presentations and plenary discussions.


It is not possible to define a specific set of SAR system that would meet all the requirements of OSSD. In particular, for ship detection, from a polarimetric SAR, HH pol is the best option, with the HV pol channel being a good choice for targets at steep incidence angles.

Synergistic use of different SAR images together with other satellite systems (visible, infrared, scatterometer, altimeter) and in situ data are highly recommended for OSSD.

For ship detection, a case has been made that a Wide Swath, even at the cost of lower resolution would be preferred for the open ocean, which normally present larger ships, however, still a resolution better than 30m would be desirable; present ASAR-WS is not good enough. For the coastal zone ship detection applications, Narrow Swath with high resolution seems to be the best option.

A set of advanced techniques of data analysis of SAR image processing for OSSD were presented. All of them seem to present good potential, but in general, they still lack an adequate validation. It is here suggested that ESA should support a benchmarking project in which all different groups which are doing research in this area would process the same set of SAR images with the support of the same ancillary environmental data, such as done in the DECLIMS project.

Regarding the Envisat data availability for operational use, at least for outside the KSAT antenna footprint, still the current delays are not adequate. Better than 30-60 minutes is essential. It has been reported that delays of more than 24 h are not uncommon. The heavy load of ESA countries data requests were reported to be a major constraint in supporting operational users of ASAR data.

The important problem of a clear discrimination between real oil spills (or oil seeps) from look-alike features, is still not resolved. Although several methods of automatic detection of oil spills have been presented, there still a need for a trained operator for final verification.

For an operational maritime surveillance system, persistence of coverage seems to be a bottleneck that cannot be resolved with a single satellite SAR system. A constellation of satellites seems to be the only solution to shorten the revisit time. Some applications demand a monitoring scale as short as few hours, or even less. The new SAR systems that were very recently launched and those which are planed to fly in the next few years from countries like Germany, Italy, Canada, China and India, for example, could guarantee the persistence and continuity of data availability for the coming years. An alternative for decreasing periods without SAR data is the synergy with airborne sensors and coastal systems like CODAR. When SAR is used in conjunction with other vessel systems like VMS (Vessel Monitoring System-mandatory for fishing vessels larger than 12 m) and AIS (Automatic Identification of Ships – mandatory after 2008), requirements on revisit periods for ship detection should be less strict. Meantime, surveillance should benefit from more flexible and faster tasking possibilities (e.g. quick scheduling of SAR data acquisition due to emergency cases).

Besides the important role in marine pollution monitoring, the use of SAR data for oil slick monitoring must be viewed as another component methodology for climate studies considering the strong impact of such films have in decreasing the rates of gas exchange between the oceans and the atmosphere.







(by H. Johnsen & W. Pichel)


Spaceborne SAR Systems can be used to produce high quality bathymetric maps in coastal waters with topography up to 30 m depth. This capability can be used to improve the navigation safety along the coast as well as to study the modifications of the coastal current regimes (i.e. due to the building of harbours).


Marine surface films, often present in coastal areas, may be related to the presence of hydrodynamic processes i.e. plumes, internal waves, eddies. The slick visibility is greatly dependent by the electromagnetic wavelength. Therefore the combined use of satellite images acquired at microwave, visible and IR frequencies can improve the discrimination capabilities between slicks and other oceanic and atmospheric phenomena. In this context the effort is to investigate on the physical mechanisms which are related to the slick signatures using auxiliary information provided by a multi-sensor approach.

Ice Applications

(by P. Clemente-Colón & Ben Holt )



Answers to Seed Questions AND RECOMMENDATIONS


Are there any pressing science issues that need to be explored?


- Ice thickness is a critical parameter not easily assessed from satellite SAR data. Based upon comparisons with EM thickness measurements, there is an apparent increase in backscatter with increasing thickness beyond 2m with HH at least for first year ice. However, a related study in the Baltic had an unclear relationship. There is little sensitivity of HV to increasing thickness. Additional investigations are required including over multiyear ice.


- Further work on sea ice characterization using polarimetric and multi-polarization data should be undertaken to improve sea ice type mapping.

Are there any new algorithms or techniques that that should be considered?


- From operational side a better handle on how to determine ice concentration through improvements in SAR automated and semi-automated objective techniques using multi-pol/polarimetric is needed


-In some Centers, the lack of automated processing may impose a limitation on the use of SAR data.


-There is a need for better L-band and fully polarimetric characterization of sea ice and how this can be used together with C-band dual pol and polarimetric mode including classification routines


-Deriving new and young ice sea ice thickness estimates from wave-ice interaction has been improved.


Are there any recommendations to modify or enhance the Envisat mission over the polar or sea ice covered regions, the ALOS mission?


- Enhancements on sea ice characterization using multi-band data should be further investigated.


- Can the daily ASAR coverage be improved over the polar regions?  This would improve derived ice motion fields.


- Particularly, increased access to ASAR Wide Swath acquisitions around Antarctica should be made available to improve the quality of NIC sea ice charts (type/stage of development) and provide estimates of sea ice thickness for PolarView and all other users.


- The limitations of using GMM for ice charting should be explored.


- The sea ice community should recommend to JAXA that sea ice coverage be added to the basic observation plan, which is presently primarily over land.


- Alternate polarization with cross-pol improves discrimination of open water from ice, smooth and rough ice, especially when used in combination with HH.  Cross-pol shows little improvement with smooth and young ice types.


- L-band data appears to improve the identification of ridges compared to C-band.  The use of Envisat together with ALOS will improve overall understanding of sea ice mass volume.



Are there any recommendations that may enhance the exploitation of the Sentinel-1 mission for polar and sea ice applications?


-There is a desire for a multi-frequency satellite mission similar to the SIR-C/X-SAR experience.  


-The need for the use of a SAR constellation to provide high revisit time over highly dynamic se ice regions was suggested (e.g. RADARSAT-2 follow-on SAR Constellation Mission [2011+] plus other missions)


  Higher level                 Last modified: 10-Feb-2010