Wave movement


  • Digital map of wave fields from L1-B level SAR images in fully automatic mode.
  • Extraction of wave fields from SAR data
    • GWW function (of e-GEOS) based on models [ Hasselmann K. and S. Hasselmann , 1991, 1996] and on inversion techniques of [ Lyzenga DR, 2002].
    • Use of Level-1B SAR data from COSMO SkyMed , Radatsat-2, Sentinel-1.
  • Algorithm
    • SAR spectrum estimation
      • de-trending by Gaussian filtering,
      • identification and removal of the SAR IPR from detected (L1B) SAR data
      • removal of noise pedestal (speckle)
      • average of independent spectrum samples
    • SAR spectrum -> Sea spectrum
      • Hasselmann’s non-linear model inversion through cost function minimization
    • Spectral partitioning
      • identification of the partitions (in ω-θ-space)
      • estimation of significant wave height (Hs), wave-length and propagation directions for each identified wave system

Operating scenario

  • Base knowledge

Technical specification

Spatial resolutionFrom 2 to 3 Km (depending on the SAR data used for estimation)
Coverage areaFrom 1600 to 60000 Km ^ 2 (depending on the SAR data used for estimation)
Geographic accuracy5 m
Information Age6 hours
Measurement uncertaintyHeight: RMS <0.5m, Bias <0.1m Direction (for waves> 0.5m): RMS <30 °, Bias <10 ° Wavelength (for waves> 0.5m): RMS <50m, Bias <10m
Measurement / projection unitHeight and wavelength: meters Direction: degrees Projection: geographic ( lat / lon )
EO data in inputSAR data from COSMO- SkyMed , Radarsat-2 and Sentinel-1 (TOPS mode excluded)
Input data usedOptional: wind and wave forecasting models (WAM) from ECMWF and NOAA
Given for validationdata from buoy or altimeter in the test area
Algorithm methodInversion of the Hasselmann model for the estimation of the sea surface spectrum starting from the estimation of the SAR spectrum. The inversion is based on the Lyzenga method which has proved effective in the Mediterranean area and does not necessarily require forecasting of input waves.
Refresh rateGiven the availability of the SAR data on the area of ​​interest: 12 h

Statistical analysis

Spatial resolutionLess than 15 km.
Coverage areaMaximum extension of the AOI: 20000km 2
Geographic accuracyComparable with spatial resolution.
Information AgeData not prior to 2008.
Measurement uncertaintyDependent on the precision of the input data for which the following errors are expected: · Wave height ≤ 4m: Bias = -0.25m; RMS = 0.5m. Wave height> 4m: Bias = -0.5m; RMS = 0.6m. Wave direction: 45 degrees. (Confidence level: 90%)
Measurement / projection unitUnit of measure · Return time: months. · Wind rose: meters for wave height and degrees for wave direction. Projection: Geographic-WGS84
EO data in input
Other input data used (not EO)Data produced by a marine wave propagation model. The model outputs include the significant height of the wave with its associated direction.
Given for validationIn situ measurements or other EO data that will be identified during the project.
FormatGeoTIFF for the product “Return time”. png for the product “Wind rose”. Metadata: meet ISO 19115 standards, and specifications provided by the INSPIRE Directive 2007/2 / EC and the relevant Decree . n. 32 of 27/01/2010
Refresh rate3 months

Example – wave movement

Wave movement field generated by SAR data amplitude

Examample – statistics on wave movement

Annual statistics of wave movement off the coast of Petacciato (Italy)