Institute of Oceanography

1
Ocean Surface Wave Imaging from Seasat to Envisat
Werner Alpers Institute of Oceanography University
of Hamburg Hamburg, Germany
Institute of Oceanography University of Hamburg
IGARSS03
2
PRELUDE
The questions about the correct SAR imaging
theory of ocean waves and the best inversion
algorithm for retrieving two-dimensional ocean
wave spectra from SAR image spectra has haunted
remote sensing scientists for the last 25 years
Institute of Oceanography University of Hamburg
IGARSS03
3
Two algorithms have been proposed to invert ERS
SAR image (Wave Mode) spectra into ocean
waveheight spectra. But they are all based on the
Hasselmann and Hasselmann (1991) nonlinear
integral transform (Krogststad, 1992).

• Max-Planck Institute (MPI) inversion scheme
• Hasselmann and Hasselmann, 1991
• Hasselmann et al., 1996 ( partion of the
  spectrum)
• Schulz Stellenfleth, 2003 (partition, rescale and
  shift algorithm (PARSA))

2) ARGOSS inversion scheme

• Mastenbroek and Valk,1996

Institute of Oceanography University of Hamburg
IGARSS03
4
1) Max-Planck Institute (MPI) inversion scheme

• starts with a first-guess surface wave spectrum
  obtained from a numerical wave forecast model
  (WAM)
• calculates from it the expected SAR image
  spectrum by using the full non-linear
  forward-mapping transformation
• compares the calculated SAR image spectrum with
  the measured one
• if both SAR image spectra do not agree, then the
  surface spectrum is changed iteratively until the
  calculated SAR image spectrum matches the
  observed SAR spectrum
• the iterations are carried out by using
  quasi-linear imaging theory for mapping the
  difference between calculated and observed SAR
  image spectra back to the difference in the
  corresponding ocean wave spectra

Institute of Oceanography University of Hamburg
IGARSS03
5
2) ARGOSS inversion scheme (semi-parametric
retrieval algorithm (SPRA))

• Ocean wave spectrum is separated into a wind sea
  and swell part
• Wind sea part is obtained by combining
  information on the sea surface wind derived from
  collocated ERS scatterometer measurements with
  information contained in the SAR image spectrum
• The swell spectrum is determined from the
  residual SAR image spectrum by using
  Hasselmanns non-linear integral transform

Institute of Oceanography University of Hamburg
IGARSS03
6
A statistical analysis carried out by Heimbach et
al. 1998 at the MPI with ERS wave mode data
acquired over a period of 3 years and
corresponding WAM data shows that only in 75 of
all cases the inversion of ERS wave mode spectra
into ocean wave spectra by using the MPI scheme
was successful.
7
PARSA Scheme (Partition Rescale and Shift
Algorithm)
Prior wave spectrum
Rescaling
Shift
Schulz-Stellenfleth, DLR
8
At present two different algorithms for
retrieving ocean wave spectra from Envisat Wave
Mode data are in competition
1) ESA Algorithm
(developed for ESA by NORUT and IFREMER)
Wave propagation directions are extracted from
the SAR data by using image cross spectral
analysis techniques
2) Max-Planck Institute (MPI) Algorithm
Wave propagation directions are taken from the
WAM model.
(based on the MPI algorithm already in use at
ECMWF for inverting ERS Wave Mode data into
ocean wave spectra)
Institute of Oceanography University of Hamburg
IGARSS03
9
MPI scheme operational with ENVISAT data at ECMWF
Global waveheight statistics for February 2003
Only modulus of cross spectra is used
10
Soon also ENVISAT Wave Mode data will be
assimilated.