Satellite SAR Remote Sensing of Ocean Internal Waves
The Detecting technology and retrieval
method of internal wave
One of the problems in detecting internal
wave signatures on the ocean surface using
satellite SAR images lies in distinguishing
internal waves from often oceanographic
and SAR phenomena. Internal wave look-
alike may include natural film, greasier,
threshold wind speed areas, wind sheltering
by land, rain cells, current shear zones, oil
slicks, eddies ships and ship wakes, and
upwelling. Among this look-alike, natural
film, rain cell, and current shears have been
see to represent the largest problems.
Therefor, the goal of the studies is to
develop an automatic method for internal
wave detection and location in which dark
qeasilinear period linear with a high
probability of being an internal wave packet
are automatically identified. The
technologies of extracting internal wave
information from SAR images are as
follows:
- Valuable SAR images was chosen. It's
ERS-1 SAR image. Acquired time is on 26
June 1995, 02:29:14GMT. The location is
21045'57"N, 120041'02"E. It's pixel internal
is 12.5m. The region size is 100km2
(ERS-1
SAR swatch width is 100km). This is
southern part of East China Sea.
- Speckles noises of SAR images are
suppressed with new filters that select
Kaiser, Hamuring, Hunning and other
windows, wavelet transform etc. In order to
get high-qualified SAR images with typical
internal wave signatures.
- To extract the characteristic parameters of
internal wave from SAR sub images using
two-dimensional Fast Fourier transforms
(2D-FFT).
Two adjoin sub images are taken from
processed ERS-1 SAR image and carried out
image spectrum analysis with 2DFFT. The
dominant wavelength, means wavelength of
internal wave and its propagation direction
etc. Geographical parameters can be
determined according to spectrum features
of internal waves.
- Studying the features of internal wave by
wavelet transform method, wavelet analysis
method is better than FFT to unsteady and
non-periodic signatures in the dimensional
analysis.
The mean wavelengths of internal wave are
estimated using wavelet analysis.
-
Comparison of the results of FFT with
that of WT.
Preliminary results of SAR to detect internal
waves some preliminary results of applying
the proposed method for internal wave
detection and location to SAR image
containing internal wave patterns are
presented.
Fig.1 shows the SAR image of internal wave.
From the SAR image, we can see obvious
dark and right band.
Fig.1 SAR image of internal wave
Fig.2 is two adjoin SAR sub images (a) &
(b).
Fig.2 SAR sub image of internal wave
Fig.3 is the results obtained by FFT
correspond to Fig.2 (a) & (b).
Fig.3 The direction spectrum of internal
waves by FFT
According to direction spectra of internal
wave, mean wavelengths are about 872m
and 816m, their dominates wavelengths are
708m and 607m respectively. Therefore we
can decide the propagating direction is from
western to eastern.
Fig.4 shows the results obtained by wavelet
transform of Symmlet and Danbedies to
Fig.2 (b). Table 1 is various layer of physical
scale and normalized deflection results
expend from wavelet basis of internal wave
section. From Table 1, we can see that 6th
layer maximizing deflection is maximum.
The mean wavelength of internal wave
corresponding to 6th layer section is about
800m.
Fig.4 Wavelet basis expend of internal wave
section
Table1.
Various expended physical scale and
normalized deflection expended by wavelet
basis of internal wave section
| Laye r(j)
| scale (m)
| Normalized
s 0 (Symmlet)
| Normalized
s 0 (Daubechies)
|
| 8
| 3200
| 0.0267
| 0.0878
|
| 7
| 1600
| 0.0044
| 0.0159
|
|
6 | 800
| 0.4565
| 0.3620
|
|
5 | 400
| 0.2166
| 0.1163
|
|
2 | 50 |
0.0280 | 0.0448 |
| 1 |
25 | 0.0375
| 0.1097 |
| 4 |
200 | 0.1748
| 0.1691 |
|
3 | 100 | 0.0555 | 0.0944
|
Conclusion and Discussion
SAR imaging mechanism of internal waves
is very complex. The imaging includes the
interactions of surface wave and waves,
waves and currents as well as current and
bottom topography. Of them, the interaction
of current fields induced by internal wave
and surface waves induced by wind fields is
mail contributed factors, particularly, high
and low of wind speed is a key factor to
detect internal wave by SAR.
Wavelet transform is a new method to study
internal wave, particularly unsteady non-
periodic signal signatures of internal wave.
FFT is also a efficient methods to
quasilimear and periodic signal signature.
The technologies combined FFT with
wavelet transform presented by us are
applied to analyze the manifestation of
internal wave. Comparing the results of both
methods gets the geophysical parameters.
For example, for choose SAR image, the
wavelength of internal wave is about 800m,
and the propagating detection is from
western to eastern.
In future, dominating research areas include
SAR imaging mechanism and information
extraction as well as determination of
characteristic parameters from SAR images.
Acknowledgment
The authors would like to thank Dr Ming K.
Hsu. Taiwan University of Oceanography,
Taiwan, China for us providing the ERS
SAR images, and would like to thank 863-
818 program.
