ERS PRI products for wave parameters estimation over the Mediterranean Sea

Abstract

A system is operative at the I-PAF which processes ERS PRI images, for the extraction sea state parameters,. The system provides sea spectra, like those supplied by the wave mode, and a map showing the waves parameters all over the imaged area. The system features and some validation results are presented.
Keywords: SAR, wave, Mediterranean

Introduction

The main limitations, for the use of wave mode in basins like the Mediterranean sea, are the minimum wavelength that can be imaged by the SAR and the measurement frequency. ERS can detect wavelengths larger than twice its range resolution. It can make wave measurements spaced at least 200 km. Moreover the wave mode is frequently unavailable because in conflict with the imaging mode. The last one is more requested over a crowded area as the Mediterranean one. Although the first constraint can’t be overcome, it makes sense to know how frequent are the sea conditions detectable by SAR over the Mediterranean region. Hence we have considered the data acquired by two buoys (one located off Alghero and the second off Catania) during a period of one year and we have found wavelengths greater than the SAR threshold in 55% and 43% of cases respectively. The second constraint can be overcome retrieving the sea parameters from the imaging mode instead of the wave one.The above considerations have been the basis for developing both a dedicated system (called "Nettuno") and a new ERS product that retreives sea state parameters from PRI images. This approach has a further advantage, the different measurements performed on a large area as wide as an ERS PRI , can be used to monitor the wave parameters evolution and to discriminate valid wave measurements from artifacts.

System description

The system ingests ERS PRI (approximately 8000 x 8000 pixels). In order to inspect the imaged region a quick look is produced. The area of interest can be selected and extracted, as shown in fig. 1.

Fig.1 - ERS 1 PRI Image acquired on 21/11/93 over the eastern Sicily (orbit 12298, frame747)

Considering that, for the retrieval of wave parameters an area of 512 x 512 pixels is processed, the scene is divided into patches of the above dimensions as shown in the following picture.

Fig.2 - Result of subarea selection procedure, the patches marked by a cross are not to be processed

The result of the processing is presented as a wave field (fig.3). The length of each arrow is proportional to the sea wave wavelength and its orientation is in the direction of the seawave propagation. The PRI images do not resolve the ambiguity in wave direction,so only one of the two possible solutions generated is displayed on the map.

Fig.3 - ERS 1 PRI wave field analysis result.

The polar diagram of a single patch is shown in fig.4. North, Azimuth and Illumination directions are reported in the center of the diagram; on the right side the wavelength ranges are indicated, whereas at the bottom the peak wavelength and its directions are shown.

Fig.4 - Polar diagram of a single patch spectrum

The algorithm adopted for the wave parameters estimation is essentially based on the following steps [C. Bruning]:

1)extraction of 9 partially overlapping 256x256 pixels subimages ;

2)auto convolution and spectrum estimation of each subimage;

3)determination and correction of the static system impulse response;

4)spectrum smoothing;

5)determination of wavelength and direction associated with the peak of the spectrum.

During the validation activity, it was observed that not all the analyzed images show a clear wave pattern. The modulation is present only on some portions of the image, moreover it comes out that the sea parameters, estimated in neighboring points, aren't drastically different, but for the case of low intensity or noisy signals. Keeping in mind the above considerations a selection procedure has been implemented to automatically reject those measurements which are very different from the neighboring ones. The main steps are as follows:

1) the wavelength measurements greater than 250 m are rejected (this assumption is quite realistic because no wavelength longer than 250 m have been detected by the Catania buoy in one year of observations, instead, the Alghero buoy recorded only two events, over more than 2800 measurements in one year). Processing results larger than 250 m, are usually associated with atmospheric phenomena interacting with sea surface. [W. Alpers], [J.A.Johannessen]

2) The average wave direction is computed and the measurements which differ from it more than are rejected.

3) The average wavelength is computed on the remaining results and accepted those differing from the average less than .

In the wave field map the invalid measurements are indicated with a dot. After the processing a quality index (Q.i.), defined as the ratio of the valid measurements over all the measures made, is annotated.

The system has been developed using a four generation software language, it can run on the following operating system: Windows, UNIX, VMS. The processing of a full ERS PRI image on P.C., equipped with a 486 CPU and a 50 MHz clock, requires four hours and a half. A DEC alpha station processes the same scene in about 30 minutes.

Results

The validation activity is carried out comparing the sea state parameters acquired by buoys deployed off the cost of Sicily (Lat. 37 °26 '.4 N; and Lon 15° 08'. 6 E) and Sardinia(Lat. 40° 32 '.9 N and Lon 08° 06'.4 E), with the data obtained processing the ERS image over the same region or in the near proximity.Since the SAR images and the buoys data available were not acquired simultaneously, we typically have a buoy data 1 h and half before and after each satellite passage, we selected those scenes relevant to parameters unchanged during two buoy recordings. The geophysical parameters compared are the peak wavelength and its direction. For those images not containing the position of the buoy, we compared the buoy measurements with the closest ones in the image. The results are reported on Table 1. The differences in wavelength estimation between SAR and buoys is within 10 %. The wave direction differences between buoys and SAR estimation are in average, with a standard deviation of

Table 1

Conclusions

The product presented has the purpose to overcome some limits of the use of ERS Wave mode over regional basins. Some features, impossible to detect by a single measurement, are made evident with this technique that permits the sea state parameters evaluation over large areas. The I-PAF is willing to cooperate with partners for an extended experimentation of this product.

References

Bruning C., Alpers W., Wilde A. 1992: Ocean wave imaging by ERS-1 SAR: first result from the Haltenbanken calibration/validation Campaign. IGARSS ‘92, pp. 169-177.

Alpers W. , Brunner, B., 1994: Atmospheric boundary layer rolls observed by SAR onboard the ERS-1 satellite. J. J. Geophy. res., 99, pp. 12613 - 12621.

Johannessen J.A. et al , 1994: SAR ocean feature catalogue. ESA SP 1174

Keywords: ESA European Space Agency - Agence spatiale europeenne, observation de la terre, earth observation, satellite remote sensing, teledetection, geophysique, altimetrie, radar, chimique atmospherique, geophysics, altimetry, radar, atmospheric chemistry