Abstract: SIR-C/X-SAR images showing natural (biogenic) surface slicks as well as man-made (anthropogenic) mineral oil spills are analyzed with the aim to study whether active radar techniques can be applied to discriminate between these two kinds of surface films. For this purpose, various surface films were deployed within the shuttle swaths during controlled slick experiments in the German Bight of the North Sea as well as in the northern part of the Sea of Japan and the Kuroshio Stream region. From this data set it is shown that the damping behavior of the same substance (i.e., the reduction of the radar backscatter caused by the surface film) is strongly dependent on wind speed. Furthermore, several SAR scenes showing biogenic and anthropogenic surface films at low to moderate wind speeds are analyzed. The damping behavior of these different kinds of oceanic surface films varies particularly at L band, where the biogenic surface films exhibit larger damping characteristics. It is concluded that at low to moderate wind speeds multifrequency radar techniques seem to be capable of discriminating between the different surface films whereas at high wind conditions a discrimination seems to be difficult. The observed differences between the damping ratios measured by SIR-C/X-SAR and by an airborne multifrequency scatterometer seem to be systematic.
SIR-C/X-SAR images showing radar signatures of rain cells over the ocean are investigated. These radar signatures usually consist of irregularly-shaped bright and dark patches that strongly depend on radar frequency and polarization. Moreover, the phase difference between the horizontally and vertically co-polarized signals is altered. The radar signatures of rain cells observed in SIR-C / X-SAR imagery of the ocean originate (1) from the scattering and attenuation of the microwaves by rain drops and ice particles in the atmosphere and (2) from the modification of the sea surface roughness induced by the impact of rain drops and by air motions (up- and downdrafts) associated with rain cells. Depending on the radar wavelength, ocean areas struck by rain can have higher or lower normalized radar cross section (NRCS) than the surrounding rain-free area: Where heavy rain is impinging on the water surface, the X and C band NRCS is usually enhanced, and the L band NRCS is reduced. Furthermore, the rain affects the phase between the HH and VV polarized backscattered radar signal at C band.
SIR-C/X-SAR images of the Atlantic Ocean showing signatures of an atmospheric front and almost range-propagating ocean waves are analyzed. The L band SAR image spectra calculated from two areas located on opposite sides of the atmospheric front are quite similar, whereas the corresponding X and C band SAR image spectra differ significantly. It is shown that this is a consequence of the SAR imaging mechanism: At L band the SAR imaging mechanism depends weakly, and at X and C band it depends strongly on the local wind field. This is in agreement with earlier results obtained from the analysis of airborne multifrequency/multipolarization SAR images acquired over the North Sea.
During the two SIR-C/X-SAR missions in April and October 1994, numerous SAR images of the ocean surface were acquired that show different oceanic and atmospheric phenomena. Following Bragg scattering theory [Valenzuela, 1978] the normalized radar cross section from the ocean surface, NRCS, is proportional to the spectral power density of short water surface waves with wavelength B = 0/2sin, where 0 is the radar wavelength and is the incidence angle. Thus, a variation of the spectral power density of these short waves results in a variation of the NRCS, which in turn is visible in SAR images from the ocean surface.
Members of the research group "Satellite Oceanography" of the Institute of Oceanography of the University of Hamburg, Germany, have analyzed SIR-C/X-SAR images with respect to (1) the imaging of rain cells over the ocean by SIR-C/X-SAR and (2) the imaging of ocean waves, particularly its dependence on radar frequency and on local wind conditions. Furthermore, together with researchers from the Communication Research Laboratory of the Ministry of Posts and Telecommunications in Tokyo, Japan, they investigated the imaging of anthropogenic and biogenic surface films by SIR-C/X-SAR and the capability of discriminating between these different surface film types. This report deals with a short summary of the main results of our investigations, starting with the imaging of surface films and continuing with the imaging of rain cells and ocean surface waves.
During the two SIR-C/X-SAR missions in 1994, various surface film experiments were performed in the German Bight of the North Sea and in the Sea of Japan and the Kuroshio Stream region. Since the ocean surface waves, which are responsible for the radar backscattering, are strongly damped by both kinds of surface films, they appear as dark patches on SAR images [Alpers and Hühnerfuss, 1988]. The aim of the experiments was to investigate whether or not a multifrequency SAR system, like SIR-C/X-SAR, is capable of discriminating between biogenic and anthropogenic oceanic surface films. For this purpose, a well-defined surface-active substance, oleyl alcohol (OLA), was deployed on the ocean surface in all test sites and under different environmental conditions, particularly different wind speeds. Furthermore, several SIR-C/X-SAR images of various biogenic and anthropogenic surface films at different places of the world's oceans were analyzed.
SIR-C/X-SAR images of an OLA slick deployed during the German surface film experiment on April 18, 1994, in the German Bight are shown in Figure 1. The three images were acquired at the three radar bands (L, C, and X) and at VV polarization and show the same section of data take 143.30 of SRL-1. It can easily be seen that the gray-level contrast caused by the monomolecular slick is lowest at L band and highest at X band (although the differences between C and X band are less pronounced). In accordance, we found that the measured damping ratio, i.e., the ratio of the backscattered radar power from the slick-free and the slick-covered water surface, is lowest at L band and increases with Bragg wavenumber. Moreover, the damping ratios at L, C, and X band strongly depend on the wind, i.e., they decrease with increasing wind speed. This effect may be explained by means of the source terms of the spectral action, i.e., by the influence of the surface film on energy input into the waves by the wind, wave breaking, and nonlinear wave-wave interaction. The results of another experiment performed in the German Bight show that under high wind speed conditions (12 m/s) different surface films cause similar damping ratios [Gade et al. 1997a].
Examples for SIR-C/X-SAR images of biogenic (natural) surface slicks and for anthropogenic (mineral) oil spills are shown in Figure 2 and Figure 3, respectively. Again, for a better comparison only images acquired at VV polarization are shown herein.
In Figure 2 three subsections are marked with arrows, in order to point out peculiarities of SAR images of natural films: The small slick marked as "a" is almost invisible in the L band image, whereas slick "b" seems to cause the highest contrast at L band. Referring to the L band image area "c" seems to be slick-free, however, the X band image shows that there are several small natural slicks in this area. A freshly dumped mineral oil spill can be delineated in Figure 3 as long dark line with the spilling ship as bright spot at its western (bottom right) end. The bright areas in the image centers are caused by strong rainfall. The oil spill causes only low contrast at L band, whereas the contrast is medium at C band and highest at X band.
Damping ratios were calculated from SIR-C/X-SAR images of natural surface films and mineral oil spills and are shown in Figure 4 (left panel: natural slicks; right panel: mineral oil spills). It can be seen that biogenic (natural) surface slicks cause a strong damping at L band, whereas anthropogenic (mineral) oil spills cause only low L band damping ratios. At C and X band the damping ratios are similar, which, particularly for the C band, may be due to an insufficient signal-to-noise ratio of the SIR-C/X-SAR system.
The two examples shown in Figure 2 and Figure 3 and the presented damping curves measured by SIR-C/X-SAR show that multifrequency SAR imagery is able to yield more reliable information about the damping characteristics of oceanic surface films and, therefore, to make the discrimination between different kinds of surface films under low to moderate wind conditions easier.
The damping ratios obtained by SIR-C/X-SAR were compared with those measured during the German surface film experiments over the same surface films by the 5-frequency/multipolarization scatterometer HELISCAT of the University of Hamburg. It was shown that the damping ratios measured by the scatterometer are higher than those measured by the SAR, and that the observed differences increase with increasing Bragg wavenumber [Gade et al., 1997b].
Finally, we performed polarimetric studies using multipolarization L and C band SAR images of various oceanic surface films (particularly, of the OLA slicks deployed during the dedicated surface film experiments). The results of these studies (not presented herein, see Gade et al. [1997a]) show that the co-polarization signatures for film-free and film-covered water surfaces are similar in most cases and for both radar bands.
SAR images of rain cells acquired during the two SIR-C/X-SAR missions have been analyzed, in order to investigate how rain cells over the ocean surface are imaged by synthetic aperture radars of different frequencies and polarizations. In particular, the influence of strong rain on the backscattered radar power as well as on the phase difference between the HH and VV polarized backscattered radar signals at L and C band have been studied. The aim of this investigation was to obtain better knowledge of the different scattering mechanisms that give rise to radar signatures of rain cells over the ocean. Such knowledge is required for a better interpretation of SAR images of the ocean and for an estimate of the bias in the wind field retrieved from wind scatterometers flown on satellites. For this purpose, a large number of SIR-C/X-SAR scenes acquired in polarimetric mode over the ocean have been analyzed qualitatively and quantitatively. A detailed study is given in Melsheimer et al. [1997].
Figure 5 shows a typical example of radar signatures of rain cells on SIR-C/X-SAR images acquired over the ocean: On this image of the Gulf of Mexico, taken at an incidence angle of 28.3°, irregular bright patches at X band, smaller bright patches at C band, and dark patches at L band are visible. At C band, the brightness contrast between the white patches and their surroundings is most pronounced at cross-polarization. Moreover, at X band, a dark, shadow-like area is seen on the far-range side of the bright patch. The variation of the NRCS in the various frequencies and polarizations as well as the variation of the phase difference between HH and VV polarized signal along a scan line are depicted in Figure 6.
Radar signatures of rain cells over the ocean are caused by a combination of two different effects: The modification of the sea surface roughness and thus the NRCS of the sea surface, and the scattering and attenuation of the microwaves by rain drops and ice particles in the atmosphere (see Figure 7). At L band, damping of the Bragg waves (wavelength in the decimeter range) by rain-induced turbulence in the upper water layer results in areas of reduced NRCS. This was already noticed previously in the L band images from Seasat and JERS-1 [Atlas, 1994; Iguchi et al., 1995]. The effect of rain in the atmosphere on L band microwaves is negligible. At C and X band (Bragg wavelength in the centimeter range), areas of enhanced NRCS are coincident with the areas of reduced NRCS at L band. They are caused by the rain-induced roughening of the sea surface: Laboratory measurements suggest that the net effect of rain drops impinging on the water surface is an increase of the spectral energy waves of X and C band Bragg waves [Bliven et al., 1995]. Our analysis has shown that the NRCS enhancement at C band is much more pronounced at cross-polarization than at co-polarization. At present, we have no explanation for this experimental result.
At C band, and even more at X band, the effect of rain in the atmosphere on microwaves is significant: The attenuation of microwaves propagating through the rain causes dark areas in the SAR image towards the far-range side of the rain cell signatures. The NRCS reduction due to this attenuation in the atmosphere can reach 5 dB at C band and can exceed 10 dB at X band for rain rates above 50 mm/h, which is not uncommon in tropical rain cells.
Since, in SAR images, rain cells give rise to areas of enhanced and reduced NRCS, the wind speed retrieval of wind scatterometers can be biased when the rain cell area fills a large portion of the scatterometer resolution cell. In case of tropical storms, which are associated with very large areas (hundreds of km2) of intense rain fall, this bias could be particularly important. Calculating spatial mean values of the NRCS of 20 km by 20 km patches (about the size of the NSCAT resolution cell), we found that the areas of enhanced NRCS dominate at C band: In areas showing radar signatures of rain cells, the mean NRCS is enhanced by 1 to 4 dB. This results in a positive bias of the wind speed retrieved from a C band scatterometer by 1 to 3 m/s. At X band, this enhancement is weaker (approximately 1 dB). The reason for this is very likely that areas of reduced backscatter (which are more pronounced at X than at C band) partially compensate for areas of enhanced NRCS. Since the attenuation coefficient for rain increases with the radar frequency, the reduction of the NRCS at Ku band due to microwave attenuation by rain is stronger than at X band. As the Ku band Bragg wave frequency is larger than the frequency of maximum ring wave generation by rain, the NRCS enhancement due to these ring waves is not expected to be stronger than at X band. The net effect is then a reduction of the NRCS in the Ku band scatterometer resolution cells in the presence of rain, resulting in a negative bias of the retrieved wind speed.
Furthermore, the propagation of C band microwaves through rain causes a phase shift between the HH and VV polarized backscattered radar signals. At rain rates above 50 mm/h, this phase difference can reach 10° at C band. Since this phase difference is caused exclusively by volume effects, it is this quantity which is most suitable for estimating rain rates. The propagation differential phase shift is larger at X band than at C band, therefore, a fully polarimetric X band SAR would be desirable for measuring the X band HH-VV phase difference and, thus, for obtaining a more reliable estimate of the rain rate.
In addition, we expect that the phase of the X and C band SAR signal is strongly affected by tropical rain cells. Thus the phase information required in repeat-track interferometry may be destroyed when rain cells are present during one overpass. In this case, repeat-track interferometry would not work.
SIR-C/X-SAR images of the ocean acquired at the location of an atmospheric front can be used to study the dependence of the SAR imaging mechanism on the wind, because the wind direction changes abruptly across such fronts.
Figure 8 shows the X, C and L band, VV polarization, SAR images of a scene in the North Atlantic in the vicinity of the North Atlantic Ridge at 49.2° N, 31.0° W, covering an area of 47.7 km 12.6 km. They were acquired simultaneously during the first SIR-C/X-SAR flight on April 15, 1994, at 07:54:58 UTC (data take 96.3). The antenna was looking to the right-hand side of the flight direction and the incidence angle in the center of this SAR swath was = 51.3°. The irregularly-shaped dark band in the center, which is an area of strongly reduced radar backscattering and is particularly pronounced in the X and C band SAR images, is interpreted as the sea surface manifestation of an atmospheric front. This is confirmed by surface weather charts which show a cold front at the location and time of image acquisition. In the X and C band images, the waves on both sides of the front seem to propagate in quite different directions, whereas in the L band image they seem to propagate almost in the same direction. On the right-hand side (area B) of the front, in the X and C band images, the ocean waves seem to propagate approximately in range direction, while on the left-hand side (area A) they seem to propagate approximately at an angle of 33° off the range direction.
The corresponding SAR image spectra for area A and B are shown in Figure 9. Such a large rotation of spectral peaks on both sides of the atmospheric front as observed in the X and C band SAR images cannot be attributed to wave dynamical effects. We therefore consider that this rotation in the SAR image is caused by the SAR imaging mechanism. Indeed, at L band the SAR imaging mechanism depends weakly, and at X and C band it depends strongly on the local wind field [Plant et al., 1983; Brüning et al., 1994; Zurk and Plant, 1996] which changes abruptly across such fronts. This is in agreement with earlier results obtained from the analysis of airborne multifrequency/multipolarization SAR images acquired over the North Sea during the SAXON-FPN experiment [Brüning et al., 1994]. In this investigation it was found that at X and C band the phase of the real aperture radar modulation transfer function (RAR MTF) changes by almost 90°, when, in the reference system moving with the group velocity of the dominant wave, the component of the wind velocity in the direction of the wave propagation changes sign. However, at L band, such a change in local wind direction affects the phase of the RAR MTF only slightly. When using this phase behavior of the RAR MTF in simulations of SAR imaging of ocean waves, we are able to reproduce the SAR image spectra measured on both sides of the atmospheric front by using a Monte-Carlo simulation method. Furthermore, our analyses of the SAR imaging mechanism based on the quasi-linear transform also confirm that this rotation results mainly from the changes in the phase of the RAR MTF across the front. For more details, the reader is referred to Bao et al. [1997]
We have shown that multifrequency/multipolarization SAR imagery is advantageous for detection of oceanic surface films, i.e., important additional information can be inferred from these SAR data. At low to moderate wind speeds multifrequency radar techniques seem to be capable of discriminating between the different surface films whereas at high wind conditions a discrimination seems to be difficult.
The various mechanisms that give rise to radar signatures of rain cells in SAR images of the ocean surface have been investigated. We have shown that the HH-VV phase difference can be used to estimate rain rates. Furthermore, we have pointed out that in tropical ocean areas spaceborne C and Ku band wind scatterometers (as flown presently on the ERS and ADEOS satellites) could yield biased sea surface wind fields in the presence of rain cells.
We have shown that, for inverting X or C band SAR image spectra into ocean wave spectra that contain range or near-range propagating ocean waves, it is quite important to have a good knowledge of the local wind field, whereas for L band SAR image spectra, this is of minor importance.
Further studies of the SAR imaging of atmospheric and oceanic phenomena have been performed, e.g., the imaging of internal waves by SIR-C/X-SAR (herein not presented in detail). It was found that the variation of the radar backscatter, which is caused by internal waves, depends on radar band and polarization as well as on the propagation direction of the internal waves, the wind speed and direction, and the radar look direction.
Summarizing, multifrequency/multipolarization SAR techniques, like SIR-C/X-SAR, are a powerful tool for obtaining additional information on the kind of various atmospheric and oceanic phenomena.
Alpers, W., and H. Hühnerfuss, Radar signatures of oil films floating on the sea and the Marangoni effect, J. Geophys. Res., 93, 3642-3648, 1988.
Atlas, D., Footprints of storms on the sea: A view from spaceborne synthetic aperture radar, J. Geophys. Res., 99, 7961-7969, 1994.
Bao, M, C. Melsheimer, and W. Alpers, Imaging of ocean waves on both sides of an atmospheric front by the SIR-C/X-SAR multifrequency synthetic aperture radar, submitted to J. Geophys. Res., 1997.
Bliven, L. F., P. Sobieski, and T. Elfouhaily, Ring wave frequency spectra, measurements and model, Proceed. Int. Geosci. Remote Sens. Sympos. (IGARSS), 15: Florence/Italy, 830, 1995.
Brüning, C., R. Schmidt, and W. Alpers, Estimation of the ocean wave-radar modulation transfer function from synthetic aperture radar imagery, J. Geophys. Res., 99, 9803-9816, 1994.
Gade, M., W. Alpers, H. Hühnerfuss, H. Masuko, and T. Kobayashi, The imaging of biogenic and anthropogenic ocean surface films by a multifrequency/multipolarization SAR measured during the SIR-C/X-SAR missions, submitted to J. Geophys. Res., 1997a.
Gade, M., W. Alpers, H. Hühnerfuss, V.R. Wismann, and P.A. Lange, On the reduction of the radar backscatter by oceanic surface films: scatterometer measurements and their theoretical interpretation, submitted to J. Geophys. Res., 1997b.
Iguchi, T., D. Atlas, K. Okamoto, and A. Sumi, Footprints of storms on the sea in the JERS-1 SAR image, IEICE Trans. Commun., E-78 B, 1580-1584, 1995.
Melsheimer, C., W. Alpers, and M. Gade, Investigation of multifrequency/multipolarization radar signatures of rain cells over the ocean using SIR-C/X-SAR data, submitted to J. Geophys. Res., 1997.
Plant, W. J., W. C. Keller, and A. Cross, Parametric dependence of ocean wave-radar modulation transfer functions, J. Geophys. Res., 88, 9747-9756, 1983.
Valenzuela, G.R., Theories for the interaction of electromagnetic and oceanic waves - a review, Boundary Layer Meteorol., 13, 61-85, 1978.
Zurk, L. M. and W. J. Plant, Comparison of actual and simulated synthetic aperture radar image spectra of ocean waves, J. Geophys. Res., 101, 8913-8931, 1996.
Figure 1: SIR-C/X-SAR images showing a monomolecular surface film consisting of OLA during a controlled surface film experiment in the German Bight. The images were acquired at L, C, and X band, VV polarization, on April 18, 1994, at 05:26 UTC (SRL-1, data take 143.30; position 54°54' N, 7°50' E; dimensions 5 km by 5 km).
Figure 2: SIR-C/X-SAR images showing biogenic surface films in the North Sea off North Jutland, Denmark. The images were acquired at L, C, and X band, VV polarization, on April 12, 1994, at 07:28 UTC (SRL-1, data take 47.10; position 56°47' N, 7°24' E; dimensions 9 km by 7 km). For a description of the marked areas Ña", Ñb", and Ñc" see text.
Figure 3: SIR-C/X-SAR images of a mineral oil spill in the East Atlantic. The spill is visible as elongated dark line with the oil spilling ship as bright spot at its western end. The large bright areas in the C and X band image centers are due to rain cells. The images were acquired at L, C, and X band, VV polarization, on April 12, 1994, at 07:21 UTC (SRL-1, data take 47.10; position 42°49' N, 26°12' W; dimensions 9 km by 6 km).
Figure 4: Damping ratios calculated from SIR-C/X-SAR images of several biogenic and anthropogenic surface films (i.e., natural slicks and mineral oils spills). The SAR images were acquired during both shuttle missions in 1994 at different places of the world's oceans and at low to moderate wind speeds. The horizontal bars included into both panels denote the Bragg wavenumber coverage of the three radar bands.
Figure 5: SIR-C/X-SAR images of rain cells in the Gulf of Mexico acquired at different radar frequencies and polarizations on 18 April, 1994, at 08:11:22 UTC (incidence angle: 28.3°, position: 25°45' N, 88°35' W). They show radar signatures of several rain cells. The frequencies and polarizations are, from left to right and from top to bottom: XVV, CVV, LVV, CHH, LHH, CHV, LHV.
Figure 6: Range profiles from scene shown in Figure 5, acquired on 18 April, 1994, 08:11:22 UTC. The profiles show the effective NRCS at VV (Panel A), HH (Panel B), and HV polarization (Panel C), and the HH-VV phase difference (Panel D) along the scan line in the X band image on top. The scan line runs in ground range direction from near range to far range. Solid, dotted, and dashed lines denote L, C, and X band data, respectively.
Figure 7: Panel A: Sketch of the scattering geometry of a rain cell imaged by a side-looking radar. Panel B: Variation of the radar backscatter at X band due to volume scattering and attenuation by hydrometeors; Panel C: Variation of the radar backscatter at X band due to the impact of rain on the sea surface; Panel D: Variation of the effective NRCS due to both effects; Panel E: Variation of the effective NRCS at L band; Panel F: Variation of the phase difference between HH and VV polarized signal.
Figure 8: SIR-C/X-SAR images of the same scene over the North Atlantic acquired at different radar frequencies on April 4, 1994, at 7:54:58 UTC. The dark band in the center is the signature of an atmospheric front. The frequencies and polarizations are, from top to bottom, X band, VV polarization, C band, VV polarization, and L band, VV polarization.
Figure 9:
Measured SAR image spectra calculated from the areas A (left panels)
and B (right panels) marked in Figure 8. Panels (a) and (b) show
X band, (c) and (d) show C band, and (e) and (f) show L band SAR
image spectra. The wavenumber in flight direction is kx
, that in ground-range
direction is ky
. Inserted in all plots
are three circles corresponding to wavelengths of 300 m,
200 m and 100 m (outwards from the center).