Giacomo De Carolis, Francesca De Santi
Institute for Electromagnetic Sensing of the Environment, National Research Council of Italy (CNR-IREA), Milan (Italy)
Research theme aims
The integration of EO SST, back-scattering and wave motion products is an important contribution to the mesoscale circulation modelling [1]. The capability of SAR systems to detect the spectral and directional properties of ocean waves has been widely demonstrated over the past decades [2]. Much less has been done for sea wave monitoring in coastal areas because of the complex geophysical interactions that may occur in the coastal environment. However, SAR observations of surface waves at a proper distance from the shoreline can be usefully employed as input to wave models to better describe coastal wave transformation due to complex bathymetry changes and refraction effects that may affect coastal structures [3].
This research theme aims to assess SAR inversion methodologies developed for the open ocean for semi-enclosed basins such as the Mediterranean Sea and the Italian seas in particular.
Materials and Methods
Study areas
Two case studies have been considered for this research theme.
The first one, in the vicinity of La Spezia (Ligurian Sea), exploits the availability of wave buoys data synchronous to SAR as benchmarks to validate the wave inversion procedure. Indeed, since 1989, all around the Italian coasts a wave buoy network named RON (Rete Ondametrica Nazionale) is routinely operating to monitor the sea state. Managed by ISPRA, RON operationally collects the directional wave spectrum, the sea surface temperature, and meteorological parameters like wind speed and direction, air temperature and barometric pressure.
To be synergic with theme 2.2 and to provide the wave state impinging on shorelines particularly exposed to coastal erosion, we considered the area of Piscinas (Sardinia) as the second case study. It represents a further example of SAR as a tool to evaluate the impact of wind/waves storms on the observed sandy coastlines changes.
Remote sensing data
For the Piscinas test site, the Sentinel-1B SAR image gathered on 6 May 2019 at 05:28 UTC has been processed. The image acquisition is concomitant to a severe wave state with, according to the wave model (WAM), a significant height up to about 5 m.
Methodology
It is assumed that the wave spectrum primarily consists of the wind-generated waves and, if present, of swells, which are waves coming from remote regions other than the area of wind-wave generation. The estimation of wind-generated ocean spectrum is based on the inversion procedure proposed in [4]. To this end, preliminary knowledge of the wind vector should be available. This task can be carried out by exploiting the SAR potential to estimate the wind speed over the sea surface provided that the SAR image intensity is radiometrically calibrated. According to a Bayesian inversion procedure as described in [5] wind information is retrieved from the SAR image by using the ECMWF wind vector as the first guess.
The parametric representation of the wind sea spectrum is given after Donelan et al. (1985) [6]. Such wave spectrum depends on the inverse age of the dominant wave. The residual wave spectrum is estimated by assuming again a parametric representation of directional swells according to the JONSWAP-Glenn spectral shape coupled with directional spreading function of the Mitsuyasu type, properly extended to account for swell propagation [7].
The corresponding wavenumber SAR image spectrum is estimated from a tile, typically 512 x 512 pixels in range and azimuth. The simulated SAR image spectrum is computed using the closed-form expression of the non-linear ocean-to-SAR spectral form described in [8] and later modified for the SAR image cross-spectral transform [9].
The implemented SAR retrieval adopted a non-linear retrieval scheme which minimizes a cost function with the truncated Newton method implemented in IDL© for the following parameters: the dominant wind wavenumber vector; the dominant swell wavenumber vector; the swell wave height; the shape parameters and the directional spread of the swell distribution.
Results and Discussion
Comparison with wave buoy data
The wind speed in a region surrounding the buoy location was estimated by using the ERS–1 SAR image. SAR backscattering data relevant to the selected area including the buoy were inverted concerning the wind speed using the empirical models CMOD–4 and CMOD–IFREMER. The estimated wind speed at buoy location ranges from 11m/s, as predicted by CMOD–4, to about 12.5m/s, the latter being the CMOD–IFREMER prediction. SAR inversion runs were carried out using as input the SAR estimated wind speed and direction. The search of the best fit inverse wave age W was constrained in the range [0.1, 5.0] to reproduce the field observations by using a constrained optimization scheme implemented in IDL. Figure 1 shows the results. Taking the SAR–retrieved wind speed as the average value between CMOD’s predictions (11.5m/s) and the dominant wave direction equal to buoy observation, the inverted wave height spectrum was 2.87m, a value consistent with that observed at 09:00 (3.00m); in contrast, the inverse wave age was W=0.95, which is a value close to the mature stage of growth but higher than the buoy observations W£0.69).
Figure 1: Wave spectra measured and retrieved at La Spezia site on 16 September 1994. SAR wave spectra inversion results for La Spezia site using the averaged SAR estimated wind speed (11.50m/s). The best fit inverse wave age value is 0.95. All spectra refer to location (43.93°N, 9.83°E).
Piscinas case study
A severe wave state with significant height up to about 5 m with wind field as high as 14 m/s were predicted by the wave model (WAM) and atmospheric model running at ECMWF on 6 May 2019. The SAR inversion procedure was run on this tile by assuming a wind speed of 16 m/s and direction 35 deg relative to the SAR look direction. The upper right panel in Figure 2 shows the best fit simulated SAR spectrum. the SAR analysis conducted on the analysed case study confirms the WAM prediction of severe wave storm impinging on the shoreline. SAR retrieved significant wave height was strikingly in accord with WAM model within few centimetres as well as the retrieved parameters of the dominant wave consistently confirmed the WAM results within the SAR sampling variability.
Figure 2: (upper left) two-dimensional wavenumber S1 SAR spectrum; (upper right) retrieved wavenumber Sar spectrum; (bottom left) two-dimensional wavenumber wave spectrum; (bottom right) uni-dimensional wave frequency spectrum (red) retrieved from the SAR inversion procedure and (black) WAM output. All spectra refer to location (39.5N, 8.25E)
Future perspectives
The proposed SAR inversion procedure seems appropriate to capture wave features occurring in the Italian seas. As a future activity, we plan to carry out a further comparison with wave buoy data over the Mediterranean Sea to assess the methodology for a selected study area. The goal is to provide SAR inverted wave spectra as input to an ocean model to improve the performance predictions on the impact of waves on shorelines.
Results on the Piscinas site obtained with different approaches within this project (see themes 2.2) highlighted a temporal variability of the coastline morphology. Systematic SAR wave observations can be used to assess the impact of wave storms on the changes of the sandy shoreline observed by optical images and using the interferometric coherency from multi-temporal SAR imagery. At the moment the only conclusion that can be drawn from the described cases study is that SAR can reliably observe stormy waves around the Italian coasts and thus provide a useful tool for effective monitoring of sea/landscape changes.
Publications and Presentations
- N. Ghirardi, M.Bresciani, G. Luciani, G. Fornaro, V. Zamparelli, F. De Santi, G. De Carolis, C. Giardino, “Mapping of the risk of coastal erosion for two case studies: Pianosa island (Tuscany) and Piscinas (Sardinia)” In book Eighth International Symposium “Monitoring of Mediterranean Coastal Areas. Problems and Measurement Techniques”, Vol.126, pag. 713-722, Editore Firenze, 2020.
- M. Bresciani, N. Ghirardi, G. Fornaro, V. Zamparelli, F. De Santi, G. De Carolis, D. Tapete, M. Palandri, C. Giardino, “Mapping Of The Risk Of Coastal Erosion For The Case Study Of Piscinas (Sardinia)”, In Proc. IGARSS 2021
References
- Falcini, F., Khan, N. S., Macelloni, L., Horton, B. P., Lutken, C. B., McKee, K. L., … & D’Emidio, M. (2012). Linking the historic 2011 Mississippi River flood to coastal wetland sedimentation. Nature Geoscience, 5(11), 803.
- Hasselmann, K., Chapron, B., Aouf, L., Ardhuin, F., Collard, F., Engen, G., Hasselmann, S., Heimbach, P., Janssen, P., Johnsen, H., Krogstad, H., Lehner, S., Li, J.-G., Li, X.-M., Rosenthal, W., and Schulz-Stellenfleth, J. (2012). The ERS SAR Wave Mode – A Breakthrough in global ocean wave observations. In: ERS Missions – 20 Years of Observing Earth. ESA Scientific Publications, SP-1326. ESA., 1-38.
- Collard, F., Ardhuin, F., and Chapron, B. (2005). Extraction of coastal ocean wave fields from SAR images. IEEE J. Oceanic. Eng., 30 (3), 526—533.
- Mastenbroek, C. D., & De Valk, C. F. (2000). A semiparametric algorithm to retrieve ocean wave spectra from synthetic aperture radar. Journal of Geophysical Research: Oceans, 105(C2), 3497-3516.
- Portabella, M., Stoffelen, A., & Johannessen, J. A. (2002). Toward an optimal inversion method for synthetic aperture radar wind retrieval. Journal of Geophysical Research: Oceans, 107(C8).
- Donelan, M. A., Hamilton, J., & Hui, W. (1985). Directional spectra of wind-generated ocean waves. Phil. Trans. R. Soc. Lond. A, 315(1534), 509-562.
- Goda, Y. (2010). Random seas and design of maritime structures (Vol. 33). World Scientific Publishing Company.
- Hasselmann, K., & Hasselmann, S. (1991). On the nonlinear mapping of an ocean wave spectrum into a synthetic aperture radar image spectrum and its inversion. Journal of Geophysical Research: Oceans, 96(C6), 10713-10729.
- Engen, G., and Johnsen, H. (1995). SAR-ocean wave inversion using image cross spectra. IEEE Trans. Geosci. Rem. Sens., 33, 4, 1047-1056.1