Mapping of seabed topography to and from Synthetic Aperture Radar

ABSTRACT

Models have been developed to simulate Synthetic Aperture Radar images from seabed topography. These models have been inverted and implemented in a so-called Bathymetry Assessment System. Result of research on the radar imaging mechanism as well as the application of the Bathymetry Assessment System in the Dutch coastal waters are presented

1. INTRODUCTION

Synthetic Aperture Radar systems on the ERS-1 and ERS-2 have already shown their capabilities for detecting ships, oil slicks and ocean structures related to spatial variability in the current, with promising results in several investigations on detecting and monitoring eddies, fronts, seabed topography, natural and man-made slicks and other surface phenomena.

Detailed information on seabed topography is important for a lot of coastal and offshore applications like for instance the construction of effective coastal defences. This information is normally obtained by means of advanced ship based echo sounders. However, this technique can be very time consuming and expensive.

The coming decade new satellites will be launched and Synthetic Aperture Radar (SAR) imagery will come available on an almost routine basis. An important application of such data could be the mapping of seabed topography.

To produce a viable SAR based seabed topography product which can be offered to potential users in an operational context quantitative processing and interpretation techniques, and integration with other earth observation data, field data as well as numerical results is required.

With the support of the Netherlands Remote Sensing Program and the European Community new collaborative research and product development programmes, bringing together research organisations, industries and potential users, have been initiated to investigate the use of Synthetic Aperture Radar for the mapping of seabed topography on an operational basis.

In this paper research as well as product development activities will be presented. Emphasis will be put on the development of ocean interaction models describing the imaging mechanism of mapping seabed topography to SAR imagery. Furthermore, the development of the Bathymetry Assessment System (BAS) will be presented in which these ocean interaction models are integrated with model inversion methods and a limited number of field data to map seabed topography from SAR imagery on an operational basis. Finally, results of the BAS will be shown and an outlook to the future will be made.

2. IMAGING MECHANISM

Under favourable meteorological and hydrodynamic conditions (moderate winds of 3 to 5 m/s and significant tidal currents of about 0.5 m/s), air- or space borne Synthetic Aperture Radar (SAR) imagery shows features of the bottom topography of shallow seas (Alpers and Hennings 1984, Vogelzang et al., 1989).

Figure 1. Imaging mechanism from bottom topography to radar backscatter.

The imaging mechanism of mapping sea bottom topography by imaging radar consists of three stages (a more detailed mathematical formulation is given in Calkoen et al., 1993):

Interaction between (tidal) flow and bottom topography results in modulations in the (surface) flow velocity. This relation can be described by several models with an increasing level of complexity: continuity equation, shallow water equations, and the Navier Stokes equations.

Modulations in surface flow velocity cause variations in the surface wave spectrum. This is modelled with the help of the action balance equation, using a relaxation source term to simulate the restoring forces of wind input and wave breaking.

Variations in the surface wave spectrum cause modulations in the level of radar backscatter. To compute the backscatter variations a simple Bragg model can be used, but also available are two-scale and first iteration Kirchoff (Holliday) models.

Based on the above three stage mechanisms, which is also visualised in Figure 1, a suite of computer models has been developed and operationalized at ARGOSS. Models with different levels of complexity and physical detail are available for each step. These models describe the flows, waves and electromagnetic scatter and can be used for a quantitative analysis of radar imagery.

3. EXAMPLE: HELIGOLAND

Within the EU sponsored MAST-II project "Advanced Mapping of Sea Bottom Topography in a Multi-Sensor Approach for Morphodynamic Studies (AMS)**3" a test site called "Heligoland" was selected for inter-comparison of existing imaging models and the validation of model improvements. The test area is located north of the German island of Heligoland in the German Bight. An extensive set of data is available of this site, including depth maps, SAR data and, flow data and hydro-meteorological conditions at the times of image acquisition.

A C-band VV radar image of the Heligoland site has been recorded on November 14, 1990, 06:44 UTC (see Figure 3). At the time of image acquisition the wind speed was approximately 9 m/s from the South-West and the tidal current speed ranged between 0.50 and 0.70 m/s from the East, the exact current velocity depending on the local depth. The depth in the area varies between 1.0 m and 31.3 m. Based on the depth map, the available hydro-meteorological information, and the flight specifications of the aircraft the radar backscatter was simulated by ARGOSS. The result is shown in Figure 2.

The most obvious difference between Figures 2 and 3 is the smooth nature of the simulation and the noisy character of the measurements. Apart from this the figures show a good overall resemblance. Closer inspection of Figure 3 shows some additional features above the deepest areas, which are not present in the simulated image. Given the hydro-meteorological conditions it is concluded that these features are not related to bottomtopography and therefore cannot be explained by the models.

Figure 2. Simulated radar backscatter image of the Heligoland test area. The image covers an area of 1200 m x 1415 m.

Figure 3. Radar backscatter image (C-band VV) of the Heligoland test area recorded November 14, 1990, 06:44 UTC. The image covers an area of 1200 m x 1411.2 m.

4. BATHYMETRY ASSESSMENT SYSTEM

The aim of the Bathymetry Assessment System is to calculate the depth given a radar image and a limited number of sounding data. To achieve this aim the models describing the imaging mechanism have been inverted, using a data-assimilation approach. The assimilation scheme minimises a penalty function, which consists of a term describing the difference between the measured and the simulated radar backscatter and a term describing the difference between the measured and the estimated depth.

Recently the operational level of the Bathymetry Assessment System has been increased considerably by embedding the "tool kit" version, which could only be used by and expert, in a graphical user interface. This user interface guides the processing of the data and eliminates the need for typing complex user commands, thereby eliminating a major source of errors. The graphical user interface has been developed using the fourth generation programming and system development language PV-WAVE. Figure 6 shows one of the screens of the system.

5. EXAMPLE: AMELAND

On the west site of the Dutch Wadden island Ameland two coastal sections are subject to severe erosion. Given the current rate of erosion there is some doubt as to whether or not all sections can be preserved effectively by supplying additional sand. Maintaining the current coastline requires an optimal use of available resources.

Several alternative options has been proposed which may solve the management problems of Ameland efficiently. These options will be analysed by morphological studies, literature searches, field measurements and model calculations. For the execution of the morphological and hydrodynamic model calculations, detailed knowledge of the seabed topography within the project area in the neighbourhood of the tidal channel of Ameland is required. Traditionally, a compilation of sounding data is projected in to the calculation grid. In order to cover the whole grid, sounding measurements of several years are used. In a dynamic area such as the one that contains the tidal channel of Ameland, results of such an approach turn out to be unrealistic. The thus constructed bathymetric map is not only outdated but inconsistent as well. The poor quality of such a bottom topography assessment has a major influence on the quality of the model calculations.

The Bathymetry Assessment System has be applied to radar backscatter images (SAR) and combined with available sounding data of the project area to obtain the complete seabed topography of the project area representing the situation in April/May 1996 (Hesselmans 1997).

For the BAS calculations five ERS SAR images were obtained (see Figure 6). Sounding data collected in 1996 (see Figure 4) were used for calibration, whereas older sounding data were used to construct an initial depth map, which serves as a starting point for the assimilation scheme. The resulting depth map is shown in Figure 6. A analysis of the difference between calculations and measurements shows that the overall error is of the same order of magnitude as the measuring error in the sounding data. In complex two-dimensional situations, such as a bifurcation in a tidal channel, the forward, models described in section 2 fail to simulate the radar backscatter correctly. The the depth estimates may be incorrect.

Figure 4 Calibration lines in the area of Ameland as sailed in 1996.

Figure 5. Bathymetric map of the project area based on ERS SAR imagery and sounding data. The depth range is between -30 m and +5 m.

6. CONCLUSIONS

A comprehensive set of models has been developed and implement to simulate the radar backscatter from the sea bottom topography given the hydro-dynamic conditions. These models have been embedded in a data-assimilation scheme capable of inverting the imaging mechanism. This scheme has been implemented in the Bathymetry Assessment System. A first version of the system has been implemented by Rijkswaterstaat (part of the Dutch ministry of Transport and Public Works) and is currently being tested for routine monitoring of the sea bed topography of the Dutch coastal waters. Steps have been taken to market the BAS system abroad. For example initiatives have been undertaken by a Dutch consortium to demonstrate and implement the system on an operational basis in Indonesia.

REFERENCES

Alpers, W., and Hennings, I., A Theory of the Imaging of Underwater Bottom Topography by Real and Synthetic Aperture Radar. Journal of Geophysical Research, Vol. 89, C6, pp 10,529-10,546, November 20, 1984.

Calkoen, C.J., Wensink, G.J. and Hesselmans, G.H.F.M., ERS-1 SAR imagery to optimize the NOURTEC ship-based bathymetric survey: feasibility study. DELFT HYDRAULICS report H1875, November 1993.

Hesselmans, G.H.F.M., SAR survey Zeegat van Ameland, ARGOSS report A43, November 1996.

Vogelzang, J., Wensink, G.J., de Loor, G.P., Peters, H.C. Pouwels, H., and van Gein, W.A., Sea bottom topography with X-band SLAR, BCRS report, BCRS-89-25.

Figure 6. Three ERS SAR images recorded in the summer of 1996 in quicklook format (100 m x 100 m) of the Ameland project site.

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