A Ku-band ATI SAR Mission for Total Ocean Surface Current Vector Retrieval: System Concept and Performance

Kurzfassung

Direct measurement of ocean currents at sub-mesoscale resolutions is one of the main observational gaps left in the quest to understand the dynamics of ocean ocean processes and ocean-atmosphere interactions. Working towards closing this gap, ESA has initiated a number of studies in order to develop a mission concept, based on a dual-beam along-track SAR interferometer. One of the visible outcomes of these efforts has been the development of the Wavemill concept. In this line, a team consisting of OHB System (Germany/Bremen), DLR (Microwaves and Radar Institute, Germany/Oberpfaffenhofen), MDA (Satellite Systems Unit, Canada/Montreal), and NERSC (Mohn-Sverdrup Center, Norway/Bergen), has been awarded a contract to develop a instrument and mission concept for the measurement of the Total Ocean Surface Current Vector (TOSCV), based on a single-platform dual (or triple) beam ATI-SAR system. This paper will report on some of the outcomes of this study. Like in any other mission design, it is necessary to start with a good understanding of the main challenges appearing when the scientific requirements are confronted with technological and even fundamental physical limitations. Therefore, the first task is to understand what contributes to the total error budget. In general, and thus also specifically for the case of a SAR interferometer, errors can be divided in noise-like random errors, and systematic errors, where in many cases studies tend to, unfortunately, focus the attention on the former ones and underestimating the importance of the later ones. A combination of short baselines (necessarily associated to a single-platform system) and the huge dynamic range of the ocean backscattering (in particular its fast decay with increasing incident angles) makes it difficult to design a system that provides the desired performance (total error in the range of 5 cm/s) for scientifically required swath widths above 100 km. Even assuming an instrument-noise-free measurement, a main factor limiting the sensitivity is the so-called geophysical noise. For ocean measurements, this geophysical-noise is the result of the dispersion of radial velocities associated to the surface waves, which have a typically wind-driven directional spectrum. Engineering studies on ATI system tend to model the performance assuming that the variance of the phase (or velocity) error will be inversely proportional to the number of looks and this looks depend directly on the ratio of product to measurement resolution. However, due to the presence of low wavenumber components in the directional wave-spectrum, the effective number of looks will be eventually upper limited by the product resolution and the wave-spectrum. Systematic measurement errors can be related to time-varying phase offsets between the interferometric channels and by uncertainty in the position of the phase centers due to attitude knowledge uncertainties, mechanical vibrations, and thermal distortions. For electronic phase unbalances the requirement is inversely proportional to the baseline. With baselines in the order of 10 m, phase unbalances must be in the order of 0.2 degree. Small rotations of the platform will result in uncharacterised relative line-of-sight offsets between the phase centers, which directly translate into a measured phase difference. This translates into pointing knowledge requirements more precise than 1 microrad. Moreover, due to the use of squinted beams, and the resulting very large Doppler centroids, small along-track baseline error (caused by either an attitude error or thermal dilation) translate into significant ATI phase errors. An interesting and challenging aspect of the error budget is to go from an analysis in terms of scalars (e.g. the RMS velocity error) to one in the 2-D spectral domain. This approach has been followed, for example, in the SWOT mission analysis. Indeed, the scientific justification of an ocean currents mission is centered around the need to resolve the high wavenumber (small scales) components of the surface currents, as the low wavenumber components are generally well understood and observed with currently available sensors. Also, the spectral power density has a clear wavenumber dependency, with the power density falling quickly as the wavenumber increases. It seems therefore natural to specify and analyse the error budget in the wavenumber domain, allowing larger errors at lower wavenumbers and probably ignoring altogether scales larger than 1000 km. This has a deep and positive impact on the system design, as systematic errors tend to be dominated by low frequency terms. Moreover, systematic errors will vary randomly in azimuth but show up as constant offsets or nearly linear slopes in range. As a result, the spectral support of systematic errors will be concentrated in a small region of the 2-D (directional) wave spectrum. The complex coupling between instantaneous local backscattering and the radial velocity field, results in a sea-state dependent geophysical bias in the observed Doppler velocity. In fact, one of the main challenges is the separation the wind-related signature from the current component. Recent literature suggest that the resulting uncertainty can be strongly mitigated by acquiring dual-polarized measurements. Our baseline measurement concept assumes the use of compact-polarimetry. Under the assumption that the cross-polarized return can be (in most cases) ignored, this allows separation of the desired vertically and horizontally polarized elements of the scattering matrix.