Analysis and interpretation of SAR data for the English Channel

Abstract

The launch of the ERS-1 SAR in 1991 has provided oceanographers with an opportunity to obtain repeatable direct measurements of small scale topography (roughness) of the sea surface. This thesis investigates the application of SAR for monitoring coastal oceanography in the English Channel. That a radar type instrument can detect small scale features, such as tidal fronts and nearshore shoals and banks, was demonstrated by a preliminary qualitative examination of Seasat SAR images. Further investigations of a small coastal front using an airborne altimeter, combined with results from an existing monitoring programme, verified that frontal features can be associated with variations in surface roughness. The quantitative analysis of ERS-1 SAR measures of backscatter in terms of the correspondence of the data to the modulation of short waves is the main subject of this work. The multi-temporal capabilities of ERS-1 SAR have enabled a rigorous analysis of these data, enabling both the performance of the calibration routine and the backscatter relationships with dynamic features to be examined for varied wind and tide conditions. A power loss caused by saturation of the analogue to digital converter (ADC) was found to result in underestimates of backscatter values by up to 3 dB, not previously accounted for. The significance and effectiveness of using the fullest possible correction was demonstrated by comparing SAR-derived estimates of backscatter across range with predictions from an empirical wind retrieval model (CM0D4), which describes the dependence of backscatter on incidence angle. It was found that in the coastal zone the backscatter signal is dominated by transitional wind effects, which often confuse or obscure any dynamic sea surface signatures. Estimates of the wind speed contribution to surface roughness were successfully achieved using the CM0D4 to accuracies within the error bands of the model (±2 ms" ). Comparisons of SAR-derived wind speeds between images demonstrated the importance of correcting for the effects of the SAR viewing geometry. It was empirically shown that a normalised SAR-derived apparent wind speed residual gave a proportional representation of surface roughness that could not be achieved using a direct measure of backscatter. It is shown that the non-linearities introduced into the data by these effects should be removed if comparisons of the backscatter signature of surface roughness variations are to be made between images, with physical variables, such as currents, or with model predictions.

More information

Identifiers

Catalogue record

Export record

ASCII CitationAtomBibTeXData Cite XMLDublin CoreDublin CoreEP3 XMLEndNoteHTML CitationHTML CitationHTML ListJSONMETSMODSMPEG-21 DIDLOpenURL ContextObjectOpenURL ContextObject in SpanRDF+N-TriplesRDF+N3RDF+XMLRIOXX2 XMLReferReference ManagerSimple Metadata