On the remote sensing of oceanic and atmospheric convection in the Greenland Sea by synthetic aperture radar

摘要

In this paper we discuss characteristic properties of radar signatures of oceanic and atmospheric convection features in the Greenland Sea. If the water surface is clean (no surface films or ice coverage), oceanic and atmospheric features can become visible in radar images via a modulation of the surface roughness, and their radar signatures can be very similar. For an unambiguous interpretation and for the retrieval of quantitative information on current and wind variations from radar imagery with such signatures, theoretical models of current and wind phenomena and their radar imaging mechanisms must be utilized. We demonstrate this approach with the analysis of some synthetic aperture radar (SAR) images acquired by the satellites ERS-2 and RADARSAT-1. In one case, an ERS-2 SAR image and a RADARSAT-1 ScanSAR image exhibit pronounced cell-like signatures with length scales on the order of 10-20 km and modulation depths of about 5-6 dB and 9-10 dB, respectively. Simulations with a numerical SAR imaging model and various input current and wind fields reveal that the signatures in both images can be explained consistently by wind variations on the order of ±2.5 m/s, but not by surface current variations on realistic orders of magnitude. Accordingly, the observed features must be atmospheric convection cells. This is confirmed by visible typical cloud patterns in a NOAA AVHRR image of the test scenario. In another case, the presence of an oceanic convective chimney is obvious from in situ data, but no signatures of it are visible in an ERS-2 SAR image. We show by numerical simulations with an oceanic convection model and our SAR imaging model that this is consistent with theoretical predictions, since the current gradients associated with the observed chimney are not sufficiently strong to give rise to significant signatures in an ERS-2 SAR image under the given conditions. Further model results indicate that it should be generally difficult to observe oceanic convection features in the Greenland Sea with ERS-2 or RADARSAT-1 SAR, since their signatures resulting from pure wave-current interaction will be too weak to become visible in the noisy SAR images in most cases. This situation will improve with the availability of future high-resolution SARs such as RADARSAT-2 SAR in fine resolution mode (2004) and TerraSAR-X (2005), which will offer significantly reduced speckle noise fluctuations at comparable spatial resolutions and thus a much better visibility of small image intensity variations on spatial scales on the order of a few hundred meters. Copyright 2004 by the American Geophysical Union.