SEA SURFACE CURRENT FIELDS IN THE BALTIC SEA DERIVED FROM MULTI-SENSOR SATELLITE DATA

摘要

Mesoscale dynamic sea surface features, such as eddies, fronts, or dipoles, are of key importance for our understanding of local dynamics of the marine coastal environment. However, they are often not fully resolved by numerical models currently in use. Series of satellite images (with resolutions ranging from a few meters to hundreds of meters), acquired within a short time period (from less than an hour to a day), can be used to close this gap, if the spatial and temporal extent of those dynamic surface features fits to the spatial and temporal resolution of the sensors and of the data acquisitions, respectively. Moreover, current tracers that are detectable by all applied sensors, need to be present during the whole time of image acquisitions. In this paper we demonstrate the use of multisensor / multi-channel satellite images for the computation of mesoscale surface currents in the Central Baltic Sea. The images were acquired by the Thematic Mapper (TM), the ERS-2 Synthetic Aperture Radar (SAR), the Envisat Advanced SAR, the Wide-Field Scanner (WiFS), and the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) during extensive summer algae (cyanobacterial) blooms in July 1997 (Northern Baltic Proper) and in July / August 1999 (Southern Baltic Proper), and after an oil spillage in May 2005 (north of the Bay of Gdansk). Both natural and man-made surface films affect the sea surface and thus are visible on satellite imagery. We show that, in some cases, data from sensors working at different electromagnetic frequency bands (e.g., TM and SAR) can be used to apply high-speed feature-matching (cross-correlation) techniques for motion detection. In other cases, best results were obtained through the calculation of the optical flow between subsequent images acquired by the same sensor (e.g., WiFS, SeaWiFS, or ASAR). Our computed two-dimensional surface current fields show good agreement with, and they also complement, results from numerical model runs. However, limitations of the proposed methods are due to the strong dependence of the visibility of marine surface films on local weather conditions and to the low availability of satellite data.