Abstract : The Danube River and its contributors cross nineteen European countries before reaching the West coast of the Black Sea. In the context of the Black Sea drainage basin, the Danube River is the most important source of liquid and solid discharges. The fluvial-marine contact zone is quite large because Danube splits up itself into multiple distributaries across a wide deltaic plain (4,142 km²) shared among Romania and Ukraine. Vast areas of compact reedbeds usually surround the three hundred freshwater lakes of the delta. Many of these shallow lakes are connected to channels or canals providing turbid water from the Danube, while others depend on flood events to be supplied by the turbid waters coming from the river. Important seasonal growth of floating and submerged macrophytes also occurs in the majority of the lakes. Thus, Danube Delta presents a complex hydrosystem which has been greatly modified since the first navigation works in the nineteenth century. Since the beginning of the nineties, the Danube Delta became a Biosphere Reserve (MAB-UNESCO), a RAMSAR site and was included on the list of the World Natural Heritage (UNESCO). Analyzing spatial patterns in complex environments, like Danube Delta and its coastal zone, requires non traditional approaches. Remote sensing multi-sensor techniques offer reliable advantages to observe and understand intricate processes operating on different space-time scales and especially in large areas with difficult and sometimes restricted access (e.g. Biosphere Reserve core areas). The main methodological objective of this thesis was to create a procedure for processing heterogeneous optical satellite images. By studying the Danube Delta area (hydrosystems principally) through different space-time scales, we worked with several types of high and moderate spatial resolution images. Based on Danube water levels, we analyzed 85 satellite images from the period 2006-2009 (52 medium resolution images - MERIS and 33 high resolution images - ALOS AVNIR-2, SPOT HVIR, Landsat TM / ETM+). This main satellite data set was completed with more than sixty "historical" images (1972-2005) acquired with the Landsat sensors (MSS, TM and ETM+). Three distinct, but complementary, approaches were used to process the satellite images. The integrated analysis of the satellite results with hydrological and meteo-oceanographical data series enabled us to bring new elements of explanation about the past and present dynamics of Danube Delta.