FLOODPLAIN WETLAND DYNAMICS IN THE DANUBE DELTA FROM REMOTELY SENSED IMAGERY

Simona Niculescu, Cédric Lardeux, Grégoire Mercier, Laurence David
Laboratoire Géomer/ Institut Universitaire Européen de la Mer / UBO, France

Abstract

Hypotheses on global warming and its consequences are pessimistic in terms of the increase of intense pluviometric phenomena and numerous uncertainties affect the pluviometric response to warming. In this context, during the coming years, climatic change will increase the number and diversity of flood risk events and other phenomena related to the water cycle. Extreme meteorological phenomena such as heavy rains causing flash floods could become more frequent in all Europe. In the Danube delta, flood risk is a major natural risk. Since 2000, this coastal wetland has been seriously hit by floods in 2002, 2005, 2006, 2009 and 2010. Based on hydrological and satellite observations of the bicentennial floods of summer 2010 and the drought of summer/autumn 2011, this paper deals with the issue of forecasting by knowledge and assessment of the flood risk in Danube delta. One of the major objectives of this methodological work consists in the exploration and evaluation of several data types relevant to this type of analysis, i.e. optical and radar satellite data from the Japanese ALOS and Canadian RADARSAT-2 sensors in order to propose regional indicators that can be used for the forecasting of flood risk in the floodplain wetland of the Danube delta. This procedure will provide a framework for flood mapping, monitoring flood area potential and, the detection and characterization of the floodable. Using 20 ALOS and RADARSAT-2 radar and optical images, this paper proposes first a series of change indicators for land cover during the floods of 2010, before the floods (of 2009) and afterwards, and during the 2011 drought. Using the two polarization modes (dual and full) of the radar sensors, these change indicators are applied first to the polarimetric indicators (intensity, entropy, span, etc.) derived from the different significant data of extreme events in the delta (2010 flood and 2011 drought). By taking advantage of the complementary of the data of the two sensors and by associating the different spatial resolutions (from 10 m to 3m), we establish wavelength change comparisons depending on the correspondence between the images acquired by the two satellites. Using a multi-modality methodology based on radar image processing (dual versus full polarimetric data, multi-angular, multi-frequency and multi-resolution data) it is possible to analyze the impact of the wavelength, the incidence angle or the polarimetric mode, but also the impact of the resolution in SAR image analysis of floodable areas. The second category of indicators to be suggested for the Regional library will be composed of the spectroradiometric measurements of the primary objects characteristic of floodable areas (the main species that compose the vegetation cover, the main crops, the main construction materials in urban areas, etc.) in the delta. The portable spectroradiometer permits the measurement of in situ reflectance, significantly improving the spectral characterization of the spectral classes detected on the optical images (ALOS). The risk indicators resulting from this research will contribute to the implementation of regional models and scenarios in terms of flood risk thus providing a useful tool for decision-making.