Sar Amplitude and Insar Coherence for Flood Monitoring: Examples with Cosmo-skymed Data

Flood recurrence frequency is increasing in recent times due to various causes, including climate and land use change. Monitoring of flood events evolution through satellite data is a task of maximal importance to timely provide information about actually flooded areas, escape pathways, or infrastructure damages. Moreover, accurate knowledge of extents and location of flooded areas in past events could help calibrate and improve hydro-geomorphological models, which are used to forecast flood hazard and risk (e.g. [1] and references therein). Thanks to their day/night and all-weather imaging capability, remote sensing radars are a precious tool in flood monitoring. The general working principle is based on the active nature of the sensor: smooth surfaces reflect impinging radiation mainly in the specular direction, while rougher surfaces have a more diffuse scattering pattern. As a consequence, backscatter level is negligible on calm water bodies, as compared to other non-flooded regions [2]. This simple scheme is complicated by the wealth of other factors influencing the actual backscattering level of the terrain, which include surface roughness and soil moisture, for land areas, and the presence of waves due to wind, for water bodies. All these factors get usually critical during flood events, so that correct identification of flooded areas often becomes a difficult task. An aid in resolving these issues comes from interferometric SAR (InSAR) techniques. SAR interferometry is used traditionally to retrieve information about the earth topography, by processing two or more SAR images taken from two close positions in space. An important InSAR product is the so-called coherence, defined as the correlation between the two interferometric images. InSAR coherence carries information about the temporal stability of the microscopic arrangement of the individual scatterers on the terrain. It can therefore be used as an additional information layer in cases in which radar amplitude alone is not sufficient to discriminate flooded and non-flooded areas [3, 4, 5, 6], In fact, InSAR coherence is a valid tool in identifying “false alarms” in flood area monitoring. For example, coherence is expected to be low on areas which have been inundated for some time during the interval between the two InSAR acquisitions, even if they are not covered by water at the time of the first or the second date. Other potential application cases include vegetated fields in which the water level does not exceed vegetation height, and thus gives rise to enhanced backscattering due to double-bounce phenomena [7].