Detection of water seepage in river embankments: a case study along the Secchia River (Modena, Italy)

Multispectral technology has been widely used for several decades in the mapping of areas with low vegetative vigor. In particular, the purpose is to detect drier crop areas which will have a lower economic yield, or to identify extended areas in water crisis that can be hazardous due to possible fires. This thesis is born with the intent to use this same technology, but with the opposite purpose: to analyze areas with a higher vegetative vigor, which can be indicative of the availability of higher groundwater content as a consequence of seepage phenomena. The problem of water seepage in natural and constructed contexts has been analyzed with different technologies, radar techniques, geophones, georadar, especially to determine seepages from water pipes. For example, Hadjimitsis et al. (2013) studied the problem of water seepage from pipes, both through hyperspectral and multispectral technology, obtaining interesting results. In this work, instead, we propose to analyze the seepage phenomena below or through a levee, whose cause presumably depends on the presence of an ancient riverbed, which acts as a preferential path for water when the hydrometric level rises as consequence of flood events. The study is tackled by three different perspectives: analysis of pore pressure data along the embankment measured by piezometers in different positions and installed at different depths, analysis of multispectral precision data acquired during three surveys (May and July 2017, April 2018) and multispectral satellite images through Vegetation Indexes (VIs). The data are then examined jointly to obtain a complete view of the phenomenon and to find correlations between them. In particular, a relationship was found between areas with high values of vegetative indices, typical of vigorous vegetation, and therefore of portions of soil with higher moisture content, with the interstitial pressures measured by piezometers following flood events.