GNGTS 2013 - Atti del 32° Convegno Nazionale

The morphostructural analysis has been performed for the Po plain, following to the specific formal criteria applied to plain areas where, besides topography, special attention is paid to drainage pattern. A preliminary correlation analysis between the epicenters of past of earthquakes with M ≥ 5.0 and the intersections of MZ lineaments has been performed. The information about earthquakes was selected from the UCI catalog (Peresan and Panza, 2002), covering the time span from 1000 year up to 1 September 2012. We found that the considered events well correlate with intersections of the morphostructural lineaments, namely the morphostructural nodes, thus allowing us to apply pattern recognition algorithms for the identification of nodes prone to earthquakes with M ≥ 5.0 in the Po Plain. Identification of earthquake-prone areas (M³5.0) in the Po Plain by the pattern recog- nition technique. Following a procedure similar to that developed and applied in the Rhone Valley, France (Gorshkov and Gaudemer, 2012), a preliminary identification of the areas prone to earthquakes with magnitude larger or equal to 5.0 (referred as M5+ hereinafter) has been carried out for the Po plain. The nodes have been defined based on the Morphostructural Zoning map (MZ) (Alekseevskaya et al. , 1979; Gorshkov et al. , 2003) shown in Fig. 1. In total, MZ delineated 102 lineament intersections and each of them is treated as a node. Formally the node is defined as circle of radius R=20 km surrounding each point of intersection of lineaments. Such node dimension is consistent with the dimension of earthquake sources with M ≥ 5.0 (Wells and Coppersmith, 1994). The scope of the recognition analysis is to classify all the nodes, delineated within the study region, into one of the following two classes: 1. class D containing the nodes where earthquakes M5+ may occur (namely, the earthquake prone areas); 2. class N including the nodes where only smaller earthquakes may occur. In this work, the nodes prone to earthquakes with M ≥ 5.0 are recognized by a two steps process, composed by a learning stage and a recognition stage. We recall that at the stage of recognition of the seismogenic nodes no use is made of the information about past seismicity, since the recognition is based on morphological, geological and geophysical parameters. At the learning stage, instead, past earthquakes are taken into account so as to identify the most informative parameters to be used for the recognition; in some cases, however, the learning stage is skipped, and the parameters identified for other regions are directly used (e.g. Gorshkov et al. , 2004). Seismicity data. To select the sample nodes for the learning stage, the information on the recorded events with M ≥ 5.0 is taken into account, considering the UCI catalog (Peresan and Panza, 2002). This earthquake catalog covers the time span from 1000 year up to August 2012 and it is representative for events with M ≥ 5.0 from 1500 (Panza et al. , 2010; Vorobieva and Panza, 1993). The location and intensity of the selected earthquakes has been cross-checked with information provided by the NT4.1 (Camassi e Stucchi, 1997) and CPTI11 (Rovida et al. , 2011) earthquake catalogs. Specifically, 71 earthquakes have been considered to select the nodes for the learning stage; 18 earthquakes are situated within the Po plain, while 53 ones are located along the boundaries between the Po plain and surrounding mountains. Their epicenters are mapped in Fig. 2, showing that the considered earthquakes well correlate with intersections of morphostructural lineaments, which are defined neglecting seismicity data. There is just one earthquake, occurred in 1894 (M=5.1), which cannot be associated with a specific node; its epicenter is located between nodes 41 and 42. Definition of the parameters for the recognition process. In order to apply the pattern recognition algorithms, each node is described by a set of parameters, the list of which is provided in Tab. 1. The values of the parameters have been measured for each node using topographic, geological and gravity maps, as well as the MZ map shown in Fig. 1. 89 GNGTS 2013 S essione 2.1