GNGTS 2015 - Atti del 34° Convegno Nazionale

the figure shows those horizons which were automatically identified as part of the first three phases of reflections with at least two identified phases, which means that each horizon must have at least one recognized sub-parallel event in order to be displayed. These requirements remove noise-related horizons, but they can also affect areas characterized by interference or overlapping. For example, some recognizable structures in the shallow region (0-15 ns), clearly marked by the algorithm in Fig. 2B, are not automatic ally displayed in Fig. 2C, because the local quality of the recorded signal prevents such horizons from meeting the thresholds set for the phase-based grouping method. The horizons shown in Fig. 2C are the result of a totally automated procedure, with only a few required control parameters and thresholds being set by the interpreter. However, the algorithm is able to identify the base of the runway pavement (H1), made by concrete plates, as well as other features interpreted as natural sedimentary contacts, with the two most apparent layer interfaces marked as H2 and H3. Fig. 2 – Application of the picking procedure to an airport runway 500 MHz CO GPR profile. The figure shows: A) the processed profile; B) all the resulting picked horizons; and C) those horizons automatically identified as one of the first three phases of reflections with at least two recognized phases. Positive amplitudes are marked in green, negative amplitudes in red. The labels H1, H2 and H3 highlight the three main layer interfaces described in the text. Glacier monitoring (GPR) . In this section, the picking algorithm is applied to a CO GPR data set acquired over the Prevala glacieret (Forte et al. , 2014a), which is located at an elevation of 1830-1960 m a.s.l. in the Julian Alps, near the border with Slovenia. The recording system was a ProEx Malå Geoscience GPR equipped with 250 MHz shielded antennas, with a 0.4 ns sampling interval and an average 0.2 m trace spacing. The latter parameter was measured with an odometer that was also used as a triggering device. The recorded profile is shown in Fig. 3A, and contains several features, including a well stratified surface layer (L), a central debris layer (D) producing several hyperbolic diffractions, and older deeper ice reaching the glacieret base 144 GNGTS 2015 S essione 3.3