GNGTS 2021 - Atti del 39° Convegno Nazionale

GNGTS 2021 S essione 3.2 458 to the seismic ones, which are defined as any component extracted from geophysical data that can be analysed in order to enhance hidden information ( Chopra and Marfurt, 2005 ). Application of GPR attributes can be found in archaeology and for hydrocarbon reservoirs characterization, while Zhao et al., 2016 and Lu et al., 2020 focused on glaciers. We chose amulti-attribute approach including different attribute categories, such as signal (first derivative), complex (sweetness, envelope, cosine of phase, dominant frequency) and geometrical (variance) attributes. After the interpretation, we applied a Kriging algorithm to obtain a 3D map of the glacier bedrock. Results and discussion FIG. 2 shows six attributes, grouped in three different classes, on an exemplary GPR profile showing both clear detectable and hidden ice-bedrock contact on amplitude display. Amplitude-related attributes, especially first derivative and envelope, proved to be the most effective to our focus, as they both highlight interfaces and discontinuities, enhancing the targets detectability, in comparison with the reflection amplitude. The first derivative is defined as the time rate change of the input trace, while envelope as the total instantaneous energy of the entire analytical trace. FIG. 2B and 2C show that, after the calculation of envelope and first derivative attributes, the ice-bedrock contact can be interpreted more easily even below the scattering facies. Also in sweetness display (FIG. 2D), defined as instantaneous amplitude divided by square root of instantaneous frequency, the ice-bedrock contact can be better identified along almost its whole continuity, thanks to the enhancement of the boundaries between zones with different physical characteristics, stressing on high amplitude ones. Variance is a geometrical attribute, which evaluates the similarity of waveforms in adjacent traces. Usually it is helpful to reveal discontinuities, either related to stratigraphic terminations or structural lineaments, in our case variance does not make a significant contribution to the ice-bedrock contact detection, as it remains hidden in the central part of the high scattered zone (FIG. 2E). Strength of frequency-related attributes, as the dominant frequency attribute (FIG. 2F), lies in their theoretical sensitivity to the intrinsic attenuation of the EM signal, which allows to highlight the changes in spectral characteristics of the EM signal. Therefore, frequency-related attributes are more helpful in the discrimination of different EM facies and their physical meaning ( Forte et al., 2020) , rather than in the ice-bedrock contact identification. Regarding phase-related attributes, we calculated the cosine of phase, which tends to highlight lateral continuity of horizons even with low signal/noise ratio, thanks to its independency from amplitude reflection. However, in the EGZ glacier, the scattered area has such a high number of scattering events that prevents the identification of the ice-bedrock contact by thickly splitting its lateral continuity with scattering phenomena. So, the cosine of phase is not actually helpful to our investigation. Fig. 1 - (A) Location map and an orthophoto of the EGZ glacier, with superimposed the GPR datasets. (B) Processed (not migrated) GPR profile showing clear ice-bedrock contact (black solid line) and hidden ice- bedrock contact (black dotted line). This figure was modified from Forte et al..