Presentación de resultados

Despite the large magnitude of the M7.8 Gorkha earthquake, and the very short epicentral distance of downtown Kathmandu, the ground motions recorded in the middle of the basin (Figure 3-7) had a very low peak ground acceleration PGA = 0.16g and a very long period (5s) predominant pulse (Figure 3-8). Unfortunately, the earthquake sequence was poorly recorded, and only one strong motion instrument (KATNP) managed by the USGS has so far provided ground motion time-series over the bandwidth of engineering interest. As a consequence, the factors that shaped the amplitude and frequency content of the mainshock are still poorly understood, particularly in the short period range of engineering interest (>0.5 Hz). At longer periods, continuous GPS stations of the Caltech Tectonic Observatory provided an excellent approximation of the rock outcrop ground shaking, and have revealed the extent to which source and path effects contributed in shaping the amplitude and frequency characteristics of the strong motion record at KATNP. The data were used by JPL/ARIA and are archived at UNAVCO

Figure 3-7 shows the location of the USGS KATNP strong motion station (27°42'43.20"N, 85°18'57.60"E). As it will be shown in Chapter 4, KATNP is located on top of the thickest layer of the Katmandu lake deposits, estimated between 200-300m.

Figure 3-7 Location of the strong motion station KATNP (27.730 N, 85.336 E), the only instrument that has so far provided acceleration time series, and two nearby 5 Hz GPS sites

KKN4 (27.8007 N, 85.2788 E) and NAST (27.6567 N, 85.3277 E).

In Figure 3-8 we show the 3-component acceleration time series of the mainshock at KATNP, where one can clearly see the long period predominant pulse of the two horizontal components; the comparatively higher frequency content of the vertical component; and what appears to be a very early aftershock at 270s (S. Hough, personal communication). In Figure 3-9, we next show the Fourier amplitude spectra of the series, which reveal that the vertical component also has significant energy at 0.25Hz, and a lower rate of high frequency decay compared to the horizontal components. To further analyze the evolution of frequency content of the strong motions as a function of time, we also plot in Figure 3-10 the spectrograms of the three mainshock components, where we can clearly see the long period 5sec pulse reverberating for several cycles (denoted by white dashed line on the vertical axis); and the secondary pulse that arrives approximately at 270s.

Figure 3-8 Strong motion time series of the 04/25/2015 M7.8 Gorkha mainshock recorded at KATNP.

Figure 3-9 Acceleration Fourier spectra of the 04/25/2015 M7.8 Gorkha mainshock recorded at KATNP.

In Figure 3-11 we compare the response spectra of the mainshock, the M7.3 05/12/2015 aftershock, and the 6 next strongest aftershocks listed in Table 3-1. We here observe two features: (i) in the low period (high frequency) range of the response spectra, the mainshock peaks at 0.47s, the M7.3 aftershock at 0.43s and the weaker aftershocks within the range 0.2-0.3s, a period elongation suggestive of nonlinear site response; (ii) the horizontal components of the mainshock appear to be the only ones with energy in the 5s period range, while the aftershock long period peaks are at much shorter periods (1-3s). From the observation alone, one could assume that the 5s pulse is purely a source characteristic of the maishock. However, as we will show in Chapter 4, source and near-surface path (basin) effects most likely contributed to the frequency content of the mainshock, although further research is required to verify our hypothesis.

Table 3-1 Six aftershocks analyzed in Figure 3-11, in addition to the Mw7.8 mainshock and the strongest Mw7.3 aftershock.

Figure 3-10 EW, NS and UD component spectrogram of the 04/25/2015 M7.8 Gorkha mainshock at KATNP.

Figure 3-11 Response spectra of mainshock and 7 strongest aftershocks of the 04/25/2015 M7.8 Gorkha mainshock sequence recorded at KATNP.

It is interesting to compare the records of stations KKN4, KATNP and NAST, located 68, 77, and 82 km from the epicenter respectively. KKN4 is situated in the western hills outside of Kathmandu basin (Figure 3.6), and KATNP and NAST are situated within the basin. KKN4 and NAST are 5 Hz GPS time series processed from ALOS-2 data (Galetzka et al., 2015). The raw GPS data is provided by the Nepal Geodetic Array run by the California Institute of Technology, and the Department of Mines and Geology (Nepal). The Scripps Permanent Array and Observation Center (SOPAC) processed the raw GPS data to displacement time histories, which is included in ascii format in the appendix of this report. Station KATNP is the USGS strong motion station at Kathmandu discussed previously. Figure 3.11 compares the three

component velocity records from the three sites. Station KKN4 has the simplest and lowest amplitude waveforms, consisting of the pulse-like S-wave with a duration of 5-6 seconds. The other stations show amplification and extension of the duration due to the low velocity sediments of the Kathmandu basin. KATNP shows secondary pulses that are basin generated surface waves, that judging from the lack of similar motions on the vertical components are Love waves, although the particle motions in the horizontal plane are not linearly polarized indicating complex 3D propagation of the basin generated surface waves. At station NAST, at the southern edge of the basin the surface wave duration lasts more than 40 seconds with a predominant period of approximately 4 seconds. There are 5 cycles with motions at or above 50 cm/s. There is a second packet of surface wave energy that likely comes from interactions of the wavefield with the edges of the basin.

Figure 3-12 Comparison of three component velocity records for the sites outside and within the Kathmandu basin. Strong Love wave signals are generated by basin edge effects and 3D

propagation within the basin sediments.

As the three Kathmandu sites are located approximately 10 km from the rupture plane the long-period displacement records are complicated through the interference of near-, intermediate- and far-field waves. As shown in Figure 3-12 the later long-period reverberations are likely surface waves propagating in the basin. In displacement all sites exhibit large static deformation, including the accelerometer site KATNP although recovery of the static is difficult due to the double integration. Figure 3-13 compares the three-component KKN4 displacement time series from 5 Hz GPS observations. The horizontal records have been rotated into fault parallel (FP, azi=295) and fault normal (FN, azi=25) components. All three components show large static offsets in displacement. The static displacement is controlled by the intermediate-field term, which is proportional to the moment time history, or the slip time history on the fault.

From the vertical and FN components the slip rise time may be directly determined from the initiation of the development of the static offset to the time that the stable offset is achieved.

The dashed lines show the inferred rise time, which also correlates with the beginning and end

be approximately 7.4 seconds. A minimum estimate can be inferred from the times that vertical velocity pulse rises from and falls to zero. The measured rise time is consistent with the interpretation that the long-period energy observed in the mainshock response spectra is primarily due to source.

Figure 3-13 Comparison of 5 Hz GPS vertical, fault-normal (FN) and fault-parallel (FP) displacement time histories. The blue trace compares the vertical component of velocity. The

dashed lines show the inferred slip rise time from zero to static offset.

The long-period slip velocity pulse in of itself is not an explanation for the low shorter period ground motions at KATNP (Figure 3-11) though it does explain the high long-period response.

The short-period motions could be low for a variety of reasons including; 1) the aforementioned subdued nature of directivity focusing in dip-slip earthquakes; 2) the possibility of a relatively smooth rupture process leading to weak radiation at short-periods; and 3) seismic attenuation properties of the region. Future work is needed to reconcile these possibilities.