Attenuation Characteristics of Northwestern Himalaya and Western Tibet Using Lg-coda Waves

We use 611 Lg-coda waveforms from regional earthquakes (epicentral distance<1000 km and Mw>5) recorded by JAKSNET to characterize attenuation in Northwestern Himalaya and Hindukush region. Lg-coda waves are modelled as single back-scattered energy, sampling an ellipsoidal volume. The attenuation of Lg-coda is quantified using the quality factor (Q), which is frequency dependent. We use the stacked spectral ratio (SSR) method to calculate the single-trace Lg-coda Q at 1 Hz (Qo) and its frequency dependence (). A moving-window stack of scaled-logarithmic ratios of spectral amplitudes, for window length of 25.6 s and different central lapse time, is computed for each frequency. We perform a linear regression of log(stacked spectral ratio) and log(frequency), using least-squares fitting, to obtain the slope and intercept as (1-) and log(Qo), respectively. Lg-coda is selected in a frequency range of 0.2-6 Hz, with coda window starting at 2.8 km/s and coda length of 256.0-307.2 s. We obtain Qo and values of 276-882 (mean 42883) and 0.077-1.311 (mean 0.750.16), respectively. We plot the Qo and as a function of epicentral distance to observe no systematic variations. Our dataset also contains a significant number of deep earthquakes (depth>60 km) constituting 247 ray paths. These deep events produce an average Qo and of 39567 and 0.740.09, respectively, which lie within the bounds of our overall measurements. Our results show that the ray paths with low Qo (<350) pass through the Hindukush region mostly, while those with high Qo (>550) are from the Tarim Basin or the Himalaya. We intend to add more measurements from western Tibet and central Himalaya, using data from different networks and will perform a back-projection tomography to study the lateral variation of Qo and across northwest Himalaya and western Tibet.