Comments on the Paper “Evidence for high velocity in Koyna Seismic Zone from P‐wave teleseismic imaging” by Srinagesh et al.

Seismicity near Koyna Reservoir, India has been studied for over three decades and it is regarded as a classical example of earthquakes triggered by impoundment of artificial reservoirs [Gupta, 1992]. The paper by Srinagesh et al. [2000] referred hereafter as SN, relates the seismicity at Koyna to intersecting fault geometry inferred from epicentral distribution and high velocity (high rigidity) of rocks, inferred from teleseismic P-wave velocity imaging. However, their results are not convincing due to a variety of reasons including a poor starting velocity model. I elaborate these points here to show that their method oes not seem to hold up against rigorous scrutiny. 1. The abstract states that the P-waves were recorded by a 20-station digital network, but it is mentioned elsewhere in the paper that the stations were used in a leap-frog mode, which means that at any given time, the network consisted only of 10 stations. Since the geometry keeps changing with time, it is not clear how many phases have been recorded by each of these stations at a given location. 2. I find some problems with the starting velocity model. In page 2738 of SN, it is stated that "the velocity of the first layer was determined by inverting the P-wave arrival time data of local earthquakes". Although no previous work is cited, I presume they are referring to the paper by Rai et al. [1999], which uses a velocity model with six layers between 0-12 km and values ranging from 2.9 km/ s to 6.4 km/s. It is not clear how these layers are replaced by a single, 12-km layer of velocity 6.45 km/s, a value much higher than the average. For the second layer (1235 km), a velocity of 6.75 km/s is used, with a reference to Krishna et al. [1989]. However, Krishna et al. suggest, "constant velocity layers with velocities of 6.3 km/s from 11.5 to 19.5 km, 6.7 km/s from 20.5 to 26 km and 6.9 km/ s from 28 to 35.5 km depth" (see page 156 of Krishna et al.). The velocity model by SN, which ignores even the top basalt layer, replaces these multiple layers by a 23km-layer of velocity 6.75 km/so There are two fundamental issues here: one, the authors are not using the model by Krishna et al. although they refer to it and two, the model they use is not representative because it ignores the multiple layers based on Deep Seismic Sounding (DSS) and synthetic modeling.