Microseismic events occur as a result of both natural and man-made processes, and their spatial distribution is important to understanding the mechanisms involved in creating this seismicity. In order to accurately locate microseismic events, a calibrated velocity model is required, but acquiring one is often a challenge. Since both the microseismic source location problem and velocity model inversion problem are wave-based, full waveform inversion (FWI) can be used to solve both problems. In this report, we propose a modified FWI scheme that iteratively solves for both source distribution and the velocity model.
We show numeric examples in a 2D time-domain acoustic framework developed in Python. Examples of the gradients for both source location and velocity model are shown in a variety of scenarios in order to begin to understand their behaviour on their own, as well as the potential cross-talk between them. We see that low frequencies are required for the source-term gradient to produce a correct update. This update has the form of a dipole, with a negative lobe at the incorrect source location and a positive lobe in the direction of the correct source location.
View full article as PDF (5.65 Mb)