5 Results and Discussion
We calculate stress drop for earthquakes at Parkfield using a variety of different approaches to correct for spatially and temporally varying material properties, in an attempt to resolve the real spatial and temporal variation in earthquake sources. We use the shallow borehole (HRSN) recordings, and only include earthquakes recorded by at least 5 stations with high signal to noise ratio over the frequency range 2-60 Hz; this is a relatively wide frequency range compared to many previous spectral-decomposition studies (e.g., AS2007; Trugman, 2020; Trugman & Shearer, 2017). We apply the self-adaptive SNSS method, that does not assume any magnitude scaling, to calculate the empirical correction spectra (ECS) and isolate the source spectra. We compare our results to those of recent studies of stress drop at Parkfield including the large-scale study of AS2007 and the smaller scale analysis of Abercrombie (2014). We begin by discussing the trends, resolution and uncertainties in the data set as a whole, and then consider the spatio-temporal patterns and their reliability.