Refining Active Source Seismic Refraction Field Configurations for Applications in Critical Zone Studies

The critical zone (CZ) is the area from the top of the weathered bedrock to the tree canopy and is important because of its ability to store water and support ecosystems. The subsurface shape of the CZ has recently been imaged using shallow seismic refraction at several field sites with different lithologies, climates, and tectonic stresses. While efforts are underway to create generalized models for CZ development, a central challenge lies in identifying CZ features that are emergent across different hillslopes and field sites. This challenge exists, in part, because of the uncertainties associated with seismic inversions and the varied methods used to collect seismic refraction data. Here we conduct systematic synthetic tests to determine the optimal field survey configuration to resolve submeter to 30 m depth velocity changes that often indicate soil, regolith, and bedrock. We first use a Fast-Marching method to calculate p-wave arrival times of synthetic velocity structures with ~2 ms Gaussian noise. We then use these arrival times in Bayesian inversions with Markov chain Monte Carlo (MCMC) to generate posterior distributions of velocity structure. Once the misfit of the predicted distributions stabilizes, we are able to gauge the reliability of the velocity structure model by performing further iterations of the inversion and examining model variances. We conduct 1D and 2D synthetic tests varying the minimum receiver spacing, minimum end-shot offset, and arrival time uncertainty. We find that our model is not always able to sharply identify deeper velocity boundaries. Despite this, we are still able to recover the overall structure of our input. In general, a three-meter receiver spacing can resolve sub-meter velocity structure near ground surface, and a minimum of 36 m end-shot offset with 24 receivers is needed to resolve velocity structure at ~30 m depth. By implementing this configuration, we can obtain field data with sufficient seismic ray-path coverage. For arrival time inversion, MCMC with Bayesian approach is ideal for inferring velocity structure and understanding model uncertainty at depth. This work contributes to our ability to reliably image and interpret CZ structure.