Tomographic Inversion Of Crosshole Radar Data: Confidence In Results

Crosshole radar tomography is increasingly being used to characterize the shallow subsurface<br>and to monitor hydrologic processes. Although tomographic inversion provides a subsurface<br>model, confidently interpreting the resulting image can be challenging. We conducted a simple<br>modeling study to better understand the capabilities and limitations of tomographic inversion. We<br>start with a known earth model, simulate a tomography experiment, and invert the synthetic data.<br>We investigate the effects of straight and curved ray approximations to wave propagation, regularization,<br>grid size, and starting model. We also investigate the effects of limited ray coverage<br>through the earth model and noise in the data. Understanding the effects of these different methods<br>and parameterizations will help us place confidence limits on modeled features to more accurately<br>reflect our knowledge of the subsurface.