Adaptive Sampling of Potential-Field Data: A Direct Approach to Compressive Inversion

Foks, Leon (Colorado School of Mines) | Krahenbuhl, Richard (Colorado School of Mines) | Li, Yaoguo (Colorado School of Mines)



In this paper, we present a direct and reliable approach to the adaptive down-sampling of potential-field data for large inversion problems. In contrast to traditional down-sampling methods, the approach significantly reduces the number of data parameters in relatively smooth/quiet regions of the data, while preserving the signal anomalies that contain the relevant target information. This allows for a simple and effective approach for compressive inversion of large datasets, without the need for large computing power, while maintaining the resolution of the recovered structures. The formulation has the flexibility to decimate large data sets for an optimal balance between data number and signal shape, or data can be decimated more conservatively if desired. We first present the data adaptive down-sampling technique, and then demonstrate the approach, applied to both synthetic and field data.