Conclusion

A new method to estimate sparse radon panels has been presented. This method is based on (1) the decomposition of the model space into positive and negative values with a bound-constrained optimization technique, and (2) the summation of the two estimated models. This decomposition has the property of reducing the null space and its effects. As illustrated with synthetic and field data examples, this method yields sparse radon panels and compares favorably with the Cauchy regularization technique. Compared to the Cauchy regularization, the proposed method is simpler to parametrize where, for instance, no Lagrange multiplier is estimated. However, more iterations are needed for the bound-constrained approach because two models are estimated independently.