2) One example of this is the regularization procedure, which is not very clear. One particular issue concerns the tradeoffs between resolution (in z), and the degree of regularization. Do we really have 10 m resolution in z (as is implied by the 10 m bins)?
3) The paper needs to discuss the differences in approach and in results between the current analysis and the previous ice-model paper. The differences between the two models seem larger than the estimated systematic errors in the two methods. This implies that the systematic errors for at least one of the two papers are underestimated, or some systematic difference is not included. This needs to be discussed; we cannot just ignore the differences from a paper that most of us (you included) signed.