Reply to comment by Melsen et al. on “Most computational hydrology is not reproducible, so is it really science?”

Abstract

In this article, we reply to a comment made by Melsen et al. [2017] on our previous commentary regarding reproducibility in computational hydrology. Re-executing someone else's code and workflow to derive a set of published results does not by itself constitute reproducibility. However, it forms a key part of the process: it demonstrates that all the degrees of freedom and choices made by the scientist in running the experiment are contained within that code and workflow. This does not only allow us to build and extend directly from the original work, but with full knowledge of decisions made in the original experimental setup, we can then focus our attention to the degrees of freedom of interest: those that occur in hydrological systems that are ultimately our subject of study.