Programming and machine learning play an increasingly important role as part of research methods in most scientific disciplines, including geosciences, as the volume of data collected increases exponentially. In hydrogeology, it is used for data pre-processing, analysis, prediction, or visualization. However, when trying to apply or adapt promising methods to one’s own data, it is often difficult or even impossible because of poor documentation. Therefore, it is essential to make use of state-of the-art methods for the documentation of research methods that include code to maximize reproducibility, transparency and the ability of collaboration in hydrogeology. The quality of the documentation is also fundamental for a fast transfer of knowledge, methods and related errors.
This state-of-the-art documentation has been developed and applied for several years mainly in computer science and data processing. At the lowest level, this includes the supplementary publication of well-named and structured script files according to ideally existing conventions and at the highest level the publication of a socalled Jupyter notebook on a development platform such as Bitbucket, Github or others. Easily reproducible research not only requires high-quality documentation, it begins with a shift from hard-to-document analysis tools such as spreadsheets to programming languages for data science and geospatial data processing. The most common documentation tools from neighbouring disciplines will be presented and discussed to make hydrogeological research more accessible and accelerate knowledge transfer. Especially because speed in gaining new insights into the consequences of climate change is a key component.