MEDUSA is a time-dependent one-dimensional numerical model of coupled early diagenetic processes in the surface sea-floor sediment. In the vertical, the sediment is subdivided into two different zones. Solids (biogenic, mineral, etc.) raining down from the surface of the ocean are collected by the reactive mixed layer at the top. This is where chemical reactions take place. Solids are transported by bioturbation and advection, solutes by diffusion and bioirrigation. The classical coupled time-dependent early diagenesis equations (advection-diffusion reaction equations) are used to describe the evolutions of the solid and solute components here. Solids that get transported deeper than the bottom boundary of the reactive mixed layer enter the second zone underneath, where reactions and mixing are neglected. Gradually as solid material gets transferred here from the overlying reactive layer, it is buried and preserved in a stack of layers that make up a synthetic sediment core. The composition of the two phases, the processes (chemical reactions) and chemical equilibria between solutes are not fixed any more, but get assembled from a set of XML based description files that are processed by a code generator to produce the required Fortran code. 1D, 2D and 2Dx2D interfaces have been introduced to facilitate the coupling to common grid configurations and material compositions used in biogeochemical models. MEDUSA can also be run in parallel computing environments using the Message Passing Interface (MPI). MEDUSA requires two external libraries to compile, both available from Zenodo: THDYCT (doi:10.5281/zenodo.4677790) µXML (doi:10.5281/zenodo.4677788) This upload provides a bug fix and code sanitization release of the previous version (v. 2.0) which was published as part of the supplement to the paper: Guy Munhoven (2021) Model of Early Diagenesis in the Upper Sediment with Adaptable complexity – MEDUSA (v. 2): a time-dependent biogeochemical sediment module for Earth system models, process analysis and teaching. Geosci. Model Dev., 14, 3603–3631, 2021 (URL: https://gmd.copernicus.org/articles/14/3603/2021 or doi:10.5194/gmd-14-3603-2021). Recent Fortran 2008 complying compilers (such as gfortran 11) consider it an error to substitute a scalar variable for a rank-1 argument of length 1 in subroutine call argument lists (here, NetCDF inquiries for variable dimensions). With previous compiler versions (e.g., gfortran 8.3), this never has been an issue. Most utilities under uti suffered from those shortcomings and have been amended. In addition, new compilers choked on NAMELISTs with non fixed-length (CHARACTER(LEN=*)) character string members (encountered again in several utilities under uti, in their GET_MENU internal subroutines). These shortcomings could all be addressed. Finally, several other minor bugs (not diagnosed by any compiler before and without consequences on the model results) have been corrected as well.

Financial support for this work was provided by the Belgian Fund for Scientific Research -- F.R.S.-FNRS (project SERENATA, grant CDR J.0123.19). The author is a Research Associate with the Belgian Fund for Scientific Research -- F.R.S.-FNRS.