Consolidate Simstrat-AED2/AED text output into a single netCDF file

Simstrat writes one plain-text .dat file per output variable (see strat_outputfile.f90 in the Simstrat source), unlike GLM-AED and GOTM-WET which write netCDF directly. This function reads every <var>_out.dat file in the simulation's output directory and writes them into a single compressed output.nc, so that Simstrat output can be read with the same netCDF-based tooling (read_model_outputs, get_model_outfile, ...) used for the other models, and so the on-disk output is much smaller than the raw text files.

Usage

write_simstrat_nc(sim_folder, config_file = "simstrat.par", remove_dat = FALSE)

Arguments

sim_folder

character; path to the simstrat_aed2/simstrat_aed simulation directory (containing simstrat.par and the output directory referenced by its Output.Path).

remove_dat

logical; delete the source <var>_out.dat files after they have been written to output.nc. Default FALSE – deleting them is the actual disk-space saving (keeping both uses more space, not less), but it also removes what read_simstrat_dat reads.

Each netCDF variable's units and long_name attributes are looked up from the package's key_naming table (matched on the variable's native Simstrat-AED2/AED name, sediment-zone variables under their base name with the _zone suffix stripped). A variable not present in key_naming is written with empty units and long_name equal to its netCDF variable name.

Value

Invisibly returns the path to the written output.nc file, or NULL if no output files were found.

Details

AED's sediment-zone output (Simstrat-AED only, not AED2) is written as a second, separate family of files, <var>_zone_out.dat – one column per benthic zone (labelled by that zone's reference depth), alongside the regular <var>_out.dat for the same variable (a single lake-bottom or whole-lake value). Because both exist side by side for the same variable name and _zone_out.dat itself ends in _out.dat, these can't share the water-column z dimension/grid the way regular depth-profile variables do – doing so silently wrote zone values against the wrong depths (or failed outright on a column-count mismatch) whenever a zone file's column count didn't happen to match the shared z grid's length. They get their own zone netCDF dimension instead, coordinate-valued by each zone's reference depth (m). Variable names keep their _zone suffix (stripping only the trailing _out.dat, as for every other file), so e.g. NIT_amm_dsf_out.dat and NIT_amm_dsf_zone_out.dat become the distinct netCDF variables NIT_amm_dsf (dims time) and NIT_amm_dsf_zone (dims zone, time) – never colliding. Reading them back needs no extra code: read_simstrat_output's load_all sweep already routes any variable whose dimensions aren't (time) or (z, time) through .read_glm_grouped_var into an new_grouped_var object, the same generic path GLM-AED's own nzones-dimensioned output uses.