Hi Willem,
Thanks so much for your help!
Hmm, I've found that when return_state=T, values are actually returned for the model states and fluxes - even though the help page for the Sacramento function says that this is not supported. However, I'm just not sure whether the values returned are trustworthy because of the help page saying that this behavior is not supported. I tried taking a look at the C implementation of Sacramento in hydromad in the repo (as you suggested) and noticed that there appears to be a separate script for running the Sacramento model when states are requested (sacramento_state.c). The notes for this script indicate that it was adapted (from the University of Arizona MOSCEM project) to export states by Joseph Guillaume on 2013-10-24. So, perhaps it is possible that this updated script returns states and the help page is simply out of date? However, I don't know C, so it is hard for me to independently evaluate exactly what these scripts are doing, how they are being called by the R script, and whether the values generated are trustworthy.
I'll try to take a closer look soon, but, in the meantime, if anyone else has already looked into this, I would love to hear your thoughts!
Thanks,
Saalem