Restart Concentration Field from ATS-PFLOTRAN Checkpoint File
Xueyuan Kang
Hello all,
I am trying to restart a reactive transport simulation in ATS–PFLOTRAN from a previously saved checkpoint file, but I am encountering an issue when loading the concentration fields.
When restarting the simulation, the log file reports the following error:
what(): No such variable "total_component_concentration.cell.0" in file "../RTM/checkpoint00575.h5"
However, upon inspecting the checkpoint file (checkpoint00575.h5), I found that the variables are stored with names such as: total_component_concentration.cell.Al+++, instead of indexed names like cell.0.
This makes me wonder:
- Does ATS expect component concentrations to be indexed numerically (e.g., cell.0, cell.1, etc.) during restart?
- If so, is there a required mapping between component names (e.g., Al+++) and indices for the checkpoint.h5 file?
- Or is there an additional configuration needed to ensure ATS correctly recognizes the named variables in the checkpoint file?
For initialization, I have specified the restart file for all PKs (canopy, snow, flow, overland flow, subsurface transport, surface transport, subsurface chemistry, and surface chemistry) as follows:
<ParameterList name="initial
condition" type="ParameterList">
<Parameter name="restart
file" type="string" value="../RTM/checkpoint00575.h5"/>
</ParameterList>
For the subsurface and surface chemistry PKs, I have also set the "initial conditions time" to match the checkpoint time.
Despite this, it seems that the solute concentrations (and potentially mineral-related fields such as volume fractions) are not being correctly read from the checkpoint.
I would greatly appreciate any suggestions. Thank you very much for your help!
Best regards,
Xueyuan
Xueyuan Kang
Afolabi Femi
--
You received this message because you are subscribed to the Google Groups "Amanzi-ATS Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-users+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ats-users/7fc5632f-f802-4500-a2b8-f70b90b539fdn%40googlegroups.com.
Ethan Coon
-------------------------------------------------------------------
Ethan Coon
917-969-6831
https://www.ornl.gov/staff-profile/ethan-t-coon
-------------------------------------------------------------------
To view this discussion visit https://groups.google.com/d/msgid/ats-users/CAAofpt0aRAUx0KuoGBYABVZ%2BkgK%3Dn6jOK73hmRfPd9Z%3Dko0jMg%40mail.gmail.com.
Xueyuan Kang
Hi Ethan,
Yes, the checkpoint file was generated using the same version of the code as the restart. I will go ahead and submit a bug report.
Best,
Xueyuan
Xueyuan Kang
Hi all,
I have a follow-up question regarding restart of reactive transport runs in ATS.
I am currently trying to restart a simulation from an existing checkpoint HDF5 file. Before the ATS-side issue is fully resolved, I would like to ask whether there is a practical workaround to make restart work with the current version.
My current understanding is that the restart file seems to have a naming inconsistency for chemistry component fields. For example, ATS appears to expect both species-based names and index-based names in different places, such as:
surface-total_component_concentration.cell.H+
surface-total_component_concentration.cell.0
and similarly for subsurface:
total_component_concentration.cell.H+
total_component_concentration.cell.0
To test this, I tried editing the checkpoint HDF5 file manually:
first by renaming fields from species names to indices;
then by keeping the original species-based fields and adding duplicated index-based fields (for both surface-total_component_concentration.cell.* and total_component_concentration.cell.*).
However, this still does not resolve the restart issue. The run can read the restart file and begin advancing, but then crashes in surface transport with PETSc errors / segmentation violation. A typical sequence looks like:
restart is read successfully;
overland flow advances;
surface transport starts;
then PETSc throws errors and crashes, possibly because some fields are still not correctly initialized or the restart state is incomplete/inconsistent.
So my questions are:
Is there a correct way to prepare the checkpoint HDF5 so that restart works with the current ATS version?
Are there additional restart fields, metadata, or initialization steps required for surface transport / chemistry beyond surface-total_component_concentration.cell.* and total_component_concentration.cell.*?
At this point I suspect the PETSc crash may be caused by incomplete or inconsistent initialization of the chemistry/transport state, rather than just a simple missing field name.
Any guidance on how to work around this with the current code version would be greatly appreciated.
Thank you very much.
Xueyuan
Coon, Ethan
Date: Monday, April 13, 2026 at 3:36 PM
To: Amanzi-ATS Users <ats-...@googlegroups.com>
Subject: [EXTERNAL] Re: Restart Concentration Field from ATS-PFLOTRAN Checkpoint File