Workflow on DesignSafe / TACC

Cascadia CoPes Hub researchers (and others doing data-intensive research related to natural hazards) can get an account on DesignSafe.

Once logged in, you can start a JupyterHub instance from the Tools & Applications page. The notebooks in this respository should run on the JupyterHub instance this opens.

You can also request an allocation of computing time on TACC, which is useful if you want to run GeoClaw or other codes there. It is also useful if you need to use ssh to scp or rsync files elsewhere, or to push commits to Github, since a terminal in the JupyterHub does not support ssh tools.

Using the JupyterHub on DesignSafe¶

In your top level directory you should see

• CommunityData (a read-only directory of data available to any user)

• MyData (the primary directory for your own files)

• MyProjects (any DesignSafe projects you are a collaborator on)

• NHERI-Published (DesignSafe projects that have been published)

• Work (Temporary storage accessible also from TACC)

and possibly other subdirectories.

Of particular interest for the CoPes Hub:¶

• CommunityData/geoclaw has data posted by rjl for general access, including some tsunami simulation results (with more to come). Notebooks in this repository may read data from CommunityData/geoclaw/CHTdata.

• MyProjects/PRJ-6005 is a DesignSafe project that is not published publicly but is intended for our collaboration, as a place to put data that you want to share with collaborators. You only have access to this project if your DesignSafe username has been added to the list of collaborators for this project (contact Randy). If you are a collaborator, you should also be able to view the files from the Data Depot webpage for PRJ-6005.

• NHERI-Published/PRJ-5885 contains all the CoPes Hub ground motion data from project PRJ-5885, Dunham et al. (2025).

Running Jupyter notebooks on DesignSafe¶

You can create and run your own notebooks, or you can clone this repository and run the notebooks included:

git clone https://github.com/rjleveque/CHTuser.git

Running these notebooks generally requires a number of Python packages that are not installed in the default Python kernel. See Python Environment.

Downloading data¶

The data found in CommunityData/geoclaw can be downloaded to your own computer if you want to run the notebooks locally, or do your own analysis/plotting with different tools. This requires an account on TACC, and then you can do, e.g.

scp username@stampede3.tacc.utexas.edu:/corral/projects/NHERI/community/geoclaw/filename ./

Accessing files from TACC¶

If you have an account/allocation on TACC, e.g. on stampede3, then you can access the CommunityData/geoclaw folder mentioned above as /corral/projects/NHERI/community/geoclaw. Files in the directory MyProjects/PRJ-6005 on DesignSafe can be accessed from TACC as /corral/projects/NHERI/projects/7f2e74be-d7ca-4e0e-b69a-22c24840b078

You can access corral from a login node, but not from a computational job run on an interactive shell created with the idev command, or submitted via Slurm, since compute nodes do not have access to /corral. So you will first have to copy files to your $SCRATCH or $WORK directory to use them. See the TACC corral documentation for more information.

Files in your Work directory on DesignSafe are in your $WORK directory on TACC. When running large jobs on TACC you probably want to direct the output to your $SCRATCH directory, but then you might want to copy some output to $WORK to make it available for analysis in a Jupyter notebook running on DesignSafe.

Some files to be shared among collaborators can be found on TACC in

/work2/04137/rjl/CHTshare

If you plan to run GeoClaw yourself, see Running GeoClaw on TACC for more about these shared files.

Files (such as topofiles or dtopofiles) found in this directory can also be downloaded to your own computer (if you have an account at TACC) using e.g. rsync or scp.