Code and data

The dashboard is public on the CGD project site, the code is in a GitHub repository, and all model output and restarts are archived on Campaign storage.

Repository layout¶

github.com/MozhganFarahani/ctsm_conus.

setup/                  case-creation scripts, one per config, each with spinup and eval modes
cases/                  the six CIME case folders, complete (archive; never edited by hand)
analysis/config/        paths.py: every GLADE path the analysis code uses, and the evaluation period
analysis/lib/           shared helpers: reading output and observations, metrics, basin averaging, map style
analysis/qc/            control QC and water-balance checks
analysis/spinup/        spinup drift and restart-boundary checks
analysis/streamflow/    streamflow evaluation against observed flow, HUC02 report figures
analysis/climatology/   monthly climatology maps
analysis/diagnostics/   time-alignment check and input-data reference
analysis/animation/     CONUS discharge animation
dashboard/              multi-config streamflow dashboard: builder, template, PBS job, web/ landing page
tools/                  scripts that maintain and test this repository
environment.yml         the Python packages the code imports, at the npl-2026a versions

The repository holds no model source and no model output.

Scripts¶

Setup scripts (setup/; run on Derecho from the CIME scripts folder, $CIME_SCRIPTS)

Every script takes --help and --dry-run. A dry run writes only the case’s final user_nl_clm to $DRY_RUN_DIR/<case>/ and runs no CIME command; all six match their archived cases/<case>/user_nl_clm once comments and whitespace are ignored.

bash setup/setup_control.sh spinup            # create and configure the spinup case
bash setup/setup_control.sh eval 2002-01-01   # switch it to the eval run from that restart date

Paths¶

Paths and settings. Every path is defined once: in analysis/config/paths.py for Python, in setup/ctsm.env for shell. ctsm.env repeats four of them (SCRATCH, ARCHIVE, OUTPUT_ROOT, and CIME_SCRIPTS, which is CASES_ROOT in paths.py): keep the two files in sync.

Output root. Every script writes under /glade/work/mozhgana/ctsm_conus_output/ by default: qc/<config>/, spinup/<case>/, streamflow/, report/, climatology/, dashboard/ (with its weight cache), animation/ and case_inputs.csv. --outdir overrides it.

PBS submission¶

environment.yml lists the packages and versions for use elsewhere.

Submitting jobs. PBS scripts carry no #PBS -A or #PBS -o line; the account and log folder come from ctsm.env at submit time. Analysis jobs run on Casper (from a Derecho login node, add -q casper@casper-pbs).

source setup/ctsm.env   # from the repo root: sets $PROJECT and $OUTPUT_ROOT
cd dashboard && qsub -A "$PROJECT" -o "$OUTPUT_ROOT/dashboard/" run_dashboard.pbs

Smoke test¶

Smoke test. bash tools/smoke_test.sh runs every script on its smallest input (one year, one gauge), the six setup dry runs and a PBS syntax check, and reports PASS/FAIL per test. All 22 tests passed. It writes only under <output root>/smoke_test/.

Directories¶

Data availability¶