Set up Salinas and Modesto cal/val sites — IC + met via magic-ensemble

[AritraDey-Dev]

Summary

Set up the two cal/val sites for SIPNET runs by creating IC and met files using the magic-ensemble CLI as per discussion with David and the Slack thread.

Sites

• White/Salinas - USDA-ARS Salinas SOCS, 36.617°N, -121.533°W, 2003–2011, annual veg (lettuce + broccoli), 8 socs_sys treatments
• Nichols 2024 - Modesto almond orchard, 37.6273°N, -121.0893°W, 2018–2019, perennial woody, 2 treatments (compost / no_compost)

Approach

Try the magic-ensemble CLI first per to surface rough edges in the new verb-based interface. Fall back to running the underlying R scripts directly only if the CLI blocks. Either way, log every rough edge encountered for follow-up.

CLI path:

magic-ensemble prepare

which under the hood calls:

./workflow/01_ERA5_nc_to_clim.R
./workflow/02_ic_build.R
./workflow/03_xml_build.R

Tasks

• Add Salinas + Modesto rows to ccmmf/downscaling/anchor_sites.csv (separate small PR there)
• Build met files per site via 01_ERA5_nc_to_clim.R (or CLI equivalent)
  • Salinas: ERA5 pull for 2003–2011
  • Modesto: ERA5 pull for 2018–2019
• Build IC files per site via 02_ic_build.R (or CLI equivalent)
  • Salinas: t=0 from cal/val Year 0 SOC + bulk density (Oct 2003)
  • Modesto: t=0 from Aug 2019 soil C/N + bulk density (back-fill caveat — only post-study sample available)
• Build PEcAn XML per site via 03_xml_build.R (or CLI equivalent)
• Sanity-check all outputs against an existing successful run (e.g. anchor sites currently in anchor_sites.csv)
• Wire in the events.json files from ccmmf/uncertainty:feature/calval-to-events once that lands
• Capture and file rough-edge feedback for the magic-ensemble CLI / docs as new issues or comments here