DRAFT: ycx_build_rasters.R emits AGC values (unblocks carbon-by-ecoregion)

[holoros]

Status: DRAFT — unrun, needs verification. Do not merge as-is.

What

Makes ycx_build_rasters.R emit the values behind the carbon overlay, not just the display PNG, so carbon (and the 100-yr change) can be summarized by EPA L3 ecoregion the same way harvest / structure / CSPI already are. Added after the existing PNG render block:

1. hybrid_agc_perplot.csv — per-plot AGC density (Mg C/ha) with lon/lat/state. Enables dependency-free point-in-polygon zonal stats by ecoregion (the same approach used for the other ecoregion layers), no GDAL needed.
2. conus_hybrid_agc2022.tif + conus_hybrid_dagc100.tif — georeferenced GeoTIFFs of the binned grids, written only when terra is available (guarded so the no-terra path is unchanged).

It is faithful to the existing agc0/dagc computation — it saves the same numbers the deployed PNG already draws. Nothing is fabricated and no public/api/ data changes.

Why it's a draft (two things to resolve before running/merging)

1. Blocked input: the script reads membership_slim.csv (the per-plot FIA/TreeMap table), which is not currently present anywhere in yield_curves_conus on Cardinal. hybrid_carbon_fits.csv and product_fractions_by_cell_age.csv are present; membership_slim.csv must be regenerated before the script can run.
2. CRS caveat: this grid uses the explorer's spherical Albers (lat_0 = 38, +R sphere), which differs ~0.5° from the lat_0 = 37.5 ellipsoidal grid of the TreeMap-derived ph_*/structure rasters. The GeoTIFF CRS is set to match the script's own projection; verify alignment against a state outline before trusting any spatial join, and keep each raster in its own CRS when zonal-summarizing.

Verification done / not done

• R parses clean (parse() check). Not executed (blocked on membership_slim.csv and terra/GDAL on a compute node).

Next steps for whoever picks this up

Regenerate membership_slim.csv → run Rscript ycx_build_rasters.R <build_dir> <product_fractions_csv> <out_dir> on Cardinal with the geo modules loaded → confirm the AGC tif aligns to a state outline → zonal-average hybrid_agc_perplot.csv (or the tif) by EPA L3 → add agc_2022/dagc_100 fields to ecoregion_harvest_summary.json via the existing generator.