Extract land-use history from RGB images of tree crowns.

Crops the NEON RGB camera mosaic (AOP product DP3.30010.001) at the HARV site to a 25 m grid, saving one GeoTIFF per cell, and writes a nested 16 × 16 sub-grid (1.5625 m sub-cells) as a vector file. One preprocessing step in a larger forest-scaling project.

• R (tested with 4.4.0) with sf and terra

Rscript Grid_and_Crop_HARV_RGB.R

Paths, site, cell_size (25 m), and n_sub (16) are set at the top of the script. Two toggles control the work:

• do_crops — write the 25 m RGB crops
• do_subgrid — write the 16 × 16 sub-grid vector

The sub-grid vector is large (~256 × number of cells). If the time usage is tight, run once with do_subgrid = FALSE, then again with do_crops = FALSE; grid_id is deterministic, so IDs match across passes.

• HARV_Tiles_NorthBoundary.shp — AOP tiles for the northern boundary (defines the grid extent)
• NEON RGB mosaic tiles (1 km mosaic tiles from 2022 at HARV)

Written to out_dir (set in the script):

• Shapefiles/HARV_grid_25m.shp — 25 m grid
• Shapefiles/HARV_grid_25m_sub16.gpkg — nested 16 × 16 sub-grid
• Imagery/NEON/HARV/RGB_25m_crops/*.tif — one RGB crop per 25 m cell