| English | 中文 |
Use sf and ggplot2 for the implemented functions, with terra/exactextractr only for separately adapted raster tasks. Follow the common configuration and result contract. The grid helpers accept projected rectangles with explicit cell dimensions, not arbitrary curvilinear or geographic raster cells.
Select a dedicated R environment with sf and ggplot2 installed. If setup is needed, run the following explicitly; Linux source builds may also require the native libraries described in the sf installation guide.
install.packages(c("sf", "ggplot2"))
From repository root:
Rscript --vanilla skills/model/shared/spatial_analysis/examples/r/synthetic_spatial.R
The synthetic script checks topology/CRS and builds four plot types, without reading business data or exporting by default. An explicitly supplied fresh output directory is only for reviewed synthetic image checks.
| Original pattern | Current reusable form | Adaptation |
|---|---|---|
sf_polygon_plot / sf_point_plot |
Same function names in the reviewed module | Explicit value label and limits, modern column evaluation and namespaced calls |
grid_plot / grid_facet_plot |
Same function names | Required CRS and rectangular cell dimensions; no implicit projection or input loading |
dfproj |
Not copied as a general adapter | Select resampling from quantity meaning; historical sum/projection defaults require adaptation |
| Boundary loaders, global labels and insets | Task-level inputs/design | No fixed source directories, global objects, automatic country bounds or source-time demonstration |
| Historical export helper | Explicit export call after review | Physical units and DPI cannot be exchanged with pixel dimensions |
The user’s original files remain unchanged. The extracted functions preserve the plotting pattern, with generalized interfaces and styling, rather than promising byte-identical legacy figures. No source paths or private map layers are copied.
Load the module from repository root, with already authorized sf objects sites and areas:
spatial <- new.env(parent = baseenv())
sys.source("skills/model/shared/spatial_analysis/scripts/r/spatial_helpers.R",
envir = spatial)
sf::sf_proj_network(FALSE)
sites_work <- spatial$transform_sf(sites, working_crs)
areas_work <- spatial$transform_sf(areas, working_crs)
matched <- spatial$point_polygon_matches(
sites_work, areas_work, "sample_id", "area_id", predicate = "intersects"
)
plot <- spatial$sf_polygon_plot(
areas_work, "value", "Study value (declared unit)", limits = shared_limits
)
working_crs and shared_limits are task choices, not implicit defaults. Coordinate transformation refuses enabled PROJ networking and ballpark operations; review the available pipeline and its accuracy for the task. The matching helper requires a reviewed projected CRS, unique keys and valid point/polygon geometries. It returns one row per point, a unique match only when available, match count/status and a list of all candidate IDs. It does not resolve borders or perform nearest matching. Callers must preserve the list when exporting; do not silently drop it into an unsupported CSV representation.
For a fresh authorized output target_plot, an explicit export can use:
if (file.exists(target_plot)) stop("fresh_output_required")
ggplot2::ggsave(target_plot, plot, width = 14, height = 10,
units = "cm", dpi = 300)
Verify the destination under the external project root before this call; it is a direct library operation, not a runner-protected export. Grid plotting draws declared rectangles only; it does not reconstruct curvilinear cell boundaries, reproject rasters or validate totals.