Changelog

Source: NEWS.md

sciSpatialR (development version)

Catalogue

• New download_layer() copies a catalogued layer from the share to a local directory, data and metadata together: every file in the layer folder, shapefile sidecars included, and — where a product splits its record — the parent product’s readme and documentation alongside. The copy is rooted at the product folder, which lands directly in the destination, so a variant’s product readme sits one level up from it exactly as it does on the share; the ISO topic category is left behind unless keep_theme = TRUE. Sibling variants are separate layers and are not copied. Every copy leaves a receipt beside the data: _download_log.txt records what was copied, from where, by whom, when, how long it took, and each file’s size and md5 checksum, with runs appending rather than replacing so a layer topped up over several sessions keeps its whole history. Each copied file is read back and checked against its source on size and checksum before the log calls it sound; pass verify = FALSE to skip that on a very large copy. Files already present and unchanged are skipped, so an interrupted copy resumes rather than restarts, and dry_run = TRUE reports what would be copied and how much it comes to without writing anything. A copy manifest naming every file, its size, and whether it was copied or skipped is returned invisibly.

• New list_variables() reports the bands each product documents — what they measure, their units, measurement scale, and valid range — so the variables available for extraction can be browsed without opening a readme. Defaults to every product; narrow with theme or product. Bands are a product-level fact, so a product with several variants contributes one set of rows rather than one per variant, and a product documenting no bands is listed with band NA so the gap stays visible.

• The scan follows the share’s product/variant layout. Products sit directly under their ISO topic category, so the manifest’s sub_theme column is gone — there is no longer an intermediate level to derive it from.

• A dataset’s readme may be split in two, as the data management guide describes: a product record holding the title, licence, and citation, and a variant record holding the measured resolution, CRS, extent, and derivation of one processed copy. Splitting is optional; a product documented in a single readme is catalogued exactly as before.

• Where a product holds variant subfolders it contributes one row per variant rather than a row of its own, since the data lives in the variant. The two records are merged for that row, the variant winning on any field it fills in, so a variant row carries its own geometry alongside the product’s identity. Previously a variant readme was catalogued as a separate, near-empty layer and the product row reported n_files = 0.

• New manifest columns product_id, variant (NA when the product has no variant folders), and product_readme (equal to readme unless the record is split).

• layer_meta() and check_metadata() read both halves of a split record, so a variant is no longer scored as missing the title and licence its product record carries.

• A variant folder holding data but no variant readme is now reported by check_metadata() instead of being treated as covered by its product’s readme.

• Esri file geodatabases are counted as one vector dataset rather than as the internal tables a recursive listing finds inside them. A .gdb previously reported n_files = 0 and data_type = NA, and get_layer() could not open one; an undocumented .gdb also went unreported, since its internals carry no recognised extension. size_mb still totals the bundle’s real contents.

sciSpatialR 0.1.0

First version. Everything below is new.

Reference layers

• ab_crs(), ab_boundary(), and ab_grid() return the project CRS (NAD83 / Alberta 10-TM, EPSG:3400), the 2020 provincial boundary, and the ABMI 1 km reference grid. The boundary and grid ship with the package, so the harmonization functions need only the layer being harmonized.
• The grid template is snapped to the lattice of the ABMI’s 1 km grid polygons, so a cell’s raster row and column reproduce its source GRID_LABEL.

Harmonization

• check_alignment() compares a layer against the reference on CRS, extent, resolution, and origin, and returns the result invisibly as a named logical vector so it can gate a pipeline.
• resample_to_grid() snaps a layer onto the reference lattice, choosing the resampling method from the resolution ratio and whether the layer is categorical, and reporting what it picked. It errors rather than silently returning an empty layer when the CRS does not already match.
• mask_to_boundary() clips to Alberta by default, or to a supplied polygon; inverse = TRUE keeps the outside instead.
• harmonize_crs() transforms points to the raster CRS and warns when raster reprojection would be implied.

Extraction

• extract_points() for point-in-cell values, extract_buffer() for summary statistics within one or more radii, extract_proportion() for class composition within a buffer, and extract_vector() for point-in-polygon attribute joins.
• All four return columns in input row order, so bind = FALSE results can be combined with a single cbind().

Inspection

• raster_stats() reports per-layer cell counts, missingness, and value summaries for a SpatRaster, a path, or a directory of GeoTIFFs. Only cells inside aoi are summarised, which defaults to the Alberta boundary the reference grid covers, so n_total counts the cells of the study area rather than the cells of the rectangle an export arrived on; aoi = NULL summarises the whole layer.
• plot_raster() and plot_hist() return ggplot objects, with the colour scale clamped to central quantiles by default so a handful of extreme cells cannot flatten the pattern.
• theme_science_map() is exported for other figures in a project.

Catalogue and metadata

• build_catalogue() scans the ABMI spatial data share and parses every dataset readme into a manifest; list_themes(), list_layers(), find_layer(), get_layer(), and layer_files() query and retrieve from it.
• spatial_root() reports where the catalogue reads from and is overridable by option or environment variable.
• The share’s top-level _temp folder is skipped by every scan, so work in progress and staging copies are neither catalogued as layers nor reported as missing readmes by check_metadata().
• read_metadata(), as_metadata_row(), layer_meta(), and check_metadata() parse and audit the readme files stored beside each layer.

Documentation

• Four vignettes: harmonizing layers to the reference grid, inspecting layers, extracting covariates at survey locations, and working with the catalogue.
• A pkgdown site published at https://ABbiodiversity.github.io/sciSpatialR/.