calculate_spectral_metrics() now calculates CV, SV and CHV directly from rasters and AOI files, replacing extract_pixel_values(), which is removed. Results match the previous pipeline (see tests/testthat/test-reference_metrics.R).
AOI files can be GeoJSON or GeoPackage as well as shapefiles. AOIs in a different coordinate system from the imagery are reprojected, with a message.
AOI files can hold several polygons (one row of metrics each). An AOI outside the image, or with too few pixels for a convex hull, now gives NA metrics and a warning instead of stopping the run; duplicate aoi_id_col values warn.
Rarefaction now checks min_points against each AOI rather than the whole site, uses the same subsample for CV and CHV in each draw, and sets seed once per call so sites get independent draws.
create_multiband_image() checks that all bands share extent, resolution and CRS before stacking, and lists any that differ.
create_multiband_image() matches band files whose names end with the band name, so Pix4D output (e.g. NSABHC0009_transparent_reflectance_blue.tif) works without renaming. Other TIFs in the folder are ignored.
The data-frame helpers calculate_cv(), calculate_sv() and calculate_chv_nopca() are removed; use calculate_spectral_metrics().
calculate_field_diversity(): group_by_cols is renamed aoi_cols; the old name still works with a deprecation warning.
calculate_field_diversity() gains species_col, exclude and abundance_col, so it works on survey data beyond AusPlots point-intercepts (e.g. cover or count data). Defaults reproduce the previous results exactly.
data.table and stringr are no longer dependencies.
saltbush 0.1.0
Initial release.
From-the-sky spectral diversity: build multiband images from single-band drone TIFs (create_multiband_image()), mask non-vegetation pixels with NDVI/NIR thresholds (create_masked_raster(), find_optimum_thresholds()), extract per-AOI pixel values (extract_pixel_values()), and compute spectral metrics — coefficient of variation, spectral variance, and convex hull volume (calculate_spectral_metrics() and the underlying calculate_cv(), calculate_sv(), calculate_chv_nopca()).
From-the-ground taxonomic diversity from AusPlots point-intercept surveys (calculate_field_diversity()): species richness, Shannon, Simpson, Pielou’s evenness, exponential Shannon, and inverse Simpson.
Bundled example data (synthetic_survey, ausplots_NSABHC0009, and example drone rasters) so examples, tests, and the vignette run offline.