SSIMmap 0.4.3

Polygon component inference

• Add local two-sided permutation tests for SIM, SIV, and SIP alongside SSIM in ssim_polygon(), using a separate permutation mean for each polygon and statistic and the existing draws that independently permute both maps.
• Return _p, _q, and _sig columns for all four local statistics. Apply BH correction across polygons separately for each statistic. Preserve p_value, q_value, and sig as aliases for the SSIM results; with fdr = FALSE, _q values equal the unadjusted p-values.
• Use q <= alpha for local polygon significance, matching rasters; without FDR adjustment use p <= alpha. The observed statistics, existing SSIM p/q values, global tests, and random-number sequence are unchanged for the same inputs and seed.
• Extend reproducibility output to include the four local inference results and add independent permutation and boundary regression tests.

SSIMmap 0.4.2

Defaults and raster inference

• Change the default transform to "none" in ssim_raster(), ssim_polygon(), and ssim_bandwidth(). Existing analyses using an explicit transformation retain that transformation.
• Change global and local raster permutation tests from upper-tail tests to two-sided tests centered on the permutation mean, matching the polygon definition. Map A remains fixed and map B is shuffled over common valid cells.
• Keep observed raster SSIM components unchanged for the same explicit transformation. Recalculate inference reference results for the existing raw FWI comparisons with the same inputs, windows, seeds, and 1,000 permutations.
• Compute local null centers and tail counts in two passes through the same draws, avoiding storage proportional to the permutation count and retaining the one-pass RNG state.
• Add regression tests for defaults, upper and lower departures, local and global null distributions, ties, missing-cell masks, BH adjustment, and RNG replay. Update help and reproduction documentation.

SSIMmap 0.4.1

Data and reproducibility

• Require R >= 3.6.0 so the bundled fixed-seed reproduction uses the documented Rejection sampling RNG consistently. The exported functions retain their numerical calculations and defaults.

• Include portable, offline manuscript reproduction scripts and compact reference values accessible with system.file("reproduction", package = "SSIMmap"). Fixed seeds reproduce the tables and local significance maps; the commuting comparison has 124 of 279 areas with q < 0.01.

• Document and independently verify the November 11 FWI against the official CWFIS historical coverage, including source attribution and Open Government Licence - Canada information.

• Add fwi_1111_bc, the actual 11 November 2023 FWI raster used in the revised manuscript. The original fwi_1101_bc remains the unchanged 1 November 2023 dataset for compatibility.

• Update the FWI examples to use the November 11 data with explicit transform = "none" and 3x3, 5x5, or 7x7 windows. The default transformation and all three numerical algorithms are unchanged.

• Correct the Toronto polygon documentation to 279 Aggregate Dissemination Areas (2021) and identify the CIMD source dimension as residential instability, while preserving the numeric example data. Document the preserved commuting measure using Census characteristic 2601, total Census population as denominator, two-decimal DA rounding, unweighted intersecting-DA averaging, and the suppressed-source-value limitation.

• Use the manuscript’s maximum bandwidth of 200 in example selection, yielding midpoint bandwidths 87 and 91.

• Make vignette examples portable, use regular panel tags, and correct raster significance-layer names.

• Use American spelling in authored documentation and display labels; preserve official source titles, license names, and API argument names. Numerical algorithms are unchanged.

SSIMmap 0.4.0

Major changes

• ssim_bandwidth(): redesigned with transform, option arguments and bias-variance trade-off plots.
• ssim_polygon() and ssim_raster(): added permutation testing (do_test, R, fdr, alpha, seed) and FDR-corrected local p-values.

Breaking changes

• Removed ssim_constant(). Constants k1 and k2 are now arguments of ssim_polygon().

New data

• Added Toronto_SSIM, fwi_0816_bc, fwi_0818_bc, fwi_1101_bc.

New data

• Added Toronto_SSIM (sf polygon object).
• Added fwi_0816_bc, fwi_0818_bc, fwi_1101_bc (packed terra rasters of BC Fire Weather Index for Aug 16, Aug 18, and Nov 1, 2023).