Changelog • dhw

dhw 3.0.1

Added calculate_dhm() for exact reproduction of the rolling monthly Mason et al. workflow, including MATLAB window alignment, strict anomaly thresholds, three- or four-month windows, and optional degree-week conversion.
Added the scientifically distinct calculate_dhmi_lough() annual standardized Degree Heating Month Index and deprecated the ambiguous legacy calculate_dhmi() interface.
Added NOAA-compatible final-product encoding via encoding = "noaa" and encode_noaa().
Added NOAA’s 7-day maximum BAA composite (calculate_baa_7d() and product "baa_7d") plus post-15 December 2023 Alert Level 3–5 fixtures.
Added a reproducible, all-cell regional NOAA tile validation workflow and vignette covering coastlines, masks and controlled missing-pixel propagation.
Added release CI across Linux, macOS and Windows plus scheduled/manual differential checks of CRAN and Bioconductor reverse dependencies.
Organised the package into core, optional data-access, and analysis layers; moved provider, plotting, and specialised analysis packages to Suggests.
Replaced packaged multi-megabyte validation rasters with a compact complete one-cell baseline fixture and an optional external validation-cache policy.
Fixed missing-DHW conversion, completeness boundary detection, annual SST metric grouping, elapsed-time trends, atomic downloads, raster-preserving reef extraction, processing defaults, and hindcast overwrite handling.

Reimplemented the NOAA Coral Reef Watch v3.1 monthly climatology exactly: daily SST is averaged by year-month for 1985–2012, each calendar month’s annual means are regressed against year, and the regression is evaluated at 1988.286.
Daily climatologies now interpolate cyclically through monthly values on the 15th day, including the December–January and leap-day transitions.
calculate_hotspots() now returns the signed SST-minus-MMM product. The 1 degree C accumulation threshold is applied only by calculate_dhw().
calculate_dhw() now uses complete 84-calendar-day windows, includes the current day, and propagates missing dates and values instead of treating them as zero heat stress.
calculate_baa() implements NOAA’s December 2023 single-day categories from No Stress through Alert Level 5.
Added offline NOAA v3.1 regression fixtures spanning Lizard Island, Hawaii, the Florida Keys, and Belize. The complete Lizard 1985–2012 baseline reproduces the published monthly climatology within its 0.01 degree C storage precision; downstream SSTA, HotSpot, and DHW products are also checked against independent NOAA output.

Removed the baa logical argument from create_climatology(). Request BAA explicitly with products = "baa" or include it in a product vector. With products = NULL, the historical default remains all products except BAA.
Missing or non-consecutive dates now produce NA DHW values for every affected 84-day window. Code that previously relied on implicit zero filling must repair or explicitly handle the input gaps.
HotSpots below zero are retained, matching NOAA’s signed product.

Monthly annual means are produced in one terra::tapp() pass and fitted with centered analytical OLS, avoiding per-cell model objects.
Daily climatology interpolation reuses exact common-year and leap-year templates.
Calculation functions accept filename, overwrite, and wopt so large outputs can be written directly to disk.
create_climatology() calculates dependencies lazily and returns/writes only requested products.

added hindcast_qm(), anomaly-based quantile mapping to reconstruct a hindcast SST time series by bias-correcting a long model dataset (e.g. ERA5) to a shorter observed reference dataset (e.g. OISST), enabling extension of historical SST records.

fixed downloadCoralTemp() and download_OISST() to handle date vectors
fixed calculate_baa() to handle NaN in dhw input
fixed calculate_dhw() so that if anomaly >= 1, only daily hotspot values greater than or equal to that threshold are included, and if anomaly < 1, all non-NA hotspot values are included without thresholding
updated process_CoralTemp() to use futures::future_lapply() instead of parallel::mclapply() to make OS compatible and fix new mc issues in MacOS (to do: process_OISST())
added isComplete() function that checks for completeness of download_OISST() and download_CoralTemp() timeseries
added plot_sst_timeseries() function for visualising SST time series
added plot_max_dhw() function for visualising annual max DHW over a timeseries
added plot_annual_dhw function for visualising daily DHW over an annual timeseries
added visualising_outputs vignette to display plot/mapping options for outputs