What's new in v1.3.0 Two new published quality artefact families plus a metadata restructure: CAL CSVs (new): per-phase wide-format tables of in-sample BAM model-fit diagnostics from mgcv::bam — BAM_K, AIC, BIC, EDF, DEV_EXPLAINED, one row per year (32 rows per phase). Complements the existing VAM holdout cross-validation tables. GEM CSVs (new): per-phase long-format tables of spatial quantiles of the BSE uncertainty raster — YEAR, Quantile, Value at the {0, 25, 50, 75, 100}% levels (160 rows per phase). Replaces the MEAN_BSE scalar column previously carried inline in VAM. DQV quality metadata restructured to the JSON-LD-correct shape: per-year measurements (RMSE, MAE, R², training/validation N, basis dimension k) become first-class @graph entities with their own @id; per-phase Datasets reference them by @id only; one shared dqv:Metric per measure name. Per-phase Dataset @id switched from phase_202-15/ to #phase-202-15 (hash-style contextual entity). The deposit's README.md records a build-time rocrate-validator PASSED verdict at REQUIRED severity against ro-crate-1.1. Data Fitness-for-Purpose (DFFP) reviews embedded where applicable. Produced by PhenoPhaseR v1.5.0 (DOI: 10.5281/zenodo.18743008). The crate's overall layout (per-phase multi-band COGs, one band per year; per-phase wide-format CSVs) is unchanged from v1.3.0. Downstream code that parses the manifest needs to follow @id references for quality measurements; see README.md inside the deposit for migration notes.

This dataset provides outputs of the PhenoPhaseR processing pipeline (https://doi.org/10.5281/zenodo.18743008) forwinter wheat (Triticum aestivum L., DWD Plant ID 202), covering observed phenological phases (sowing to harvest) for the entire territory of Germany for the period 1993–2024. PhenoPhaseR is an open-source R workflow that couples the PHASE model - a growing-degree-day (GDD) approach with adaptive outlier filtering - to derive spatially consistent, quality-controlled point observations of phenological phase-start dates from the DWD volunteer observer network. The dataset component provided here corresponds to processing Step 7 (spatial interpolation) of the seven-step pipeline.