SwissPhenoCamDataset

The SwissPhenoCam dataset provides individual tree-level chromatic coordinatetime series and phenological transition dates extracted from a network ofhigh-resolution outdoor webcams across Switzerland. The dataset is structuredat the tree-year level: each monitored tree in each year of observationconstitutes one unit. From 14,436 candidate tree-years (every annotated tree in every observed year)we release the data at two quality-control levels, both indexed by a master CSVat the root of the dataset: - FULL set (lenient quality control) — phenology_records_full.csv 8,067 tree-years, 2,567 unique trees. A permissive selection that maximises coverage. Intended for users who are prepared to apply their own quality filtering before analysis. - CURATED set (strict quality control) — phenology_records_curated.csv 5,944 tree-years, 2,204 unique trees. An analysis-ready subset of the full set (curated is a strict subset of full). Recommended as the default entry point: take these tree-years if you want data that is ready to use without further filtering. Both sets were obtained by applying the same battery of automated qualitymetrics and a manual visual check to the candidate tree-years; they differ onlyin how strict the acceptance thresholds are (see AUXILIARY DATA below). If youare unsure which to use, start from the curated set and reach into the full setonly when you need the extra coverage and are willing to filter it yourself. Key statistics: - 34 camera sites - 97 site-sublocation directories - 2,567 unique tree ROIs (regions of interest) in the full set (2,204 in the curated subset) - 8,067 tree-year directories on disk (the full set), of which 5,944 are in the curated subset (phenology_records_curated.csv) - Observation years 2010-2024 Chromatic coordinates (GCC, RCC, BCC) are derived from RGB pixel valueswithin manually annotated polygon regions. Transition dates mark the timingof seasonal green-up and green-down events, estimated from smoothed p90time series with Monte Carlo uncertainty quantification. See readme.txt and related journal article for additional information.