This dataset contains the global crop calendar training and validation data used in the manuscript “Enhancing WorldCereal Crop Calendars with Land Surface Phenology and Machine Learning” (Moletto-Lobos et al., 2025). It includes harmonized reference crop calendar points derived from GEOGLAM Crop Monitor, USDA-FAS, JRC ASAP, and FAO sources, together with phenology-based samples retrieved from MODIS Aqua NDVI Land Surface Phenology (LSP) time series and ERA5-Land climate variables.

The dataset is organized into two main seasonal groups that follow the nomenclature used in the manuscript:

S1 = Winter crops (equivalent to winter cereals in the paper)

Includes reference and LSP-derived Start of Season (SOS) and End of Season (EOS) dates, including explicitly modeled dormancy periods.

S2 = Summer crops (equivalent to summer cropping systems in the paper)

Includes SOS and EOS samples for rainfed and monsoon-driven summer crops, corresponding to the “summer season” definition of the study.