B&W_Disc_RO2022 Barley and Wheat Discrimination Dataset based on Copernicus and Ground Truth Data for Călărași, Romania 2022

The dataset contains Sentinel-1 Synthetic Aperture Radar (SAR), Sentinel-2 Multispectral (MS), meteorological data from ERA5-Land, ground-truth labels and Normalized Difference Vegetation Index (NDVI) time series for winter wheat and winter barley discrimination. The data were aquired between January and July 2022 over an area of interest (AoI) located in Călărași County, South Romania. The AoI contains 895 agricultural parcels, including 707 winter wheat parcels and 188 winter barley parcels, containing approximately 2.1 million pixels. The ground-truth labels were provided by Field Data Zoom SRL, Bucharest, Romania. Sentinel-2 MS Data. The Sentinel-2 images were obtained from the Harmonized Sentinel-2 MSI Level-2A dataset through Google Earth Engine (GEE). The images are atmospherically corrected and contain 12 spectral bands, with band B10 excluded from the original product. All images were visually inspected to ensure complete coverage of the AoI and to minimize cloud contamination over agricultural parcels. The dataset includes Sentinel-2 observations from 12 acquisition dates between January and July 2022, covering the main growing season of wheat and barley in Romania. Data source: https://developers.google.com/earth-engine/datasets/catalog/COPERNICUS_S2_SR_HARMONIZED Sentinel-1 SAR Data. The Sentinel-1 images were downloaded from the Sentinel-1 SAR Ground Range Detected (GRD) dataset using GEE. The data include two polarization channels: VV (vertical transmit, vertical receive) and VH (vertical transmit, horizontal receive). The images were preprocessed using thermal noise removal, radiometric calibration and terrain correction. Sentinel-1 data were obtained at a spatial resolution of 10 m. For each Sentinel-2 acquisition, the closest available Sentinel-1 acquisition was selected to ensure temporal correspondence between the optical and radar observations. The maximum temporal difference between corresponding Sentinel-2 and Sentinel-1 acquisitions was four days. Data source: https://developers.google.com/earth-engine/datasets/catalog/COPERNICUS_S1_GRD ERA5-Land Meteorological Data. The dataset includes three meteorological variables obtained from the ERA5-Land reanalysis dataset developed by the European Center for Medium-Range Weather Forecasts (ECMWF): air temperature: measured at 2m above the Earth’s surface; soil temperature: measured in the upper soil layer (0-7cm); total precipitation: daily accumulated rainfall and snowfall reaching the Earth’s surface. ERA5-Land provides hourly meteorological observations at a spatial resolution of approximately 9 km. Temperature values were converted from Kelvin to degrees Celsius. For each agricultural parcel, the corresponding meteorological values were extracted using the geographical centroid of the parcel. Air and soil temperature values were selected according to the satellite acquisition times, while precipitation was calculated as the total daily accumulation for each acquisition date. Data source: https://developers.google.com/earth-engine/datasets/catalog/ECMWF_ERA5_LAND_HOURLY Ground-truth Data. The dataset includes ground-truth information provided as shapefiles containing the geographical boundaries and crop labels of 895 agricultural parcels. The ground-truth data comprise two agricultural crop classes: Winter wheat: 707 agricultural parcels, representing approximately 1.6 million pixels; Winter barley: 188 agricultural parcels, representing approximately 0.57 million pixels. NDVI Time Series. The dataset also includes NDVI time series derived from Sentinel-2 imagery for both winter wheat and winter barley. NDVI was computed using Sentinel-2 bands B8 (near-infrared) and B4 (red), according to the following equation: NDVI = (B8 – B4) / (B8 - B4). NDVI values were computed for valid pixels within the agricultural parcels for each available Sentinel-2 acquisition date, generating temporal vegetation profiles covering the January–July 2022 period. The NDVI time series were also used to assess crop development throughout the growing season. Parcels exhibiting unexpected NDVI temporal behavior were excluded from the analysis to improve the reliability of the dataset. Dataset Structure. The dataset is organized into separate directories according to the data source and data type. The sentinel1_2022 and sentinel2_2022 directories each contain four subdirectories corresponding to the satellite tiles: 35TMJ, 35TMK, 35TNJ and 35TNK. The images from these four tiles collectively cover the entire area of interest (AoI). All images and associated files follow specific naming conventions to facilitate their identification and provide information about their characteristics. The main directory contains the following subdirectories and file: sentinel1_2022: Contains the Sentinel-1 SAR images acquired using VV and VH polarizations, temporally matched with the corresponding Sentinel-2 acquisitions. The images are stored in GeoTIFF format; sentinel2_2022: Contains the Sentinel-2 multispectral images acquired on the 12 selected dates in 2022. All spectral bands have a spatial resolution of 10 m, and the images are stored in GeoTIFF format; meteorological_variables_2022: Contains the meteorological data associated with the satellite acquisition dates, including air temperature, soil temperature and total precipitation. The data are stored in CSV files; ground-truth_2022: Contains shapefiles representing the geographical boundaries and corresponding crop labels of the agricultural parcels; ndvi_timeseries_2022: Contains the pixel-level NDVI time series calculated from Sentinel-2 imagery for winter wheat and winter barley. The data are stored in CSV files; overlay_all_tiles_parcels.png - Provides a visual representation of the AoI, including the agricultural parcels and satellite tile coverage. Winter wheat parcels are displayed in brown, while winter barley parcels are displayed in magenta. Acknowledgment. This work was supported by a grant of the Ministry of Research, Innovation and Digitization, CCDI-UEFISCDI, ELIAC (Early Identification of Agricultural Crops) project no. PN-IV-P7-7.1-PED-2024-0375 in PNCDI IV.