Validation of sunflower anthesis period estimates from Sentinel optical and radar bands with beehive weight as reference data

Satellite-derived estimates for the onset of sunflower flowering are important resources for crop management as well as for agroclimatological research. Apiaries widely exploit sunflower pastures for honey production. Although in situ reference data on sunflower anthesis are rare, beehive weight time series emerge as widely applied bioindicator proxies of blooming. This study presents an approach using beehive scale time series as reference data to validate and optimize satellite-based estimates of the sunflower anthesis period. Apiaries were selected at 17 different locations in Hungary to derive the blooming period in 2021–2024. Using Sentinel-1 and Sentinel-2 satellite imagery, parcel-averaged and smoothed daily time series were produced for each study area. Vegetation indices obtained from optical bands failed to detect the sunflower blooming period. However, the local minimum and inflection point of the Synthetic Aperture Radar (SAR) VV/VH channel ratio identified the flowering start and end days, respectively. Descending (morning) satellite pass directions provided results superior to ascending pass directions, which can be explained with the orientation of sunflower heads. SAR-derived blooming period estimates outperformed those derived from growth degree days (GDD) in 10 and 16 of the 20 investigated cases for the start and the end of the blooming season, respectively. Satellite-based methods proved to be particularly useful in reducing the general underestimation of the pasture-wide flowering duration, as well as for the detection of blooming period anomalies in years with higher mean temperatures. This analysis allows us to track the temporal dynamics of sunflower flowering based on the SAR dataset and facilitates the exploitation of beehive scales to provide in situ reference data for flowering estimates over bee pastures.