Flowering dynamics of nectar and pollen sources in Hungary based on Sentinel-2 satellite data

Knowledge of the phenological dynamics of various flowering plants is important for food supply and essential for analysing and modelling adaption of pollinators, especially the honey bees, Apis melifera. The increased frequency of extreme events associated with climate change has resulted in new challenges for farmers as well as beekeepers, and has made research on this issue a priority for adaptation planning. Satellite products can be used for phenological monitoring and detection with good spatial coverage providing valuable information for agriculture and beekeeping. In this study, we investigated the flowering period and dynamics of the two most important honey crops in Hungary, Sunflower (Helianthus annuus L.) and Oil Seed Rape (Brassica napus L.). To generate flowering time series data from Sentinel-2 we used the MSIL2A product in 20-meter resolution, which contains 10 spectral bands with 5 days revisit days. This product also contains additions mapping data helping the classification of the studied area. The research has been created using the Ecosystem Map of Hungary, which also a 20-meter-resolution map helping us to describe the agricultural land developed in 2020. To find the honey producing crops, we used machine learning applications for classification of the arable lands and these desired parcels. Subsequently, flowering timeline was estimated from NDVI and NDYI fields obtained from Sentinel-2 satellite data. Hive weight time series was measured locally with automatic hive scales during the flowering seasons of 2021-2022. The scales grants the apiarists supplementary information like weight in a 10-minute interval. This provide essential information about the health and productivity of the colony and enables information to be gathered before winter season or swarming events. We examined the results of the satellite data against the hive weight data during the honey collecting period and investigated with the actual meteorological conditions to determine a multiplex relationship between the parameters. A method is presented to apply local honey collection data to evaluate and downscale satellite-based phenological estimates for applications in beekeeping.