A Framework for Mapping Cropping Pattern Using Time Series Earth Observation Data

Cropping patterns, defined by the spatial and temporal arrangement of crops, can be classified into homogeneous and heterogeneous systems. Homogeneous cropping systems involve the cultivation of a single crop over large areas, offering advantages such as simplicity in management, efficient machinery use, and uniform harvesting. In contrast, heterogeneous cropping systems feature multiple crops on the same land, promoting biodiversity, reducing crop failure risks, and enhancing soil health. To conserve the environment, farmers are encouraged to adopt sustainable agricultural practices. To support this transition, various incentives are offered across different regions, motivating farmers to implement practices like Agroforestry and managed hedgerows, which can be effectively identified through the analysis of cropping patterns. This study focuses on assessing cropping patterns in the South Ostrobothnia region in western Finland. We propose a framework that uses Earth Observation (EO) imagery, Sentinel-2 NDVI time series data to understand the cropping pattern. Open source property boundary from the National Land Survey of Finland (NLS) and crop level information from the Finnish National Food Authority were used to create homogeneous or heterogeneous cropping pattern labels. A Random Forest algorithm was then applied to classify the property as either homogeneous or heterogeneous. A random sampling approach was employed to select properties for accuracy assessment, ensuring unbiased evaluation of the classification results. The model achieved an overall accuracy 77-80% for year 2020, 2021 and 2022. This study shows that satellite-based Earth observations effectively analyze cropping patterns, offering valuable insights for policymaking, scheme compliance monitoring, and environmental conservation. Future research could focus on mapping specific crops within cropping patterns to better understand agricultural intensity. Also, integrating multi-sensor data, linking cropping patterns with sustainability indicators, can help us to validate the Sustainable Agriculture Practices.