Detecting anomalous crop development with multispectral and SAR time series using unsupervised outlier detection at the parcel-level: application to wheat and rapeseed crops
This paper proposes a generic approach for detecting anomalous crop development at the parcel-level based on unsupervised outlier detection techniques. This approach consists of four sequential steps: preprocessing of synthetic aperture radar (SAR) and multispectral images acquired using Sentinel-1 and Sentinel-2 satellites, extraction of SAR and multispectral indicators, computation of zonal statistics at the parcel-level and outlier detection. This paper analyzes different factors that can affect the relevance of the outlier detection results for crop monitoring, such as the considered features and the outlier detection algorithm used. The proposed method is validated on rapeseed and wheat crops located in Beauce (France).