The study investigates a time series approach using Sentinel-2 images and the suitability of the BFAST (Breaks for Additive Season and Trend) Monitor method to detect changes that correspond to land use anomaly observations in the assessment of agricultural parcel management activities. We focus on identifying certain signs of ineligible use in permanent meadows and crop fields in one growing season that can be associated with time-defined greenness (vegetation vigour). Depending on the requirements of the BFAST Monitor method and currently time-limited Sentinel-2 dataset for the reliable anomaly study, we introduce customized procedures to support and verify the BFAST Monitor anomaly detection results. We propose using the analysis of NDVI object-based temporal profiles and time-series standard deviation output, where geographical objects of interest are parcels of particular land use. The validation of land use candidate anomalies was performed with evidence on declared land annual use and field controls of subsidy granting in Slovenia. The results confirm the proposed combined approach proves efficient to deal with short time series and provides high accuracy rates in monitoring agricultural parcel greenness.
COBISS.SI-ID: 43706157
The paper deals with the production of large scale vegetation products derived from PROBA-V 100 m satellite images. Image compositing and mosaicking are needed to create seamless products on a larger, global or continental scale. In the study, we have analysed and compared two compositing methods. Evaluation of the methods was focused on the visual inspection of the product seamlessness, and efficiency in eliminating (reducing) the influence of inadequate pixels that were misclassified by a cloud detection algorithm. During the compositing procedure, we also created a raster quality map and calculated statistical properties of the mosaics in order to assess the quality of the product, as well as input imagery characteristics.
COBISS.SI-ID: 45702189