Sentinel-2 data are widely used to estimate phenological parameters for agronomic applications, although processing methods are often difficult to reproduce and the efficiency of estimations is often not verified. In this contribution we describe a workflow used in the framework of E-crops project (https://bit.ly/cnr-ecrops) for the analysis of Bonifiche Ferraresi farm in Arborea (Sardinia): creation of the data archive (2018-2020), extraction and smoothing of MSAVI2 time series, identification and interpolation of relevant crop cycles and computation of phenological metrics. Algorithms were developed as R functions and publicly released in the R package "sen2rts" (https://sen2rts.ranghetti.info), so to facilitate reproducibility and methodology generalisation. Estimates were validated using farm information: good agreement was found between detected seasons and cultivated crops and among SOS estimates and sowing dates. These findings are encouraging to further develop "sen2rts" in order to manage a wider range of crop types and to extract additional information (e.g. crop categories).
A Reproducible Workflow to Derive Crop Phenology and Agro-Practice Information from Sentinel-2 Time Series: a Case Study for Sardinia Cropping Systems
Luigi Ranghetti;Francesco Nutini;Mirco Boschetti
2021
Abstract
Sentinel-2 data are widely used to estimate phenological parameters for agronomic applications, although processing methods are often difficult to reproduce and the efficiency of estimations is often not verified. In this contribution we describe a workflow used in the framework of E-crops project (https://bit.ly/cnr-ecrops) for the analysis of Bonifiche Ferraresi farm in Arborea (Sardinia): creation of the data archive (2018-2020), extraction and smoothing of MSAVI2 time series, identification and interpolation of relevant crop cycles and computation of phenological metrics. Algorithms were developed as R functions and publicly released in the R package "sen2rts" (https://sen2rts.ranghetti.info), so to facilitate reproducibility and methodology generalisation. Estimates were validated using farm information: good agreement was found between detected seasons and cultivated crops and among SOS estimates and sowing dates. These findings are encouraging to further develop "sen2rts" in order to manage a wider range of crop types and to extract additional information (e.g. crop categories).I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.