Scalable Parcel-Based Crop Identification Scheme Using Sentinel-2 Data Time-Series for the Monitoring of the Common Agricultural Policy

Abstract

This work investigates a Sentinel-2 based crop identification methodology for the monitoring of the Common Agricultural Policy's (CAP) Cross Compliance (CC) and Greening obligations. In this regard, we implemented and evaluated a parcel-based supervised classification scheme to produce accurate crop type mapping in a smallholder agricultural zone in Navarra, Spain. The scheme makes use of supervised classifiers Support Vector Machines (SVMs) and Random Forest (RF) to discriminate among the various crop types, based on a large variable space of Sentinel-2 imagery and Vegetation Index (VI) time-series. The classifiers are separately applied at three different levels of crop nomenclature hierarchy, comparing their performance with respect to accuracy and execution time. SVM provides optimal performance and proves significantly superior to RF for the lowest level of the nomenclature, resulting in 0.87 Cohen's kappa coefficient. Experiments were carried out to assess the importance of input variables, where top contributors are the Near Infrared (NIR), vegetation red-edge, and ShortWave Infrared (SWIR) multispectral bands, and the Normalized Difference Vegetation (NDVI) and Plant Senescence Reflectance (PSRI) indices, sensed during advanced crop phenology stages. The scheme is finally applied to a Lansat-8 OLI based equivalent variable space, offering 0.70 Cohen's kappa coefficient for the SVM classification, highlighting the superior performance of Sentinel-2 for this type of application. This is credited to Sentinel-2's spatial, spectral, and temporal characteristics.

Crop Classification Based on Temporal Signatures of Sentinel-1 Observations over Navarre Province, Spain

“…These modifications recommend the establishment of procedures to check and track all eligibility criteria using Copernicus Sentinels data or similar data. Therefore, several initiatives and research efforts are being conducted at present to fulfill this aim, including EU-funded projects like Sen2-Agri [6], RECAP [7] or SEN4CAP [8].…”

“…Yet, the unavailability of multi-temporal datasets with the sufficient revisit time (i.e., 16 days for Landsat) precluded the further development of this type of method until sensors such as MODIS became available [12,13]. With the availability of S2 data, similar approaches but with a higher spatial resolution have been developed [7,14,15].…”

Decision Support on Monitoring and Disaster Management in Agriculture with Copernicus Sentinel Applications

“…In DiAS, in order to achieve a reliable and accurate crop discrimination, non-parametric image classification algorithms are used, because they do not require a-priori information on the data distribution. Specifically, two image classification algorithms are implemented in the DSS, i.e., the Random Forests (RF) [41] and the Support Vector Machine (SVM) algorithm [42], which are selected because of their capability and efficiency in handling long time series [4]. The RF algorithm is reported to have good accuracy results when a lot of different crop types are expected in the scene, and it has relatively short computation time.…”

“…Lately, the control efforts involve the examination of satellite imagery, but this refers to very high-resolution commercial imagery and visual image interpretation or rare case image classification of single-date images [3]. These procedures improve the control efforts in some cases, but they have a high cost and their effectiveness relies mostly on the skills of the photo-interpreter and therefore cannot always offer timely, reliable, and robust results [4].There are several attempts for automating the mapping procedures in agriculture-related applications. For example, in terms of crop types mapping, the EU is coordinating the Monitoring Agricultural Resources (MARS) initiative [5] with the help of the Joint Research Center (JRC) in order to facilitate the coordination of the CAP.…”

“…Lately, the control efforts involve the examination of satellite imagery, but this refers to very high-resolution commercial imagery and visual image interpretation or rare case image classification of single-date images [3]. These procedures improve the control efforts in some cases, but they have a high cost and their effectiveness relies mostly on the skills of the photo-interpreter and therefore cannot always offer timely, reliable, and robust results [4].…”