Assessment of the benefit of a single sentinel-2 satellite image to small crop parcels mapping

Abstract

The present study demonstrates the feasibility of producing crop type maps using a single high spatial resolution image from Sentinel-2A (S-2A). Our main objectives focus on the evaluation of the potential of S-2A data for crop mapping in the Tadla Plain and the determination of the most suitable period for discriminating different crops using a single S-2A image. In order to achieve these objectives, we have acquired a set of seven images of the S-2A and twelve vegetation indices. Then, we produced crop type maps at a spatial resolution of 10 m using (SVM) and (RF) classifiers for each image. The accuracy and performance of classification were assessed using field data. The optimal period for mapping was done during March with an overall accuracy and kappa coefficient of 91.49%, 0.90 respectively. The results show that a single S-2A image should be sufficient for the discrimination of different crops.

Keywords:

• Remote sensing
• Sentinel 2
• crop type mapping
• SVM
• random forest

Acknowledgement

The authors thank the Faculty of Sciences and Techniques of Béni-Mellal for its financial and logistical support. In addition, we acknowledge the ESA for providing; free of charge the S-2 data.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The authors thank the Faculty of Sciences and Techniques of Béni-Mellal for its financial and logistical support. In addition, we acknowledge the ESA for providing; free of charge the S-2 data.