REMOTE SENSING APPLICATIONS IN CROP MONITORING: A SYSTEMATIC LITERATURE REVIEW

Abstract

The dynamic development of the Earth Observation (EO) data and the development of remote sensing technologies have radically transformed the process of monitoring crops on local and regional, as well as global scales. Remote sensing has become a very instrumental tool in crop monitoring and in aiding food security, yield estimation and agro climatic analysis and in agricultural policy planning. Despite significant proliferation of the satellite platforms and machine learning techniques, there is no synthesis between the pixel-wise crop mapping processes and functionality crop-specific land cover products. This literature gap is filled in with the systematic literature review where over 60 open-access operation datasets and peer-reviewed articles are analyzed in accordance with a systematic PRISMA-based method. It is analyzed in terms of satellite platforms, data fusion plans, classification algorithms and operation crop mapping products. Findings have indicated a high reliance on the multispectral sensors such as Landsat and Sentinel-2 and an increased use of radar and hyperspectral sensors is on the increase. The popular machine learning algorithms (including the Random Forest and Support Vector machines) are not as effective as deep learning models on large scale and high-resolution problems. Multi-source data fusion is highly significant in the enhancement of classification strength, and model generalization. However, ground truth data availability, inter-agro ecological zone model transferability and multi sensor dataset harmonization remain a problem. The paper also exposes the current trends that are taking place towards scalable deep learning systems and products of operational crop specific land cover but mentions the significant research gaps. In general, the review is a systemic summary of the history of technologies, evolution of approaches, and working implementation, which will inform the further research and the widespread implementation of the remote sensing-based crop surveillance systems.