Quantitative remote sensing for agricultural monitoring

Abstract

Agriculture is one of the main essential economic activities of the area of material production. It is the main food supplier for the population and provides a number of industries with raw materials of vegetable and animal origin. Consequently, improving the productivity of crop yield is essential. An effective solution to this is smart agriculture which is possible only on the basis of the creation and implementation of information technologies, including technology for agricultural production. The most promising and scalable resource of information support for agricultural production are methods and means of remote sensing of the Earth. The satellite technologies provide users with opportunities not only to search and obtain objective primary information but also its processing, as well as analyzing the state of agro-ecosystems using distributed computing resources. Crop mapping and yield prediction are two key aspects of agricultural monitoring that remain challenging due to a lack of timely, accurate, and qualitative data availability. Existing methods depend on ground surveys means time-consuming, expensive; requiring more men power and laborious. Time series of satellite imagery provides continuous and regular information to farmers about crop growth as well as suggests proper crop management to obtain better yield production. This research work present Sentinel-2 data utilization for crop classification and yield production by machine learning methodology. The Sentinel-2 is the latest satellite of earth observation, which provides high spatial resolution observations on Earth's surfaces and new possibilities for agricultural monitoring. Different vegetation indices were selected to show the variances within an experimental rice crop yield. Statistical graphs and functional data analysis were used in this study for crop yield estimation. Based on vegetation indices and environmental data, under the random forest model, crop yield estimation and prediction were successfully recognized as well as identifying which regions of the study area have the lowest yield and highest yield at the early stage of the crop. The main aim of this research work is to illustrate how to use earth observation satellite data and secondary data such as environmental data for agriculture mapping, monitoring, prediction, sustainable development of agriculture fields, and suggestions. The aims were achieved by the combination of primary and secondary data under the latest algorithms such as machine learning for accurate yield estimation. This analysis work provides key information about crop management, farming practices, and suggestions for policy and decision-making about better yield production. Finally, this research work demonstrated the latest and innovative methods based on satellite techniques in place of old and existing methods under augmented environmental management for improving farmer's life and communities to compete with future food demand in a sustainable way.