Deep learning methodologies for crop models and decision support systems: A comprehensive review

Recently, the world has been facing challenges, such as a substantial increase in food demand and supply. Smart irrigation is an efficient and economical method to increase productivity (agriculture) and to meet the increase in food demand. A smart irrigation system ensures uniform water distribution, reduces water wastage and fuel consumption for irrigation, and improves crop production. A smart system monitors soil moisture and weather conditions and accordingly controls the irrigation valves to ensure optimal water usage. Smart irrigation system built using sensors or hyperspectral imaging and machine learning techniques has a remote monitoring unit for monitoring the irrigated fields. Hyperspectral imaging employed in smart irrigation possesses many advantages and overcomes the limitations of other contact sensing methods. In this method, the spectral reflectance of the plants for various soil moisture levels is monitored using hyperspectral imaging, and the soil moisture level is predicted using a machine learning technique. The machine learning algorithms predict the amount of water needed and guide the user on when to irrigate from the record of weather patterns and present soil moisture data. According to the predicted results, optimal soil moisture is maintained to ensure maximum crop production by realizing smart irrigation. This paper states the necessity and importance of smart farming. It presents an overview of the various contact and non-contact sensing methods and machine learning models employed to achieve smart farming. A detailed and comprehensive review of various smart irrigation practices' advantages, disadvantages, and feasibility is highlighted.