Farm-Wise Estimation of Crop Water Requirement of Major Crops Using Deep Learning Architecture

Abstract

Each crop has different cultivation practices with various phases including seed treatment, soil management, land preparation, sowing of seeds, irrigation, application of fertilizers, etc. Irrigation is a very important phase of any crop cultivation practice. Irrigation scheduling, water management, crop forecasting, and demands precise crop-specific water requirements (CWR) which nowadays become extremely important for various crops grown under irrigation, especially in arid and semi-arid regions. This study will enable efficient use of water and better irrigation practices like scheduling as the supply of water through rainfall is limited in some areas. In growing crops, irrigation scheduling is a critical management input to ensure optimum soil moisture status for proper plant growth and development as well as for optimum yield, water use efficiency, and economic benefits. Operational CWR methods in India are mainly based on sparsely located in situ measurements and high-resolution remote sensing data, which limit the overall precision. To overcome the mentioned challenge, the deep learning architecture and soil moisture techniques have been used in this study to generate high-resolution farm boundaries, followed by the generation of crop maps and then generated soil moisture at the parcel level using our company’s own algorithms to estimate the farm-specific CWR. Over most of the farms, a direct positive relationship has been observed between the crop growing period and its particular CWR. The irrigation scheduling module of the Agrogate platform developed is currently used in many states by different stakeholders for the proper management of water resources.