Coupling AquaCrop and machine learning approaches for cotton yield simulation

Estimating crop yield helps farmers plan for equipment needs, labor and fuel requirements, and cash-flow budgeting. Forecasting crop yield is also useful for analyzing weather-related variability to guide decisions such as irrigation scheduling and crop management. Decision support systems have previously been developed using deep learning models with success in several fields. The objective of this paper is to use multiple deep-learning techniques to determine the expected seasonal yield to guide farmers in their daily decision tasks. We used multiple machine learning algorithms to simulate cotton yield values predicted by Aquacrop using weather, irrigation, soil water content and crop growth stage data. Machine learning algorithms such as Random Forest (RF), Gated Recurrent Unit (GRU) and Multi Linear Perceptron (MLP) were used. These approaches were implemented and validated on a 7-acre cotton crop field irrigated by a center pivot located at Clemson University Edisto Research and Education Centre (REC), near Blackville, South Carolina, USA. Analysis suggested that the three algorithms performed satisfactorily with very good to excellent performances. We conclude that machine learning algorithms are useful tools that can provide insights into how much yield to expect in an upcoming season and help farmers optimize energy, water, and fertilizers applications accordingly.