A NeuralFAO56 Python Package for data-driven Irrigation Demand Calculation

The accurate estimation of crop evapotranspiration (ETc), root zone soil moisture depletion, and irrigation demands is critical for optimizing water resource management and enhancing sustainability in precision agriculture. The FAO-56 model serves as a foundational tool for these predictions; however, its conventional workflow necessitates the manual acquisition of essential inputs such as climatic data and soil moisture from disparate external sources. This process can be time-intensive, cost-prohibitive, and susceptible to human error. Furthermore, the deterministic nature of FAO56 can lead to inaccuracies if reference evapotranspiration and crop coefficients are not meticulously estimated. This study introduces NeuralFAO56, a Python package that integrates advanced machine learning models and real-time data acquisition with the FAO-56 framework to automate and improve the estimation of ETc and irrigation demands. By leveraging application programming interfaces (APIs) to automatically collect real-time climatic data from meteorological stations and NASA’s Soil Moisture Active Passive (SMAP), NeuralFAO56 dynamically updates model inputs. The package incorporates a range of machine learning models, including Long Short-Term Memory (LSTM) and transformer architectures, to generate data-driven ETc estimations, thereby enhancing the accuracy and adaptability of irrigation predictions. NeuralFAO56 is designed with a modular architecture, enabling users to customize its functionalities for diverse agro-hydrological contexts. This tool provides a robust, user-friendly platform for researchers, water resource managers, and agricultural professionals, facilitating intelligent irrigation decision-making, improving water-use efficiency, and contributing to sustainable agricultural practices.