Indian Wheat Yield & Irrigation Datasets for Machine-Learning Surrogate Modelling

This repository contains datasets associated with the study “Integrating AquaCrop-OSPy and machine learning to develop transferable surrogates for balancing wheat yield–irrigation trade-offs under deficit irrigation in semi-arid NW India.” The datasets were developed to investigate the use of machine-learning surrogates to emulate the yield and irrigation responses of the AquaCrop-OSPy (ACOSP) crop-water model under different soil-moisture-threshold-based (SMT) deficit-irrigation strategies. The Mahendragarh and Bhilwara datasets are synthetic datasets generated using calibrated AquaCrop-OSPy simulations. The ACOSP model was first calibrated for wheat using observed seasonal yields over 26 growing seasons. The calibrated model was then used to generate synthetic irrigation-management scenarios by varying combinations of SMT and maximum irrigation (MaxIrr), thereby generating corresponding wheat yield and irrigation responses. The datasets cover 26 wheat growing seasons from 1997–98 to 2022–23. The Sangrur dataset has a different purpose and provenance. It was generated from observed seasonal data only and was not produced using AquaCrop-OSPy. It provides the input features required for transferring the developed surrogate models to an unseen district and demonstrating their potential application for irrigation decision support. These datasets are simulation-derived and should not be interpreted as direct field observations.