Global Synthetic Crop Yield, Meteorological, and Climate Teleconnection Dataset for Machine Learning Benchmarking

This dataset contains high-fidelity synthetic data representing global agricultural production, local meteorological conditions, and large-scale climate teleconnection indices spanning the period 1990–2023.

The dataset was generated to benchmark HESE-GNN-CP, a machine learning architecture designed to capture "teleconnections" (long-distance climate links) using Graph Neural Networks. Unlike standard agricultural datasets, this dataset explicitly models the physical correlations between global climate drivers (ENSO, NAO) and regional weather patterns, making it ideal for testing Spatial-Temporal Graph Neural Networks (ST-GNNs).