Comprehensive Climate, Soil, and Management Dataset for Cotton Irrigation in Xinjiang, China

This study selected one representative research region in southern Xinjiang and one in northern Xinjiang. Using soil sensors and UAV technology, we systematically collected and compiled a data‑driven integrated dataset. The dataset primarily includes climate and weather data, real‑time soil sensor data (moisture, temperature, fertility), crop management records, visible and multispectral images of cotton plants, spore and insect images, as well as agronomic and yield data. By integrating publicly available historical meteorological records with observations from on‑site weather stations, we constructed a multi‑year dataset that elucidates the long‑term relationships among climatic conditions, soil properties, and crop yields at the experimental sites. To validate the utility of this dataset, we employed various machine learning methods (such as reinforcement learning, deep neural networks, and XGBoost) to model the dynamic relationship between the data and crop yield, thereby optimizing irrigation management strategies.