Data from: County-Level Irrigation Water Demand Estimation Using Machine Learning: Case Study of California

This study presents a machine learning approach to estimate annual irrigation water demand at the county level in California, using Gaussian Process Regression (GPR) for improved predictive accuracy. Key input variables include meteorological parameters, geographical characteristics, and irrigated cropped area. The GPR model demonstrated high predictive accuracy (R² > 0.97, RMSE as low as 0.06 km³), identifying temperature, vapor pressure deficit, and irrigated area as the most influential factors. This research offers a robust tool for decision support in regional agricultural water management, enabling efficient evaluation of climatic and agricultural scenarios to optimize water resource allocation.

本研究提出了一种机器学习方法，旨在估算加利福尼亚州县级行政区每年的灌溉用水需求。该方法采用高斯过程回归（Gaussian Process Regression，GPR）以提高预测精度。关键输入变量包括气象参数、地理特征以及灌溉作物面积。GPR模型展现了卓越的预测准确性（R²值超过0.97，均方根误差（RMSE）低至0.06 km³），并确定了温度、蒸汽压亏缺以及灌溉面积是最为显著的影响因素。本研究提供了一种稳固的工具，以支持区域农业水资源管理的决策制定，并允许对气候和农业情景进行高效评估，以优化水资源分配。