<p>This paper examines the associations between climate variability, geopolitical uncertainty, and agricultural commodity price dynamics under increasing environmental instability and compound shocks. Understanding how climate-related pressures are linked to commodity market fluctuations is essential for climate-sensitive market assessment and the design of resilient policy responses. To address this issue, the study develops a structure-regularized neural framework inspired by the regularization logic of PINN-based learning, which integrates high-frequency environmental indicators and geopolitical risk measures while imposing dynamic constraints on price trajectories. Using daily data on major agricultural commodities in China, the proposed framework captures central price dynamics and periods of heightened volatility associated with climate stress and geopolitical disruption. The empirical results indicate that periods of climate stress and geopolitical tension are linked to stronger price fluctuations, particularly during episodes of environmental or political instability. Compared with conventional benchmarks, the proposed model delivers competitive and often superior forecasting performance, while producing smoother and more structurally consistent price trajectories. By combining machine learning with structural regularization, this study contributes to the literature on climate-sensitive agricultural market forecasting. The proposed framework may support market monitoring and risk assessment by helping identify periods in which climate and geopolitical pressures are associated with increased commodity price instability.</p>