利用廣義概似不確定性估計法校正 DNDC 模式

Abstract

本研究利用廣義概似不確定性估計法 (Generalized Likelihood Uncertainty Estimation, GLUE) 對 DeNitrification-DeComposition (DNDC) 模型進行校正，以模擬 2023 年在安康農場種植的二期作水稻試驗的產量與溫室氣體排放量。研究以敏感度分析探討參數與感興趣的模擬值之間的關係，並挑選出對模擬值敏感的參數，包括作物相關參數 (如最大產量與積溫) 及土壤參數 (如有機碳含量與孔隙率)。 研究結果顯示，GLUE 方法可以提升模型的準確性，也提供一種自動化且系統化的參數校正流程，可應用於 DNDC 模型的參數優化。未來研究應結合不同的氣候條件與田間數據進行校正，並採用貝式迭代方法來估計參數，以提升模型的普遍性與預測能力。

This study uses the Generalized Likelihood Uncertainty Estimation (GLUE) method to calibrate the DeNitrification-DeComposition (DNDC) model, simulating the yield and greenhouse gas emissions of the second-cropping rice experiment conducted at the AnKang Farm in 2023. Sensitivity analysis was employed to explore the relationship between parameters and the simulation values of interest, identifying the parameters that are sensitive to the simulation outcomes, including crop parameters (e.g., maximum yield and growing degree days) and soil parameters (e.g., organic carbon content and porosity). The results demonstrate that the GLUE method not only improves the accuracy of model simulations but also provides an automatic and systematic parameter calibration process, which could be applied to optimize DNDC model parameters. Future studies should incorporate different climate conditions and field data for calibration and adopt Bayesian iterative methods to estimate parameters, thereby enhancing the model’s generalizability and predictive capability.