Abstract: Root zone water quality model is widely used to describe the influence of soil hydrological cycle on crop growth, and guide the agriculture production management through model simulation. However, once calibration needs long time, it is a difficult task to find a suitable model parameters in an acceptable time; besides, traditional trial-and-error calibration method depends largely on user’s experience and expertise, and needs to try more times to get satisfactory simulation results. A sparse grid method is employed to construct surrogate model of RZWQM and random drift particle swarm optimization algorithm is applied in parameters optimization of surrogate model. The optimized parameters are used in practical simulation of RZWQM. The surrogate model has high precision and calibration speed, which greatly saves computational cost in parameters optimization. Five years of yield, drain flow, and [NO-3]-N loss data from a subsurface-
drained corn-soybean field in Iowa are employed in empirical analysis of sparse grid surrogate model with random drift particle swarm optimization. The results show that proposed method can greatly improve the efficiency of model parameter optimization and save manpower, and can get better performance of RZWQM model through evaluation index with PBIAS, NSE and RSR.

Key words: random drift particle swarm optimization algorithm, sparse grid, Root Zone Water Quality Model（RZWQM）, surrogate model

摘要： 根区水质量模型（Root Zone Water Quality Model，RZWQM）被广泛应用于刻画土壤水文循环过程对作物生长的影响，并通过模型率定模拟指导农业生产管理。然而RZWQM模型的一次率定需要较长时间，在可接受时间范围内找到一组合适的模型参数是一件较困难的工作；同时传统的模型参数试错法依赖于使用者的专业知识和经验，也需要多次尝试才能达到较满意的模拟效果。提出使用稀疏网格方法建立RZWQM模型的近似替代模型，并使用随机漂移粒子群优化算法对替代模型进行自动参数优化，将优化后的参数用于RZWQM模型的实际应用模拟。替代模型近似精度高，率定速度快，大大节省了模型参数优化的计算开销。最后将提出的稀疏网格近似替代模型方法结合随机漂移粒子群优化算法使用美国爱荷华州5年玉米-大豆间中的作物产量、排水流量、[NO-3]-N流失量田间实测数据进行了验证分析。结果显示该方法能够极大地提高模型参数优化效率和节省人力；同时，通过模型性能评价指标PBIAS、NSE和RSR的数值比较也表明该方法优化后的RZWQM模型性能要好于传统试错法的模型性能。

关键词: 随机漂移粒子群算法, 稀疏网格, 根区水质量模型（RZWQM）, 替代模型