珠江三角洲洪潮实时预报关键技术

均一化迭代是基于栅格数据的洪水演进模拟常采用的迭代策略，存在设定迭代步长与栅格单元实际所需时间不一致的问题，直接影响到栅格单元水位、流量的计算精度。针对该问题，提出了一种双层异步迭代策略。外层迭代控制洪水演进时刻，内层迭代则通过分析不同栅格单元洪水的流速特征，实现迭代步长及迭代次数的自适应设定。采用商用FloodArea软件、均一化迭代算法及双层异步迭代算法，分别对福建省万安流域暴雨洪水历史数据进行模拟。该算法模拟结果的平均误差比FloodArea小0.361 m，比均一化迭代算法小0.654 m，说明该算法能有效提高洪水演进模拟的整体精度。

Key Technologies of Real-time Flood and Storm Surge Forecasting for Pearl River Delta

In the uniform iterative algorithm for hydrodynamic flood routing model, the iteration step and the actual required time of each grid are inconsistent, and directly affects the precision of water level and discharge estimation for each grid cell. To solve this problem, a double-layer asynchronous iterative algorithm is proposed. The outer iterative process controls the flood routing time, while the inner iterative process adaptively selects the step size and the times for iterations by analyzing the flood velocity characteristics of different grid cells. FloodArea, the uniform iterative algorithm and the double-layer asynchronous iterative algorithm are applied to simulate historical heavy rain flood processes in the Wanan basin in Fujian province. Experimental results show that the simulated results of the double-layer asynchronous iterative algorithm are most consistent with actual disaster data. The average error in the proposed algorithm is less than FloodArea, 0.361 m and less than the uniform iterative algorithm, 0.654 m. The proposed method effectively improves the integral accuracy of flood routing simulation.