High-precision flood simulation can effectively give early warning or assess the corresponding risk of urban flood disaster. Recently, high-resolution terrain data have been used to completely solve the two-dimensional (2D) shallow water equations for urban flood simulations, thus improving the reliability of the results. With the development of refined terrain data, the number of calculations required for model simulations has increased significantly, and traditional parallel computing methods are becoming relatively time-consuming. This article describes the application of a parallel algorithm in the compute unified device architecture framework. A 2D hydrodynamic parallel model is established and applied to the simulation of rainfall-induced flooding in Nangang District, Harbin, China. Compared with the serial calculation method, an acceleration ratio of 42 times is obtained. Finally, the influence of different grid resolutions on the simulation accuracy and speed is discussed. The research results provide a reference for the parallel calculation of high-precision flood simulations and the formulation of simulation schemes.

中文翻译：

---

基于计算统一设备架构的城市二维雨洪模型并行计算方法

高精度洪水模拟可以有效地预警或评估城市洪水灾害的相应风险。最近，高分辨率地形数据已被用于完全求解城市洪水模拟的二维（2D）浅水方程，从而提高了结果的可靠性。随着精细化地形数据的发展，模型模拟所需的计算量显着增加，传统的并行计算方法变得相对耗时。本文介绍了一种并行算法在计算统一设备架构框架中的应用。建立了二维水动力并行模型，并将其应用于哈尔滨市南岗区降雨诱发洪水的模拟。与串行计算方法相比，得到42倍的加速比。最后，讨论了不同网格分辨率对仿真精度和速度的影响。研究结果为高精度洪水模拟的并行计算和模拟方案的制定提供了参考。