In this paper, the simulation and modeling of the turbulent atmospheric boundary layers (ABLs) in the presence of forests are studied using a lattice Boltzmann method with large eddy simulation, which was implemented in the open-source program GASCANS with the use of Graphic Processing Units (GPU). A method of modeling forests in the form of body forces injected near the wall is revisited, while the effects of leaf area density (LAD) on the model accuracy is further addressed. Since a uniform cell size is applied throughout the computational domain, the wall-normal height of the near-wall cells is very large, theoretically requiring a wall function to model the boundary layer. However, the wall function is disregarded here when the forest is modeled. This approximation is validated based on the comparison with previous experimental and numerical data. It concludes that for the ABL conditions specified in this study as well as a large body of literature, the forest forces overwhelm the wall friction so that the modeling of the latter effect is trivial. Constant and varying LAD profiles across the forest zone are defined with the same total leaf area despite the varying one being studied previously. It is found that the two LAD profiles provide consistent predictions. The present forest modeling can therefore be simplified with the use of the constant LAD without degrading the model accuracy remarkably.

在本文中，使用具有大涡模拟的格子玻尔兹曼方法研究了森林存在的湍流大气边界层（ABLs）的模拟和建模，该方法在开源程序GASCANS中使用图形处理实现单位 (GPU)。重新审视了一种以壁附近注入的体力形式模拟森林的方法，同时进一步解决了叶面积密度 (LAD) 对模型精度的影响。由于在整个计算域中应用了统一的单元大小，因此近壁单元的壁法线高度非常大，理论上需要一个壁函数来模拟边界层。但是，在对森林进行建模时，此处忽略了壁函数。基于与以前的实验和数值数据的比较，验证了该近似值。它得出的结论是，对于本研究中指定的 ABL 条件以及大量文献，森林力压倒了墙壁摩擦，因此后一种效应的建模是微不足道的。尽管先前研究了不同的叶面积，但整个森林区域的恒定和变化的 LAD 剖面定义为相同的总叶面积。发现两个 LAD 配置文件提供了一致的预测。因此，可以使用常数 LAD 简化当前的森林建模，而不会显着降低模型的准确性。森林力压倒了墙壁的摩擦力，因此后一种效果的建模是微不足道的。尽管先前研究了不同的叶面积，但整个森林区域的恒定和变化的 LAD 剖面定义为相同的总叶面积。发现两个 LAD 配置文件提供了一致的预测。因此，可以使用常数 LAD 简化当前的森林建模，而不会显着降低模型的准确性。森林力压倒了墙壁的摩擦力，因此后一种效果的建模是微不足道的。尽管先前研究了不同的叶面积，但整个森林区域的恒定和变化的 LAD 剖面定义为相同的总叶面积。发现两个 LAD 配置文件提供了一致的预测。因此，可以使用常数 LAD 简化当前的森林建模，而不会显着降低模型的准确性。