A Modulated-Gradient Parametrization for the Large-Eddy Simulation of the Atmospheric Boundary Layer Using the Weather Research and Forecasting Model
A Modulated-Gradient Parametrization for the Large-Eddy Simulation of the Atmospheric Boundary Layer Using the Weather Research and Forecasting Model
复制标题
DOI：
10.1007/s10546-017-0287-5
复制
发表时间：
2017-08
影响因子：
4.3
通讯作者：
S. Khani;F. Porté-Agel
中科院分区：
文献类型：
--
作者：
S. Khani;F. Porté-Agel
使用天气研究和预报模型中的中性大气边界层的大涡模拟（LES）来研究调制梯度亚网格尺度（SGS）模型的性能。由于模型包含空间导数的有限差分格式，因此离散化误差可能会影响模拟结果。我们在这里重点了解有限差分格式对动量平衡和平均速度分布的影响，以及特设冠层模型的要求（或不要求）。我们发现，与 Smagorinsky 和湍流动能 (TKE) 模型不同，使用调制梯度模型计算的平均速度和垂直剪切与 Monin-Obukhov 相似理论非常一致，而不需要额外的近壁冠层模型。与经典的 Smagorinsky 和 TKE 模型相比，使用调制梯度模型可以更好地解析近壁湍流涡流的结构，而经典的 Smagorinsky 和 TKE 模型耗散太大，并且对最小解析尺度产生不切实际的平滑。此外，与从常规 Smagorinsky 和 TKE 模型获得的 SGS 通量相比，从调制梯度模型获得的 SGS 通量在壁面附近要小得多。 LES 模型明显无法使用表面附近的总应力（已解析加上 SGS）来再现动量平衡的平均流向分量，这可能是由于离散化误差的影响，可以使用已解析速度场的泰勒级数展开来进行后验计算。总的来说，我们证明了与经典的 Smagorinsky 模型相比，调制梯度模型耗散更小，并且产生更准确的结果，并且计算成本相似。
The performance of the modulated-gradient subgrid-scale (SGS) model is investigated using large-eddy simulation (LES) of the neutral atmospheric boundary layer within the weather research and forecasting model. Since the model includes a finite-difference scheme for spatial derivatives, the discretization errors may affect the simulation results. We focus here on understanding the effects of finite-difference schemes on the momentum balance and the mean velocity distribution, and the requirement (or not) of the ad hoc canopy model. We find that, unlike the Smagorinsky and turbulent kinetic energy (TKE) models, the calculated mean velocity and vertical shear using the modulated-gradient model, are in good agreement with Monin–Obukhov similarity theory, without the need for an extra near-wall canopy model. The structure of the near-wall turbulent eddies is better resolved using the modulated-gradient model in comparison with the classical Smagorinsky and TKE models, which are too dissipative and yield unrealistic smoothing of the smallest resolved scales. Moreover, the SGS fluxes obtained from the modulated-gradient model are much smaller near the wall in comparison with those obtained from the regular Smagorinsky and TKE models. The apparent inability of the LES model in reproducing the mean streamwise component of the momentum balance using the total (resolved plus SGS) stress near the surface is probably due to the effect of the discretization errors, which can be calculated a posteriori using the Taylor-series expansion of the resolved velocity field. Overall, we demonstrate that the modulated-gradient model is less dissipative and yields more accurate results in comparison with the classical Smagorinsky model, with similar computational costs.