|Review of the revised manuscript GMD-2019-165: “Implementation of a synthetic inflow turbulence generator in idealised WRF v3.6.1 large eddy simulations under neutral atmospheric conditions” by Jian Zhong, Xiaoming Cai1 and Zheng-Tong Xie submitted for publication in the journal Geoscientific Model Development.|
In the revised manuscript “Implementation of a synthetic inflow turbulence generator in idealised WRF v3.6.1 large eddy simulations under neutral atmospheric conditions” the authors have made relatively minor changes to the exposition of the methodology for generation of inflow turbulence for large-eddy simulations based on synthetic turbulence generation approach developed by Xie and Castro (2008) implemented in the Weather Research and Forecasting model. In addition, the results presented in the revised manuscript are improved in comparison to the original manuscript.
While, the revised manuscript includes improved results, the main deficiency of the manuscript remains – the synthetic inflow turbulence generation methodology developed by Xie and Castro (2008) was previously implemented in the Weather Research and Forecasting model by Muñoz-Esparza et al. (2015) and extensively tested. It is therefore not clear what new research is presented in this manuscript. One element that is explored in greater detail is integral length scale, however, the authors do not clearly articulate any potential differences in implementation or results of simulations. Furthermore, the authors claim that the simulations presented in the manuscript are of neutral atmospheric conditions, however, Coriolis force was not used in the simulations and therefore important characteristic of a real atmospheric boundary layer, namely wind veering, could not be reproduced. Finally, although minor changes to the background material and exposition were introduced, some of these include misrepresentation of previous work. Examples are given below under Specific Remarks.
Taking all the above into account I do not recommend the manuscript for publication in the journal Goescientific Model Development.
Page 2, line 13 – It is stated that: “Therefore such periodic WRF-LES simulations are restricted to studies of the atmospheric boundary layer flow with a single domain (e.g. Zhu et al., 2016; Kirkil et al., 2012; Kang and Lenschow, 2014; Ma and Liu, 2017) or the outmost domain for the nested 15 cases (e.g. Moeng et al., 2007; Khani and Porte-Agel, 2017; Nunalee et al., 2014).” However, the simulation listed here are quite different. While Moeng et al. (2007) present simulations included two-way nested domains where periodicity on the outer domain impacts the flow on the inner domain, Nunalee et al. (2014) present a one-way nested simulations where inner domain is not impacted by periodicity on the outer domain, so that such setup can be used for simulation of heterogeneous boundary layers.
Page 3, line 21 – It is stated that: “Generally speaking, these methods impose “white-noise” perturbations, thus having a flat spectrum, to a variable (e.g. temperature) at the inlet, and the model dynamics will “process” the signals once these signals are advected into the domain, e.g. to dissipate high-wavenumber signals quickly and to adjust low-wavenumber signals gradually.” This statement misrepresents the temperature perturbation methodology. Temperature perturbations are introduced at a specific length and time scale related to the highest well resolved wave number in LES and therefore they cannot be considered “white noise.” White noise can be defined as “random signal having equal intensity at different frequencies, giving it a constant power spectral density.”
Page 3, line 25 – It is stated: “It is thus not surprising that a large distance of about 20-40 boundary-layer depths 25 (Munoz-Esparza et al., 2015; Mazzaro et al., 2019) is normally required to allow a transition to fully-developed turbulence.” However, Munoz-Esparza et al. (2015) demonstrated that “From those results, it is evident that the performance of the cell perturbation method is not affected by these factors, rather induces turbulent structures that become fully developed at x/zi0 ≈ 15.” See also Figure 18 in Munoz-Esparza et al. 2015.
Furthermore, the difference between the application of synthetic inflow turbulence generator presented in the manuscript and the temperature perturbation methodology presented in Munoz-Esparza et al. (2015) is not recognized by the authors. Temperature perturbation is introduced for mesoscale to microscale coupling approach where smooth mesoscale flow (no resolved turbulence) is forcing microscale flow and for that purpose one-way nesting approach is used in WRF. The nesting approach necessarily represents a different challenge for transition to fully-developed turbulence due to nesting compared to just specifying inflow turbulence, e.g., there is significant difference in inflow turbulence levels between the results presented in the manuscript and those in Munoz-Esparza et al. (2015).
Page 10, line 6 – It is stated: “The spectrum in Munoz-Esparza et al. (2015) drops steeper at high wave numbers, mainly due to a coarser resolution…” The drop in the spectra is related to the implicit filter associated with the numerical discretization and drops at lower wave numbers due to coarser resolution, but it does not drop steeper. Steepness of the drop is related to the implicit filter. It has been determined that the effective resolution of WRF is ~7 dx (Skamarock 2004).