Toward Low‐Level Turbulence Forecasting at Eddy‐Resolving Scales

Abstract

Microscale turbulence in the atmospheric boundary layer (ABL) is characterized by significant spatiotemporal variability. Consequently, a change in the turbulence forecasting paradigm needs to occur, moving beyond average turbulence estimates at mesoscale grid resolutions (several kilometers) to eddy‐resolving forecasts. To that end, the viability of dynamic downscaling to large‐eddy simulation scales is evaluated. We present for the first time, multiday dynamic downscaling from currently available numerical weather prediction forecasts to a high‐resolution grid spacing of 25 m. It is found that these eddy‐resolving forecasts can realistically reproduce turbulence levels and peak events in the bulk of the daytime ABL, adequately capturing turbulence variability at subminute intervals. Moreover, probability distributions of turbulence quantities are in very good agreement when compared to in situ sonic‐anemometer observations. These results demonstrate the feasibility of eddy‐resolving forecasts to derive accurate probabilistic estimates of turbulence in the ABL and provide a path toward real‐time large‐eddy simulation scale prediction.

Simulating Real Atmospheric Boundary Layers at Gray-Zone Resolutions: How Do Currently Available Turbulence Parameterizations Perform?

“…These measurements were used to validate the simulations performed for this month [23], and 5 days were chosen based on the model performance in simulating wind speed, wind direction and air temperature near the ground: 20, 21, 28, 29 and 30 March. The simulations analyzed here consider the convective portion of the diurnal cycle for these 5 days.…”

“…For each convective period, the mesoscale (i.e., the two largest domains) is run for 36 h. The smaller nests are initialized at hour 24, running for a total of 12 h: 2 h for spin up and 10 h for the final analysis. This setup has been previously demonstrated in a validation study that compared resolved turbulence quantities against sonic anemometer data from the XPIA campaign [23]. The cell-perturbation method [30,31] is used in RLES to ensure the development of small-scale turbulence as the mesoscale flow structures in the second domain (1-km grid spacing) are passed to the microscale LES in the third domain (100-m grid spacing).…”

“…In scientific literature, it is possible to find studies that treat the gray zone in various ways: fully parameterizing it and treating it like the mesoscale [17,18]; not parameterizing it and treating it like a very large-eddy simulation (VLES) [19,20]; using scale-aware PBL parameterizations, which can be modified versions of the traditional mesoscale schemes [21,22]; and skipping it altogether and going from coarser, mesoscale domains directly into finer, microscale ones [23]. Previous studies concerned with model development have investigated the ability of different parameterizations to model specific aspects of the flow for idealized convective conditions [12,[24][25][26].…”

“…Performing these studies is difficult because of the lack of reference datasets against which to quantify simulation performance; highly resolved meteorological observations are rare, and high-resolution, real-case WRF-LES require high-performance computing, massive data storage and unique expertise. Here, we leverage a highly resolved, WRF-LES dataset that was previously validated against meteorological observations [23] and can be used to evaluate the performance of other modeling approaches.…”

Formation of Wind Gusts in an Extratropical Cyclone in Light of Doppler Lidar Observations and Large-Eddy Simulations

“…On the modeling side, realistic large-eddy simulations are now possible for case studies. They can be used for the dynamical downscaling of numerical weather predictions and provide a framework to forecast turbulence at the 10-100-m scale (Muñoz Esparza et al 2018). The increase in computational power further allows running large-eddy simulations for a day over a domain extending over several 100 km, thus encompassing synoptic-scale systems (Heinze et al 2017).…”