Evaluating planetary boundary-layer schemes and large-eddy simulations with measurements from a 250-m meteorological mast

Alfredo Peña,Andrea N. Hahmann

doi:10.1088/1742-6596/1618/6/062001

Abstract

We evaluate the performance of six planetary boundary layer (PBL) schemes and a large-eddy simulation (LES) for the characterization of the neutral PBL through idealized simulations using the Weather Research and Forecasting model. The evaluation is performed by comparison with sonic anemometer measurements from a 250-m tall meteorological tower that observes close-to-homogeneous winds under the predominant westerlies. All simulations show similar behavior for the vertical temperature profile except for one that uses a non-local PBL scheme, which is known to produce excessive vertical mixing. Within the range of measurements of the tower and except for a PBL scheme that produces a too low jet, the simulations using PBL schemes, which were developed to simulate the nighttime atmosphere, generally show the largest deviations from the observed wind speeds. As expected within the surface layer, the LES shows excessive vertical shear. More importantly, we find that two PBL schemes that use their own surface-layer scheme are effectively reducing the surface roughness. When looking at the vertical profile of momentum exchange coefficient, we find very good agreement between a local and a non-local PBL scheme within the bulk of the PBL, the highest values for the simulation with the PBL scheme showing excessive vertical mixing and the lowest values for that most suitable for nighttime conditions. The comparison with the observed turbulent kinetic energy reveals a good match between the LES and a simulation using a local PBL scheme and a general underestimation (overestimation) of turbulence within the range of measurements by the PBL schemes (LES).

Highlights

Planetary boundary layer (PBL) schemes are used in numerical weather prediction (NWP) models to simulate mesoscale processes as accurately as possible, because it is still not computationally affordable to predict weather with higher fidelity models, such as those that have the ability to perform large-eddy simulations (LESs)
For the LES, we select the outputs at 11 h because, at this hour, we find the highest value of spatial-averaged horizontal wind speed, U = ux 2 + uy 2 1/2, where represents a spatial average, within the 24-h simulation period
When looking at the whole planetary boundary layer (PBL), we find very good agreement for the potential temperature between all simulations, with Medium-Range Forecast (MRF)-MOST1 being the most different among them

Summary

Introduction

Planetary boundary layer (PBL) schemes are used in numerical weather prediction (NWP) models to simulate mesoscale processes as accurately as possible, because it is still not computationally affordable to predict weather with higher fidelity models, such as those that have the ability to perform large-eddy simulations (LESs). Turbulent eddies are responsible for the exchanges of moisture, heat, and momentum occurring within the PBL. They operate on spatio-temporal scales that cannot be explicitly represented using the grid resolution and time steps of most NWP models [1]. The effect on the wind speed of simulations using different PBL schemes has been intercompared and evaluated against observations of atmospheric parameters in many different studies [2, 3]. In the majority of these studies, it is difficult to adequately evaluate the ability of the PBL scheme to characterize the PBL processes because the PBL schemes have been

Methods

Results

Conclusion