Investigation of Turbulence Parametrization Schemes with Reference to the Atmospheric Boundary Layer Over the Aegean Sea During Etesian Winds

Abstract

The spatial structure of the marine atmospheric boundary layer (MABL) over the Aegean Sea is investigated using the Weather Research and Forecasting (WRF) mesoscale model. Two ‘first-order’ non-local and five ‘1.5-order’ local planetary boundary-layer (PBL) parametrization schemes are used. The predictions from the WRF model are evaluated against airborne observations obtained by the UK Facility for Airborne Atmospheric Measurements BAe-14 research aircraft during the Aegean-GAME field campaign. Statistical analysis shows good agreement between measurements and simulations especially at low altitude. Despite the differences between the predicted and measured wind speeds, they reach an agreement index of 0.76. The simulated wind-speed fields close to the surface differ substantially among the schemes (maximum values range from 13 to 18ms-1 at 150-m height), but the differences become marginal at higher levels. In contrast, all schemes show similar spatial variation patterns in potential temperature fields. A warmer (1–2 K) and drier (2–3gkg-1) layer than is observed, is predicted by almost all schemes under stable conditions (eastern Aegean Sea), whereas a cooler (up to 2 K) and moister (1–2gkg-1) layer is simulated under near-neutral to nearly unstable conditions (western Aegean Sea). Almost all schemes reproduce the vertical structure of the PBL and the shallow MABL (up to 300 m) well, including the low-level jet in the eastern Aegean Sea, with non-local schemes being closer to observations. The simulated PBL depths diverge (up to 500 m) due to the different criteria applied by the schemes for their calculation. Under stable conditions, the observed MABL depth corresponds to the height above the sea surface where the simulated eddy viscosity reaches a minimum; under neutral to slightly unstable conditions this is close to the top of the simulated entrainment layer. The observed sensible heat fluxes vary from −40 to 25Wm-2, while the simulated fluxes range from −40 to 40Wm-2; however, all of the schemes’ predictions are close to the observations under unstable conditions. Finally, all schemes overestimate the friction velocity, although the simulated range (from 0.2 to 0.5ms-1) is narrower than that observed (from 0.1 to 0.7ms-1).