Evaluating forecasts of the evolution of the cloudy boundary layer using radar and lidar observations Andrew Barrett, Robin Hogan and Ewan O’Connor Submitted to Geophys. Res. Lett.
Introduction Stratocumulus interacts strongly with radiation –Important for forecasting surface temperature –A key uncertainty in climate prediction Very difficult to forecast because of many factors: –Surface sensible and latent heat fluxes: to first order, sensible heat flux grows the boundary layer while latent heat flux moistens it –Turbulent mixing, which transports heat, moisture and momentum vertically –Entrainment rate at cloud top –Drizzle rate, which depletes the cloud of liquid water Use Chilbolton observations to evaluate the diurnal evolution of stratocumulus in six models
Models Used
Longwave cooling Different mixing schemes Virtual potential temp. ( v ) Height ( z ) d v /dz<0 Eddy diffusivity ( K m ) (strength of the mixing) Local mixing scheme (e.g. Meteo France) Local schemes known to produce boundary layers that are too shallow, moist and cold, because they don’t entrain enough dry, warm air from above (Beljaars and Betts 1992) Define Richardson Number: Eddy diffusivity is a function of Ri and is usually zero for Ri>0.25
Longwave cooling Different mixing schemes Use a “test parcel” to locate the unstable regions of the atmosphere Eddy diffusivity is positive over this region with a strength determined by the cloud-top cooling rate (Lock 1998) Virtual potential temp. ( v ) Height ( z ) Eddy diffusivity ( K m ) (strength of the mixing) Non-local mixing scheme (e.g. Met Office, ECMWF, RACMO) Entrainment velocity w e is the rate of conversion of free-troposphere air to boundary-layer air, and is parameterized explicitly
Longwave cooling Different mixing schemes Model carries an explicit variable for TKE Eddy diffusivity parameterized as K m ~TKE 1/2, where is a typical eddy size Virtual potential temp. ( v ) Height ( z ) Prognostic turbulent kinetic energy (TKE) scheme (e.g. SMHI-RCA) d v /dz<0 d v /dz>0 TKE generated TKE destroyed TKE transported downwards by turbulence itself
Observed Radar And Lidar Figures from cloud-net.org
Cloud values compared Cloud Existence Cloud Top Cloud Base Cloud Thickness Liquid Water Path Observed Cloud Fraction ECMWF Model Cloud Fraction
Results UKMO- Meso UKMO- Global ECMWFMeteo France RACMOSHMI- RCA Cloud Clear Sky Cloud Clear Sky Cloud Clear Sky Cloud Clear Sky Cloud Clear Sky Cloud Clear Sky Observed Cloud Observed Clear Sky Log of Odds Ratio, ln θ ln θ = ln (A D / B C)
Cloud Height Comparison UKMO- Meso UKMO- Global ECMWFMeteo France RACMOSMHI-RCA Height (km) Observations Model
Composite over diurnal cycle
Liquid Water Composite
Biases and random errors Worst two models in terms of bias and random error Tendency for all models to place cloud too low
Fluxes of Latent And Sensible Heat
Profiles of Model Temperatures
Conclusions Met Office Mes best at placing clouds at right time Met Office, ECMWF & RACMO best diurnal cycle –All use non-local mixing with explicit entrainment –Met Office and ECMWF clouds too low by 1 model level –RACMO height good: ECMWF physics but higher res. Meteo-France clouds too low and thin –Local mixing scheme underestimates growth SMHI-RCA clouds too thick and evolve little through the day –Only model to use prognostic turbulent kinetic energy