Transcript Evaluating forecasts of the evolution of the cloudy

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
Institute
Model
Horizontal
resolution
(km)
Met Office
Mesoscale
12
38 (12)
277
Met Office
Global
60
38 (12)
277
ECMWF
Integrated
Forecast
39
60 (16)
235
Météo France
ARPEGE
24
41 (14)
227
18
40 (14)
235
44
24 (8)
358
Royal Netherlands
Meteorological
Institute (KNMI)
Swedish
Meteorological and
Hydrological Institute
(SMHI)
Regional
Atmospheric
Climate Model
(RACMO)
Rossby Centre
Regional
Atmospheric
Model (RCA)
Vertical
levels
(BL, <3km)
Grid-box
depth at
1km (m)
Different mixing schemes
Local mixing scheme (e.g. Meteo France)
Height (z)
Longwave cooling
• Define Richardson
Number:
g dq v
q v dz
Ri 
shear 2
dqv/dz<0
Virtual potential temp. (qv)
Eddy diffusivity (Km)
(strength of the mixing)
• Eddy diffusivity is a
function of Ri and
is usually zero for
Ri>0.25
• 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)
Different mixing schemes
Non-local mixing scheme (e.g. Met Office, ECMWF, RACMO)
Height (z)
Longwave cooling
Virtual potential temp. (qv)
Eddy diffusivity (Km)
(strength of the mixing)
• 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)
• Entrainment velocity we is the rate of conversion of free-troposphere
air to boundary-layer air, and is parameterized explicitly
Different mixing schemes
Prognostic turbulent kinetic energy (TKE) scheme (e.g. SMHI-RCA)
Height (z)
Longwave cooling
TKE generated
dqv/dz<0
dqv/dz>0
TKE transported
downwards by
turbulence itself
• Model carries an explicit
variable for TKE
• Eddy diffusivity parameterized
as Km~TKE1/2l, where l is a
typical eddy size
TKE destroyed
Virtual potential temp. (qv)
dTKE
 shear production  buoyancy production  transport  dissipation
dz
Observed Radar And Lidar
Figures from
cloud-net.org
Cloud values compared
Observed
Cloud
Fraction
ECMWF
Model
Cloud
Fraction
• Cloud Existence
• Cloud Top
• Cloud Base
• Cloud Thickness
• Liquid Water Path
Composite over diurnal cycle
Liquid Water Composite
Biases and random errors
Model
Cloud
Cloud Top Cloud Base
Thickness
(m)
(m)
(m)
Met Office Mesoscale
-213 ± 370
-190 ± 307
-22 ± 429
Met Office Global
-270 ± 416
-287 ± 365
+18 ± 485
ECMWF
-126 ± 378
-84 ± 354
-42 ± 416
Météo France
-567 ± 415
-326 ± 366
-241 ± 443
KNMI – RACMO
63 ± 432
-115 ± 357
178 ± 495
SMHI-RCA
-46 ± 778
-387 ± 436
341 ± 823
Worst two models in terms of bias and random error
• Tendency for all models to place cloud too low
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