ECMWF Cloud and Radiation Parametrization: Recent Activities

1
ECMWF Cloud and Radiation Parametrization Recent
Activities

• Richard Forbes,
• Maike Ahlgrimm,
• Jean-Jacques Morcrette,
• Martin Köhler

Evaluation of models University of Reading,
17-18 Nov 2009
2
Some ECMWF Cloud/Radiation Recent Parametrization
Activities

1. Development of cloud and precipitation
   parametrization (prognostic variables and
   microphysical processes..)
2. Evaluation of cloud/precip with
   CloudSat/CALIPSO (Radar reflectivity)
3. Evaluation of cloud regimes (TCu - new
   dual-mass flux shallow convection scheme)
4. Representation of aerosol and radiative
   impacts (GEMS/MACC)

3
1. Cloud Scheme Developments
4
ECMWF Cloud Scheme Developments
Current Cloud Scheme
New Cloud Scheme

• 2 prognostic cloud variables (condensate cloud
  fraction) water vapour.
• Diagnostic liquid/ice split as a function of
  temperature between 0C and -23C.
• Diagnostic representation of precipitation.

• 5 prognostic cloud variables (liquid, ice,
  snow, rain, cloud fraction).
• Additional sources/sinks for new processes.
• New explicit/implicit solver

5
New 5-prognostic cloud microphysicsLiquid vs Ice
Fraction
New prognostic scheme
Current diagnostic scheme
1.0
Liquid Water Fraction
-23ºC
0ºC
0.0
Temperature
Temperature
6
New 5 prognostic cloud microphysics Ice vs. Snow
Model Ice Water Path (IWP) (1 year climate)
Current scheme (IWC)
Observed Ice Water Path (IWP)
CloudSat 1 year climatology From Waliser et al.
(2008)
g m-2
New scheme (IWCSWC)
7
VerificationAnnual average Ice Water Path from
Satellite
Widely varying estimates of IWP from different
satellite datasets!
CloudSat
8
2. Evaluation with CloudSat
9
Radar ReflectivityAlong-track model vs. CloudSat
comparison
Spatial distribution of cloud/precipitation
reflectivities generally very good!
However, there are some discrepancies that are
highlighted by the radar reflectivity comparison
10
Radar Reflectivity Statistics
Radar Reflectivity vs. Height Frequency of
Occurrence Tropics over ocean 30S to 30N for
February 2007
Peak reflectivities too high altitude (from
convective snow)
Relatively too frequent low-level high
reflectivity convective rainfall
Lack of low reflectivity mid-level and low-level
cloud ?
Significantly higher occurrence of cloud in model
but is this due to overestimating
the precipitation fraction?
11
3. Regime Evaluation (Maike Ahlgrimm)
12
Regime evaluation
Zonal cross-section of frequency of
cloud/precipitation occurrence

• Defining a regime
• Use criteria like cloud top height, cloud
  thickness, cloud fraction.
• Geographical region
• Use model (dynamical) quantities.
• Different issues for ground based, satellite
  (vertical profile vs, 2D view).

• Compositing
• To avoid focussing on potentially
  unrepresentative individual cases.
• To get large enough sample size without losing
  characteristics of cloud type.

Maike Ahlgrimm
13
Example Trade cumulus using CALIPSO
CALIPSO
Control

• Criteria
• Cloud top height lt4km
• Over ocean
• 30S to 30N
• Cloud fraction lt50

CALIPSO
DualM
Control
DualM
Compensating errors Model cloud occurs too
often, but has too little cloud fraction when it
occurs.
CALIPSO
Maike Ahlgrimm
14
Example Mid-latitude cold air outbreak

• Criteria from model
• Surface pressure 1015 hPa
• Potential temperature difference 700 hPa to
  lowest model level 9K
• Over ocean
• Add criteria from satellite, such as cloud top
  height.

15
4. Radiation and aerosol J-J Morcrette
16
Recent developments in aerosol representation in
the ECMWF IFS (GEMS)

• ECMWF IFS model including prognostic aerosols has
  been run in two configurations
• In aerosol free-wheeling mode aerosol advection
  and full (but simplified) aerosol physics using
  temperature, humidity, winds etc. from the
  analyses/forecasts every 12 hours
• In analysis mode with subsequent forecasts
• In both configurations, what is included is
• Sea salt aerosols (3 bins, 0.030.5520 mm)
• Dust aerosols (3 bins, 0.030.550.920 mm)
• Organic matter (hydrophilic, hydrophobic)
• Black carbon (hydrophilic, hydrophobic)
• Sulphate aerosols (SO4 from SO2 sources)

MISR AOD Jul 2003
Model AOD analysis Jul 2003
Morcrette et al. (2008) Benedetti et al. (2009)
17
AATSR
MERIS
SEVERI
MISR
GEMS
MODIS
18
Comparisons AERONET, ECMWF climatology, GEMS-AER,
GlobAEROSOL-SEVIRI (Azores)
Azores/Cabo Verde 500nm
19
To improve model parametrizations

• The challenge is to determine real differences
  between the model and observations, identify the
  most important physical processes, understand
  their interactions and improve their
  representation in the model.

20
Some Questions to Highlight

• How do we compare incompatible model and obs
  ? (different quantities, spatial and temporal
  scales, obs limitations/errors)
• Forward models/simulators/emulators
• Sub-columns or appropriate averaging
• Understand the observation limitations/errors
• How do we evaluate physical processes ?
• Regime-dependent evaluation (where particular
  processes dominate)
• Model sensitivity studies.
• Combining different observations to evaluate
  physical relationships?
• How do we disentangle model compensating errors ?
• Exploit synergy of different observations (to
  provide information on clouds, radiation,
  aerosol, water vapour all at the same time!)
• How important is variability on different spatial
  and temporal scales ?
• Need temporal and spatial heterogeneity from
  observations
• Cloud cover, cloud condensate, humidity,
  aerosols..

21
Questions ?
22
ECMWF Cloud Parametrization Representing
sub-grid variability
ECMWF cloud parametrization
In the real world
Cloud cover is integral under supersaturated part
of PDF
G(qt)
qt
A mixed uniform-delta total water distribution
is assumed
23
Radar ReflectivityCross-section through tropical
convection
CloudSat Radar Reflectivity
Model Radar Reflectivity (Ice, Liq, Snow, Rain)
Model Radar Reflectivity (Ice, Liq only)