TAWEPI

1
TAWEPI Thorpex Arctic Weather and Environmental
Prediction Initiative summary of modelling
data assimilation activities
Ayrton Zadra Meteorological Research
Division Environment Canada on behalf of
TAWEPIs team of researchers
Arctic System Model Workshop III July 16th,
2009 UQAM, Montreal, Quebec
2
TAWEPI projects team
Primary objective develop validate Polar- GEM,
an experimental regional Numerical Weather
Prediction model over the Arctic
RPN
CCCma
modelling snow over sea-ice
sea-ice modelling
modelling polar clouds
stratospheric analyses
singular vector sensitivity studies
validation assimilation of polar-orbiting
satellite data
ARMA
__________________________________________________
_________ http//collaboration.cmc.ec.gc.ca/scienc
e/rpn/tawepi/en/index.html
_______________________ for 1- to 2-day
forecasts
3
Collaboration Status of extended regional model
at CMC

• Polar extension of CMCs regional NWP model
• global, rotated, variable-resolution
• lat-lon grid
• core 15-km resolution
• 58 hybrid vertical levels, top 10 hPa
• timestep 7.5 min
• 4 runs per day
• - 00Z 12Z runs 48h forecasts
• - 06Z 18Z runs 54h forecasts
• Implementation
• implemented in March 2009
• has become the CMCs
• operational regional model
• extension into the middle
• atmosphere (raising of model lid

Figure Grid of CMCs new regional model (Note
Only every 5 grid-point is shown)
__________________________________________________
___________________________ Project partly
funded by IPY-LIEP. Grid parameters kindly
provided by A. Patoine (CMC).
4
Collaboration Status of extended regional model
at CMC

• A continental
• GEM-LAM version
• for the CMCs
• regional NWP model
• project led by L. Fillion
• (RPN)
• may become operational
• in the next 12 months

Figure Grid of REGLAM15 (in red) and CMCs
currently operational regional model (in blue)
________________________________________ Image
kindly provided by E. Lapalme (CMC).
5
Subproject 1 Coupling blowing snow, snow and
iceY.-C. Chung, S. Bélair, J. Mailhot
For more details, see poster The impact of
blowing snow on Arctic sea ice and snow results
from an improved sea-ice / snow / blowing snow
coupled system, by Y.-C. Chung et al., in
session 21-J (Tue Jul 21).

• Goal
• Examine the effect of blowing snow on the
  simulation of snow and sea ice in the Arctic
  Ocean.
• Blowing snow effect
• saltation and suspension
• sublimation from blowing snow
• - 32 of snowfall on the Arctic coast of
  Alaska
• (Benson et al. 1982)
• - 37-85 mm of snow water equivalent (SWE)
• over Canadian Arctic tundra (Essery et
  al.1999)
• accumulation and erosion

Blowing snow on the ice shelf edge near Rampen
(72S, 16W), Antarctica, from Dr. R. Bintanja.
Institute for Marine and Atmospheric Research
Utrecht (IMAU), Utrecht University, the
Netherlands

• 1-D, blowing snow model, PIEKTUK (Déry, 2001)

snow
1-D, multi-layer snow model SNTHERM (Jordan, 1991)
ocean
Multi-layer, thermodynamic sea ice model from
Meteorological Service of Canada (MSC)
operational forecasting system run 1-D, offline
model
6
Fig 1 - Observed wind speed versus
PIEKTUK-simulated blowing snow sublimation

• 1- Sublimation due to blowing snow
• The accumulated sublimation is significant during
  SHEBA year (total of 56mm SWE)

Fig 2 - Simulated evolution of snow grain size
mm in snowpack

• 2- Impact of blowing snow on snow and ice
  properties
• Blowing snow improves the estimates of snow depth
  and temperature at snow/ice interface
• Blowing snow has a small impact on ice thickness
  and snow structure (e.g. grain size and density)

7
Subproject 2 P. Vaillancourt (PI), J.
Milbrandt, F. Chosson Polar-GEM clouds

• Motivation
• Especially in Arctic, clouds are poorly
  represented in NWP and RCM, e.g.
• too low cloud cover and condensed water
• too high albedo, too low absorptivity
• water phase change (mixed-phase cloud)
• multi-layered cloud systems
• Arctic haze or clear sky ice precipitation

Objective Improve representation of clouds
precipitation and surface radiative energy
budget in GEM, specially over the Arctic.
GEM GLOBAL"
Polar-GEM LAM"
Boundary conditions
MICROPHYSICAL SCHEMES
RADIATIVE TRANSFER SCHEME
8

• Ongoing work
• Improve consistency/link between microphysical
  and radiative schemes in GEM
• (e.g. effective size, shape, type of hydrometeors
  and their optical properties)
• Assess 2-moment microphysical scheme in
  POLAR-GEM and compare with the other schemes
  using
• - AIRS (IR) satellite data (subproject 5) and
  CO2-slicing simulator
• - CALIPSO (lidar), CLOUDSAT (radar) satellite
  data and COSP simulator

COSP CFMIP Observational Simulator Package
CFMIP Cloud Feedback Model Intercomparison
Project COSP simulates the signal that
satellites (e.g. CloudSat) would see in a
model-generated world. Initiative of the UK Met
Office, LMD/IPSL (Paris), LLNL, Colorado State
Univ. and Univ. of Washington (USA).
CLOUD FRACTION GEM
CLOUD FRACTION SIMULATED
CLOUD FRACTION OBSERVED
AIRS
AIRS
CALIPSO
Polar-GEM
COSP
CLOUDSAT
9
Subproject 3 Sea-ice modellingN. Steiner, G.
Flato, Y. Lu

• Implement expand latest version of Los Alamos
  CICE model (used in several GCMs and US Navy's
  ice-ocean forecast model)?
• Apply test model in various settings
  (operational sea-ice, ocean atmosphere
  forecasting couple to Polar-GEM/GEM-LAM,
  coupled climate studies)?
• Develop a Canadian community sea-ice model to be
  used in climate mode (GCM,RCM) and forecast mode
  (weather, sea-ice)?

Photo N. Steiner
10
Status
(Due to hiring issues and funding cuts, work on
subproject 3 has been significantly slowed down).

• Global CICE4.0 installed on CCCma machines as
  standalone sea-ice model.
• Adjustments to CCCma grid format have been
  performed.
• Model currently tested with climatological daily
  forcing from a 20 year GCM run (atmosphere) and
  monthly Polar Science Center Hydrographic
  Climatology (PHC) (ocean).
• An option to run as a regional model is now
  included in CICE4.0 and will be attempted for the
  MSC (Meteorological Service of Canada) RCM grid
  as soon as global model testing is complete

11
TAWEPI subproject 4 Sensitivity studies in the
Arctic using singular vectors A. Mahidjiba, M.
Buehner, A. Zadra
SV calculation Optimization Time Interval (OTI)
48h Norms Initial Total Energy (TE) over the
globe Final TE over Arctic region (60ºN lat
85º N) Vertical domain for final-time norm
0.1044 ? 1 (from 100hPa to surface) GEM
3.3.0 model resolution for SVs calculation
240x120x58 for NL integration
800x600x58 Number of SVs 15 Analysis CMC
operational 4D-Var high resol.

• Objectives
• quantify fraction of forecast error explained by
  errors in initial conditions
• examine sensitivity of weather forecast over the
  Arctic due to analysis errors
• Singular Vectors (SV)
• patterns of greatest instability in initial
  uncertainty
• in early stages of development, error growth is
  governed by linear dynamics
• Methodology
• daily calculation of SVs for IPY period
• compute combinations of SVs that best explain
  forecast error

12
Summer 2007 (11 Jul to 20 Sep 2007)
Timevertical average of the total energy
(J/m2) of the 48-h forecast error
80N
60N
40N
180E
90W
0
90E
0
13
Summer 2007 (11 Jul to 20 Sep 2007)
Timevertical average of the total energy
(J/m2) of the forecast-error projection on SVs
80N
60N
40N
180E
90W
90E
0
0
14
Summer 2007 (11 Jul to 20 Sep 2007)
Timevertical average of the total energy
(J/m2) of SV projection at initial time
(pseudo-inverse)
80N
60N
40N
180E
90W
90E
0
0
15
Summer 2007 (11 Jul to 20 Sep 2007)
(c)
(a)
(b)
(a) Total energy (J/m2) of pseudo-inverse, (b) of
its propagation with TLM at 48h, and (c) 48-h
forecast error over the Arctic, vertically
integrated and averaged over the period of 11
July to 20 September 2007.
Growth rate given by SV1 over the Arctic for the
summer
Fraction of forecast error explained by SVs
Average 362
Average 17.6
16
Winter 2007-2008 (21 Dec 2007 to 20 Mar 2008)
(c)
(b)
(a)
(a) Total energy (J/m2) of pseudo-inverse, (b) of
its propagation with TLM at 48h, and (c) 48-h
forecast error over the Arctic, vertically
integrated and averaged over the period of 21
December 2007 to 20 March 2008.
Fraction of forecast error explained by SVs
Growth rate given by SV1 over the Arctic for the
autumn
Average 522
Average 15
17

• TAWEPI subproject 5
• Hyperspectral IR assimilation in cloudy
  atmospheres global and IPY applications
• O. Pancrati, L. Garand, S. Heilliette

• Background
• AIRS radiances assimilated operationally
• (since June 2008)
• - 87 channels
• - radiances not sensitive to lower clouds
• assimilated
• Therefore need to validate cloud
• height/cover determination for
• improved quality control
• By extension, interest in validating trial
• fields of cloud parameters and more
• generally cloudy radiance spectra to infer
• model deficiencies
• Specific problems found in
• Arctic/Antarctic region linked to cloud

AIRS Satellite (image taken from NASA Airs
Satellite Homepage)
AIRS Atmospheric Infrared Sounder MODIS
Moderate Resolution Imaging Spectroradiometer CALI
PSO Cloud-Aerosol Lidar and Infrared
Pathfinder MISR Multi-angle Imaging
SpectroRadiometer
18
Cloud height/amount from CO2-slicing technique
Observed CTP (CO2-slicing)

• CO2 slicing 12 estimates of cloud height from as
  many channels coupled with a reference profile
  peaking near the surface. Mean of valid estimates
  used.

Direct model output CTP
Calculated CTP (CO2-slicing) 6h forecast
_____________________________________________ For
more details, see poster Validation of model
cloud parameters using AIRS radiances by O.
Pancrati et al., session 21-J (Tue, Jul 21).
Note Good overall agreement differences most
notable for low clouds west of continents.
19
Focus on Arctic areas (July 2008) Cloud
parameters comparison with independent data
sources
Note The differences depend on input data and
retrieval methodology
AIRS (CO2-slicing)
AIRS (official product)
MODIS
Cloud Top Pressure
Cloud Fraction
Source AIRS science team
Source MODIS science team
20
TAWEPI subproject 6 GEM-BACH Stratospheric
Analyses for IPY M. Reszka, J. DeGrandpré, A.
Robichaud, C. Charette, M. Roch, S. Polavarapu

• Main results to date
• Dynamics and chemistry analyses for March 1,
  2007 Feb 28, 2009 have been generated and
  provided to SPARC IPY database (NetCDF)
• Dynamics fields are produced using Canadian
  Meteorological Centre's 3D-Var global
  assimilation scheme and GEM forecast model
• Chemistry fields are produced using an online
  stratospheric chemistry package (Belgian
  Institute for Space Aeronomy)
• Data set is being used to study several
  processes, including
• fine structure of polar temperatures during
  2007/2008 stratospheric sudden warming
• trace gas distribution as compared with
  spectrometer measurements
• deep stratospheric intrusions as revealed by the
  ozone field
• Current activities
• Documentation is in preparation
• For access, see http//www.sparc.sunysb.edu/html/
  user_ipy.html

21
Comparison of trace-gas measurements from a
Fourier transform spectrometer with GEM-BACH IPY
analyses at Eureka, Nunavut Mar 01 to Oct 30,
2007 R. Batchelor (U. of Toronto)
O3
N2O

• Total (and partial) columns derived from
  spectrometer data (blue) and analyses (red)
• are in very good agreement for most gases
  measured (e.g. O3, N2O, HCl, HNO3)
• ClONO2 and HF columns exhibit a bias, but
  variability is captured quite well
• CO and CH4 less satisfactory (probably due to
  lack of tropospheric chemistry)
• See e.g. Batchelor et al., CSPARC Workshop
  presentation (Toronto, 2008)
• See also talk by Rebecca Batchelor,
  Characterizing the spring-time Arctic
  stratosphere
• during IPY, J03 IPY symposium

22
Thank you
and thanks to all TAWEPI investigators
and collaborators
www.ec.gc.ca