http:saf'met'noSAF on Ocean and Sea Ice

1
Søren Andersen, Morten Lind Rasmus Tonboe, DMI
Steinar Eastwood, Vibeke Thyness, Harald
Schyberg, Lars-Anders Breivik, Øystein Godøy,
Jørg Haarpainter met.no

• SAFSatellite Application Facility
• Distributed MSG/Metop ground segment
• Other SAFS NWP, Nowcasting, Climate, Land
• To provide satellite derived products for use in
  the meteorological/oceanographic communities.

2
OSI SAF organisation

• Consortium led by Meteo-France
• High-latitude center led by met.no
• 2 ice services
• Backup / complementary production scheme
• Shared responsibility for validation

3
General Objectives

• Satellite derived Parameters describing the
  interface between atmosphere and ocean
• 10 km resolution, covering Atlantic Ocean
• Quantities to be calculated
• Ice concentration
• Ice edge
• Ice type
• Radiative Fluxes
• Sea Surface Temperature

4
Sea Ice Objectives

• Final products
• Ice edge retrieved using SSM/I, scatterometer and
  AVHRR expressed as probability of ice
  infestation.
• Dominant ice type using SSM/I, scatterometer and
  AVHRR expressed as probability of ice infestation
  with given type of ice.
• Total sea ice concentration from SSM/I and
  possibly AVHRR.

5
Availability

• Downloaded from the HL center homepage
• Distributed via EUMETCAST
• Pushed by internet or RMDCN.

6
Odden Jan-Feb 2004
7
Validation

• Objective validation is done routinely against
  DMI and met.no ice analyses based on AVHRR and/or
  SAR data
• Summer conditions are problematic
• East Greenland concentrations are consistently low

8
Ongoing development

• Internal development
• Global coverage (summer 2004)
• Quikscat/Seawinds (2004)
• Use of SSMIS sounding channels (2005)
• Improved use of AVHRR (METOP/2007)
• ASCAT (METOP/2007)
• EU funded projects
• IOMASA, improved ice conc. (2003-2006)
• Mersea, ASAR global mode, AMSR (Apr. 2004)

9
Operational Methods

• Atmospheric correction of passive microwave obs.
• Uses NWP model estimates of water vapour and wind

10
Global coverage
25 km resolution Northern and southern
hemisphere products Atmospheric corrections from
ECMWF global analysis/forecast
11
Multi sensor approach
Bayesian approach for ice detection

Extended to combine several satellite observed
parameters into an optimal ice class analysis
Estimate ice class including measure of
uncertainty
12
QuikScat

• A priori statistics compiled from
• colocation with DMI ice charts
• Quantities used
• Polarization ratio
• Mean horizontal backscatter
• Mean vertical backscatter

QuikScat
SSM/I

• Better determination of MY ice than SSM/I
• More details in ice edge
• Will be added to multi-sensor scheme in 2004

13
Quikscat example
Ice type 18 Febr 4 March 2004
14
Ice detection using AVHRR channel 3A (1.6
µm)Examples from April 2003
15
Combinations of AVHRR channels
16
Training the system
Cloud
Cloud
Ice
Ice
Channel A3/A1
RGB (124)

• Manually picking out areas where we are certain
  of the surface type.
• Creating dataset based on these areas
• Make sure that the dataset contains different
  types of ice clouds, and different sun angles.

17
Ice/Cloud/Sea separabilityResults for April 2003
18
Further plans

• Analyze one full year of AVHRR passages over
  North Atlantic / Arctic
• Create PDFs for the different surface signatures
• Include new AVHRR method in OSI-SAF ice cover
  product

19
Collaboration

• Visiting scientist opportunities
• Can be found on EUMETSAT web page
• New projects can be proposed
• A way to cooperate internationally
• Previous visitors/visits
• University of Bremen (1-2), NIC (3), CIS (1),
  University of Dundee (1)
• International Ice Charting Working Group
• Cooperation agreements with NIC and CIS on
  validation
• Scientific coordination