Evaluation of CONCEPTS Ice-Ocean Forecasting Systems

1
Evaluation of CONCEPTS Ice-Ocean Forecasting
Systems

• Greg Smith1, Christiane Beaudoin1, Alain Caya1,
  Mark Buehner1, Francois Roy2 , Jean-Marc
  Belanger1, Frederic Dupont1, Fraser Davidson3,
  Jennifer Wells3, Tom Carrieres4 , Hal Ritchie5,
  Youyu Lu6 , Charles-Emmanuel Testut7 and Gilles
  Garric7
• 1 Meteorological Research Division, Environment
  Canada, Dorval, CANADA
• 2 Canadian Meteorological Centre, Environment
  Canada, Dorval, CANADA
• 3 Northwest Atlantic Fisheries Centre, Fisheries
  and Oceans Canada, CANADA
• 4 Canadian Ice Service, Environment Canada,
  Ottawa, CANADA
• 5 Meteorological Research Division, Environment
  Canada, Dartmouth, CANADA
• 6 Bedford Institute of Oceanography, Fisheries
  and Oceans Canada, Bedford, CANADA
• 7Mercator-Océan, Toulouse, FRANCE

GODAE Oceanview Workshop, Santa Cruz, USA (Jun
13-17, 2011)
2
Outline

• Operational Gulf of St. Lawrence Coupled
  Atmosphere-Ice-Ocean Forecasting System
• CONCEPTS Objectives
• Global Forecasting System
• Evaluation with AVHRR and Radarsat
• NW Atlantic/Arctic Forecasting System
• Evaluation against IMS analyses
• Future Work and Interests

3
The Gulf of St. Lawrence (GSL) Coupled Regional
Deterministic Prediction System (RDPS-CGSL)

• Operational regional forecasting system (GEM-Ops)
  has tendency to overestimate cold events in
  winter.
• Increased heat fluxes in coupled system buffers
  air temperatures and improves forecasts
• Demonstrates importance of air-sea-ice coupling
  even for short-range weather forecasts
• Coupled GSL system now operational at CMC
• as of last week!

-5C
-15C
-25C
S. Desjardins
4
CONCEPTS

• Canadian Operational Network of Coupled
  Environmental Prediction Systems
• Aim
• Development of Regional and Global Coupled
  Atmosphere-Ice-Ocean Forecasting Systems
• Global coupled medium-to-monthly forecasting
  system
• GEM atmospheric model and 4DVAR/EnKF analysis
  system
• Coupled to 1/4 resolution (ORCA025) NEMO
  ice-ocean model
• Ocean initialized using Mercator analysis system
  (PSY3)
• Initially produce 10 day uncoupled ice-ocean
  forecasts
• Regional short-term forecasting system
• Preparation and issuing service for MET/NAVAREAS
  1718
• Build on developments made by CNOOFS (F. Davidson
  et al.)
• Coupled to CMC regional forecasting system (RDPS)
• Based on subdomain of ORCA12 for NW Atlantic
  Arctic

5
CONCEPTS Global - V0

• Ice-ocean model
• NEMO v3.1 OPA9 ocean model and LIM2-EVP sea ice
• ORCA025 Global tri-polar 1/4 resolution
• Atmospheric forcing from GEM Global (GDPS 33km)
• Forced using CORE bulk formula
• 3hrly forcing frequency (including diurnal cycle)
• Initialization
• Ice and ocean fields taken from Mercator (PSY3V2)
  analysis
• Output
• Weekly 10-day forecasts of ocean and ice fields

6
Comparison with AVHRR SST observations
F. Roy
Mean

• Differences taken between AVHRR SST data and
  hourly output from weekly forecasts.
• Statistics accumulated for each day for forecasts
  made from May 20, 2009 to Mar 23, 2010.
• Results shown for day 10 of forecasts
• Poor coverage in polar regions and due to cloud
  cover

Std. dev.
Number of comparisons
7
Comparison with AVHRR SST observations
Day 1
Day 10
Mean
Mean
Development of warm bias
Std. dev.
Std. dev.
8
Comparison with AVHRR SST observations
Day 1
Day 10
Mean
Mean
Cold bias present in analysis
Std. dev.
Std. dev.
9
Regional RMS differences with AVHRR SST
Forecasts Persistence of SAM2 analyses Persistence
of CMC analyses

• CMC SST analysis has smaller RMS diff for day1
• Similar error growth in both persistence curves
• Forecast beat persistence of analysis for most
  regions
• Forecasts show smaller diff as compared to
  persistence of CMC SST analyses for N. Atl,
  N. Pac and T. Ind.

10
CONCEPTS Global V1

• AIM Produce daily analyses and 10day forecasts.
• Based on PSY3V2, with following modifications
• Updated SAM2 to NEMOv3.1, with LIM2-EVP(done)
• Assimilate CMC-SST analysis (in place of
  RTG)(done)
• Ocean analysis merged with 3DVAR-FGAT ice
  analysis (done)
• Daily analysis updates (planned)
• Status
• Modifications to SAM2 ongoing
• Routine production of ice-ocean analyses since
  Dec. 2010
• Evaluation of ice-ocean forecasts underway
• Starting initial trials of coupled runs.

11
CMC/CIS Sea-ice Analysis System
M. Buehner

• Uses 3DVAR-FGAT, with covariances obtained from
  EnKF
• North American Analysis
• Four analyses per day of ice concentration at 5
  km resolution
• Global Analysis
• two analyses per day on 10km grid
• Currently assimilates
• SSM/I, AMSR-E, CIS daily charts, RadarSAT image
  analyses
• Work in progress to add
• SSMIS, scatterometer, visible-infrared, SAR and
  ice thickness satellite-based observations

SSM/I
AMSR-E
CIS Chart
RadarSAT
12
Verification against NOAA IMS analyses
M. Buehner

• Evaluation of 5km North American analyses
• Based on contingency tables values
• Uses threshold of 0.4 for ice/noice
• Overestimation of ice cover in CMC operational
  analysis during melt

13
CONCEPTS Global V1

• AIM Produce daily analyses and 10day forecasts.
• Based on PSY3V2, with following modifications
• Updated SAM2 to NEMOv3.1, with LIM2-EVP(done)
• Assimilate CMC-SST analysis (in place of
  RTG)(done)
• Ocean analysis merged with 3DVAR-FGAT ice
  analysis (done)
• Daily analysis updates (planned)
• Status
• Modifications to SAM2 ongoing
• Routine production of ice-ocean analyses since
  Dec. 2010
• Evaluation of ice-ocean forecasts underway
• Starting initial trials of coupled runs.

14
Comparison with PSY3V2R2

• Difference after 28 cycles (valid 20091125)
• Cycle started 20090513
• Impact of changes to SST assimilation
• Difference in SSS due to relaxation timescale,
  atm forcing and ice assimilation

15
Verification against Radarsat

• Evaluation of 30 day ice forecasts
• Model appears to have some skill in predicting
  mean ice cover, but ice dynamics is still a
  challenge
• Careful analysis required to understand
  small-scale details represented in Radarsat image
  analyses

Model (mean error) Model (std. dev.) Persistence
of CMCICE (mean error) Persistence of CMCICE (std
dev)
CIS Radarsat image analysis
Labrador Sea
16
CONCEPTS Regional forecasting system

• C-NOOFS Canadian-Newfoundland Operational
  Oceanographic Forecasting System
• Lead F. Davidson (NAFC)
• Produces daily 10-day forecasts at 1/12
  resolution for the Northwest Atlantic
• Initialized using Mercator data assimilation
  system (PSY2).
• Merged with 3DVAR-FGAT ice analysis
• Designed to meet needs of Coast Guard and Navy,
  as well as variety of applications influenced by
  sea ice

C-NOOFS
17
CNOOFS Comparison with Spring Survey
F. Davidson

• Comparison of bottom temperature from 2010 Spring
  Survey of Grand Banks for
• NWA025 (PSY3V2R2)
• NWA12 (PSY2V3R1)

18
EC/MSCs involvement in METAREAs

• Development of an integrated marine Arctic
  prediction system in support of METAREA
  monitoring and warnings.
• Development of short-term marine forecast system
  using a regional high resolution coupled
  multi-component modelling (atmosphere, land,
  snow, ice, ocean, wave) and data assimilation
  system
• To predict- Near Surface atmospheric
  conditions,- Sea ice (concentration, pressure,
  drift, ice edge) - Freezing spray,- Waves,
  and- Ocean conditions (temperature and
  currents)
• Improved Arctic monitoring

19
Regional Coupled Forecasting System

• Build on CNOOFS and Coupled GSL
• Develop coupled forecasting system for N.
  America/Arctic
• Couple NEMO to GEM regional (10km)
• 5km LAM over METAREAS 1718
• with 5km Atm 4DVAR/EnKF
• 1/12th regional SAM2
• Produce 48hr weather and marine forecasts

C-NOOFS
1/12
GEM RDPS
10km
20
Plans and interests

• CONCEPTS Global
• Running 1/4 global 10day forecasting system
  since Dec. 2010
• Operational transfer in coming year
• Next steps
• produce daily analyses and
• improve consistency of ice and ocean analyses
• CONCEPTS Regional
• Develop/evaluate N.Atl/Arctic coastal 1/12th NEMO
• Begin work on 1/12th regional data assimilation
  system
• Evaluation and intercomparison of ice-covered
  waters
• Ice thickness (Radarsat, Cryosat, AVHRR)
• Marginal ice zone (MIZ)
• How can ice rheologies be improved to better
  represent fine-scale ice deformations over short
  lead times?
• How do we constrain the ocean under-ice and in
  the MIZ?