Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
GOVST IV-TT meeting
5th November 2012
Rio de Janeiro
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Agenda, scope
• Class 4 intercomparison
• Multi-model ensemble
• GSOP2012 intercomparison
• Other
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Class 4 intercomparison
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Background + goals • A verification intercomparison
• Following similar methodology to JCOMM wave
intercomparison:
– share model counterparts in observation space (cls4)
– any centre can download and intercompare results
– one centre (UKMO) computes statistics and makes plots
available to participants
– results not released further without the permission of all
participants
• If successful, should be adopted by JCOMM as “official”
intercomparison methodology for OOFS
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
1. Class 4 inter-comparison
Overview
UKMO download obs files and produce
model analysis, forecast, persistence
and climatological information at the obs
locations in netCDF format.
US GODAE server
Each OOFS downloads the UKMO files
and replaces the model data with those
from their system.
Any participating group can download
files and inter-compare the results.
Produce a core set of plots for the GOV
web-site (password protected).
Observation inputs
~5 days behind real time
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Present status • 3 centres contributing class 4 files in near real
time: UKMO, BLUELink, NCEP (profiles)
• Obs are:
– SST – surface drifters from USGODAE (+ AATSR)
– profiles – Argo from Coriolis/MyOcean (5 day delay)
– SLA – Jason-1+2 + Cryosat from Aviso (+ UKMO obs
processing)
• UKMO production not yet operational - some
gaps in the record
• Demonstration intercomparison results produced
by UKMO – participants now checking these
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Experiences / feedback from
BLUELink, NCEP, REMO
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Preliminary results • As a demonstration of what could be shared
• These plots are anonymised because only
participants get to see results
• Plots are for profiles only (3 participants)
• Important caveats:
– Very short period (~2 weeks) so may not be
representative
– Participants have not yet checked the results, so
there could be mistakes
– These simple metrics do not tell the whole story
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
RMSE Bias
Day 1 Day 5
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Anomaly correlation
Day 1 Day 5
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Best estimate Day 5
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
RMSE Bias
Forecast Persistence
Climatology
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Title
Climatology
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
1. Class 4 inter-comparison
Prototype web pages
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Issues + next steps
• Robustness of UKMO production
• Participants to verify results / method
• Other participants to join
• Verification areas (proposal on next slide)
• Post-processing “rules” (esp. SLA)
• More observation types (longer term?):
– satellite SST (SLSTR, or sooner?)
– currents (drifters?)
– sea-ice
• More advance verification methods
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
GODAE class 1 regions
Overlaps
causing
problems for
interpretation of
verification
results – e.g.
large Tropical
Pacific errors
contaminate
North and
South Pacific
stats
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Proposed verification regions
TAT
NPA
TPA
ARC
NAT NPA
TPA
IND
SPA SAT
ACC
MED
SPA
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Proposed verification regions:
non-overlapping (mostly)
Name
Short
name
Latitude
range
Longitude
range
Excluded basins
North Atlantic NAT 20 - 66N 100W - 31E Mediterranean, Pacific
South Atlantic SAT 60 - 20S 70W - 20E
Tropical Atlantic TAT 20S - 20N 90W - 15E
North Pacific NPA 20 - 66N 100E - 77W
Atlantic, Gulf of Mexico,
Caribbean Sea
South Pacific SPA 60 - 20S 150E - 70W
Tropical Pacific TPA 20S - 20N 90E - 70W
Caribbean Sea,
Area south of Indonesia
and west of 120E
Indian Ocean IND 40S - 31N 20E - 120E
Gulf of Thailand and area
north of Indonesia
Arctic Ocean ARC 66 - 90N 180W - 180E
Southern
Ocean ACC 89S - 35S 180W - 180E
Mediterranean MED 30N - 48N 6W - 42E Atlantic
Global GLO 89S - 90N 180W - 180E
Others (specific to participants: Australia, Brazil, ...) [100m isobath?]
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Multi-model ensemble
demonstrations
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Multi-model ensemble
demonstration • Participating centres make surface fields available for
download
– native grid if possible
– daily mean
– 4 centres contributing (NCEP, NRL, Bluelink, UKMO)
• 2 partners proposed to produce demonstrations (NCEP-
global, CSIRO-Australia)
– Results from partners: NCEP, Mercator, ...
• Lessons learned?
• Ideas for 2013...
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Other topics
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
Updates to membership?
BoM, Australia Gary Brassington, Prasanth Divakaran
CSIRO, Australia Peter Oke
Environment Canada Hal Richie, Greg Smith
INGV, Italy Marina Tonani
Mercator-Ocean, France Fabrice Hernandez (co-chair)
NERSC, Norway Pavel Sakov
NMEFC, China Liying Wan
NOAA-NCEP, USA Avichal Mehra, Todd Spindler
NRL, USA Jim Cummings
UFBA, Brazil Clemente Tanajura
UK Met Office Alistair Sellar (co-chair), Matt Martin
GOVST-III Meeting, Rio de Janeiro, 5-9 November 2012
WWRP/WGNE Joint Working Group
on Forecast Verification Research
• Approximately the NWP equivalent of IV-TT
• Very active in exploring new verification methods
• Their remit includes promotion of good practice in
verification (as does ours)
• Can we collaborate / piggy-back to help achieve some of
our aims?
• 2 initial ideas:
– Raise awareness of (and contribute to) the very good CAWCR
verification website
– JWGFVR run verification tutorial workshops. Explore the
possibility of opening these up to ocean forecasting students,
and/or borrowing some material for GODAE summer schools.