Meteorological Service of Canada Status Report Gilles Verner Chief, Data Assimilation and Quality Control Canadian Meteorological Centre Meteorological.

Meteorological Service of Canada Status Report

Gilles VernerGilles VernerChief, Data Assimilation and Quality ControlChief, Data Assimilation and Quality ControlCanadian Meteorological CentreCanadian Meteorological CentreMeteorological Service of CanadaMeteorological Service of Canada

www.ec.gc.ca

NAEDEX-21 MeetingAsheville, North Carolina.17-19 Septembre 2008

Mike ManoreMike ManoreDirector, Monitoring Strategies andDirector, Monitoring Strategies andNetwork DesignNetwork DesignMeteorological Service of CanadaMeteorological Service of Canada

NAEDEX-21 Asheville, NC. – September 17-19, 2008Meteorological Service of Canada Status Report

Outline

• Current Operational Status:

– Space Based Monitoring and Satellite reception facilities– Computers and Telecoms – Canadian AMDAR and other MSC observations– Data issues and problems– Operational NWP system at CMC and data usage

• Recent changes and Future Developments:

– ISO-9001 certification : forecast system– New assimilation set-up for regional system and 06/18Z runs– Modification to EPS– Incorporation of additional observations in assimilation– Next version of global system (raised lid and new obs.)– New SST analysis


Organizational Structure

‘Space-Based Monitoring’ - Environment Canada lead: Mike Manore1. understanding satellite data requirements

▪ Environment Canada wide▪ meteorological, climate, ice, marine oil spill, wildlife habitat, etc,….

2. ensuring data access & exchange▪ coordination/arrangements with satellite operators and other users

– gov’t-to-gov’t, gov’t-to-agency, commercial– MoU with NOAA, Agreement with Eumetsat…– all appropriate missions and sensors

▪ operation of EC satellite data reception and processing systems– within Gov’t of Canada ‘Shared Federal Infrastructure’– EC leadership for operational meteorological missions– operate as a national network: GOES- Alfheim; HRPT - Bradley

▪ future missions– influence future missions to meet EC requirements – Canada (CSA), International– E.g. CSA PCW mission


CSA PCW Mission

• Cooperation CSA, DND, EC• Concept of 2 satellites in HEO (Molniya 12-hr

orbit)• ABI-like imager, 2015 timeframe• Proto-operational, North Polar areas only (Nof60)

(2 other sat. needed for South Polar areas)• Phase 0 just completed• Phase A kick-off next week, opportunities for

cooperation to be examined• Mission funding not yet secured• EC: Louis Garand, Mike Manore, Gilles Verner

Video Clip


MSC Satellite Data Reception Facilities

Dartmouth

Toronto

Vancouver

WinnipegCMC (3)

Edmonton (2)

Ottawa

Resolute

EARS - Eumetsat

GOES

HRPT

EARS

Dartmouth HRPT site moved to Gander

Upgrade to Resolute HRPT site and telecoms is underway

Gander


Canada-EUMETSAT Cooperation Agreement

• Signed Fall 2007

• 8 areas of cooperation– science and product development

▪ participation in SAFs, visiting scientists

– optimize data acquisition and exchange of satellite data and products▪ specific agreements –e.g., EARS

– define requirements and opportunities to improve high-latitude observations

– explore options for Canada to contribute to EUMETSAT mission instruments▪ build on CSA associate membership in ESA

– share training experience and materials

– cooperation in GEO, including projects in developing countries▪ GEONETCast

– exchange views on matters of satellites and satellite meteorology

– explore possibilities for longer term cooperation


Computers and Telecoms

• Final upgrade to IBM: 2 clusters, upgrade is underway:– current is OPS: 40 nodes, and R&D: 60 nodes– to reach OPS/DEV: 80/120 nodes by Dec08, and– to reach 80/180 nodes by June 09– Total of 4512 P5+ processors (9024 virtual) by mid 2009

• New procurement for 2010-11• Upgrade to SGI-FE’s: 2 x capacity, replacement with Linux

clusters underway• CFS: 2 PB tape capacity• Links to NOAA-TOC (GTS), UKMetO (Sat data + GTS back-

up), EUMETSAT (EARS)• Link to NOAA-NESDIS soon (may be done by now)• Internet still used for much satellite data• EumetCast / GeoNetCast being considered


Canadian AMDAR coverage


Canadian AMDAR - Status

• AC Jazz: 46 CRJ and 42 DHC-8 on GTS, 2 CRJ and 14 DHC-8 retained due to problems with data quality. More CRJ to be added. CRJ data very good. DHC-8 may be removed.

• Still issues with DHC-8 TT data, although improved. Problem is related to data averaging aspects in avionics. Not corrected.

• No progress with AMS and First Air. TAMDAR approach does not seem viable for these aircraft.

• Still trying to obtain data from NavCanada CRJ’s (2), hopefully to be this fall. Some data samples just obtained.

• Negotiations on-going with WestJet, slow progress.• MSC-ADAS has replaced former C-ADAS software, in-house

Linux based system.• Monitoring of data is on-going and showing importance of good

monitoring before distribution of data.


TT data quality – JAZZ CRJ and DHC-8

For DHC-8 Aircraft, significant difference in bias between ascent /

descent data.

Related to smoothing algorithm in avionics.

CRJ Aircraft data are good.

April 2007


Other Canadian observations

• “Synoptic” observations from CTBTO network, using SYNOP MOBIL code, header: SNCN19 CWAO. Observations every 10-min, 10 stations for now but should increase, received at CMC by e-mail: data problems as reported by MetOffice : surface elevations not available or wrong.

• Canadian radar data - now available centrally at CMC, NRP (National Radar Processor) operational, full volume scans available. Provided to NOAA.

• Forestry and Road weather stations in British Columbia - data received at CMC, in old SA format. Data redistribution restrictions currently apply…should eventually be available in BUFR.

• Co-operative network in Quebec (Province, Hydro-Quebec, etc), received at CMC, also redistribution restrictions…problems with BUFR data.

• Some research wind profilers in Southern Ontario and Quebec: operated by Universities, data now available to CMC and should be inserted on GTS in the near future, in BUFR code: headers to be communicated as soon as available.

• Some soil TT data now available, eventually in BUFR.• New equipment for radiosondes. Should improve wind quality, also includes provision

for 4D data in BUFR. Data available at stations but not yet on GTS.• Ozone soundings on GTS (KULA01 CWAO) and total column ozone.• Major data management project in MSC, should lead to all data available in BUFR.• Potential development of surface GPS Network. Lead: Godelieve Deblonde.• Surface Weather stations for 2010 Olympics (Whistler) already on GTS.


Canadian Profilers

Resolute Bayalso probable


McGill VHF Profiler (Montreal)


Clovar Profiler (London, Ont.)


Resolute Bay profiler (Arctic)


Availability of aircraft data : status

ADS data collected by NavCanada: UANT01 CWAO – on GTSSome ADS data now collected in Northern Canada

ADS data East of 30° W now available.Thanks to UK-Nats.


Reception of ADS & AIREP data

Occasionl drop in ADS data


Some data issues

• SSM/I data : any chance to get F15 back, status of F14?

• SSM/I data quality issues – some stray scans and orbit identification and/or positioning/labelling

• SSM/IS – status to be discussed at this meeting, F17-18

• « Slower » access to Metop data

• Access to some data on GTS has been problematic to CMC: IASI (300 ch subset) due to link saturation, dropsondes over Western Pacific (obtained from MetO),

• AMSU – timeliness of N15, N16 (not primary)

• Need for a Eumetcast-Europe-like system for Americas – GeoNetcast with similar content as Eumetcast?


SSM/I reception


AMSU/MHS Assimilated


Stray scans in SSM/I

Can be removed byproper filtering


Mislabelled SSM/I Orbits

• Visual Check - 21/06/2006 - 00Z• ===============================

• 1. Two orbits appear in a BURP file before they should exist: 58036F13, 58037F13

• ^58017F13 0 0 55 0 000400 20060620 1608 168 521324 1060• ^58018F13 0 0 19 0 000400 20060620 1749 168 507960 522384• ^58019F13 0 0 35 0 000400 20060620 1929 168 549274 1030344• ^58020F13 0 0 55 0 000400 20060620 2115 168 437448 1579618• ^58021F13 0 0 2 0 000400 20060620 2303 168 481674 2017066• ^58022F13 0 0 40 0 000400 20060621 33 168 523274 2498740• ^58023F13 0 0 40 0 000400 20060621 212 168 475824 3022014• ^58036F13 0 0 55 0 000400 20060621 17 168 518364 3497838• ^58037F13 0 0 11 0 000400 20060621 158 168 479724 4016202• ^58024F13 0 0 27 0 000400 20060621 344 168 537574 4495926• ^58025F13 0 0 15 0 000400 20060621 526 168 644824 5033500• ^58026F13 0 0 42 0 000400 20060621 730 168 538224 5678324• ^58038F13 0 0 6 0 000400 20060621 330 168 514200 6216548• ^58039F13 0 0 53 0 000400 20060621 511 168 643524 6730748• ^58040F13 0 0 20 0 000400 20060621 716 168 514824 7374272

• Orbits 36 and 37 received before they should exist! Wrong date?• Also some serious orbit orvelapping


^58020F13

^58021F13

^58022F13

Orbits 20 & 21 overlap but should not,look misplaced compared to orbit 22


^58023F13

^58036F13

^58037F13

Orbit 36 is before 23, 37 overlaps 23;Looks like mislabelled orbits and wrong location!

No easy QC solution, would need knowledge about orbit identification scheme


CMC OPERATIONAL RUNS: MAIN COMPONENTS

GEM modelGlobal (100km) : 0-10/15 days Regional (15km) : 0-48 hoursHIMAP (10km) : 0-24 hours

air quality (CHRONOS)

Environmental EmergencyTransport model (CANERM)

Trajectory model

monthly and seasonalforecasts

global assimilation cycle

4D-Var Analysis (data collection and

assimilation)

Models,9-hour forecast

(trial field)

Wave model (WAM)

regional data assimilation

ensemble forecasts(20 members)

surface analysesGEM model

Global (35 km) : 0-10/15 days Regional (15 km) : 0-48 hours

LAM (2.5 km) : 0-24 hours


• Atmosphere (4D-Var)– Wind (U,V)– temperature + Ps– Moisture (ln Q)

• Surface (OI)– air temperature– dew point depression– precipitation amount– sea surface temperature (also lakes)– mean sea level pressure– snow depth– ice cover– ice thickness– deep soil temperature– soil moisture– albedo– and other geophysical fields based

on climatology

Data Assimilation – Fields Analyzed

4-D analyses fromsurface up to 10 hPa

(58 Levels)

2-D analyses at surface

Global and Regional grids (some on LAM)


200kmx200km/time step7 MW channels

U,V at 10 meter over ocean

250kmx250km/time step87 IR channels

SSM/I DMSP13-14

QUIKSCAT

AIRS

Observations assimilated in 4D-Var

(750 m) Vertical hourlyU,VProfiler (NOAA Network)

~180 km boxes11 layers, per time step

U,VMODIS polar winds

(Aqua, Terra)

1.5o x 1.5o

11 layers, per time step

U,V

(IR, WV, VI, 3.9μ channels)

AMV’s(METEOSAT 7-9, GOES 11-12, MTSAT-1R)

2o x 2o 3-hourlyIM3 (6.7 m)Water vapor channel GOES 11-12

250 km x 250 km

per time step

Ocean Land

AMSU-A 4-10 6-10

AMSU-B / MHS 2-5 3-4

ATOVS

NOAA 15-16-17-18, AQUA

1o x 1o x 50 hPaper time step

U, V, TAircraft

(BUFR, AIREP, AMDAR, ADS)

1 report / 6hT, (T-Td), ps, (U, V over water)Surface report

(SYNOP, SHIP, BUOYs)

28 levelsU, V, T, (T-Td), psRadiosonde/dropsonde

ThinningVariablesType

100kmx100km/time step


Horizontal Grid of meso global GEMPrevious model Global-meso

400X200 -> 100 km, L28 800X600 -> 33 km at 45o N, L58


575 X 641 grid (66% in 15-km uniformarea), 58 levels

Regionalmodel


Currently 4 GEM LAM 2.5 km domains, plus new higher resolution domain for the Winter-2010 Olympics

Topography at 2.5 km


Changes to CMC NWP System since last meeting

New assimilation setup for regional system on July 4, 2007• 6 hour spinup cycle (previously 12)• 3D FGAT analysis 58 levels• Extension to 06/18Z forecast runs on July 3, 2008

Major upgrade of the ensemble prediction system on July 10, 2007• 20 members up to 16 days at 0.9 deg (400x200) and 28 levels• Some stochastic physics

Major changes to Global & Regional Systems on May 28, 2008 :• 50% increase in amount of assimilated data• AIRS, QuikScat, more AMV`s and AMSU• Removal of AMSUA-ch3• RTTOV-8 and new vertical interpolation (Rochon), being incorporated into the

RTTOV library• Off-line dynamical bias correction for all radiance data


New Assimilation set-up for Regional System

• Update to 58 Levels and B Stats of meso-global GEM (from 28L, B stats from 100-km Global GEM)

• Spin-up shortened to 6 hours (from 12) with 3D-Var FGAT, more benefit from Global 4D-Var

• Snow depth anal. on 15-km model grid• New ice climatology, forcing a southern limit to the ice

line (monthly)• Implemented July 4th, 2007• Easy extension to 4 forecast cycles per day,

implemented on July 3rd, 2008


Regional 12 hr spin-up cycle

L’analyse est transmise

Champ d’essai est produit

L’analyse est produiteG200

G218

G212

G206

R200 R206 R112

R212R218R100

R312 R118

R106 R300

G100

G112


Regional 6 hr spin-up cycle,allows extension to 06/18Z runs

L’analyse est transmise

Champ d’essai est produit

L’analyse est produite

R206

R218R100

R312 R118

R106 R300

R112

G200

G218

G212

G206

R200

R112

R212

G100

G112

R206

R312 R118

R218R100

R106 R300

R318

R306


Summer O-P 48h Winter(42 cases) (42 cases


42 winter cases– synop and shef (00-24h)


Gain obtained with 4 regional runs per day,updates at 06/18Z (42 winter cases)

zonal wind std

temperature

zonal wind


Changes to EPS implemented July 10, 2007

a) New physics / New dynamics / Horizontal resolution

• 20 GEM members (SEF not used in the new EPS)

• Horizontal resolution: 0.9°

• Model lid: 10 hPa

• Forecasts up to 16 days

• Inclusion of two stochastic components in the physics

• Multi-parameterization approach...

• ENKF – True 4D with FGAT like innovations, 0.9°

• Harmonization with NCEP for NAEFS


New Set-up for EPS

b) Stochastic physical tendency perturbations

• All physical tendencies on horizontal winds, temperature, and humidity of each member are multiplied by a random function:– Defined in the range [0.5 , 1.5]– With a decorrelation time scale of 3 hours– And a decorrelation length scale of ~1700 km


New Set-up for EPS

c) Stochastic kinetic energy backscatter algorithm

• Numerical models are over-dissipative near the truncation limit

• This likely inhibits upscale energy transfer that can affect the large scale flow

• It is thought that this phenomenon can be a cause of under-dispersion in EPSs

• Parameterization: Inject energy near the truncation limit to compensate for the over-dissipation


Comparison between the former and new EPS

Mean of errors

Error of the mean

Ensemble spread

Tro

pica

l RM

S S

core

s

Thin lines : New EPSThick lines: Old EPS

Tropics


Major increment to assimilation systems:more observations in 4D-Var – May 2008

AIRS & SSMI radiancesDynamic bias correction for all radiance data✓More low level AMV’s✓QuikSCAT✓Additional AMSU/MHS at larger scan angles

Removed due to time:

GPS Occultations from COSMIC, CHAMP and GRACE Metop ATOVS: AMSUA & MHS SSMI/S radiances 3 hourly sfc data More levels from radiosondes and aircraft Revised background & observation error statistics


Summary of the new components

• AIRS radiance data (87 channels).• SSM/I radiance data (+ eliminating AMSU-A ch3 and new cloud mask for

AMSU-B).• Going from RTTOV7 to RTTOV8.7 (and new vertical interpolation) • Inclusion of high scan angle AMSU data.• QuikScat oceanic surface winds (from KNMI).• 3.9 micron AMV (nighttime low level winds).• Dynamical bias correction for all radiance data.

Current operational assimilation and forecast system • ~35 km resolution global forecast model (GEM with 800x600 points and 58 eta

levels) .• 4DVAR global assimilation with 6 hour window .• 15 km regional forecast model over North America (58 eta levels). • 3DVAR FGAT regional assimilation with 6 hour spin-up cycle launched from

global cycle every 12 hours (or 6).


Data volumes (typical for 6 hour)

~50% Increase of assimilated data

Operational System New data

RAOBS 50000

Aircraft 55000

Profilers 8000

Surface 13000

GOES 5000

SatWind 40000

AMSU-A 50000

AMSU-B 20000

SSMI 14000

QuikScat 10000

AIRS 75000

SatWind 1500

AMSU-A 14000

AMSU-B 5000

Total ~240000 Total ~120000


Global model 500 hPa geopotential forecast anomaly correlation

156 summer cases 160 winter cases 160 winter cases

hours hours hours

Global model 500 hPa geopotential forecast anomaly correlation

156 summer cases 160 winter cases 160 winter cases

hours hours hours


Verification against Radiosondes for winter 2007

5 day global forecast – 160 cases verified over Southern Hemisphere

Pre

ssur

e hP

a

m/s m/s

degrees

degrees

dam

2 day regional forecast – 40 cases verified over North America

UUUV UU

UV

GZ TT

ESES

GZTT

degreesdam

Pre

ssur

e hP

a

m/s m/s

degrees

OPE in blue versus NEW in red

BIASSTD DEV STD DEV

BIAS

Verification against Radiosondes for winter 2007

5 day global forecast – 160 cases verified over Southern Hemisphere

Pre

ssur

e hP

a

m/s m/s

degrees

degrees

dam

2 day regional forecast – 40 cases verified over North America

UUUV UU

UV

GZ TT

ESES

GZTT

degreesdam

Pre

ssur

e hP

a

m/s m/s

degrees

OPE in blue versus NEW in red

BIASSTD DEV STD DEV

BIAS


Recent NWP verification


Recent NWP verification


Next version of global system

• Hybrid vertical coordinate, 80L, lid 0.1hPa• Li & Barker Radiation scheme• Hines non-orographic GWD• New error statistics (Buehner)• Additional AMSU and AIRS channels, IASI ?• AMSU/MHS from Metop• GPS-RO: CHAMP, GRACE, COSMIC, GRAS• ASCAT• SSM/IS• CSR data with RTTOV – 5 GEO• Bias Correction above 10 hPa• Target: Spring 2009


GEM STRATO global modelGEM-

OperationalGEM-Strato

No. of pointshorizontal

800 x 600 (33 km)

800 x 600 (33 km)

Vertical coordinate

Eta (normalized sigma)

Hybrid

No. of levels in vertical

58 80

Top 10 hPa 0.1 hPa

Sponge del2 at top (wind and temp)

4 levels 6 levels (to 1 hPa)No diffusion on

mean zonal wind

Tropical sponge layer

4 levelsFrom top to 50

hPa

8 levels From top to 3 hPa


Impact of GPS-RO (with, without) at 300hPa

Gain of ~2 h NH and ~3-4 h SH at days 3 to 6 70-day winter cycle

SHNH


GEM-STRATO (0.1 hPa) vs. Operational (10 hPa) (both with new data types and GPS-RO)

NH Winter 115 120H forecasts

U V

GZ T

T-Td

U V

GZ T

T-Td

SH Winter


• Currently assimilate GOES W,E WV channel at 3h intervals

• Testing 1h interval WV chan. data for 5 geo satellites (2 GOES, 2 Meteosat, MT-SAT)

• Additional channels may be included

(New) To be added this year (2008)

Geostationary Radiances

O-P (K) Meteosat W

Std

Mean

Mean (uncorrected)

2

1

0


• Currently assimilate QuikScat

• Monitoring of ASAT is underway

• Assimilation cycles are starting

• QuikScat and ASCAT from KNMI

Metop ASCAT

New CMC SST ANALYSIS

Bruce BrasnettData Assimilation and Quality Control DivisionCanadian Meteorological CentreSeptember, 2008


Data Sources

• AVHRR17 & AVHRR18 via ftp from PO-DAAC (retrievals from Naval Oceanographic Office)

• AMSR-E via ftp from Remote Sensing Systems Inc.

• A/ATSR via ftp from ESA (ATS MET data)• Ships, drifters and moored buoys

• Using OI on 1080x540 grid (0.33 deg)• Once a day• Published in Sept. 08 QJRMS, 134


Analysis error estimates using Argo floats. HRDOI is the National Climatic Data Center product, RTG is the NCEP product and MGDSST is the JMA product


Multi-Product Ensemble - Departure from median of 10-member ensemble (CMC vs. RTG(NCEP), Sept. 1 2008) from:ghrsst-pp.metoffice.com/pages/latest_analysis/sst_monitor/daily/ens/index.html


Multi-Product Ensemble - Departure from median of 10-member ensemble CMC vs. OSTIA (UKMO) Sept. 1 2008 from: ghrsst-pp.metoffice.com/pages/latest_analysis/sst_monitor/daily/ens/index.html


Monitoring Web sites

• http://collaboration.cmc.eg.gc.ca/cmc/data_monitoring

• http://collaboration.cmc.eg.gc.ca/cmc/data_monitoring/CAMDAR

• http://collaboration.cmc.eg.gc.ca/cmc/data_monitoring/SAR

• User: monitoring PW: CMC


www.ec.gc.ca

Meteorological Service of Canada Status Report Gilles Verner Chief, Data Assimilation and Quality Control Canadian Meteorological Centre Meteorological.

Documents