Presented by Caio Coelho

Co-Chair of the Inter-Programme Expert Team on Operational Predictions

from sub-seasonal to longer time scales (IPET-OPSLS)

Most slides provided by Sook Young, Hyun-Ju Lee,

Soonjo Yoon and the WMO LC-LRFMME Team

NOAA-USAID 11th International Training Workshop

Climate Variability and Predictions (11ITWCVP),
Second Symposium on the Variability and
Predictability of the Global Climate System,
Ankara, Turkey, 15 – 22 April, 2019
Global Producing Centres
1
 Background
• In recent years WMO (through CBS and CCl) designated
centres responsible for generating and delivering
long-range forecasts and other climate information
• This infrastructure is part of the Global Data-Processing
and Forecast System (GDPFS)
• It is also the core for the Climate Service Information
System (CSIS) of the Global Framework for Climate
Services (GFCS)
• The designated centres and their roles and
responsibilities are defined in the Manual on the GDPFS
(WMO-No. 485)
 WMO infrastructure for seasonal and
inter-annual prediction
• GFCS three-tier structure: global, regional and national
For long-range forecasting:
Global domain:
• Global Producing Centres for Long-Range Forecasts (GPC-LRFs) responsible for
producing and disseminating long-range forecasts and associated verification with
global coverage;
• A Lead Centre (LC) for Long-Range Forecast Multi-Model Ensembles (LC-LRFMME)
responsible for collecting GPC forecasts, displaying the forecast information,
generating multi-model products, and more recently generating verification
information for GPC-LRFs delegating score computation to the LC
• A Lead Centre for Standardised Verification of Long-Range Forecasts (LC-SVSLRF)
responsible for providing documentation on verification scores, software and data
sets for long-range forecast verification
Regional domain:
• Regional Climate Centres (RCCs) and RCC Networks covering WMO
RAs and Polar Regions are either fully desig. or in demons. phase
National domain:
• National Met. and Hydrol. Serv. (NMHSs) play central role at national level,
including coordination of National Climate Outlook Forums (NCOFs)
 Motivation for establishing a LC-LRFMME

4
 LC-LRFMME Historical timeline

5
 LC-LRFMME History
Historical timeline

6
7
8
 LC-LRFMME operational activities on LRF
in support of WMO members
• Collection of retrospective and real-time forecasts from GPCs
• Production of real time forecast products from the collected GPCs
• Production of verification products from the collected GPCs retrospective
forecasts for GPCs that delegate score computation to the LC
• Production of multi-model ensemble (MME) forecast products from the
collected GPCs
• Production of verification products for the MME forecast
products from the collected GPCs
• Dissemination of all forecast and verification products listed above in the
LC-LRFMME web site for NMHSs, RCCs and RCOFs
• Dissemination of retrospective forecast data from GPCs (that allow data
distribution) to RCCs, NMHSs and RCOFs
• Production of forecast and verification products for the Global Seasonal
Climate Update (GSCU)
• More recently, development of pilot sub-seasonal forecast products
9
 WMO Lead Centre for Long-Range Forecast
Multi-Model Ensemble
http://www.wmolc.org

Target audience:
NMHSs, RCCs and RCOFs

Jointly operated by Korean Meteorological Administration (KMA) and

10 NOAA NCEP
11
12
13
 24 15 41 42 33 20 10 30 40 42 51 41 40

1991- 1979- 1981- 1993- 1981- 1981- 1986- 1981- 1981- 1991- 1981- 1979- 1982-
2010 2010 2010 2015 2011 2010 2010 2010 2009 2010 2010 2007 2010

24 10 15 28 99 20 10 15 10 12 10 11 20

14
 •

•
•

•
•
• •
•
• •

•
•
•

•
•
•
15
16
 Date of the Month
Activity
2 4 6 8 10 12 14 16 18 20 22 24

Collection of GPCs’ model forecasts

Standardization of GPC’s forecasts to common format

Calculation of DMME and PMME

Displaying seasonal forecasts on the homepage

Calculating relevant products of GSCU

17
18
19
20
 Global Seasonal Climate Update

21
 Global Seasonal Climate Update

22
 Global Seasonal Climate Update

23
 Surface temperature products in support of GSCU
- Ensemble mean forecasts of 2m temperature anomalies wrt 1993-2009
example
1) Each GPC 2) Consistency Map 3) DMME

4) PMME

24
 Surface temperature products in support of GSCU
- Verification measure: correlation btw hindcast and obs anomalies

1) DMME 2) Each GPC

25
 Surface temperature products in support of GSCU
- Verification measure : ROC
1) PMME 2) Each GPC

26
 Surface temperature products in support of GSCU
- Verification measure: Relative Operation Characteristics(ROC) Curve and Score
Reliability Diagram over globe
1) PMME 2) Each GPC

27
 Precipitation products in support of GSCU
- Ensemble mean forecasts of precipitation anomalies wrt 1993-2009

1) Each GPC 2) Consistency Map 3) DMME

4) PMME

28
 Precipitation products in support of GSCU
- Verification measure : correlation btw hindcast and obs anomalies

1) DMME 2) Each GPC

29
 Precipitation products in support of GSCU
- Verification measure : Relative Operating Characteristics(ROC)
1) PMME 2) Each GPC

30
 Precipitation products in support of GSCU
- Verification measure: Relative Operation Characteristics(ROC) Curve and Score
Reliability Diagram over globe
1) PMME 2) Each GPC

31
 Summary of GSCU

Summary

2. Potential evolution of the state of the climate over the next three months
2.1 Large-scale SST-based indices, upcoming season
2.2 Predicted temperature, upcoming season
2.3 Predicted precipitation, upcoming season

32
 Summary of GSCU

 Large-scale multi-model forecasts of SST-based lndices, June 2018 –

August 2018

Month Nino 1+2 Nino 3 Nino 4 Nino3.4 IOD NTA STA

June 2018 -0.38 ± 0.43 0.03 ± 0.37 0.05 ± 0.15 0.00 ± 0.29 -0.01 ± 0.28 -0.40 ± 0.15 0.22 ± 0.24

July 2018 -0.20 ± 0.47 0.15 ± 0.49 0.12 ± 0.19 0.14 ± 0.39 0.02 ± 0.27 -0.28 ± 0.24 0.16 ± 0.36

August 2018 -0.07 ± 0.58 0.23 ± 0.60 0.24 ± 0.25 0.33 ± 0.51 -0.04 ± 0.28 -0.21 ± 0.23 0.16 ± 0.31

June-
-0.22 ± 0.44 0.14 ± 0.47 0.14 ± 0.18 0.16 ± 0.39 -0.01 ± 0.26 -0.30 ± 0.19 0.18 ± 0.29
August 2018

33
 Summary of GSCU

 A summary of outlook for 2m Temperature and precipitation for

the subsequent season

34
 Pilot products development

Global Producing Centres

35
 Sub-Seasonal prediction context

36
 Historical time line: LC-LRFMME sub-seasonal forecasts

 Meeting of the CBS Expert Team on Extended and Long-Range

Forecasting (ET-ELRF), 2012
- The meeting recognized the need to coordinate this initiative with
activities proposed as part of the WWRP-THORPEX/WCRP research
project on sub-seasonal to seasonal prediction.
- In response to the above request from Cg-XVI, the meeting prepared
a preliminary list of exchange variables and related sub-seasonal
forecast products.

37
 Historical time line: LC-LRFMME sub-seasonal forecasts

pilot

38
 Input data : List of contributing centres

Time Forecast Forecast Hindcast Hindcast Hindcast

GPC name Center Resolution Hindcast
Range Ens. Size Frequency Length Frequency Ens. Size

Beijing CMA d 0-60 T106L40 4 daily fix 1994-2014 daily 4

Tco639/319 2/week 2/week

ECMWF ECMWF d 0-46 51 on the fly past 20 years 11
L91 (Sun, Thu) (Mon, Thu)
4/month
Exeter UKMO d 0-60 N216L85 4 daily on the fly 1993-2015 7
(1, 9, 17, 25)
2/week 6/month
Melbourne BoM d 0-62 T47L17 33 fix 1981-2013 33
(Sun, Thu) (1, 6, 11, 16, 21, 26)
weekly Weekly
Montreal ECCC d 0-32 0.45x0.45 L40 21 on the fly 1995-2014 4
(Thu) (Thu)
4/month
Seoul KMA d 0-60 N216L85 4 Daily on the fly 1991-2010 3
(1, 9, 17, 25)
Tl479/319 weekly 3/month
Tokyo JMA d 0-33 50 fix 1981-2012 5
L100 (Wed) (10, 20, last day)
NCEP
Washington /CPC
d 0-44 T126L64 16 daily fix 1999-2010 Daily 4

39
 Operational setup challenge
Time Forecast Forecast Hindcast Hindcast Hindcast
GPC name Center Resolution Hindcast
Range Ens. Size Frequency Length Frequency Ens. Size

Beijing CMA d 0-60 T106L40 4 daily fix 1994-2014 daily 4

Tco639/319 2/week 2/week

ECMWF ECMWF d 0-46 51 on the fly past 20 years 11
L91 (Sun, Thu) (Mon, Thu)
4/month
Exeter UKMO d 0-60 N216L85 4 daily on the fly 1993-2015 7
(1, 9, 17, 25)
2/week 6/month
Melbourne BoM d 0-62 T47L17 33 fix 1981-2013 33
(Sun, Thu) (1, 6, 11, 16, 21, 26)
weekly Weekly
Montreal ECCC d 0-32 0.45x0.45 L40 21 on the fly 1995-2014 4
(Thu) (Thu)
4/month
Seoul KMA d 0-60 N216L85 4 Daily on the fly 1991-2010 3
(1, 9, 17, 25)
Tl479/319 weekly 3/month
Tokyo JMA d 0-33 50 fix 1981-2012 5
L100 (Wed) (10, 20, last day)
NCEP
Washington /CPC
d 0-44 T126L64 16 daily fix 1999-2010 Daily 4

Heterogeneity Heterogeneity Heterogeneity Heterogeneity

of Time Range of ens size of fcst & hcst init. date of hcst period

40
 Operational setup challenges

→ Every Mondays

→ 4-6 weeks (Ens. members depend on start/lead time)

→ 1.5 degree

→ 1999~2010

→ Same date as forecast or Closest date to forecast

41
 Operational setup challenge

Forecast Forecast Forecast Hindcast Hindcast Common Hind.

GPC name Center
Init. Date Time range Ens. Size Init. Date Ens. Size period

Beijing CMA 2 weeks ago Sat – last Fri 4(10)-45(51) days 28 (4*7) Same date as fcst 4 1999-2010

ECMWF ECMWF last Thu 5-46 days 51 Same date as fcst 11 1999-2010

Exeter UKMO 2 weeks ago Sat – last Fri 4(10)-45(51) days 28 (4*7) Closest date to fcst 7 1999-2010

Melbourne BoM last Thu 5-46 days 33 Closest date to fcst 33 1999-2010

Montreal ECCC last Thu 5-32 days 21 Same date as fcst 4 1999-2010

Seoul KMA 2 weeks ago Sat – last Fri 4(10)-45(51) days 28 (4*7) Closest date to fcst 5 1999-2010

Tokyo JMA last Wed 5-32 days 50 Closest date to fcst 5 1999-2010

Washington NCEP last Thu 5-39 days 16 Same date as fcst 4 1999-2010

42
 Operational setup challenge

Fcst. Ens. Avg. Tot.

WED THU FRI SAT SUN MON TUE GPC / MME Time range
Week Size Day Size

Merge 7days data Beijing 4(10)-45(51) days 6 4 7 28

ECMWF 5-46 days 6 51 1 51

Merge 7days data Exeter 4(10)-45(51) days 6 4 7 28

Melbourne 5-46 days 6 33 1 33

Montreal 5-32 days 4 21 1 21

Merge 7days data Seoul 4(10)-45(51) days 6 4 7 28

Tokyo 6-33 days 4 25 1 25

Washington 5-39 days 5 16 1 16

MME - - - - 229

43
 MME method

44
 Output data for sub-seasonal prediction

File format Grib 1

Data period 2016 to present
MME method Only SCM forecast data
Lead time 6 weeks (Ensemble members are depend on lead time)
Spatial resolution 1.5° x 1.5°

- Sea Surface Temperature (SST)

- 2m Temperature (T2M)
- Mean Sea Level Pressure (MSLP)
- Total Precipitation (PREC)
- Outgoing Longwave Radiation (OLR)
Variables
- Geopotential Height at 500 hPa (Z500)
- Zonal wind at 850 hPa (U850)
- Meridional wind at 850 hPa (V850)
- Zonal wind at 200 hPa (U200)
- Meridional wind at 200 hPa (V200)

45
 Output data for sub-seasonal prediction

Products/Variable Covering Periods Charts Verification scores

〮 ROC Curve
Probabilistic 〮 Precipitation 〮 Weeks 1, 2, 3, 4, 5, 6,
〮 Probabilistic maps 〮 ROC Score map
MME 〮 2m Temperature 3-4, 5-6, 3-6
〮 Reliability Diagram

〮 Precipitation
〮 2m Temperature 〮 Pattern Correlation
〮 Mean Sea Level Pressure
Coefficient
〮 Geopotential Height
Deterministic 〮 Weeks 1, 2, 3, 4, 5, 6, 〮 Each variable 〮 Root Mean Square
at 500hPa
MME 3-4, 5-6, 3-6 anomaly map Error
〮 850hPa Wind
〮 200hPa Wind 〮 Time Correlation
〮 200hPa Velocity Potential Coefficient
〮 200hPa Stream Function

〮 MJO Need:
OLR, U850, U200 〮 Root Mean Square
〮 Hendon and Wheeler
Intraseasonal 〮 BSISO Need: Error
〮 42 days Diagram
Oscillations OLR, U850
〮 Hovmöller diagram 〮 Correlation
〮 Outgoing Longwave Coefficient
Radiation

46
 Products : Probabilistic MME

- Temp: Tercile probabilistic forecast

using parametric method
(Gaussian fitting)

- Precip: Tercile probabilistic forecast

using parametric method
(Gamma fitting)
47
 Products : Deterministic MME

<T2M> <PREC> <MSLP> <Z500>

<850hPa wind> <200hPa wind> <200hPa velocity potential> <200hPa stream function>

48
 Products : Prediction map over various regions

Europe Russia North America

South Asia East Asia

South America
Africa

Australia

49
 Products : Intraseasonal Oscillation

<OLR, Hovmöller diagram> <MJO phase diagram>

50
 Products : Intraseasonal Oscillation

* Lee, J.-Y., B. Wang, M. C. Wheeler, X. Fu, D.E. Waliser, and I.-S. Kang, 2013: Real-time multivariate indices for the boreal summer
intraseasonal oscillation over the Asian summer monsoon region. Clim. Dyn., 40, 493-509.
51
 Products : Probabilistic verification


- Probabilistic verification : ROC curve, Reliability Diagram

<ROC curve> <Reliability Diagram>

52
 Products : Probabilistic verification

- Probabilistic verification : ROC score map / T2m, Precipitation

<Temp. AN> <Temp. NN> <Temp. BN>

<Precip. AN> <Precip. NN> <Precip. BN>

53
 Products : Deterministic verification

- Deterministic verification: Pattern Correlation Coefficient, RMSE, etc

<PCC>

<TCC>

<RMSE>
54
55