Creating Diagnostics For Sub-Seasonal Forecasts

Creating Diagnostics for Sub-
Seasonal Forecasts
14ITWCVP Training Workshop
Antalya, Turkey, 5 – 15 October 2022
Endalkachew Bekele
NOAA/CPC/International Desks
Tools for Operational Sub-Seasonal
Forecasting
a) Numerical Weather and Climate Model Forecasts
b) Madden Julian Oscillation (MJO)
c) Indian Ocean Dipole (IOD)
d) Other Tools
 Tropical Cyclone Activity (mostly for week-1 outlooks)
 ENSO
 NAO
 QBO, etc
2
State of the MJO
• Much emphasis is given to the state of the MJO and its

projected phases on the Wheeler-Hendon diagram at the
moment of the forecasts.
• Refer to the MJO monitoring and prediction tools to

determine if an active MJO is present
https://
www.cpc.ncep.noaa.gov/products/precip/CWlink/MJO/mjo.sht
ml#discussion
MJO, 200-hPa Velocity Potential Anomalies
Green shades: Anomalous divergence (favorable
for precipitation).
Brown shades: Anomalous convergence

(unfavorable for precipitation).
• Eastward
propagation of
– Area of an enhanced
convection
(precipitation)
– Area of Upper-level
divergence (green
contour)
MJO, The Wheeler-Hendon Diagram
Each dot/number MJO exists when

represents a single Weak there is
day and location of MJO counterclockwise
the MJO enhanced movement on
rainfall. diagram
Strong
MJO
MJO Precip Composites source:
https://www.cpc.ncep.noaa.gov/products/precip/CWlink/MJO/Composites/Tropical/precip.shtml
Nov - Mar May - Sep

IOD Influences source: https://
www.cpc.ncep.noaa.gov/products/international/ocean_monitoring/IO_monitoring_fcsts/io_index.shtml
Negative IOD Positive IOD
• An east-west Gradient in the Indian Ocean SST

• Negative IOD brings drought to equatorial East Africa and Excessive rainfall
to eastern Asia and Australia
• Positive IOD brings floods to equatorial East Africa and droughts and
bushfires to eastern Asia and Australia
ENSO Influences (OND/DJF),
source: https://www.cpc.ncep.noaa.gov/products/precip/CWlink/ENSO/regressions/geplr.shtml
OND NDJ
NAO Influences (DJF), source: https://
www.metoffice.gov.uk/research/climate/seasonal-to-decadal/gpc-outlooks/ens-mean/nao-desc
ription
NAO Positive NAO Negative

NWP Tools
• Lower mid, and Upper-level Circulation Anomalies

– 850-hPa, 700-hPa, 200-hPa wind and Divergence
– MSLP and 500-hPa height
• Raw Rainfall Forecasts

– Rainfall Anomaly and 2-category probabilistic forecasts
• Exceedance Probabilities
– >10mm, >20mm, > 30mm, >40m, >50mm, > 75mm
• Post-processed forecasts
– Ensemble regression calibrated
NWP Tools, GEFS Precip Forecasts
• GEFS - Global Ensemble Forecast System

– A weather forecast model made up of 31 separate forecasts, or ensemble
members.
• Ensemble Mean – Average of the 31 ensemble member forecasts
• Raw Forecasts - no bias correction/calibration
• Model Climatology – Average of model reforecasts over the reference

hindcast period (2000 – 2019)
• Raw forecast anomalies are computed by removing model climatology

from the ensemble mean forecast.
NWP Tools, Post-Processing
• The skill of NWP models decreases with

forecast lead time.
– Larger model errors for forecasts beyond week-2
 A need for forecast error correction methods:

– From simple linear bias removal (bi = fi –oi) based
on recent forecasts, to
– Ensemble Regression Calibration method
NWP Tools, Ensemble Regression
Calibration Process - Rainfall
Real-time
Transformed raw
Hindcast
Observation Forecast
Observation
(4th root) Observation
Observation Std.
Anomaly Transformed
Transformed Deviation Forecast
Observation Observation Anomaly
Climatology Climatology
(4th root) Regression
Correlation Corrected
Coefficient
Forecast
Transformed
Hindcast Forecast (4th
Reforecast root) Forecast
Forecast Std. Two-
Anomaly Deviation Category
Transformed Prob.
Model Model Forecast
Climatology Climatology (normal CDF)
(4th root)
NWP Tools, Ensemble Regression
Calibration Process - Temperature
Hindcast
Observation
Observation Real-time
Observation Std. raw
Anomaly Deviation Forecast
Observation
Climatology Two-
Category
Regression Corrected Prob.
Correlation Forecast
Coefficient Forecast
(normal CDF)
Hindcast
Reforecast
Forecast
Forecast Std.
Anomaly Deviation
Model
Climatology
Week-1/2 Forecast Process
• Incorporate MJO influences incase of active MJO is present

• Incorporate IOD/ENSO influences incase of active IOD/ENSO event
• If the IOD/ENSO influence is not significant as well as the MJO is not
active, or projected to weaken considerably during the week-1/2
outlook period.
– NWP outputs are the primary tools for guiding week-1/2 forecasts.
– NWP output tools include quantitative precipitation forecasts (raw), and

post-processed precipitation forecasts.
– Examine the predicted circulation features associated with the predicted

rainfall anomalies.
– More weight will be given to the NWP precipitation outlook tools that are
more consistent with the predicted circulation anomalies.
Week-1/2 Forecast Process
Phase
Active Yes diagrams
MJO and
composites
No
Circulation Consensus
Anomalies
Forecast
Forecast,
ENSO/NAO/IOD Discussion Polygons
forecasts and and Text
composites
Description
No Raw Rainfall
Forecasts
Yes
Active
ENSO/NAO Bias
/IOD Corrected /
Reg
Calibrated
Example, Week-1 Outlook for East Africa
• Week-1 Forecast, valid 8 – 14 October 2022

• Tools
– MJO
– IOD
– ENSO
– NWP Guidance
MJO, 200-hPa Velocity Potential Anomaly
•Green shade indicates areas of upper level divergence and convection or precipitation at surface.
Brown contours indicate areas of upper level convergence or subsidence and suppressed
precipitation at surface.
MJO, Wheeler-Hendon Index - Forecasts
GEFS CFSv2 ECMWF
Forecast
Observation
BOM Canadian
MJO, GEFS MJO-Related OLR Anomalies
Red shade indicate areas
Initial date: 06 October 2022 of suppressed convection
Blue shade indicate areas

of enhanced convection
1 - 5 days ave. Forecast
6-10 days ave. Forecast
11-15 days ave. Forecast

MJO, Rainfall Composites
MJO Rainfall Composites, Africa, SON Global Tropics
State of the IOD, obs. and fcst.
IOD Index Forecast
• 7-day
SST
Anom
• 7-day SST
Tendency
• NMME forecasts suggest an

increased chance for the negative
IOD phase to increase and peak in
October 2022.
State of the ENSO, obs. and fcst.
7-day Pacific SST Anom
CPC/IRI ENSO Forecast

Season La Niña Neutral El Niño
ASO 97 3 0
SON 91 9 0
OND 89 11 0
NDJ 80 20 0
DJF 65 33 2
JFM 54 43 3 • CPC ENSO Alert System Status: La
FMA 38 56 6 Niña Advisory
MAM 22 67 11
AMJ 16 67 17
MJO Contribution?
• Do the MJO predictions suggest

enhanced/suppressed rainfall?
Eastward propagation between the Maritime
Continent (phase-5) to Western Pacific (phase-6)
>> Suppressed Rainfall over the eastern portion
of equatorial East Africa.
IOD Contribution?
• Is there any IOD contributions?

The current negative phase of the IOD is
expected to suppress rainfall in equatorial East
Africa.
ENSO Contribution?
• Is there any ENSO contributions?

The current negative phase of the ENSO (La Nina)
is expected to suppress rainfall in equatorial East
Africa.
Week-1, 7-day Circulation Total & Anomaly Forecasts
Mean Sea Level Pressure Total Mean Sea Level Pressure Anomaly
850mb Wind/Divergence Total 850mb Wind/Divergence Anomaly
500mb Height Total 500mb Height Anomaly
200mb Wind/Divergence Total 200mb Wind/Divergence Anomaly
Week-1, 7-day Circulation Anomaly Forecasts
MSLP 850-hPa wind/divergence
500-hPa height 200-hPa wind/divergence

Week-1, 7-day Precip Forecasts
Week-1 Total Week-1 Anomaly 2-catgory Calibrated
Prob. Forecast
2-catgory Calibrated Exceedance Exceedance

Prob. Forecast probability, >10mm probability, >50mm
Summary, Convergence of Evidence, Precip
• Wet
– Rainfall Model Guidance -> Suggest Above-average precip over
western Ethiopia.
– Exceedance Probability -> Higher chances for rainfall to exceed
50mm across western Ethiopia.
• Dry
– MJO signal propagation from Phase 5 to phase 6, Negative phases
of ENSO and IOD >> Suppressed rainfall in East Africa
– Lower/mid/upper-level Circulation -> suggest below-average
rainfall over parts of East Africa
– Rainfall Model Guidance -> Suggest below-average precip Near
Lake Victoria, and over parts of equatorial East Africa.
Consensus Week-1 Precip Outlook,
Valid: 8 – 14 October 2022

Creating Diagnostics For Sub-Seasonal Forecasts

Uploaded by

Copyright:

Available Formats

Creating Diagnostics For Sub-Seasonal Forecasts

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Creating Diagnostics For Sub-Seasonal Forecasts

Uploaded by

Copyright:

Available Formats

Creating Diagnostics for Sub-

a) Numerical Weather and Climate Model Forecasts

b) Madden Julian Oscillation (MJO)

c) Indian Ocean Dipole (IOD)

• Much emphasis is given to the state of the MJO and its

• Refer to the MJO monitoring and prediction tools to

Brown shades: Anomalous convergence

Each dot/number MJO exists when

Nov - Mar May - Sep

Negative IOD Positive IOD

• An east-west Gradient in the Indian Ocean SST

NAO Positive NAO Negative

• Lower mid, and Upper-level Circulation Anomalies

• Raw Rainfall Forecasts

• GEFS - Global Ensemble Forecast System

• Ensemble Mean – Average of the 31 ensemble member forecasts

• Raw Forecasts - no bias correction/calibration

• Model Climatology – Average of model reforecasts over the reference

• Raw forecast anomalies are computed by removing model climatology

• The skill of NWP models decreases with

 A need for forecast error correction methods:

• Incorporate MJO influences incase of active MJO is present

– NWP output tools include quantitative precipitation forecasts (raw), and

– Examine the predicted circulation features associated with the predicted

• Week-1 Forecast, valid 8 – 14 October 2022

Blue shade indicate areas

1 - 5 days ave. Forecast

6-10 days ave. Forecast

11-15 days ave. Forecast

• NMME forecasts suggest an

CPC/IRI ENSO Forecast

• Do the MJO predictions suggest

• Is there any IOD contributions?

• Is there any ENSO contributions?

500-hPa height 200-hPa wind/divergence

2-catgory Calibrated Exceedance Exceedance

You might also like