Variability, Predictability and Prediction of DJF season Climate in CFS

1
Variability, Predictability and Prediction of
DJF season Climate in CFS

• Peitao Peng1, Qin Zhang1, Arun Kumar1,
• Huug van den Dool1, Wanqiu Wang1,
• Suranjana Saha2 and Hualu Pan2
• 1 CPC/NCEP/NOAA
• 2 EMC/NCEP/NOAA

2
Why DJF season?

• In NDJ, ENSO reaches its peak
• In February, Atmospheric teleconnections are the
  strongest

3
Objectives

• Evaluate the performance of CFS in forecasting
  DJF climate
• Understand the CFS performance
• Estimate the potential predictability of DJF
  climate with CFS

4
Outline

• Document the CFS forecasted climatic state and
  its drift with the lead time of forecast
• Examine the variability of CFS forecasted climate
  and its dependence on the lead time of forecast
• Examine the CFS forecasted ENSO and its
  associated climate anomalies
• Document the CFS prediction skill for DJF climate
  and estimate the potential predictability of CFS

5
Data

• Model 23-year CFS hindcast dataset
• (1982-2004)
• OBS
• SST OI SST
• Surface Temperature CAMS data
• Z200 Reanalysis 2 (R2)

6
More for Model Data
DJF
May
Jun
Jul
Aug
Sep
Oct
There are 15 runs from each month
7
Climatic state and its drift with lead time of
forecast
8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
Variability of DJF mean

• Total variance
• Variance of ensemble mean (signal) Variance of
  spread (noise)

15
(No Transcript)
16
(No Transcript)
17
EOFs of Z200

• CFS (total variability) vs OBS
• EOFs of ensemble mean

18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
ENSO and its associated climate anomalies

• CFS vs OBS
• El Nino vs La Nina (linearity)
• Dependence on lead time

22
obs
OCT_IC
Aug_IC
May_IC
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
Prediction skills

• Against obs
• Against model itself Taking one member as OBS
  and the average of other 14 members as forecast
  (perfect model)

27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
Summary

• Part of the CFS climate drift in the extratropics
  is likely forced by the drift in the tropics
• Climate drift increases moderately as lead time
  of forecast increases from one to six months
• ENSO dominates the predictable component of
  interannual climate variability
• In the period of 1982-2004, ENSO-related mean
  anomalies are pretty linear in both CFS and OBS.

32
Summary continued

• CFS shows pretty high forecast skills for the
  tropics and appreciable skills for the
  extratropics with up to six-month lead time
• The decrease of forecast skills in the
  extratropics for longer lead time is partially
  due to the westward shift of the ENSO
  teleconnection patterns in forecast, which in
  turn is caused by the westward shift of tropical
  SST and precipitation patterns
• perfect model skills show us brighter future