1

Why is the Stratosphere More Predictable and What are the Implications for Seasonal

Predictions of the Troposphere?

Ming CaiDepartment of Earth, Ocean, and Atmospheric Science,

Florida State University

Acknowledgement: Huug M. van den Dool, Qin Zhang Rongcai Ren, Chul-Su Shin, and YY-Yu

Grant Support: NOAA CPO: NA10OAR4310168

2

Outline• Motivation and theoretical background.

• Observational evidence linking tropical-extratropical coupling coupling to meridional mass circulation and its variability.

• Modeling evidence of high predictability of the stratosphere beyond the two-week limit.

• Mass circulation variability and cold air outbreaks in winter seasons

33

Motivation and theoretical background

4

Sources of Prediction Skill • NWP: initial value problem. Rossby waves and

baroclinic system => chaotic nature => inherent predictability (1-2 weeks).

• NWP+lower boundary forcing: “forced” problems Internal variability often is as large as the forced anomalies, particularly in the absence of large SST anomalies (e.g. non ENSO years) and over the regions where prominent atmospheric internal modes are present at all time scale =>the forecasts of the “forced” anomalies are indecisive.

• New Source: Global mass circulation => The extratropics is connected to the tropics via stratosphere: => Much a longer time scale.

Timely and zonally averaged circulation in pressure/height coordinates

N S

From ECMWF ERA-40 Atlas

EQ

Pre

ssur

e (h

Pa)

66

Mean meridional mass circulation in winter hemisphere (Cai and Shin 2013)

7

Vertical and meridional couplings by baroclinically amplifying (westward tilting)

waves (Johnson 1989)

• A net poleward (adiabatic) transport of warm air mass aloft and a net equatorward (adiabatic) transport of cold air mass transport below.

8

Westward tilting waves and meridional mass transport

Johnson (1979) and Townsend and Johnson (1985)

Less mass

More mass

More mass

Less massA net warm air

transport polewardat upper layers

A net cold air transport

equatorward at lower layers

9

Observational evidence linking tropical-extratropical coupling to meridional mass circulation and its variability (derived from

NCEP/NCAR reanalysis:1979-2003)

Cai (2003, GRL)Cai and Ren (2006, GRL)Ren and Cai (2006, AAS)Cai and Ren (2007, JAS)Ren and Cai (2007, GRL)

1984-88

1979-83

1989-93

1994-98

1998-03

10

Daily time series of NH Polar vortex oscillation +NAM

-NAM/SSW

69% variance of NHdaily PV anomalies

Time

A see-saw pattern betweenmid-lat. and high-lat., &between polar stratosphere and polar troposphere.

11

Meridional propagation of thermal anomalies

Lead day of PVO index (-60 ~60)

10mb

20mb

50mb

100mb

150mb

200mb

250mb

300mb

500mb

850mb

925mb

1000mb

Upper layers totally in the ST

Middle layers across the Tropopause

Troposphere

10 hPa

20 hPa

50 hPa

100 hPa

150 hPa

200 hPa

250 hPa

300 hPa

500 hPa

850 hPa

925 hPa

1000 hPa

12

Downward Propagation of thermal Downward Propagation of thermal anomaliesanomalies

Lead day of PVO index (-60 ~60)

10N 20N 30N 40N

50N 60N 80N70N10 N

50 N

20 N

60 N

30 N

70 N

40 N

80 N

heig

ht

1313

Summary of theoretical background:

•Stratospheric annular mode variability is intimately associated with variability of the stratospheric part of the warm air branch mass circulation. •Positive phase corresponds to the state of a weaker meridional mass circulation whereas the negative phase results from a stronger meridional mass circulation.•Presence of large-amplitude westward titling waves leads to a stronger poleward mass transport.•Limited meridional scale of westward titling waves results in successive poleward progress of stronger poleward mass transport => slow time scale of annular mode variability.

1414

Implications for stratosphere predictions:

A numerical model may still have a good skill in predicting stratospheric annular mode variability even when it already loses its skill in predicting the exact locations of individual planetary-scale waves, as long as it can retain the amplitude and westward tilting of these waves.

1515

NCEP CFSv2 Prediction Skill for Stratospheric Variability

(Zhang, Shin, van den Dool, and Cai 2013)

Day 1 through Day 90 forecastsfor the period from January 1,

1999 to December 2010

1616

Forecast evaluation procedure•Remove annual cycle of CFSv2 reanalysis (observation) from CFSv2 daily forecasts at all lead times. •Consider Temperature anomalies at 50 hPa (T50 stratosphere) and 500 hp (T500, troposphere). •Verify anomalies of CFSv2 CFSv2 forecasts against anomalies of CFS reanalysis (observations).•Use either map correlation for temporal-spatial field or temporal correlation for time series of anomalies to measure the skill of CFSv2 (AC > 0.5 reliable forecasts and AC > 0.3 useful forecasts)•Calculate the same skill for “persistence forecasts”. The gain of CFSv2 skill over persistence forecasts measures the usefulness of CFSv2 forecasts.

Overall skill (map AC)NH SH

Forecast lead time (days)

T500

T50

T500

T50

15days 18days

18

Seasonal dependency of CFSv2 skill over polar stratosphere

NH SH

19

1st 4 EOF modes of daily T50 anomalies

23.4% 18.6%

13.4% 5.2%

NH only

20

AC

ski

ll of

pre

dict

ions

of P

Cs

of E

OF1

-4

24 36

1117

21

Temporal evolution of zonal mean of T50 anomalies

22

CFSv2 skill in predicting zonal mean of T50 anomalies

23

Gain of CFSv2 over persistence forecasts

24

Summary (1)• The NCEP CFSv2 still has a remarkable prediction skill of polar stratospheric

temperature anomalies at the lead time of 30 days whereas its counterpart in the troposphere at 500 hPa drops very quickly and below the 0.3 level at 12 days. We also prove that the CFSv2 has a high prediction skill both in an absolute sense and in terms of gain over persistence except in the equatorial region where the skill mainly comes from the persistence of the QBO signal.

• In particular, the CFSv2 is capable of predicting mid-winter polar stratosphere sudden warming events in the Northern Hemisphere and the timing of the final warming polar stratosphere warming in both hemispheres 3-4 weeks in advance.

• The remarkable skill comes from the signal of systematic poleward propagation of thermal anomalies from the equator to the pole in the stratosphere associated with the global mass circulation variability (intensity/time scale).

• As long as the westward tilting of planetary waves in the stratosphere and their overall amplitude can be captured, the CFSv2 forecasts would still be very skillful in predicting zonal mean anomalies even though it cannot do so for the exact locations of planetary waves and their spatial scales.

2525

Mass circulation variability and cold air outbreaks in winter seasons

Cai (2003, GRL)Cai and Ren (2007, JAS)

Cai and Yu (2013)

26

Meridional propagation (Cai and Ren 2007) P

olew

ard

prop

agat

ion

in

the

stra

tosp

here

Equ

ator

war

d pr

opag

atio

n in

tro

posp

here

2 Periods

40 days

70 days

60N

25N

90N

Stronger Meridional Mass Circulation

Mass circulation variability and cold air outbreaks in the mid-latitudes

Warmair

Cold air

60N

25N

90N

Weaker Meridional Mass Circulation

WarmairCold

air

Less cold air outbreaks in mid-latitudesColdness in high latitudes

More cold air outbreaks in mid-latitudesWarmness in high latitudes

2828

Indices of mass circulation (1) Total mass flux into polar cap in the warm air branch(2) Total mass flux out of polar cap in the cold air branch(3) Total mass flux into polar cap in the stratosphere

2929

Indices of coldness and warmnessColdness indices(1) Percentage of area occupied by negative temperature

anomalies below n*standard_deviation (local) in high latitude zone (60 N poleward) and mid-latitudes (between 25 and 60N), with n = 0, 0.5, 1, 1.5, 2.0 etc..

(2) The areal mean of temperature anomalies below n*standard_deviation (local) in high latitude zone (60 N poleward) and mid-latitudes (between 25 and 60N), with n = 0, 0.5, 1, 1.5, 2.0 etc..

Warmness indices: same as coldness indices except for positive temperature anomalies above n*standard_deviation (local).

The results with n = 0 are representative

3030

Exa

mpl

e of

circ

ulat

ion

and

tem

pera

ture

indi

ces

(afte

r rem

ovin

g th

eir a

nnua

l cyc

les)

Area

Amp.

Masscirculation

3131

Exa

mpl

e of

lead

/lag

corr

elat

ion

betw

een

circ

ulat

ion

and

tem

pera

ture

indi

ces

Lag days of warm air branch MV

Lag days of Stratosphere MV

High Lat. Mid-latitudes

Lag days of Stratosphere MV

Lag days of warm air branch MV

32

Stratosphere mass circulation versus warm/cold air branch circulations

Lag days of Stratosphere MV Lag days of Stratosphere MV

Warmair

branch

Coldair

branch

33

Stratosphere mass circulation versus temperature indices

Coldness area indices

Lag days of Stratosphere MV Lag days of Stratosphere MV

Warmness area indices

> 60N 25-60N > 60N 25-60N

34

Stratosphere mass circulation versus temperature indices

Coldness area indices (n =1) Warmness area indices (n=1)

> 60N 25-60N > 60N 25-60N

Lag days of Stratosphere MV Lag days of Stratosphere MV

35

Stratosphere mass circulation versus temperature indicesHigh Lat. Mid Lat.

Coldness

Warmness

36

Summary (2)• Variability of mass flux into polar stratosphere is synchronized with

that of warm air branch and cold air branch.

• Surface temperature anomalies in winter seasons are intimately coupled with the timing/intensity of global mass circulation variability.

• Lack of warm air into polar stratosphere is accompanied by weaker equatorward advancement of cold air near the surface. As a result, the cold air mass is largely imprisoned within polar circle, responsible for general warmness in mid-latitudes and below climatology temperature in high latitudes.

• Stronger warm air into polar stratosphere is accompanied by stronger equatorward advancement of cold air near the surface, resulting in massive cold air outbreaks in mid-latitudes and warmth in high latitudes.

37

A hybrid forecasting strategy for intraseasonal cold season climate predictions (14 - 60? days)

• Prognostic component: Dynamical prediction for (extratropic) stratospheric anomalies using CFS

• Diagnostic component: Statistical “instantaneous” relations between stratospheric and surface temperature anomalies (downscaling)

• Potential products:1.Likelihood of series and severe winter storms in a

time span of a few weeks over a continental domain.2.Mild versus cold winter season.3.Timing of the coldest period (“early” versus “late”

winter).