第1期業務実験の中間報告

Products of the JMA Ensemble
Prediction System (EPS)
for
Seasonal Forecast
Shuhei Maeda
Climate Prediction Division
[email protected]
Contents
1. Introduction
2. Products of the EPS for
one-month prediction
3. Products of the EPS for
three-month outlook
4. Remarks
TCC Web page
http://cpd2.kishou.go.jp/tcc/
One-month prediction
Three-month outlook
2. Products of the EPS for onemonth prediction
1) Outline of the EPS
2) Performance of the system
3) Examples of products
The EPS for one-month prediction
Analysis
Atmosphere
Atmospheric
Model
Land
Boundary
Condition
(AtmosphereLand)
Ocean
SST
persisted SST
anomaly
Products
• Map
• Verification
・GPV
Horizontal
resolution
Time integration
range
Executing
frequency
Ensemble size
SST
Physical
processes
T106 (about 1.125 º
Gaussian grid ~110km)
34 days
Once a week
26 members
persisted anomaly
The same as those of shortand medium- range forecast
model
Example of ensemble prediction
Initial
Spread
W
Ensemble mean
Each member
7 day running mean Z500 normalized anomaly
over the western North Pacific (40N,140E-170E)
Products
Kind
Forecast
range
Ensemble 2-8day
mean
9-15day
map
16-29day
2-29day
Parameter
(NH)
Z500,Z500A
T850,T850A
PSEA
PSEAA
*○○○A:
(60N-60S and Asia)
Z500,Z500A
SLP,SLPA
PSI200,PSI200A
PSI850,PSI850A
CHI200,CHI200A
T2m,T2mA
RAIN,RAINA,SSTA
anomalies of
○○○
(continue) Products
Forecast
range
Parameter
Spread map 2-8 day
Z500A
9-15 day
Large
16-29 day
anomaly
index* map 2-29 day
Time series 7-day
running
(each
mean
member)
Circulation indices (Z500A)
Area averaged T850A
Area averaged spread (Z500A)
CHI200 in the equatorial region
*Ratio of number of members whose anomaly is higher than 0.5SD to
total number of ensemble menber at each grid point
2) Performance of the system
Example of one-month prediction
verification charts
W
Anomaly correlation scores of ensemble mean
forecast in 2001
28 day (2-29) mean 500hPa height fields over the
Northern Hemisphere. Initial: 2001/1/4-2001/12/27
W
RMSE of ensemble mean forecast in 2002
28 day (2-29) mean 500hPa height fields over the
Northern Hemisphere. Initial: 2002/1/3-2002/12/27
Example of verification of probabilistic
one-month prediction
Event (E)
Z500 anomaly is
greater than 0
Frequency of occurrences
where E was forecast
within each of the
probability category bins
W
Forecast frequency
distribution
Reliability diagram for 7 day mean 500hPa field: day 9-15
2003 Spring (2003/2/27-5/29)
3) Examples of products
Prediction of
・Arctic Oscillation
・Rossby waves along Asia jet
・Intra Seasonal Oscillation in
the tropics (Madden Julian
Oscillation (MJO))
Example of one-month prediction
products(1)
Z500
Z500
analysis
forecast
AO
!
T850
T850
analysis
forecast
W
Ensemble average forecast map ( Initial: 2002/10/24)
Northern Hemisphere Z500 and T850 ; 28day mean
Many members predict
large Z500 anomaly
(above 0.5*standard
deviation)
Many members predict
large Z500 anomaly
(below 0.5*standard
deviation)
W
Large anomaly index map
Init:2002.10.24
28 day mean ( day:2-29)
W
Analysis and
ensemble average
Each
member
Analysis
7 day running mean Z500 normalized anomaly
over North western Pacific (40N,140E-170E)
Example of one-month prediction
Oct
Nov products(2)
Dec
Jan
Time sequences
of temperature
anomalies in
Japan (5 day
running mean)
2001.10-2002.1
Normalized
temperature
anomalies
2002.1.11-15
W
What?
Prediction of the Rossby wave along Asia jet
PSI200
analysis
1/12-1/18
PSI200
forecast
W
Ensemble average forecast map ( Initial: 2002/1/3)
PSI200 ; 7day mean (9-15 day forecast range)
Rossby wave guide along Asia jet
Dispersion relation of barotropic Rossby wave in U(y) is
ω=Uk-β*k/K2,
here, β* = β-∂2U/ ∂y2 , K2 =k2+l2.
Therefore, total wave number of a stationary Rossby wave is
K=Ks=( β*/U)1/2.
Hoskins and Ambrizzi(JAS,1993,1661-) showed that if Ks has
maximum at some latitude as figure 1, it works as “Rossby
wave guide”
latitude
Schematic figure of
Rossby wave guide
Climatological Ks
in DJF (shade)
and
Analyzed stream
function anomaly
at 200hPa
(2002.1.11-15)
Temperature anomalies, and Rossby wave along
Asia jet
W
W
Normalized
temperature
anomalies
Stream function
(200hPa)
anomalies
Longitude and
height cross
section of 20N-30N
averaged stream
function anomaly
2002.1.11-15
Time-longitude cross-section of stream function
anomaly (20N-30N) at 200 hPa initial 2002.1.3
(left) analysis (right) ensemble mean forecast
Spread among ensemble members
W
Spread among ensemble
members ( Initial: 2003/10/16)
Z500 ; 7day mean (9-15 day )
Predicted stream function
(200hPa) anomalies and
wave activity flux of two of
26 ensemble member
Example of one-month prediction
products(3)
96015
98618
Observed amount of
daily precipitation in
MAY 2003
What ?
?
97430
91425
1
10 20 30
MAY 2003
Eastward propagation of MJO
1-5 May
2003
6-10 May
11-15 May
16-20 May
Velocity potential
(200hPa)
OLR
1 May
day
Velocity
potential
(200hPa) in the
equatorial region
W
longitude
20 May
Schematics of MJO 3-D structure
over the Indian Ocean
Rui and Wang(1990)
Prediction of MJO in May 2003
Observation
2nd week prediction ( 7 day mean)
Velocity
potential
anomalies
(200hPa)
Stream
function
anomalies
W (200hPa)
2003/5/3-5/9
Ensemble mean
Initial:2003/4/24
Observation
2nd week prediction
Velocity
potential
anomalies
(200hPa)
Stream
function
anomalies
W (200hPa)
2003/5/10-5/16
Ensemble mean
Initial:2003/5/1
Observation
2nd week prediction
Velocity
potential
anomalies
(200hPa)
2003/5/17-5/23
W
Ensemble mean
Initial:2003/5/8
Stream
function
anomalies
(200hPa)
Prediction of MJO in Oct. 2003
Init:10/9 2nd week
20 Sep
day
Obs.
10 Oct
CHI200a
Fcst
W
Longitude
PSI850a
Velocity potential at
200hPa (CHI200)
PSI200a
Prediction of MJO in July 1984-1993
OBS
FCST
1
84
31
86
92
88
CHI200A in the equatorial region