NWP, Climate Modeling, and Observational Studies at FSU With

NWP, Climate Modeling, and
Observational Studies at FSU With
Special Emphasis on the Use of
ERA-40 Data sets from SCD/DSS
Vijay Kumar Tallapragada
Department of Meteorology
Florida State University
Tallahassee, Florida 32306-4520
Numerical Weather Prediction
(Short-to-Medium Range)
Precipitation and
Flood Forecasts
Use of
Satellite
derived/
observed
Rain-rates
Physical
Initialization
and
Assimilation
Studies
Hurricanes and
Tropical Cyclones
Multimodel
Superensemble
forecasts for
tracks and
intensity
Adaptive
Observational
Strategy for
improved
hurricane
forecasts
High-Resolution
Numerical Modeling
Experiments
Real-Time
Global/
Regional
NWP
Cumulus Parameterization,
Model
Improvements,
Error Reduction
Climate Modeling Studies
MJO/ISO
Scale
Interactions
in the
frequency
domain
Lowfrequency
filtered
initial states
for MJO
prediction
Seasonal Climate
Forecasts
Multimodel
Superensemble
forecasts from
Coupled
Models
AMIP/
DEMETER
Model Data
Sets, CMAP,
Reynolds SST
etc.
Predictability and
Statistical Skill
Measures
Brier Skills,
TSS, RMS/
ACC, FAR
Synthetic Data
Generation
using EOFs and
PC Time Series
Observational & Diagnostic
Studies
Monsoon Studies
TRMM PR
2A25 Data for
Hydro-meteor
Distribution Monsoon
Depressions
MJO/ISO;
ENSO-IOD;
Phase
Locking,
Energetics
Hurricane Intensity
Weakening of
Hurricane Lili
(October 2002)
Angular
Momentum
Budget and
Scale
Interactions
Synoptic Studies
Antecedents
of Great
Floods of
England,
2001
Advection of
Ozone; Moisture
Transport over
Monsoon
Region etc.
Resources
Data
•Historical Atmospheric Data sets from
ECMWF and NCEP Reanalyses
•Real-time High-Resolution ECMWF
operational analysis
•Real-time medium range forecasts
from several operational centers
across the world
Tools
•
•
•
•
•Real-time TRMM, SSM-I, IR, CMORPH,
Stage-IV and other rainfall products
•
•Special observations from field
campaigns (CAMEX, INDOEX, TEXMEX
etc.)
•
State-of-the-art High-Resolution
FSU Global Spectral Model
(T255L28) and Nested Regional
Spectral Model (~25km)
FSU Coupled Ocean-Atmosphere
Global Spectral Model (T63L14)
High-resolution MM5 and WRF
mesoscale models
Data Assimilation (3D-VAR, SSI)
and Physical Initialization
IBM SP4 Architecture Super
Computing; NCAR Computing
resources and Linux clusters
Visualization and graphics
software - GrADS, IDL, Vis5D, NCL
etc.
Multimodel Superensemble For NWP
Participating Models Training Data
•
•
•
•
•
•
•
ECMWF
NCEP GFS
JMA GSM
CMC GEM
FNMOC NOGAPS
BMRC GASP
FSUGSM
•
•
ECMWF Data Sets for
u, v, T and z, and the
surface parameters
TRMM based rain
rates for precipitation
Surface Temperature
RMS Error of U-850 hPa during Monsoon season (June to September 2004) over the
Indian Domain (30E-150 E, 40 S-40 N)
BMRC
NRL
FSU
ECMWF
JMA
RPN
NCEP
SUPERENSEMBLE
6.00
5.00
RMSE (m/s)
4.00
3.00
2.00
1.00
0.00
1
2
3
4
Forecast Day
5
6
RMS Error of V-850 hPa during Monsoon season (June to September 2004) over the
Indian Monsoon Domain (30E-150 E, 40 S-40 N)
BMRC
NRL
FSU
ECMWF
JMA
RPN
NCEP
SUPERENSEMBLE
5.00
4.50
4.00
RMS Error (m/s)
3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00
1
2
3
4
Forecast Day
5
6
The interactions among different processes in a NWP model (taken
from ECMWF).
!
!
"
!
#
#
Superensemble based Climate Studies
• Global Seasonal-Climate Hindcasts based on
multimodel (atmospheric and coupled oceanatmosphere) superensemble
• Variants of classical superensemble technique
(synthetic approach)
• Huge data base consisting DEMETER (forced by
ECMWF ERA-40), BMRC, CMC and 4 versions of
FSU models
• Couple Ocean-Atmosphere model for studying the
MJO, ENSO and IOD
• Observed climatology based on ERA-40, NRA, X-A
and Reynolds data sets.
Multi Model Synthetic Ensemble/
Superensemble for Seasonal
Climate
-Provides highly skillful deterministic and
probabilistic forecasts for seasonal climate.
-Potential applications include understanding
the climate variability and predictability
E(ε2)
Minimization
E(ε2)
Minimization
•
•
• EOF
Generating Synthetic Data Using
N - Actual Data Sets
F i( x , T ) =
F
i, n
(T ) .
i, n
( x )
n
Observed Analysis
O ( x ,T ) =
P
n
(T ) .
n
( x )
n
Estimating Consistent Pattern
What is matching spatial pattern in forecast
data Fi(x,t), which evolves according to PC
time series P(t) of observed data, O(x,t) ?
P (t ) =
i,n
F i, n (t ) +
α
F i reg (T ) =
i,n
n
N - Synthetic Data Sets
(t )
n ,i
Fi syn ( x, T ) =
Fi ,reg
n (T ). n ( x)
n
F i ,n (T )
Characteristics of the 13 models
RMS Error in Meridional Wind Forecasts, JJA, Indian region
Continued to next page…
RMS error from the Best Model, Ensemble Mean and the Superensemble
Seasonal SST Forecasts
SST Anomaly (K) Over Tropical Pacific Ocean, March-May 1999
Observational/Diagnostic Studies
• MJO/ISO Energy Exchanges in the
frequency domain and Predictability
studies
• Moisture Transport in the Monsoon Region
• Ozone distribution in the INDOEX region
• Hurricane Intensity issues
• Superanalysis
The Use of Scale Interactions
as a framework for the Maintenance of the
Madden Julian Oscillation
!
"
"#
$%&!
Data
• ECMWF Re-Analysis Data (ERA 40): 2.5 degree
resolution
interpolated to T63 spectral res’n
(1.875 degrees)
– Employed once daily data consisting of new in-situ and
remotely-sensed measurements (i.e. Data from
HIRS/MSU/SSU/SSMI– instrumentation available post
1979)
– 300 hPa, 850 hPa and 30 hPa (QBO)
– Variables: Temperature, U/V winds, vertical velocity
Data – Cont.
• Relevant Time-scales of Maintenance
– MJO time-scale (30-60 days) – centerpiece
–
–
–
–
–
Synoptic (3-7 days)
Semi-annual (170-190 days)
Annual
ENSO (3-7 year)
Decadal and beyond (10 years +)
• Does the ReAnalysis data
contain MJO
time-scale
feature at 300
hPa?
• 300 hPa
illustrates upper
tropospheric
structure of
MJO
• 850 hPa
signal of MJO
– clearly seen
in filtered Chi
field
($
&'$
! "
($
&'$
! "
) *
+, ' $
$-
.
/0
'
($
&'$
&
!
)
! "
'
($
) *
+, ' $
$-
.
/$ 0
'
($
&'$
&
!
)
! "
' 1
($
Eastward passage of
the Madden Julian
Oscillation. Result
taken from a 90-day
coupled model run
during
summer
1996. 200 hPa wind
anomalies
(ms-1)
shown here are
averaged between
5oS and 5oN.
East - West Circulations
Wind Stress
Over the Ocean
Im
e
MJ
OT
MJO TIme Scale
JO
M
MJO
ic
pt
no les
Sy ca
S
Constant Flux
Layer
PBL
MJO
ic
pt
no les
Sy ca
S
Sc
al
e
MJO
HADLEY
OCEAN
(Many Slow Modes)
Wind Stress
Over the Ocean
A schematic diagram illustrating regions on a frequency ‘r’ versus frequency ‘s’ space
where the synoptic time scales can interact with the MJO time scales. The shaded
area denotes where s-r or r-s can amplify the MJO time scales via triad interactions .
2
3
4
2
Some Recent Results on Monsoon
ISO using ERA-40 data sets and “daily
Data” based Superensemble forecasts
Of seasaonal climate.
Ozone Transport and Dynamical
Variability
• Pollution emitted in one area can be transported half way around
the globe in a matter of a few days and affect the air quality in
that region. Transport and chemical transformation of optically
active trace gases modify the largescale distribution of chemical
species and their global budgets, which can have important
implications for the chemical composition of the atmosphere and
its radiative properties. Along with CO, CO2, NOx, CFC'
s, and
CH4, ozone plays an important role in the climate system by
being a main atmospheric oxidizing component and opticallyactive constituent.
• Develop a four-dimensional Eulerian model to assimilate the
vertical distribution of ozone from satellite derived total ozone.
• Develop tropospheric ozone season and annual climatology.
• Assimilate the elevated ozone plume documented at Kaashidhoo
Climate Observatory, and assess the role of dynamic features in
the formation of this high ozone event.
• Use assimilated ozone fields to study the intercontinental
transport of tropospheric ozone.
• Data: TOMS, ERA-40, NRA, ECMWF 3D-VAR assimilated
Ozone (TOMS+SBUV), Nox and CO from Max-Planck
(MATCH-MPIC), GOME-NO2 and INDOEX observations,
BADC Monthly mean gridded ozone
Comparison of Vertical
Profiles of Ozone
Model computed Vertical Profile
of Ozone at Kaashidhoo
10-day backtrajectories at 500 hPa
10-day backtrajectories at 700 hPa
10-day forward
trajectories at 300 hPa
10-day forward
trajectories at 700 hPa
Phase Locking as evidenced through
ERA-40 data sets for the year 2001.
The data here consists of u,v,T,RH,W,
Z, surface and PBL fluxes of heat,
Moisture and momentum.
Chakraborty and Krishnamurti, JAS 2005
Variations in different
parameters during the
active and break phases
of monsoon activity over
the Central Indian
Region, from the ERA-40
data sets.
Hurricane Intensity Issue
• High Resolution Modeling of Hurricanes Bonnie
and Isabel
• Combining TRMM, CAMEX and ERA-40
observations to initialize the model
• Angular Momentum budgets
• Energy exchanges (scale interactions) in local
cylindrical coordinates: Clouds interacting with
Hurricane Scale Motions
• Energy exchanges during an Explosive Growth
situation (e.g., Hurricane Isabel
Observed and model predicted tracks of
Hurricane Bonnie
2
"55&
6
7+ 8
$
#
&
%
Red - Inner Area
Green - Fast Winds
Blue - Outer Area
'
" )
'*
"
("
'
!
+
Ongoing Hurricane Intensity Initiative
Using Mesoscale Models
Sensitivity Studies
em
M
od
rM
MM5- I
er
mb
Me
del
Mo
Intensity
forecasts and
interpretation
of skills
Multi Model
Meso Scale
Superensemble
od
el
M
M
em
be
r
WRF
Me
m
be
r
r
da
un
Bo ns
itio
al
ter ond
C
Me
mb
er
La
Meso Scale
Data
Assimilation
Member Model
Mo
de
l
MM5-II
el
COAMPS
Global Spectral
Model
T255
28 Layer
Angular
Momentum
Perspective
be
ry
da
n
u
Bo ition
l
d
a
ter con
a
L
CORE High
resolution
Mesoscale
microphysical
nonhydrostatic
model
Mo
de
l
Operational Data
and
Data Assimilation
Ice, groupel,
snow, cloud
liquid water,
rain water, hail
y
ARPS
Meso scale
data sets
RAMS
Scale
Interaction
among
hurricane
and cloud
scale
Krish’s lab at FSU has been
actively engaged in many areas of
research in weather and climate
for the last three or more decades,
and we largely depend on the
Datasets from NCAR SCD for
these research studies.
The support from SCD in terms of
data distribution, storage and
computing can not be overstated.
The inclusion of ERA-40 Datasets
(particularly the “corrected”
gridded u, v data sets) have
provided us with more choices and
added confidence to our efforts;
we anticipate more aggressive use
of these data sets in many of our
ongoing projects.
THANK YOU
Credit goes to my colleagues at
FSU for sharing many of the
research results presented here,
and to Krish who is the master
mind behind all these projects.