Transcript Document

Getting Ready for Winter:
An NCEP Update
Louis W. Uccellini
Director, NCEP
Northeast Regional Operational Workshop
Albany, New York
November 2, 2004
“Where America’s Climate, Weather and Ocean Services Begin”
1
Outline
• NWS Seamless Suite of Products
• New Climate Forecast System: Seasonal Climate
Prediction
• 6-10 day forecasts
• 4-7 day gridded forecasts
• 1-3 day range: Winter Weather Desk
• NCEP Update
– Computer
– Building
– Collaborative Activities/Test Beds
• Summary
2
P
LI RO
FE T
E
& CT
P R IO
O N
PE O
R F
M F TY
I
SP NA TIG LO
AC VI AT OD
E G A IO
T N
TR OP IO &
AN ER N
SP ATI
O ON
R
TA S
TI
O
H
N
YD
R
AG OP
R OW
IC
U ER
L
R
EC TU
R
R
EA E
R
E TI
C SE ON
O R
N V
T
EC RO OIR
O L
SY
ST
E
EN M
ER
H GY
ST EA
L
PL ATE TH
AN /L
N OC
I
C NG AL
O
EN M M
ER
VI
R
C
O
E
N
M
EN
T
W
A
C RN
AS T
O IN
O G
SE H R
F
R S
O
O
G
D & W
SS EA
U
U
IN
AT RE
TL
ID
TS
M
C
AT AL
O
C
A
E
A
N
O
IO E R H E
ST NT
C
KS
N T
S
E
S
S
New CFS
CD 6-10 Day Forecast Upgrade
DGEX + HPC support for Days 4 -7
HPC Winter Weather Desk Days 1-3
FI
RE
TH
A
E
W
ER
3
The NCEP coupled Climate Forecast System
(implemented August 24, 2004)
1. Atmospheric component
•
•
•
Global Forecast System 2003 (GFS03)
T62 in horizontal; 64 layers in vertical
Recent upgrades in model physics




Solar radiation (Hou, 1996)
cumulus convection (Hong and Pan, 1998)
gravity wave drag (Kim and Arakawa, 1995)
cloud water/ice (Zhao and Carr,1997)
2. Oceanic component
•
•
•
•
GFDL MOM3 (Pacanowski and Griffies, 1998)
1/3°x1° in tropics; 1°x1° in extratropics; 40 layers
Quasi-global domain (74°S to 64°N)
Free surface
3. Coupled model
•
•
Once-a-day coupling
Sea ice extent taken as observed climatology
Coupled Model
Simulation
ENSO SST cycles
Simulated
2002-2040 (top)
Observed
1965-2003
(bottom)
5
Latest CPC Forecast:
Weak El Nino conditions
are expected to continue
into early 2005.
6
7
8
CFS Seasonal
Precip Forecast
(mm/month)
Without
skill mask
9
CFS Seasonal
Precip Forecast
(mm/month)
With
skill mask
•
If anomaly
correlation between
forecast and observed
conditions over the
1982-2003 period is
below 0.3, values are
not shown
10
CFS Seasonal
Precip Forecast
(mm/month)
Without
skill mask
11
CFS Seasonal
Precip Forecast
(mm/month)
With
skill mask
•
If anomaly
correlation between
forecast and observed
conditions over the
1982-2003 period is
below 0.3, values are
not shown
12
CPC Winter Season Forecast
2004-05
13
Upgrade to the 6–10; 7–14 Day Forecast
CDC Calibration of 1998 MRF
•
•
•
•
Dynamical models of the atmosphere all contain biases
(differences between model and atmosphere climatologies) and
systematic errors which are flow-regime dependent.
These errors cause the model to make forecasts which are, for
example, too dry/wet, warm/cold in comparison with what is
observed.
The errors described above cause the probability forecasts from
ensemble models to be“uncalibrated” in comparison with nature.
This means that when we look at model forecasts, we find that
the probabilities predicted for wet or dry, warm or cold, do not
happen in nature with the same frequency.
CDC’s reforecast calibration greatly reduces this calibration
14
error.
CDC Calibration
•
•
•
•
To calibrate the MRF, CDC
re-ran the 15-member
ensemble of the model over a
23 year period and compared
the probability forecasts from
the model with the frequencies
of events (wet/dry, warm/cold)
in nature.
This leads to rules for
correcting future, real-time
probability forecasts from this
model.
CDC automated approach for
temp/precip is as good or
better than official forecast
during 1 year test period
Operational implementation
Sept 04
15
6-10 day forecast
Valid Nov 2 – 6, 2004
Temperature
Precipitation
16
4-7 Day Gridded Forecast
• The following grids are being generated daily
–
–
–
–
Maximum temperature
Minimum temperature
12 hour PoP
6 hourly dew point
– 6 hourly wind
(direction and speed)
– 6 hourly cloud cover
– 12 hourly weather
• Experimental grids can be viewed here
– http://www.hpc.ncep.noaa.gov/5km_grids/5km_gridsbody.html
• Production Methodology documentation
– http://www.hpc.ncep.noaa.gov/5km_grids/medr_5km_methodolo
gy.pdf
• All grids extend offshore to cover the coastal waters
• Verification program underway
17
4-7 Day Gridded Products
7 Day Forecast Valid November 3, 2004
Max Temp
Min Temp
18
Focus on Winter Weather
NWS Winter Weather Desk
Goals of 3 year experiment from 2001- 2004:
Improve Winter Weather Services to the public through
coordination of the winter weather watches/warnings
with National guidance products
Test Short Range Ensemble Forecast (SREF) system for
applications to winter weather forecasting
Motivation:
Jan 24-25, 2000; December 30, 2000: March 4-6, 2001
19
NWS Winter Weather Desk
Time line: Sep 15 – April 1
Participants
NCEP HPC
Provide SREF based Winter Wx guidance
Collaborate with WFOs (Chat Room Technology)
WFOs
All CONUS WFOs
Use guidance from NCEP to produce coordinated Winter Storm
Watches/Warnings
Products: http://www.hpc.ncep.noaa.gov/wwd/winter_wx.shtml
24 h probability (low, moderate, high) of meeting/exceeding
4”, 8”, 12” snow, 0.25” freezing rain (for day 1, 2, 3)
72h Low tracks graphic and discussion
20
Winter Weather Desk Products
PROBABILITY GRAPHICS FOR SNOW AND FREEZING RAIN
• Indicate the probability (potential) for a location to receive specific
thresholds of accumulated snow or ice.
• Snowfall - lines represent the probability (low, moderate, and high)
that enclosed areas will receive equal to or greater than a specific
threshold accumulation
Day 1
Prob > 4”
(4", 8" or 12") of snowfall
Prob > 8”
in a 24 hour period.
• Freezing Rain - depicts the
probability in the same manner
and time period as snowfall, but
with an accumulation threshold
of .25" (one quarter of an inch)
Prob > 12”
Prob > 0.25”
of freezing rain.
21
Winter Weather Desk Products
LOW TRACKS GRAPHIC
• Depicts the forecast location and central pressure of significant surface
lows impacting the 48 contiguous United States in 12 hour increments
out to 72 hours into the future.
• The low position and track
forecast by the meteorologist at
NCEP HPC is depicted in black.
• About 20 different computer model
forecasts of low position for a
given time period are depicted
with symbols.
• Provides a user both the preferred
position and track of the low and
a sense of the uncertainty with
the forecast.
22
Dec 6-7 2003 Event
• Ranked 15th on
all time ALY
snowstorm list
• Very little
Snow in E
NY/W NE in
NOV
• First week of
December
featured a
major event
23
NCEP HPC WWE Guidance
24
NCEP HPC WWE Guidance
25
Regional Statistics
Trend of ER WWE WFO stats
ER
WWE1
WWE2
WWE3
# WFOs
8
23
23
POD
.89
.90*
.92*
FAR
.33
.30
.32
* Broke previous ER regional record
CSI
.62
.65
.64
LT–Warn
13
15*
18*
26
Short Range Ensemble Forecast
15 members twice per day
63 hrs from 9 and 21Z
Resolution upgraded
August 2004
0.25” prcp
To 32 km from 48
To 60 levels from 45
Mean and spread charts available for forecaster use
Most intensive use is in production of probability of
snow and ice accumulations product
http://wwwt.emc.ncep.noaa.gov/mmb/SREF/SREF.html
27
Planned SREF Upgrades
Summer 2005
•
•
•
•
•
•
•
•
•
•
Run SREF 4 times per day (03, 09, 15 and 21 UTC)
Improved and new products (Convective, Aviation, Tropical, Energy)
Output SREF forecasts for Alaska
Implement Grid Based Bias Correction
Improve Probabilistic FVS verification
Develop Confidence Factors for forecasts
Add 5 WRF members
Add RSM BUFR files
Common WRF post-processor
Implement ensemble mean BUFR files
28
NCEP Update
29
Computing Capability
$20M/Year
Investment
Commissioned/Operational IBM Supercomputer in Gaithersburg, MD (June 6, 2003)
•Receives Over 123 Million Global Observations Daily
•Sustained Computational Speed: 450 Billion Calculations/Sec
•Generates More Than 5.7 Million Model Fields Each Day
•Global Models (Weather, Ocean, Climate)
•Regional Models (Aviation, Severe Weather, Fire Weather)
•Hazards Models (Hurricane, Volcanic Ash, Dispersion)
•2.4x upgrade operational by mid-January, 2005
•Backup in Fairmont, WV operational by mid-January, 2005 30
NCEP’s Future Location
New Location
NOAA Center for
Weather and Climate Prediction
UMD Research Park, College Park
(Early FY08)
Current Location
NOAA Science Center
World Weather Building
Camp Springs
31
NCEP’s Future Location
NOAA Center for
Weather and Climate Prediction
UMD Research Park, College Park
(Early FY08)
32
NOAA Center for
Weather and Climate Prediction
o Lease build-to-suit facility with 268,762 sq ft
o House 800 staff
o Federal employees, contractors, visiting scientists
o 40 spaces for visiting professors and students
o From the World Weather Building
o National Weather Service’s (NWS) National Centers for
Environmental Prediction (NCEP)
o National Environmental Satellite, Data, and Information Service’s
(NESDIS) research and satellite services
o From Silver Spring
o Office of Oceanic and Atmospheric Research’s (OAR) Air Resources
Laboratory (ARL)
33
NCEP Test Beds
•
•
•
•
•
•
Developmental Test bed Center (EMC)
Climate Test Bed (EMC, CPC)
USWRP/Joint Hurricane Test Bed (TPC)
Hazardous Weather Forecast Test Bed (SPC)
Aviation Test Bed (AWC)
USWRP/Hydrometeorological Test Bed (HPC) (in
progress)
• Joint Center for Satellite Data Assimilation
34
Summary
• NWS/NCEP’s Seamless Suite of Products from
S/I forecasts to daily support for WFO
watch/warning responsibilities have been
upgraded
– Based on new research results
– Enhanced modeling capabilities
– Systematic implementation and test of a “collaborative
forecast” approach
• NCEP’s infrastructure needs – computer, building,
staff – are being addressed
• NCEP will continue to make advanced products
readily available
35
Background Slides
36
37
38
39
40