CPC Forecasts: Current Methods and Future Requirements

1
CPC Forecasts Current Methods and Future
Requirements

• Ed OLenic
• NOAA-NWS-Climate Prediction Center
• Camp Springs, Maryland
• ed.olenic_at_noaa.gov
• 301-763-8000, ext 7528
•  

2
Outline over-view
Outline - overview

• Weather versus climate
• Current status of CPCs forecast system
• Schematic of CPCs forecast system
• Schematic of the forecast process

3
Outline 30- and 90-day
Outline 30-and 90-day long lead outlooks

• Seasonal and monthly forecasting
• Forecast schedule and leads
• Outlook category and probability formats
• Maps and bulletins
• Forecast tools
• Canonical Correlation Analysis (CCA)
• Optimal Climate Normals (OCN)
• Ocean-Atmosphere Model
• Operations Concept
• Sample seasonal forecast using actual tools

4
Outline - 6-10/week 2
Outline 6-10 day and week 2

• Forecast schedule and leads
• Outlook category and probability formats
• Maps and bulletins
• Tools
• Dynamical models
• Natural analogs
• Teleconnections
• Downscaling techniques
• Sample forecast using actual tools

5
Outline - Threats
Outline - Threats

• Forecast schedule and leads
• Definition of Threats
• Outlook formats
• Maps and bulletins
• Tools
• Dynamical models
• Composites

6
WEATHER vs. CLIMATE

• Wildly oscillating curve daily weather

• Smooth curve 30 year mean (climatology)

Subtracting the climatology and performing a
31-day running mean reveals the low-frequency
signal or short-term climate variations we are
trying to predict.
7
Climate Prediction Center Forecast System
Schematic
8
Forecast Process Schematic
Recent observations
Dynamical model forecasts/multi-model ensembles
Historical observations..
Verifications/Statistical tools Downscaling,
Analogs, Composites
WEB PAGES/AUTOMATED DATABASES Peer-reviews of the
forecast tools and of the penultimate forecast
via web/telephone conference with partners and
through local discussions (map discussions,sanity
check, conference calls, etc)
Forecaster-created or automated products
Dissemination to public
9
Monthly and Seasonal Outlooks
10
Seasonal/Monthly Outlook Schedule/Leads

• Each month, near mid-month
• (http//www.cpc.ncep.noaa.gov/products/predictions
  /schedule.html),
• CPC prepares a set of 13 outlooks for 3-month
  seasons (any set
• of 3 adjacent months) for lead times ranging from
  ½ month, 1 ½
• months, 2 ½ months, 3 ½ months, , 12 ½ months.
• http//www.cpc.ncep.noaa.gov/products/predictions/
  90day/
• The outlook for each successive/prior lead time
  overlaps the
• prior/successive one by 2 months. This overlap
  makes for a
• Smooth variation from one map to the next. A
  monthly outlook is
• prepared at the same time as the seasonal.
  However, it is for a
• single lead time, ½ month.

11
Outlook Categories and Probabilities

• Seasonal and monthly outlooks are prepared for
  average temperature and total accumulated
  precipitation category (please see
  http//www.cpc.ncep.noaa.gov/products/predictions/
  multi_season/13_seasonal_outlooks/color/seasonal_f
  orecast.html )
• Three categories are used (terciles). These are
  BELOW-,NEAR- and ABOVE-normal, for temperature,
  and BELOW-, NEAR-and ABOVE-median for
  precipitation.
• Regions where the likelihoods of the three
  categories are the same (33.33 each) are
  designated as CL, for climatology. Note that,
  at any given location and time, the sum of these
  three probabilities is 100. This is true for
  non-CL regions, too.
• In non-CL regions the labels on the contours give
  the (positive) difference between the probability
  of the indicated category and 33.33, i.e., they
  are probability anomalies.
• The simple rule that applies here is For A or
  B, the likelihood of the indicated category is
  33.33 plus the contour value, while, the
  likelihood of the opposite category (B or A) is
  33.33 minus the contour value. The value of N
  remains 33.33.
• Finally, when N is favored, its value is 33.33
  plus the contour value, while the likelihoods of
  both A and B are 33.33 minus ½ the N contour
  value.

12
Forecast Maps and Bulletins
Each month, near the 15th, CPC issues a set of 13
seasonal outlooks. These are formulated by two
forecasters. There are two maps for each of the
13 leads, one for temperature and one for
precipitation for a total of 26 maps. Each
outlook covers a 3-month season, and, for leads
2-13, overlaps the next and prior season by 2
months. The seasonal outlook is accompanied by 2
bulletins the prognostic discussion for the
seasonal outlook over North America, FXUS05, and,
for Hawaii, FXUS40. The lead seasonal forecaster
writes the first bulletin. The monthly
forecaster writes the second. The monthly
outlook is issued at the same time as the
seasonal outlook. Formulated by a single
forecaster, it consists of a temperature and
precipitation outlook for a single lead, 0.5
months. The monthly outlook is accompanied by a
single bulletin, FXUS07. All maps are sent to
AWIPS.
Sample seasonal outlook maps for the 0.5 month
lead (lead 1) forecast valid November-December-Jan
uary 2001-2.
13
Sample 90-day bulletin
PROGNOSTIC DISCUSSION FOR LONG-LEAD OUTLOOKS
CLIMATE PREDICTION CENTER NCEP NATIONAL WEATHER
SERVICE WASHINGTON DC 3 PM EDT THURSDAY OCTOBER
18 2001 BASIS AND SUMMARY OF THE CURRENT
LONG-LEAD OUTLOOK . . . . . . . . THESE OUTLOOKS
REFLECT THE EXPECTED U.S. CLIMATE ANOMALIES FOR
ENSO-NEUTRAL CONDITIONS ... THE LONG-TERM TREND
AS ESTIMATED BY OCN AND FORECASTS OF OCEANIC AND
ATMOSPHERIC CONDITIONS FROM DYNAMICAL MODELS.
EQUATORIAL PACIFIC SSTS ARE NOW SLIGHTLY ABOVE
NORMAL IN THE WEST AND CENTRAL PACIFIC AND BELOW
NORMAL NEAR SOUTH AMERICA. MOST STATISTICAL AND
DYNAMIC PREDICTIONS OF NINO 3.4 SSTS INDICATE
ANOMALIES VERY NEAR ZERO THROUGHOUT THE UPCOMING
COOL SEASON WITH SLIGHT WARM EPISODE (EL NINO)
CONDITIONS BY LATE IN THE SPRING OR EARLY IN THE
SUMMER OF 2002. WHILE IT IS LIKELY THAT
EQUATORIAL PACIFIC SSTS NEAR AND JUST WEST OF THE
DATELINE WILL BE ABOVE AVERAGE ... THE ANOMALIES
ARE EXPECTED TO BE WEAK AND NOT LIKELY TO PRODUCE
THE LARGE-SCALE ATMOSPHERIC EFFECTS ASSOCIATED
WITH STRONG EL NINO EVENTS THIS COOL SEASON. THE
ABSENCE OF SIGNIFICANT ENSO-RELATED INFLUENCES
SUGGESTS THE UPCOMING COOL SEASON MAY BE MARKED
BY SIGNIFICANT MONTH-TO-MONTH VARIABILITY...WHICH
MAY BE ENHANCED IF THE CURRENTLY STRONG SERIES OF
MADDEN AND JULIAN OSCILLATION (MJO) WAVES
CONTINUES OVER THE PACIFIC. CURRENT ATMOSPHERIC
AND OCEANIC MODELS ARE NEAR-NORMAL IN THE
TROPICAL PACIFIC IN SEPTEMBER. SST ANOMALIES
ALONG THE EQUATOR CURRENTLY ARE NEAR 1.0 DEG C
WEST OF THE DATELINE AND ARE AROUND -1.O DEG IN
THE EASTERN PACIFIC - EAST OF ABOUT 130W. ABOVE
NORMAL OCEAN TEMPERATURES EXTEND FROM THE SURFACE
OR NEAR THE SURFACE TO OVER 200 M DEPTH OVER MOST
OF THE CENTRAL PACIFIC OCEAN AND TO 100 M DEPTH
AS FAR EAST AS 130 W. ENHANCED MJO ACTIVITY HAS
BEEN NOTED DURING THE LAST SIX MONTHS. THIS IS
EXPECTED TO CONTINUE AND WILL HAVE AN IMPACT ON
SSTS IN THE CENTRAL EQUATORIAL PACIFIC. WITH WARM
SSTS FARTHER EAST THAN LAST WINTER THERE IS A
BETTER CHANCE STRONG MJO WILL MAKE IT TO THE
DATELINE AND HAVE AN OPPORTUNITY TO IMPACT
PRECIPITATION IN THE WESTERN UNITED STATES. . . .
. . . . . PROGNOSTIC DISCUSSION OF SST FORECASTS
NEAR-NEUTRAL ENSO CONDITIONS ARE PREDICTED FOR
MUCH OF THE WINTER. MOST DYNAMICAL MODELS (NCEP -
SCRIPPS) GRADUALLY INCREASE NINO 3.4 SSTS
BEGINNING IN THE SPRING. THE CONSTRUCTED ANALOG
AND INVERSE MODEL PREDICT NEUTRAL OR COOL NINO
3.4 SSTS FOR THIS COOL SEASON WHILE CCA PREDICTS
MODESTLY ABOVE NORMAL SSTS. A MARKOV SST MODEL
INDICATES PERSISTENCE OF CURRENT SSTS THROUGH THE
SPRING WITH WARM ENSO CONDITIONS DEVELOPING
DURING APRIL THROUGH JUNE 2002. A CONSOLIDATION
OF THREE NCEP TOOLS BASED ON THEIR PAST
PERFORMANCE INDICATES NEAR NORMAL CONDITIONS WILL
PREVAIL THROUGH THE WINTER WITH GRADUALLY
INCREASING ANOMALIES THAT REACH AROUND 0.5
DEGREES C BY LATE NEXT SPRING. PROGNOSTIC TOOLS
USED FOR U.S. TEMPERATURE AND PRECIPITATION
OUTLOOK THE STATISTICAL TOOLS ... CCA ... OCN AND
SMLR ... WERE CONSULTED AT ALL LEAD TIMES.
ENSO-NEUTRAL COMPOSITES WERE ALSO CONSULTED.
THESE WERE SUBJECTED TO A HIGH-PASS FILTER PRIOR
TO ISOLATE INTERANNUAL SIGNALS FROM LONGER TIME
SCALES. FINALLY DYNAMICAL MODEL FORECASTS FROM
NCEP - EC HAMBURG - NASA AND NCAR - INITIATED
FROM ATMOSPHERIC AND OCEANIC CONDITIONS FROM LATE
SEPTEMBER - WERE ALL CONSULTED. ALTHOUGH THE
INITIAL ENSO STATE IS NEAR ZERO - THERE WAS
AGREEMENT AMONG 3 OF THE 4 DYNAMICAL MODELS ON A
WAVE-TRAIN EMANATING FROM THE ABNORMALLY WARM
POOL OF WATER IN THE WESTERN TROPICAL PACIFIC
WHICH APPEARS TO CAUSE A PERSISTENT PATTERN OVER
NORTH AMERICA. THIS PATTERN FAVORS POSITIVE UPPER
AIR HEIGHT ANOMALIES OVER THE GULF OF ALASKA AND
NEGATIVE ANOMALIES OVER EASTERN NORTH AMERICA.
THE TEMPERATURE FORECASTS FROM THESE DYNAMICAL
MODELS ARE IN OVERALL AGREEMENT WITH MANY OF THE
STATISTICAL TOOLS - INCLUDING THE ENSO-NEUTRAL
COMPOSITES AND CCA - BOTH OF WHICH PREDICT BELOW
NORMAL TEMPERATURES IN THE NORTHEASTERN US AND
ABOVE NORMAL IN THE SOUTHWEST DURING THE COLD
SEASON. IN PARTS OF THE EAST WE GAVE GREATER
WEIGHT TO THESE TOOLS THAN TO THE TREND (OCN) -
LEADING TO FORECASTS OF CL IN THE SOUTHEAST AND
BELOW NORMAL IN THE NORTH-CENTRAL AND
NORTHEASTERN U.S. . . . . . . . . PROGNOSTIC
DISCUSSION OF OUTLOOKS - NDJ 2001-2 TO NDJ 2002-3
TEMPERATURE MOST FORECAST TOOLS INDICATE
WARMER-THAN-NORMAL TEMPERATURES FOR NDJ 2001-2
THROUGH NDJ 2002-3 IN THE SOUTHWEST U.S. THIS
SIGNAL IS THE STRONGEST TREND SIGNAL ON THE MAP
AND THE PROBABILITIES ARE ACCORDINGLY HIGH. MIXED
SIGNALS APPEAR OVER ALASKA AND CLIMATOLOGICAL
PROBABILITIES ARE OFTEN PREDICTED THERE. ONE
MAJOR FACTOR INFLUENCING THE SEASONAL CLIMATE
ANOMALIES OVER THE EASTERN U.S. THIS WINTER WILL
BE THE PHASE OF THE NAO- WHICH IS NOT YET
PREDICTABLE BEYOND ABOUT 1-2 WEEKS IN ADVANCE.
ENSO-NEUTRAL COMPOSITES INDICATE COOLER THAN
NORMAL TEMPERATURES OVER MUCH OF THE CENTRAL AND
EASTERN U.S. - WHICH OPPOSES RECENT WARM TRENDS.
BELOW-NORMAL TEMPERATURE WHICH CCA PREDICTED FOR
ALASKA IN NDJ AND DJF IN LAST MONTHS FORECAST
HAVE BEEN REMOVED - SINCE OCN AND CMP BOTH
PREDICT WARM. BEYOND FMA 2002 CCA AND OCN ARE THE
PRIMARY TOOLS. THESE SHOW THE WARM TREND IN THE
WESTERN AND SOUTHERN U.S. DURING THE WARM SEASON
AND A COOL TREND IN THE NORTH-CENTRAL U.S. DURING
THE LATE SUMMER. . . . . . . . . . PRECIPITATION
CCA AND OCN PREDICT ABOVE-MEDIAN PRECIPITATION
OVER TEXAS FOR NDJ THROUGH JFM AND SUB-NORMAL
RAINFALL OVER FLORIDA AND PARTS OF THE SOUTHEAST
FOR NDJ THROUGH DJF. WHILE MJO ACTIVITY IS
EXPECTED TO BE STRONG DURING THE COLD SEASON AND
SUCH ACTIVITY CAN BE ASSOCIATED WITH HEAVY
PRECIPITATION EVENTS ALONG THE WEST COAST - THE
TOOLS CURRENTLY GIVE MIXED INDICATIONS IN THAT
REGION AND CL IS PREDICTED THERE. ODDS FOR
ABOVE-MEDIAN PRECIPITATION INCREASE IN THE
NORTHWEST IN AMJ AND MJJ DUE TO LONG-TERM TRENDS.
BEYOND MJJ PREDICTABILITY OF PRECIPITATION IS
LIMITED TO SCATTERED SMALL REGIONS ASSOCIATED
WITH LONG-TERM TRENDS. THE BIGGEST CHANGES FROM
LAST MONTHS OUTLOOK ARE REMOVING ABOVE-MEDIAN
PRECIPITATION FROM THE NDJ THROUGH JFM FORECASTS
FOR THE PANHANDLE OF FLORIDA AND FORECASTING
SUB-MEDIAN PRECIPITATION FOR NDJ AND DJF FOR
FLORIDA AND PARTS OF THE SOUTHEAST. THIS SIGNAL
COMES FROM CCA AND THE DYNAMICAL MODELS - WHICH
INDICATE THAT THE SIGNAL FADES RAPIDLY AFTER DJF.
LAST MONTHS FORECAST OF BELOW-MEDIAN IN THE
SOUTHWEST FOR NDJ AND DJF HAS BEEN MODIFIED TO CL
IN LIGHT OF DYNAMICAL MODEL FORECASTS OF
ABOVE-MEDIAN RAINFALL. ABOVE MEDIAN PRECIPITATION
PREDICTED FOR ALASKA IN DJF THIS MONTH COMES FROM
THE DYNAMICAL MODEL FORECASTS. FORECASTERS ED
OLENIC AND HUUG VAN DEN DOOL
14
30-,90-day Forecast Tools

• Examples of the variety of tools used in
  preparing the 30-day 0.5 month
• lead 30-day outlook and the multi-lead 90-day
  outlooks may be found at
• http//www.cpc.ncep.noaa.gov/products/predictions/
  90day/tools/briefing
• These forecast tools consist of 3 types
• 1. Statistical CCA, OCN, MLR
• 2. Dynamical Coupled GCM, AGCM, ensembles
• 3. Physically-based composites.

15
Statistical tools Canonical Correlation
Analysis (CCA)

• CCA is used to find predictive relationships
  among variables. For long-range
• forecasting the variables used include tropical
  Pacific Ocean sea-surface
• temperatures (SSTs), 700 hPa heights, (the
  predictors) and U.S. surface
• temperatures (T) and precipitation (P) (the
  predictands) ( The SST fields are
• weighted twice a much as the 700 hPa heights in
  the CCA)
• When CCA is developed, relationships are found
  between observed U.S. T and
• P for a given season, say, January-February-March
  (JFM) and the predictors for
• the prior four non-overlapping seasons, in this
  case, OND, JAS, AMJ and JFM
• of the prior year. This is done for the entire
  available historical record of data,
• roughly 1950-95.
• To make a forecast, the relationships found in
  the developmental dataset are
• applied to the most recent observations of the
  predictors to produce T and P
• forecasts for the target season.

16
Statistical toolsOptimal Climate Normals (OCN)

• OCN, as it is used as a tool at CPC is, quite
  simply, a measure of the
• trend. For a given station and season, the OCN
  forecast is the difference
• between the seasonal mean temperature during the
  last 10 years and the 30
• year climatology. As the amount of time between
  the last year in the
• climatology and the current year increases the
  difference trend gets larger
• (aging climatology). This continues until a new
  climatology is
• implemented, at which time, the impact of the
  trend suddenly drops and
• begins anew its growth as the climatology ages.
• The Optimal in OCN refers to optimal number of
  years to use in the
• trend calculation to optimize skill when OCN is
  used to make forecasts.
• Very often this number, called k, is less than 30
  years. When it is, the tool
• has value in making forecasts.

17
Dynamical tools Operations Concept for
Ocean/Atmosphere Model

• NCEP currently uses dynamical coupled
  ocean-atmosphere models in combination with
  statistical models to produce seasonal outlooks
  with ½ to 5 ½ month leads and, to a lesser
  extent, monthly outlooks with ½ month lead.
  Enhanced model operations which include increased
  numbers of ensemble members, more frequent model
  runs and enhanced capability to include the
  influence of within-season variations in SST and
  OLR will be used to
• - Produce more highly resolved distributions of
  predicted
• variables,
• - Produce forecasts which increasingly and more
  appropriately
• reflect the influence of intra-seasonal
  variability on middle
• latitude climate,
• - Produce improved week 2 and monthly outlooks
  and develop
• and implement new outlook products for
  the week 3-4 period.
• - Develop and implement new products to predict
  seasonal
• variations in frequency of extreme events,
  primarily during
• ENSO.

18
Detailed operations concept for ocean-atmosphere
model

• Currently, coupled dynamical model forecasts are
  one of several tools
• used in preparing long-range outlooks. NCEPs
  model is run to produce
• one set of ensemble forecasts per month during
  the first week of the
• month. This is done in a two-tiered system, in
  which first, an ensemble of
• 16 ocean forecasts are created using a coupled
  GCM. The average of
• these is used as the official SST forecast.
• This SST forecast is then used as the lower
  boundary for an AGCM to
• create a set of 20 atmosphere ensemble members.
  The forecasts are run
• out to 9 months. A 20-year run of the AGCM is
  created each month. The
• seasonal means from this run are used as the
  climatology to create
• anomaly maps from each of the ensemble members.
  The means of these
• anomaly maps are used as the forecast tools which
  are presented to the
• forecasters.
• The forecasters use the NCEP model tools,
  together with other model
• tools to subjectively create outlook maps of the
  probability of monthly and
• seasonal mean temperature and total precipitation
  category.

19
NCEP Two-Tier Climate Modeling System
OCEAN INITIAL CONDITIONS
AGCM FORECASTS
SST
TOPEX
XBT
TAO
STATISTICAL TOOLS
INTEGRATED OCEAN MODEL-DATA ASSIMILATION SYSTEM
COUPLED OCEAN-ATMOSPHERE GCM
CDC
IRI
SURFACE T, P ANOMALIES
SSTA
STATISTICAL TOOLS CCA, CA
EVAP-PRECIP FLUX
HEAT FLUXES
STRESS
FORECASTERS
SSMI/ERS-2
OFFICIAL SST FCST
OFFICIAL PROBABILISTIC T,P OUTLOOKS
20
Forecast tools web page
21
Forecast tools page
Web-based tools
Web-based display of forecast tools gives
ready-access to a wide variety of tools. A
bi-weekly telephone conference call brings
together participants to assess the current
status of the climate system. Participants and
contributors include CPC, CDC, IRI, GFDL,
University of Washington, Scripps, FSU, COLA and
others
22
Sample Outlook Using Actual Tools
23
Atmospheric GCM (AGCM) T Forecasts
24
Atmospheric GCM (AGCM) NDJ 2000-01 T Verification
25
CCA 0.5 Mo lead NDJ T Outlook
26
OCN 0.5 Mo lead NDJ T Outlook
27
AGCM 0.5 Mo lead NDJ T Outlook
28
AO-ENSO-T composites
29
OFFICIAL NDJ 2001-02 T Outlook
30
AGCM P Forecasts
31
AGCM NDJ 2000-01 P Verification
32
CCA 0.5 Mo lead NDJ P Outlook
33
OCN 0.5 Mo lead NDJ P Outlook
34
AGCM 0.5 Mo lead NDJ P Outlook
35
AO-ENSO-P composites
36
ENSO-Neutral Composite
37
Official NDJ 2001-02 P Outlook
38
Temp Skill
39
Precipitation Skill
40
El Nino Global Impacts
Current dynamical models accurately forecast
gross ENSO properties. Future models must be
able to predict more subtle event-to-event
differences and underlying statistics of weather,
including extreme events.
41
Summary Seasonal/Monthly Outlooks

• CPC long-lead seasonal outlooks are produced
  using, in order of perceived reliability
• ENSO composites, trend, soil moisture,
  statistical models,
• dynamical models (DM).
• Frequent model changes hamper perceived
  reliability of DMs.
• Lack of information about model biases and
  calibration reduce perceived reliability of DM
  forecasts.
• Weather/climate impacts of AO, MJO, PDO are
  known.
• Prediction of AO, MJO, PDO is not yet possible.
• DMs cannot yet predict subtle differences among
  ENSOs.
• Stratospheric annular mode (SAM) is a link to AO
  prediction.
• DMs ability to predict SAM is unknown/doubtful.
• Relationship of Indian Ocean to predictable
  signals is unclear.
• Relationship of QBO, Southern Hemisphere
  mid-latitude circulations to predictable signals
  is unclear.

42
Six-to-Ten Day and Week 2 Outlooks
43
6-10 DAY/WEEK 2 OUTLOOK SCHEDULE/LEADS

• Each day, CPC prepares and disseminates outlooks
  for 6-10 days (lead time of 5 days) and week 2
  (days 8-14, lead time of 7 days).
• Each of these outlooks is accompanied by forecast
  maps of
• 1) 500 mb height and 500 mb height anomalies over
  the Northern Hemisphere, forecast maps of
• 2) surface temperature and
• 3) precipitation for the continental U.S. and
  Alaska and a single bulletin, FEUS40, giving a
  prognostic discussion for both forecasts and a
  table of state-by-state forecasts for each
  forecast.
• On Monday through Friday, including holidays, the
  outlooks are prepared by a
• forecaster who draws the maps, writes the
  bulletin and composes the table.
• On Saturday and Sunday only automated versions of
  the maps and the tables are
• disseminated.

44
6-10 day/week 2 Outlook Categories, Probabilities

• 6-10 day/week 2 outlooks are prepared for
  5-day/7-day
• average temperature and total accumulated
  precipitation
• category. Three categories are used (terciles).
  These are
• BELOW-,NEAR- and ABOVE-normal, for temperature,
  and
• BELOW-, NEAR- and ABOVE-median for precipitation.
  The
• contours on the maps depict the TOTAL probability
  of the
• occurrence of the indicated category. Contours
  of the
• climatological normals are also shown.

45
Recent Changes to Procedures

• From 3 times/week to daily in October 2000
• Automated weekend forecasts from October 2000
• Percent probability format from October 2000
• Alaska and week 2 added October 2000
• Automated weekend forecasts improved October
  2001neural net tool omitted and consistency with
  weekday forecasts added
• Bias-corrected precipitation forecast tool and
  other improvements added in the fall of 2001

46
6-10 day/week 2 process schematic
Downscale get surface weather from 500 mb height
via analogs, regression, neural network.
Forecaster formulates maps of predicted T, P, PMD
bulletin
Multi-model ensemble 900 AM
Weighted average of model 500 hPa height
R
R
Disseminate via web, AWIPS, FOS 3-4 PM
R
Forecaster reconciliation of tools required
47
Forecast Maps and Bulletins
Each day,between 3 and 4 PM Eastern Time, CPC
issues a set of 6-10 day and week 2 outlooks.
These are formulated by a forecaster (Monday
through Friday) and are automated on weekends.
There are two 500 mb height maps, two surface
maps and a single bulletin.
Sample 6-10 day outlook 500 mb height and anomaly
forecast map from CPC web page.
48
Sample 6-10 day average T outlook
49
Sample 6-10 day T outlook
Sample 6-10 day 5-day total precipitation outlook
from CPC web page.
50
Sample 6-10day/week 2 bulletin
PROGNOSTIC DISCUSSIONS FOR 6 TO 10 AND 8 TO 14
DAY OUTLOOKS CLIMATE PREDICTION CENTER NCEP
NATIONAL WEATHER SERVICE WASHINGTON DC 3 PM EDT
FRI OCT 26 2001 6-10 DAY OUTLOOK NOV 1 - 5 2001
. . . . . . . . MODEL COMPARISONS 6-10 DAY
MODELS - MRF - MRF ENSEMBLE MEAN ARE IN VERY GOOD
AGREEMENT TODAY ... WITH THE ECMWF SOMEWHAT
DIFFERENT AS IT BUILDS THE RIDGE OVER THE GREAT
LAKES AND SOUTH CENTRAL CANADA MUCH STRONGER AND
HAS A SLIGHTLY DIFFERENT SOLUTION ALONG THE WEST
COAST. WEATHER ACROSS MUCH OF THE CONUS WITH
EITHER SOLUTION WILL BE MAINLY MILD AND DRY FOR
EARLY NOVEMBER EVERYWHERE EXCEPT ALONG THE WEST
COAST ... WHERE NORMAL TEMPERATURES AND
BENEFICIAL RAINS ARE STILL EXPECTED. ALL MODELS
CONTINUE TO FORECAST A DEEP LOW OVER ALASKA ...
EXTENDING INTO WESTERN CANADA. THE BLENDED
PATTERN ... WHICH RELIES MORE STRONGLY ON THE MRF
ENSEMBLE MEAN HAS ABOVE NORMAL HEIGHTS COVERING
THE ENTIRE CONUS EXCEPT FOR THE PACIFIC NORTHWEST
AND FLORIDA. THIS PATTERN IS VERY WELL
TELECONNECTED ... AND THE COMPOSITE ANALOG
CORRELATES AT A VERY HIGH 0.94 WITH THE FORECAST
HEIGHT PATTERN. CONFIDENCE IN TODAYS FORECAST IS
ABOVE AVERAGE FOR THE SEASON... RATED 4 ON A
SCALE OF 1 TO 5. THE OFFICIAL 500-MB HEIGHT PROG
USES 60 PERCENT OF TODAYS MRF ENSEMBLE MEAN
CENTERED ON DAY 8... 15 PERCENT OF THE ECMWF
MODEL CENTERED ON DAY 7... AND 25 PERCENT OF
TODAYS MRF CENTERED ON DAY 8. MODEL OF THE DAY
ENSEMBLE MEAN THE TEMPERATURE OUTLOOK IS BASED
MAINLY ON THE KLEIN SPECIFICATIONS FROM THE
500-MB BLEND AND FORECAST TEMPERATURES BASED ON
THE 10 BEST HISTORICAL ANALOGS. THE PRECIPITATION
OUTLOOK IS BASED ON THE MRF MODEL OUTPUT
CORRECTED FOR RECENT BIASES ... SPECIFICATIONS
FROM THE NEURAL NET AND PRECIPITATION BASED ON
THE COMPOSITE OF THE 10 BEST HISTORICAL ANALOGS.
. . . . . . . . 8-14 DAY OUTLOOK FOR NOV 3 - 9
2001 THE MRF AND MRF-BASED ENSEMBLE MEAN ARE
AGAIN IN GOOD AGREEMENT TODAY WITH BOTH MODELS
SHOWING A BROAD FLAT RIDGE CENTERED IN THE MIDDLE
OF THE COUNTRY EXTENDING FROM COAST TO COAST ...
WITH TROUGHS LOCATED OFF THE WEST COAST AND IN
THE WESTERN ATLANTIC. MAIN DIFFERENCE BETWEEN MRF
AND ENSEMBLE MEAN CONCERNS THE WESTERN ATLANTIC
TROUGH ... WITH THE MRF RETROGRADING THE TROUGH
INTO THE NORTHEAST. THE BLENDED MAP MAINTAINS
ABOVE NORMAL HEIGHTS ACROSS MOST OF THE CONUS ...
EXCEPT FOR RIGHT ALONG THE EAST COAST ... THE
SOUTHEAST ... AND RIGHT ALONG THE PACIFIC
NORTHWEST COAST. THIS SHOULD RESULT IN CONTINUED
QUIET EARLY NOVEMBER WEATHER ACROSS THE US ...
WITH TEMPERATURES ABOVE NORMAL AND PRECIPITATION
BELOW NORMAL. A DEEP LOW IS MAINTAINED OVER
ALASKA WITH A TROUGH EXTENDING SOUTHWARD ALONG
140W ... WHICH SHOULD RESULT IN CONTINUED
BELOW-NORMAL TEMPERATURES OVER ALASKA.
CONFIDENCE IN TODAYS FORECAST IS ABOVE AVERAGE
FOR THE SEASON... RATED 4 ON A SCALE OF 1 TO 5.
THE OFFICIAL 500-MB HEIGHT PROG USES 75 PERCENT
OF TODAYS ENSEMBLE MEAN CENTERED ON DAY 11 AND 25
PERCENT OF TODAYS MRF MODEL CENTERED ON DAY 11.
THE TEMPERATURE PROG IS BASED ON KLEIN
SPECIFICATIONS FROM THE OFFICIAL HEIGHT PROG. THE
PRECIPITATION IS BASED ON THE SPECIFICATIONS FROM
THE NEURAL NET AND THE LIKELY STORM TRACKS
INDICATED BY THE 8-14 DAY MEAN 500-MB
CIRCULATION. . . . . . . . . FORECASTER M.
HALPERT
51
6-10 day/week 2 Forecast Tools

• Examples of the variety of tools used in
  preparing the 6-10
• Day and week 2 outlooks.
• These forecast tools consist of
• 1. Statistical Klein, Neural Network, analogs,
  teleconnections
• 2. Dynamical hrMRF, MRF ensembles, ECMWF, dAVA

52
A Simple Calibration Scheme J. Whitaker, CDC
Archive of 23 years of daily forecasts (1
forecast/day) out to week 2, October 1978-present
(MRF T62)
Operations Run a 15-member ensemble in
real-time using the same model as is used to
create the archive.
Ensemble mean for 6-10, 8-14 days
Calculate anomalies using the 23-year model
climatology (bias corrected)
Compute the climatology of spread for this model,
I.e. assume the mean week 2 spread is determined
by the error variance of the 23-years of
forecasts.
53
6-10-day/Week 2 Outlook Using Actual Tools
54
ECMWF, MRF Ensembles
The first step in making a 6-10 day forecast is
to formulate a forecast of upper-air height,
which is easier to forecast accurately than
surface parameters. The forecasters main tasks
are to reconcile differences between different
models, account for biases in individual models
and become satisfied the models are performing
properly. Teleconnections are useful to determine
whether a given predicted circulation pattern has
occurred often or infrequently in the past.
Infrequent patterns are less likely to persist
and are more difficult for models to predict
accurately. The opposite is true for persistent
patterns.
55
ECMWF upper-air height forecasts, analog
56
MRF Ensemble upper-air-height forecasts, analog
57
Official 6-10 day 500 hPa forecast
58
Teleconnections (TC)

• Definition Composite of those maps, for a
  calendar month, with largest (top 10) or
  (bottom 10) 500 hPa height at a specified space
  point from 1950-1999 (150 maps).
• Forecasters compute TC on the major anomaly
  centers (base points) of 500 hPa forecast maps.
  If there is a strong relationship between the
  base point centered at the largest anomaly, and
  distant points, the TC map will display large
  correlation values at the base point and at the
  distant centers. If there is no strong
  historical relationship, only the correlations at
  the base point will be large.
• Weak TC indicate the pattern is probably
  transient and not as likely to be well predicted
  by the model as would a persistent (strong TC)
  pattern.

59
Teleconnection on 500 hPa weighted mean anomaly
center at 10W/56N ()
60
Teleconnection on 500 hPa weighted mean anomaly
center at 88W/55N ()
61
Teleconnection on 500 hPa weighted mean anomaly
center at 140W/53N(-)
62
Composite of observed T, P anomalies associated
with teleconnection on 500 hPa anomalies at 56N
10W
63
Composite of observed T, P anomalies associated
with teleconnection on 500 hPa anomaly at 55N
88W
64
Composite of observed T, P anomalies associated
with teleconnection on - 500 hPa anomaly at 53N
140W
65
T prediction analog maps
66
MRF precipitation bias correction
http//www.cpc.ncep.noaa.gov/products/fcst_eval/pl
ots/precip.6-10day_error.us_mex.gif
67
Official 6-10 day T, P forecast
The final forecasts of temperature and
precipitation. The forecasters reasoning is
given in the bulletin (next slide).
68
6-10 day Prognostic Map Discussion
PROGNOSTIC DISCUSSIONS FOR 6 TO 10 AND 8 TO 14
DAY OUTLOOKS CLIMATE PREDICTION CENTER NCEP
NATIONAL WEATHER SERVICE WASHINGTON DC 3 PM EST
THU NOV 8 20016-10 DAY OUTLOOK NOV 14 - 18
2001 MODEL COMPARISONS 6 TO 10 DAY MRF...MRF
ENSEMBLE MEAN...AND ECMWF ARE IN FAIRLY GOOD
AGREEMENT TODAY. ALL THREE MODELS HAVE A MEAN
500 MB TROUGH OVER ALASKA EXTENDING SOUTH OVER
THE EASTERN PACIFIC... A RIDGE COVERING MOST OF
CENTRAL NORTH AMERICA...A TROUGH EXTENDING
SOUTHWARD FROM EAST OF LABRADOR AND A BLOCKING
RIDGE NEAR THE BRITISH ISLES. IT IS NOW THE
ECMWF WHICH GIVES THE MOST AMPLIFIED SOLUTION...
ESPECIALLY THE RIDGE OVER NORTH AMERICA...
PLACING A STRONG POSITIVE ANOMALY CENTER NEAR THE
SOUTH SHORE OF HUDSON BAY. THE MRF DOES HAVE A
SECOND BUT NOT STRONG POSITIVE ANOMALY CENTER
OVER NORTHERN CANADA... AND AS USUAL THE MRF
ENSEMBLE MEAN WAS SIMILAR TO THE MRF BUT HAD
WEAKER ANOMALIES. ALL MODELS ARE BASICALLY
DRY... EXCEPT FOR GIVING RATHER HEAVY
PRECIPITATION ALONG THE PACIFIC NORTHWEST COAST
DUE TO STRONG ONSHORE FLOW SOUTH OF A SERIES OF
INTENSE STORMS IN THE GULF OF ALASKA. ANOMALY
CORRELATIONS OF THE MODELS WITH THE CONSENSUS OF
THE 10 BEST ANALOGS WERE ONLY IN THE RANGE OF .82
TO .86 TODAY... AND SURPRISINGLY THE ANALOG
COMPOSITE FOR THE MRF SOLUTION WAS RATHER
DIFFERENT FROM THE ENSEMBLE MEAN ANALOG
COMPOSITE. THE OFFICIAL 500 MB PROG HAS RIDGES
WITH STRONG POSITIVE ANOMALIES CENTERED BETWEEN
THE GREAT LAKES AND HUDSON BAY... AND ALSO NEAR
THE BRITISH ISLES... FLANKED BY TROUGHS EXTENDING
SOUTHWARD FROM THE GULF OF ALASKA TO JUST NORTH
OF HAWAII... WITH A NEGATIVE ANOMALY CENTER NEAR
52N/142W AND ANOTHER TROUGH EXTENDING SOUTHWARD
FROM THE GREENLAND-DAVIS STRAIT AREA INTO THE
NORTH ATLANTIC NEAR 50W. WEAKLY BELOW NORMAL
HEIGHTS COVER MOST OF MEXICO AND THE GULF OF
MEXICO. CONFIDENCE IN TODAYS FORECAST IS NOW ONLY
AVERAGE FOR THE SEASON... RATED 3 ON A SCALE OF 1
TO 5 - FOR FAIRLY GOOD MODEL AGREEMENT AND ANALOG
CORRELATIONS BUT HINTS THAT THE PATTERN MAY BE
ABOUT TO CHANGE. THE OFFICIAL 500-MB HEIGHT PROG
USES 40 PERCENT OF TODAYS MRF ENSEMBLE MEAN
CENTERED ON DAY 8... 25 PERCENT OF TODAYS ECMWF
MODEL CENTERED ON DAY 7... 25 PERCENT OF TODAYS
MRF CENTERED ON DAY 8... AND 10 PERCENT OF
YESTERDAYS ECMWF CENTERED ON DAY 6. MODEL OF THE
DAY SLIGHT PREFERENCE FOR MRF ENSEMBLE MEAN. THE
TEMPERATURE OUTLOOK IS BASED ON A CONSENSUS OF
THE KLEIN... ANALOGS... AND NEURAL NETWORK
SPECIFICATIONS FROM THE OFFICIAL HEIGHT PROG...
WHICH WERE IN GOOD AGREEMENT IN MOST AREAS. THE
PRECIPITATION OUTLOOK IS BASED MAINLY ON THE MRF
ACCUMULATED PRECIPITATION ADJUSTED FOR RECENT
BIASES - ANALOGS TO THE PREDICTED UPPER AIR
HEIGHT PATTERN AND THE NEURAL NETWORK. 8-14 DAY
OUTLOOK FOR NOV 16 - 22 2001 THE MRF-BASED
ENSEMBLE MEAN CONTINUED TO SHOW RATHER WEAK
ANOMALIES IN MOST AREAS... SUGGESTING THAT THERE
IS A LOT OF SPREAD AND UNCERTAINTY IN THE
FORECASTS OF THE INDIVIDUAL MEMBERS. THE MRF
PROG SHOWS A SIMILAR PATTERN...BUT WITH STRONGER
NEGATIVE HEIGHT ANOMALIES OVER THE SOUTHEASTERN
U.S. CONSISTENT WITH THE APPEARANCE OF THE DAILY
PROGS BEYOND THE 6-10 DAY PERIOD. THE OFFICIAL
500 MB PROG MAINTAINS ABOVE NORMAL HEIGHTS ACROSS
THE NORTHERN PART OF THE COUNTRY WITH A TREND
TOWARDS LOWER HEIGHTS OVER THE SOUTH. THIS IS
CONSISTENT WITH A TREND THAT ALL THE MODELS HAD
TO A GREATER OR LESSER EXTENT OF JOINING THE
BRITISH ISLES BLOCK WITH THE NORTHEASTWARD
BUILDING RIDGE OVER CANADA... WHILE BEGINNING TO
SEAL OFF THE BOTTOM OF THE TROUGH SOUTHEAST OF
THE CANADIAN MARITIMES. IF THIS TREND IS CORRECT
AND CONTINUES... COLDER WEATHER WILL EVENTUALLY
DEVELOP OVER THE U.S. WITH A REVERSAL IN PHASE OF
THE NAO. THE DAILY MRF PROGS NEAR THE END OF
THE FORECAST PERIOD HAVE A DECIDEDLY COOLISH
LOOK... ESPECIALLY OVER THE SOUTHEASTERN U.S.
THE RIDGE OVER THE WESTERN U.S. ALSO BUILDS
SOMEWHAT... AND THE TROUGH OVER THE EASTERN
PACIFIC RETROGRADES WITH THE ASSOCIATED NEGATIVE
ANOMALY LOCATED NEAR 40N/156W. THIS WILL ALLOW
THE TREND TOWARD MILDER WEATHER OVER ALASKA TO
CONTINUE...EVENTUALLY ENDING THE LONG COLD PERIOD
OVER THE STATE. CONFIDENCE IN TODAYS FORECAST IS
SOMEWHAT BELOW AVERAGE FOR THE SEASON... RATED 2
ON A SCALE OF 1 TO 5... DUE TO HINTS OF CHANGES
IN THE PATTERN. THE OFFICIAL 500-MB HEIGHT PROG
USES 40 PERCENT OF TODAYS MRF ENSEMBLE MEAN
CENTERED ON DAY 11... 30 PERCENT OF TODAYS MRF
CENTERED ON DAY 11... 10 PERCENT OF YESTERDAYS
MRF CENTERED ON DAY 10... AND 20 PERCENT OF
TODAYS ECMWF 5 DAY MEAN CENTERED ON DAY 7. THE
TEMPERATURE PROG IS BASED ON KLEIN AND NEURAL
NETWORK SPECIFICATIONS FROM THE OFFICIAL HEIGHT
PROG WHICH WERE IN GOOD AGREEMENT. THE
PRECIPITATION PROG IS BASED ON THE NEURAL NETWORK
SPECIFICATIONS FROM THE OFFICIAL HEIGHT PROG WITH
CONSIDERATION OF THE MRF ACCUMULATED
PRECIPITATION. BOTH FORECASTS WERE SUBJECTIVELY
MODIFIED SOMEWHAT ON THE BASIS OF THE DAILY MRF
PROGS. FORECASTER A. J. WAGNER
69
6-10 day Monthly Average Skill Scores
70
Week 2 forecast tools from CDC
71
MRF Precipitation Bias Correction Week 2
72
Z500 d8 skill
73
Z500 d11 skill
74
T850 d8 skill
75
T850 d11 skill
76
U.S. Threats Assessment
77
U.S. Threats Assessment Schedule, Leads
The U.S. Threats Assessment (USTA) is intended to
provide advance warning of extreme weather events
to emergency managers, weather forecasters and
planners at all levels of government. The USTA is
issued each week on Tuesday. It covers days
3-14. A preliminary version of the product is
prepared and placed on a web site by noon on
Monday. A web site is available to NWS field
personnel to review the preliminary product and
then send comments to CPC suggesting
modifications. The product is updated often
during the week following its release. A single
forecaster prepares the product primarily using
output from dynamical models, including the MRF,
ECMWF, Canadian model, NOGAPS, etc Each
Wednesday, CPC hosts a telephone conference call
which is open to the public. The call gives
users the opportunity to discuss the product and
its implications.
78
Definitions of Threats
Listed at left are the nominal definitions of
threats. Because prior conditions play a role in
the impact of anomalies on the level of threat,
the definitions listed are only guidelines, to be
used along with knowledge of prior conditions, by
the forecaster in assigning threats. Research is
required to objectively determine thresholds for
threats at a large number of locations.
79
U.S. Threats Assessment Web Page
80
Threats text Summary of Forecasts and Threats
81
Threats Text Discussion
82
Briefing sequence
83
Seasonal U.S. Drought Outlooks

• Prepared monthly following release of the
  long-lead outlooks

84
Seasonal U.S. Drought Outlook
85
New medium range automated products

• CPC and MDL have collaborated to create new
  MOS-based
• products which are produced and disseminated
  automatically.
• Heat index outlooks for 6-10 and 8-14 days
• Wind-chill outlooks for 6-10 days
• Average temperature outlooks for 6-10 days

86
Excessive Heat Outlooks
87
Excessive Heat probabilities
88
Wind-chill Outlooks
89
Average T Outlook
90
The End