Title: Extreme Weather Events and their Probabilistic Prediction By NCEP Ensemble Forecast System
1 Extreme Weather Events and their Probabilistic
PredictionBy NCEP Ensemble Forecast System
- Yuejian Zhu
- Environmental Modeling Center
- National Centers for Environmental Prediction
- NWS/NOAA/USA
- Contribution to 2005 Annual Meeting of CMS
- September 15-17, Zhengzhou, China
2Dedicated to International Symposium of
Rainstorms, Inundation and Disaster
Mitigation And To commemorate the 30-year
anniversary of the August 1975 flooding
rainfall and inundation that occurred in Henan
Province
3Significant modern floods around the world-
Records from encyclopedia
- Flooding in Mumbai India in July 2005 left over
700 dead. - The 2002 European flood was a flooding disaster
that affected many states including Czech
Republic, Germany and Poland. Historical cities
like Prague and Dresden were partly flooded. In
Germany the so called "Jahrhundertflut" (flood of
century) caused a 22,6 billion Euro damage. -
- In 1975 a freak typhoon destroyed over sixty dams
in China's Henan Province, killing over 200,000
people. (see Banqiao Dam) -
- The 1931 Huang He flood caused between 800,000
and 4,000,000 deaths in China, one of a series of
disastrous floods on the Huang He.
4758 flood and Banqiao Reservoir Dam
The Banqiao Reservoir Dam (Chinese ??????
Pinyin Banqiáo Shuikù Dàbà) and Shimantan
Reservoir Dam (Chinese ??????? Pinyin
Shímàntan Shuikù Dàbà) are among 62 dams in
Zhumadian Prefecture of China's Henan Province
that failed catastrophically in 1975 during a
freak typhoon. Approximately 26,000 people died
from flooding and another 145,000 died during
subsequent epidemics. In addition, about
5,960,000 buildings collapsed.
5NCEP/NCAR Reanalysis at 2.5 degree resolution
1975/08/04
6NCEP/NCAR Reanalysis at 2.5 degree resolution
1975/08/05
7NCEP/NCAR Reanalysis at 2.5 degree resolution
1975/08/06
8NCEP/NCAR Reanalysis at 2.5 degree resolution
1975/08/07
9NCEP/NCAR Reanalysis at 2.5 degree resolution
1975/08/08
10CONTENTS
- Introduction.
- Definition of Extreme Events.
- Nature of Extreme Events.
- Forecasting Extremes.
- Verifying Forecasts for Extremes.
- Use of Extreme Forecasts.
- Summary and Discussion.
- Deterministic or probabilistic?
- High predictable system.
- Economic values and uncertainties.
- Numerical production!!!
- Acknowledgments.
11Introduction
- 1. Extreme Weather Events
- . Unusual, Unexpected, rare weather
events. - . Cost loss of lives, properties,
equipments and etc. - . Forecast ( may be difficulty or may be
not ) ? - . Alarms to users ( Such as Watch, Warning
and etc... ) - . Early decision, and early protection!!!
- . Widely social impacts.
- . Always use updated forecast information
- 2. Probabilistic Forecasts
- . Ensemble model based forecasts.
- . Forecast In terms of probability or
possibility. - . Wide coverage of the weather events from
probabilistic - sense. Include extreme weather events.
- . Many variables ( temperature,
precipitation, wind and etc. ).
12Definition of Extreme Events
- 1. Climatological extremes
- . Based on climatological distributions.
- . The tails ( 5 or less ) of the
climatological distribution. - . Considering a particular meteorological
variable. - . Considering a specific time and place.
- 2. Forecast extremes
- . Similar to climatological extremes.
- . Different range and values of
distribution. - . Narrow band than climatology.
- . Conditional climatological sense.
- 3. User specific extremes
- . User defined extreme (not climatology, not
forecasting ). - . For particular user, in particular area
and in time period - . Sensitivity to particular meteorological
element. - . The combination of the temporal/spatial.
13Nature of Extreme Events
- 1. Physical system.
- . The same for extreme and non-extreme
events. - . Different from phase space of system.
- . Near the edge of the distribution.
- . Small scale system in generally.
- 2. Nonlinear process.
- . Play a crucial role to define the "edge".
- . Creating additional uncertainty.
- . Model's limitation to predict extreme by
nonlinear process. - 3. Combination of many factors
- . Snow covers, cloud covers.
- . Minimum temperature, and maximum
temperature. - . Combined high temperature and high
humidity heat index. - . Wind speed, combined cold temperature and
wind sheer. - . Precipitation amount and concentration.
- . Time, location and etc...
14Forecasting Extremes
- 1. Procedures
- . No specific tools or procedures in
generally. - . The same process for non-extreme.
- . Improving the model forecast accuracy.
- . Increase model's predictability to
extreme. - . Experiences of forecasters ( sometime ???
) - 2. Methods
- . Probabilistic forecast, such as PQPF.
- . In sense of probability by ensemble or
single forecast. - . Particular process for specific extreme
events (possible). - . Multi-methods ( relative measure of
predictability - RMOP). - ensemble spread
15Verifying Forecasts for Extremes
- 1. Systematic forecast behavior
- . Systematic model errors ( model biases,
need remove ). - . Model forecast is more likely to give
less extreme. - or moderate extreme.
- . Good example for precipitation forecast.
- Over-forecast for less, under-forecast
for extreme amount - 2. Control .vs. ensemble forecasts
- . Single/deterministic forecast ( more
difficulty to predict extreme ) - . Ensemble mean/medium forecast.
- is better for climatological extreme
events. - . Ensemble based probabilistic forecast
- is a best approach to extreme.
16Verifying Forecasts for Extremes (Continues)
- 3. RMS error
- . In general sense, large RMS error for
extreme - events, but not necessary
- . Persistence circulation forecast
(Zolton,1992). - . Surface temp. forecast (Ziehmann,2000)
- . Need more case to verify
- 4. Error in categorical forecasts
- . Climate forecast, normal forecast and
extreme forecast. - . Brier scores, Brier skill scores for
climate and ensemble. - 5. Error in probabilistic forecasts
- . Hitting rate, false alarm.
- . ROC, information content, economic values.
- . Reliability, rank probability skill scores
(RPSS). - . Brier scores for ensemble precipitation
forecast. - . Extreme forecast has a better scores.
17Use of Extreme Forecasts
- 1. Probabilistic forecast
- . Numerical model based probabilistic
forecast. - such as ensemble forecast.
- . Statistical model based probabilistic
forecast. - . Convert single forecast to probabilistic
forecast based on - climatological information.
- . Reliability, calibrated reliability (by
remove bias). - 2. Forecast skills
- . No long term skill to refer to extreme
events. - . Leading time dependence.
- . Seasonal dependence.
- . Optimize user behavior by analysis
- . Considering the ratio of cost/loss and
tolerance.
18Summary and Discussion
- 1. Observed extreme weather events
- . Mostly large impact to society, natural
and etc.. - 2. Model forecast ability
- . Limitation to predict extreme events.
- . Properly interpret probabilistic forecast.
- . RMS error is poorer than near normal
forecast - . Probabilistic/categorical evaluation is
better for extreme. - 3. Suggestions
- . Enhance the model predictability for
nonlinear process. - . Using probabilistic approach.
- . Reducing model's systematic error/bias.
- . Establishing/improving evaluation system.
19Deterministic or Probabilistic ?
Any prediction for the future is uncertainty Is
for-cast! No deterministic answer! Must be
probabilistic!
20Clear sky, no precipitation
2120mm/24hrs (0)
Precipitation
2mm/24hrs (30)
22(No Transcript)
23Current Hurricane Forecast
24High predictable heavy precipitation event
GFS ENS
February 12-13 1997 (Southern Louisiana flooding)
Location and intensity
GFS made a very good forecast, But Ensemble made
a excellent forecast.
25HIGHLY PREDICTABLE HEAVY PRECIPITATION EVENT
(20010313)
26High Predictable Heavy Precipitation Events
(20010113)
Ensemble-based precipitation forecasts gave
relatively high probability values for the half
and one inch thresholds for the 24-hr period
ending 031312 for the Gulf states with 1 through
8 days lead time. The corresponding observed
precipitation amounts indicate that the forecasts
were rather successful. The high predictability
in precipitation was associated with high
confidence (and well verifying) forecasts for 500
hPa height. The cut-off low over the SW US that
allowed Pacific air to reach the Gulf of Maxico
at low latitudes (over and south of Baja CA) was
well predicted, with high confidence, at various
lead times (see, for example, at 4, 7, and even
at 10 days). Red colors in these charts over the
cut-off low correspond to an area associated with
high predictability.
27RMOP
28DETERMINSTIC/PROBABILISTIC FORECASTQPF .vs. PQPF
- Northern California State Christmas-New Year
flooding - Winter storm last more than 10 days
- Total precipitation amount exceeding 660mm over
the huge area - The homes of 100,000 residents who has been
evacuated. - Some stranded residents has to be rescued by
helicopter. - Caused a lot of damages include road, bridge and
resident houses.
Photo from Washington Post
2924 hours observation
GFS ENS
30(No Transcript)
31Probabilistic Evaluation (cost-loss analysis)
- Based on hit rate (HR) and false alarm (FA)
analysis - .. Economic Value (EV) of forecasts
Ensemble forecast
Average 2-day advantage
Deterministic forecast
32Example of cost-loss analysis (economic values)
- Wind sheer damages the airplane
- Un-protect airplane (loss)
- 2-million dollars for each airplane
- Protect airplane (cost)
- 20,000 dollars for each airplane
- For decision makers !!!
- 1100 cost-loss ratio for this case
- Probabilistic forecast and forecast reliability
- Typhoon Mai-sha affected Beijing City
- Un-protect (may loss)
- Flooding, traffic and others.
- Protect (definitely cost)
- Activities will be cancelled
- Labors cost
- Others
- Scientific decision ???
- Anyone counts this ratio?
33Probabilistic Evaluation (useful tools)
- ... Small and large uncertainty.
- 1 day (large uncertainty) 4 days (control)
10-13 days (small uncertainty)
34Ensemble based Probabilistic Products
- Spaghatti diagram for uncertainty
- Standard/normalized spread, mean
- Relative measure of predictability (RMOP)
- Probabilistic quantitative precipitation forecast
(PQPF) - Probabilistic precipitation types
- Calibrated PQPF
- Hurricane tracks/strike probability
- Anomaly forecasts
- User specified
- Calibration !!!
35Ensemble Forecast for Uncertainty (1)
By Bill Bua
36Ensemble Forecast for Uncertainty (2)
By Bill Bua
37Ensemble Forecast for Uncertainty (3)
By Bill Bua
38Ensemble Forecast for Uncertainty (4)
39 1. By using equal climatological bins
(e.g. 10 bins, each grid points)2. Counts of
ensemble members agree with ensemble mean, (same
bin)3. Construct n1 probabilities for n
ensemble members from (2).3. Regional (NH,
weighted) Normalized Accumulated Probabilities
(n1)4. Calculate RMOP based on (3), but 30-d
decaying average.5. Verification information
(blue numbers) historical average (reliability)
40Ensemble mean
10 Climatological equally likely bins
Example of 1 grid point
10 ensemble forecasts
The value of ensemble members agree to ensemble
mean is 4/10 or 40 (probability) There are 10512
points ( values ) for global at 2.5 2.5 degree
resolution
10 ensemble members could construct 11
probabilities categories, such as 0/10 (0),
1/10(10), 2/10(20), 3/10(30), 4/10(40),
5/10(50), 6/10(60), 7/10(70), 8/10(80),
9/10(90), 10/10(100) Sum of each grid point for
above 11 probabilistic categories by area
weighted and normalized for global or specified
region Get 0/10 1/10 2/10 3/10
4/10 5/10 6/10 7/10 8/10 9/10
10/10 .029 .047 .077
.085 .100 .135 .116 .089 .081
.070 .177 sum of these 1.0
(1.007 here) 2.9 7.6 15.3 23.8
33.8 47.3 58.9 67.8 75.9 82.9 100
accumulated values There is 30-day decaying
average of above values ( last line ) in the
data-base and updated everyday. Assume these are
30-day decaying average values In this case,
point value is 4/10, RMOP value of this point is
33.8
41China
42General public for past 8 years
43Specific request
44(No Transcript)
45(No Transcript)
46(No Transcript)
47Ensemble Based Hurricane Track Plots
Karl (09/18)
Frances (08/28)
48Example of probabilistic forecast in terms of
climatology
49ENSEMBLE 10-, 50- (MEDIAN) 90-PERCENTILE
FORECAST VALUES (BLACK CONTOURS) AND
CORRESPONDING CLIMATE PERCENTILES (SHADES OF
COLOR)
50 The pre-NWP forecast accuracy
- A schematic illustration of the increase of
RMSE with forecast time. The pre-NWP forecaster
started from a persistence forecast which he
skillfully extrapolated into the future,
converging towards climate for longer ranges
A?2
persistence
A
meteorologist
- The time unit can be anything from hours to
days depending on the parameter (hours for
clouds, days for temperature)
51 NWP more accurate - but also less
persistence
A?2
- A good NWP model is able to simulate all
atmospheric scales throughout the forecast. It
has the same variance as the observations and the
persistence forecasts, which yields an error
saturation level 41 above the climate
worlds best NWP
A
meteorologist
52The art of good forecasting
- The way out of the dilemma
- Combine the high accuracy of NWP in the
short range with a filtering of the
non-predictable scales for longer ranges -
- This can be done both with and without the EPS
A?2
persistence
worlds best NWP
A
meteorologist
modified NWP forecast
53Acknowledgments
- I benefited from discussions with Drs.
Zoltan Toth, Richard Wobus and Hua-Lu Pan of
EMC/NCEP/NOAA - We acknowledge the support and
encouragement of Dr. Stephen Lord, Director of
EMC/NCEP/NOAA.