Kelvin K. Droegemeier School of Meteorology Center for Analysis and Prediction of Storms University...

Post on 28-Mar-2015

215 views 0 download

Tags:

transcript

Kelvin K. DroegemeierKelvin K. DroegemeierSchool of Meteorology School of Meteorology

Center for Analysis and Prediction of StormsCenter for Analysis and Prediction of StormsUniversity of OklahomaUniversity of Oklahoma

National Press Foundation Program “Understanding Violent National Press Foundation Program “Understanding Violent Weather”Weather”

26 October 200526 October 2005

Advances in the Advances in the Observation and Observation and

Computer Prediction of Computer Prediction of Severe StormsSevere Storms

                                                                      

                      

Copyright © 2003 WGN-TV

Everyone is Familiar With This Everyone is Familiar With This Person!!Person!!

                                                                      

                      

Copyright © 2003 WGN-TV

Computer ModelsComputer Models are the Primary Source are the Primary Source of Information for All Weather Forecastsof Information for All Weather Forecasts

Numerical Weather Numerical Weather PredictionPrediction

The use of computer models The use of computer models of the atmosphere toof the atmosphere to

predictpredict the weather given a the weather given a set of current observationsset of current observations

According to Webster…According to Webster…

pre·dictpre·dict: To state, tell about, or : To state, tell about, or make known in advance, especially make known in advance, especially on the basis of special knowledge.on the basis of special knowledge.

According to According to Meteorologists…Meteorologists…

pre·dictpre·dict: To state, tell about, or : To state, tell about, or make known in advance, trying not make known in advance, trying not to lie and always keeping the coin to lie and always keeping the coin concealed from curious onlookers.concealed from curious onlookers.

The Prediction ProcessThe Prediction Process

Analyze ResultsAnalyze Results

Com

pare

and

Ver

ify

Com

pare

and

Ver

ify

Observe the AtmosphereObserve the Atmosphere

Identify and ApplyIdentify and ApplyPhysical LawsPhysical Laws

Create a MathematicalCreate a MathematicalModelModel

Create and Run aCreate and Run aComputer ModelComputer Model

The Prediction ProcessThe Prediction Process

Analyze ResultsAnalyze Results

Com

pare

and

Ver

ify

Com

pare

and

Ver

ify

Observe the AtmosphereObserve the Atmosphere

Identify and ApplyIdentify and ApplyPhysical LawsPhysical Laws

Create a MathematicalCreate a MathematicalModelModel

Create and Run aCreate and Run aComputer ModelComputer Model

Observe the Atmosphere Observe the Atmosphere

Upper-AirUpper-AirBalloonsBalloons

SatellitesSatellites

NEXRAD NEXRAD Doppler Doppler

RadarRadar

Commercial AircraftCommercial Aircraft

AutomatedAutomatedSurface Surface

NetworksNetworks

The Prediction ProcessThe Prediction Process

Analyze ResultsAnalyze Results

Com

pare

and

Ver

ify

Com

pare

and

Ver

ify Identify and ApplyIdentify and Apply

Physical LawsPhysical Laws

Create a MathematicalCreate a MathematicalModelModel

Create and Run aCreate and Run aComputer ModelComputer Model

Observe the AtmosphereObserve the Atmosphere

Identify & Apply Physical Identify & Apply Physical LawsLaws

F=ma

The Prediction ProcessThe Prediction Process

Analyze ResultsAnalyze Results

Com

pare

and

Ver

ify

Com

pare

and

Ver

ify

Create a MathematicalCreate a MathematicalModelModel

Create and Run aCreate and Run aComputer ModelComputer Model

Observe the AtmosphereObserve the Atmosphere

Identify and ApplyIdentify and ApplyPhysical LawsPhysical Laws

Create a Mathematical ModelCreate a Mathematical Model

                                                                   

The Prediction ProcessThe Prediction Process

Analyze ResultsAnalyze Results

Com

pare

and

Ver

ify

Com

pare

and

Ver

ify

Create and Run aCreate and Run aComputer ModelComputer Model

Observe the AtmosphereObserve the Atmosphere

Identify and ApplyIdentify and ApplyPhysical LawsPhysical Laws

Create a MathematicalCreate a MathematicalModelModel

Create Computer Model Create Computer Model

Create Computer Model Create Computer Model

Solve highly nonlinear partial differential equations Solve highly nonlinear partial differential equations East/West WindEast/West Wind North/South WindNorth/South Wind Vertical WindVertical Wind TemperatureTemperature Water VaporWater Vapor Cloud WaterCloud Water Precipitating WaterPrecipitating Water Cloud IceCloud Ice GraupelGraupel HailHail Surface TemperatureSurface Temperature Surface MoistureSurface Moisture Soil TemperatureSoil Temperature Soil MoistureSoil Moisture Sub-Grid TurbulenceSub-Grid Turbulence

Run the Computer Model Run the Computer Model

Over the course of a single forecast, the Over the course of a single forecast, the computercomputermodel solves billions of model solves billions of equationsequations

Requires the fastest Requires the fastest supercomputers in the supercomputers in the world -- capable ofworld -- capable ofperforming trillions of performing trillions of calculationscalculationseach secondeach second

Run the Computer Model Run the Computer Model

Finer resolution allows the model to capture Finer resolution allows the model to capture more detailmore detail

Requires more computer power Requires more computer power – doubling the number of grid boxes in 3-D increases the doubling the number of grid boxes in 3-D increases the

computer requirements by a factor of 16! computer requirements by a factor of 16!

More Power!!! More Power!!!

The Prediction ProcessThe Prediction Process

Analyze ResultsAnalyze Results

Com

pare

and

Ver

ify

Com

pare

and

Ver

ify

Observe the AtmosphereObserve the Atmosphere

Identify and ApplyIdentify and ApplyPhysical LawsPhysical Laws

Create a MathematicalCreate a MathematicalModelModel

Create and Run aCreate and Run aComputer ModelComputer Model

Analyze the Analyze the Results/Compare/Verify Results/Compare/Verify

In the Beginning… ENIACIn the Beginning… ENIAC

ENIAC Versus TodayENIAC Versus Today

Weighed 30 tonsWeighed 30 tons Had 18,000 vacuum tubes, 1,500 Had 18,000 vacuum tubes, 1,500

relays thousands of resistors, relays thousands of resistors, capacitors, inductorscapacitors, inductors

Peak speed of 5000 adds/second Peak speed of 5000 adds/second and 300 multiplies/secand 300 multiplies/sec

A 1.2 GHz Pentium IV processor is A 1.2 GHz Pentium IV processor is 500,000 times faster than the 500,000 times faster than the ENIACENIAC

A desktop PC with 1 Gbyte of RAM A desktop PC with 1 Gbyte of RAM can store 5 million times as much can store 5 million times as much data as the ENIAC data as the ENIAC

1950: The First Computer Weather 1950: The First Computer Weather Forecast ModelForecast Model

450 Miles

Today’s ModelsToday’s Models

A Typical Forecast From Today’s A Typical Forecast From Today’s ModelsModels

What Causes the Major What Causes the Major Problems?Problems?

Why the Lack of Detail in the Why the Lack of Detail in the Model? Model?

This ThunderstormFalls Through the Cracks

Why the Lack of Detail in the Why the Lack of Detail in the Model? Model?

A Foundational QuestionA Foundational Question

Can computer forecastCan computer forecasttechnology. . .technology. . .

. . . explicitly predict this. . . explicitly predict thistype of weather?type of weather?

Would This Capability Be Would This Capability Be Useful?Useful?

Intense local weather causes economic losses in the Intense local weather causes economic losses in the US that average US that average $300 M$300 M per weekper week

Over 30% of the $10 trillion US economy is impacted Over 30% of the $10 trillion US economy is impacted each yeareach year

Commercial aviation loses Commercial aviation loses $1-2 B per year$1-2 B per year due to due to diversions, delays, and cancellations (one diverted diversions, delays, and cancellations (one diverted flight costs $150K)flight costs $150K)

Agriculture losses exceed Agriculture losses exceed $10 B/year$10 B/year

Dutton (2002)Dutton (2002)

Dutton (2002)Dutton (2002)

Actual Losses – Extreme EventsActual Losses – Extreme Events

Pielke and Carbone (2002)Pielke and Carbone (2002)

Cargo shippingCargo shipping– Most expeditious route can save $40,000 per voyage – thousands of ships travel Most expeditious route can save $40,000 per voyage – thousands of ships travel

continuously!continuously!– ExamplesExamples

High temperature and humidity can cause grain to germinate in cargo holdsHigh temperature and humidity can cause grain to germinate in cargo holds Ships affected differently by wavelength of ocean swellsShips affected differently by wavelength of ocean swells

Commercial aviation Commercial aviation – Single diversion averages $10,000 per domestic flightSingle diversion averages $10,000 per domestic flight– Not unusual for one carrier to have 70 diversions at a hub for a single weather event (1-2 Not unusual for one carrier to have 70 diversions at a hub for a single weather event (1-2

hours)hours)– Cost is $700,000 per eventCost is $700,000 per event– Industry loses $1-2 B per year due to weatherIndustry loses $1-2 B per year due to weather

Specific ExamplesSpecific Examples

Source: Weathernews, Inc.Source: Weathernews, Inc.

About About 50%50% of the loss is deemed preventable with of the loss is deemed preventable with better forecasts!better forecasts!

876876 deaths annually due to severe deaths annually due to severe weatherweather

7000+7000+ weather-related traffic fatalities weather-related traffic fatalities 450,000450,000 weather-related traffic injuries weather-related traffic injuries

A Great Toll in Human A Great Toll in Human LifeLife

Model Model TypesTypesGlobal

(2 weeks)

Model Model TypesTypesGlobal

(2 weeks)

Continental (few days)

Model Model TypesTypesGlobal

(2 weeks)

Continental (few days)

Special

Model Model TypesTypesGlobal

(2 weeks)

Continental (few days)

Special

Operational

Model Model TypesTypesGlobal

(2 weeks)

Continental (few days)

Regional (day)

Local (few hours)

Special

Experimental

Incr

easi

ng

Sk

ill

Trends in Large-Scale Forecast SkillTrends in Large-Scale Forecast Skill

10 km 1 km

Crossing the DivideCrossing the Divide

The next quantum leap in NWP The next quantum leap in NWP

will come when we start will come when we start resolving explicitly the most resolving explicitly the most energetic weather features, energetic weather features, e.g., individual convective e.g., individual convective storms in 3-Dstorms in 3-D

For global models, the For global models, the predictability increases for all predictability increases for all resolvable scales as the resolvable scales as the spatial resolution increases spatial resolution increases – The improvement is The improvement is boundedbounded– Going finer than a few 10s of km Going finer than a few 10s of km

in grid spacing gives little payoffin grid spacing gives little payoff

60 km 30 km

30 km 10 km

Importance of Finer Grid Spacing in Models

Courtesy NCAR

512 km

256 km

128 km

64 km

32 km

16 km

8 km

Each improvement requires 10X computer resources, total increase of 10,000,000!

4 km

Where Are We Today?Where Are We Today? Tremendous advances are being made in the computer-Tremendous advances are being made in the computer-

based prediction of high-impact local weather, such as based prediction of high-impact local weather, such as thunderstormsthunderstorms, owing to, owing to– Increases in computer power and networking capacityIncreases in computer power and networking capacity– Affordability of computersAffordability of computers– Availability of fine-scale observations (NEXRAD Doppler radar)Availability of fine-scale observations (NEXRAD Doppler radar)– Improved understanding of the atmosphereImproved understanding of the atmosphere– Societal need, especially that of weather impacted industries Societal need, especially that of weather impacted industries

(aviation, energy, recreation, defense)(aviation, energy, recreation, defense)

Example : March 28, 2000 Fort Example : March 28, 2000 Fort Worth Tornadic StormsWorth Tornadic Storms

Tornado

NWS 12-hr Computer Forecast Valid at 6 pm NWS 12-hr Computer Forecast Valid at 6 pm CDTCDT

No No Explicit EvidenceExplicit Evidence of Precipitation in North of Precipitation in North TexasTexas

6 pm 7 pm 8 pmR

adar

Hourly Radar Observations(Fort Worth Shown by the Pink Star)

6 pm 7 pm 8 pmR

adar

Com

pu

ter

For

ecas

t

2 hr 3 hr 4 hr

As a Forecaster As a Forecaster Worried About Worried About This Reality… This Reality…

7 pm

As a Forecaster As a Forecaster Worried About Worried About This Reality… This Reality…

How Much How Much Trust Would Trust Would You Place in You Place in This Model This Model Forecast? Forecast?

3 hr

7 pm

UncertaintyUncertainty We never know the complete state of We never know the complete state of

the atmosphere everywhere, with the atmosphere everywhere, with perfect accuracyperfect accuracy

Small observation errors can grow Small observation errors can grow with time in a forecast (chaos)with time in a forecast (chaos)

Rather than run a Rather than run a singlesingle forecast from forecast from one estimate of the current one estimate of the current conditions, we run conditions, we run severalseveral based upon based upon equally plausible initial conditions to equally plausible initial conditions to account for observational uncertaintyaccount for observational uncertainty

This is “ensemble forecasting”This is “ensemble forecasting”

Actual RadarActual Radar

Forecast #1Forecast #1 Forecast #2Forecast #2

Forecast #3Forecast #3 Forecast #5Forecast #5Forecast #4Forecast #4

Actual RadarActual Radar

Probability of Intense PrecipitationProbability of Intense Precipitation

Model Forecast Radar Observations

MUCH MORE Computing Power is MUCH MORE Computing Power is Required!!Required!!

Forecast #1Forecast #1

Forecast #50Forecast #50

Forecast #2Forecast #2

Forecast #3Forecast #3

Recent Real Time Experimental Forecasts Recent Real Time Experimental Forecasts Run by OU for the National Weather Run by OU for the National Weather

ServiceService

Actual Radar Observations

Recent Real Time Experimental Forecasts Recent Real Time Experimental Forecasts Run by OU for the National Weather Run by OU for the National Weather

ServiceService

24 Hour Forecast Actual Radar Observations

A Foundational QuestionA Foundational Question

Can computer forecastCan computer forecasttechnology. . .technology. . .

. . . explicitly predict this. . . explicitly predict thistype of weather?type of weather?

The Answer Appears to be “Yes,” But NewMethodologies May Be Needed…

Predicting Storms and Anticipating Predicting Storms and Anticipating Tornadoes Requires Fine-Scale Tornadoes Requires Fine-Scale

ObservationsObservations

NEXRAD Doppler Radar Network

Current Operational Radar System in Current Operational Radar System in USUS

NEXRAD Doppler Radar Network

#2. Earth’s curvature prevents 72% of #2. Earth’s curvature prevents 72% of the atmosphere below 1 km from being the atmosphere below 1 km from being

observedobserved

#1. Operates largely independent#1. Operates largely independentof the prevailing weather conditionsof the prevailing weather conditions

#3. Operates entirely independent from#3. Operates entirely independent fromthe models and algorithms that use its the models and algorithms that use its

datadata

The Limitations of NEXRADThe Limitations of NEXRAD

Source: NWS Office of Science and Technology

NEX

RA

DN

EX

RA

D

The ConsequenceThe Consequence

New NSF Engineering New NSF Engineering Research Center for Research Center for

Adaptive Sensing of the Adaptive Sensing of the Atmosphere (CASA)Atmosphere (CASA)

UMass/Amherst, OU, CSU, UPRMUMass/Amherst, OU, CSU, UPRM Concept: inexpensive, phased array Doppler radars on Concept: inexpensive, phased array Doppler radars on

cell towers and buildingscell towers and buildings Dynamically adaptive dynamicDynamically adaptive dynamic sensing of multiple sensing of multiple

targets while simultaneously meeting multiple end-user targets while simultaneously meeting multiple end-user needsneeds

Oklahoma Test Bed: Spring 2006Oklahoma Test Bed: Spring 2006

The $1M Question: Will Numerical Models Ever Be The $1M Question: Will Numerical Models Ever Be Able to Able to PredictPredict Tornadoes? Tornadoes?

Schematic Diagram of a Supercell Storm (C. Doswell)

The FutureThe Future The National Weather Service will begin running models The National Weather Service will begin running models

to to explicitly predictexplicitly predict thunderstorms thunderstorms Private companies will Private companies will

play a major role in play a major role in providing customized providing customized numerical forecasts for numerical forecasts for weather-sensitive weather-sensitive industries, especially industries, especially energy and aviationenergy and aviation

The FutureThe Future Human forecasters will continue to Human forecasters will continue to

be essential, though with changing be essential, though with changing rolesroles

Contact InformationContact Information

Kelvin K. DroegemeierUniversity of Oklahoma

Sarkeys Energy Center, Suite 1110100 East Boyd StreetNorman, OK 73019Email: kkd@ou.eduPhone: 405-325-0453Fax: 405-325-7614Mobile: 405-413-7847