use of radar and other data for forecasting april 07/tuesday...use of radar and other data for...

56
Use of Radar and Other Data for Forecasting Rita Roberts James Wilson Brant Foote National Center for Atmospheric Research Sahael Conference, 3 April 2007 Improving Lives by Understanding Weather 0-6 hr NOWcasting

Upload: trinhdiep

Post on 24-May-2018

214 views

Category:

Documents


1 download

TRANSCRIPT

Use of Radar and Other Data for Forecasting

Rita RobertsJames WilsonBrant Foote

National Center for Atmospheric Research

Sahael Conference, 3 April 2007Improving Lives by Understanding Weather

0-6 hr NOWcasting

How to Make the Most of Radar for 0-6 hr Nowcasting

Women in Burkina Faso prepare an earthen bund to slow rain run-off and prevent erosion of the topsoil.

•For Water Management and Agriculture

•For Prediction of Thunderstorms

•For Warning of Flash Flooding and Severe Weather

•For Avoidance of Weather-Related Aviation HazardsAnd Terminal and Enroute Aviation Planning

B. Lamptey TalkTuesday at 12:00

• Weather is a contributing or causal factor in– 90% of General Aviation accidents– 25% of commercial aircraft accidents

• Causes 75% of the airline delays• Need for Decision Support

Systems

Aviation Weather

Dallas/Ft Worth

Air Traffic Flow In and Out of New York City

OutlineThunderstorm -

• Predictability

• Basic understanding of evolution

• Nowcasting System

OutlineThunderstorm -

• Predictability

• Basic understanding of evolution

• Nowcasting System

Single cell storms live < 30 min

Thunderstorm LifetimeSingle cell storms live < 30 min Multi-cell storm systems live

> 30 min

(Henry 1993; Battan 1953; Foote and Mohr 1979)

Conv

ectiv

e St

orm

Sys

tem

Sing

le-ce

ll Thu

nder

storm

TIME (hr)

SIZ

E

1 2 3 4 5

There are frequent and rapid changes in storm size and intensity.

Example Evolution of a Single Cell and a Convective Systems

Nowcasting by extrapolation

East

Nor

th

Echo at Time-1

Time-2

Time-3

Time-4Nowcast forTime-5

TITANTITAN

Predictability

Forecast Length (hr)

Extrapolation

NWP

Fore

cast

Ski

llInitiation

Growth

Decay

2 40

1

6 8

Radar Reflectivity

5 hr elapsedtime

Predictability using Radar

OutlineThunderstorm -

• Predictability

• Basic understanding of evolution

• Nowcasting System

Basic Understanding Of Evolution Using Radar

Factors important in determining storm initiation

• Convergence lines (boundaries)

• Orography and Terrain Features

• Boundary relative cell motion

• Boundary collision

•Climatology – Preferred regions

Basic Understanding Of Evolution Using Radar

Factors important in determining storm initiation

• Convergence lines

• Orography and Terrain Features

Boundary relative cell motion

• Boundary collision

•Climatology – Preferred regions

Boundary Influences on Thunderstorm Evolution

1. Satellite cloud imagery.Note the N-S line ofcumulus associated witha sea breeze along theFlorida east coast.

2. Clear-air radar features.Note enhanced N-S lineof reflectivity associated with a boundary. Redarrows are wind directionfrom surface stations.

Boundary layer convergence lines (boundaries) frequently influence the evolution of thunderstorms. These boundaries can often be observed in:

15:41

16:48

Reflectivity

Reflectivity

Doppler Radial Velocity

Reflectivity

15:41

17:21

Basic Understanding of Evolution Using Radar

Factors important in determining storm initiation

Convergence lines

• Boundary collision

• Boundary relative cell motion

• Orography and Terrain Features

• Climatology – Preferred regions

Radar ReflectivityTime Lapse= 5 hr

120

km

Basic Understanding of Evolution Using Radar

Basic Understanding of Evolution Using Radar

AmazonBenign environment

Courtesy ofAndrea Lima

Factors important in determining storm initiation • Orography• Boundary relativecell motion• Boundary collision

Amazon (Brazil)Satellite visibleimages

Courtesy ofAndrea Lima

Basic Understanding of Evolution Using RadarFactors important in determining storm initiation • Orography• Boundary-relativecell motion• Boundary collision

Amazon (Brazil)Radar reflectivity

Courtesy ofAndrea Lima

Basic Understanding of Evolution Using RadarFactors important in determining storm initiation • Orography• Boundary-relativecell motion• Boundary collision

Basic Understanding of Evolution Using Radar

Conceptual model of storm evolution over the Amazon

Factors important in determining storm initiation • Orography• Boundary-relativecell motion• Boundary collision

29 August 2006 Cartesian Image at 1.75 km

BamakoRadar & Airport

15:57

29 August 2006 Cartesian Image at 1.75 km

16:09

29 August 2006 Cartesian Image at 1.75 km

16:22

29 August 2006 Cartesian Image at 1.75 km

16:35

29 August 2006 Cartesian Image at 1.75 km

17:09

29 August 2006 Cartesian Image at 1.75 km

17:35

29 August 2006 Cartesian Image at 1.75 km

18:09

29 August 2006 Cartesian Image at 1.75 km

18:44

Basic understanding

Factors important in determining storm initiation

Convergence lines

• Boundary collision

• Boundary relative cell motion

• Orography and Terrain Features

• Climatology – Preferred regions

Radar Climatology for a Specific Region (Diurnal Cycle of Storm Locations – 7 Years of Radar Data)

9-10 Local 13-14 Local12-13 Local11-12 Local10-11 Local

22-23 Local21-22 Local

18-19 Local

19-20 Local 20-21 Local

16-17 Local15-16 Local 17-18 Local14-15 Local

23-24 Local

Percentage of radar volumes that had greater than 35 dBZ observed.

Radar Climatology for a Specific RegionDiurnal Cycle of Storm Locations: Winds from Southerly Direction

9-10 Local

18-19 Local16-17 Local15-16 Local 17-18 Local14-15 Local

13-14 Local12-13 Local11-12 Local10-11 Local

22-23 Local21-22 Local19-20 Local 20-21 Local 23-24 Local

Percentage of radar volumes that had greater than 35 dBZ observed. Saxen, 2005

Radar Reflectivity Mosaics of 2 or more radars

David Ahijevych NCAR

Hourly average frequency of radar echo for June, July & Aug 1996-2003

Freq

uenc

y

OutlineThunderstorm Storm -

• Predictability

• Basic understanding of evolution

• Nowcasting System

Detection and extrapolation of surface convergence boundaries ….

….that trigger thunderstorm initiation and impact storm evolution.

NCAR Thunderstorm System (Auto-Nowcaster)is unique in its ability to provide nowcasts of storm initiationby…..

NCAR Thunderstorm Nowcast System (Auto-Nowcaster)

• Produces 0-1 hr time and place specific forecast

• Expert system utilizes fuzzy logic

• Ingest multiple data sets

• Forecast storm initiation, growth and dissipation

• Algorithms derive forecast parameters based on the characteristics of the boundary-layer, storms, and clouds.

• 4-D Variational Doppler Radar Analysis System (VDRAS)

• Extrapolates radar echos

Nowcasting Methodology• Boundary layer structure

– convergence line position– colliding boundaries– strength of the

convergence– low-level shear– boundary-relative steering

flow– stability

Lifting Zone

Convergence Line

Convergence strength

Boundary Influences on Thunderstorm EvolutionBoundary Characteristics That Influence Storm Evolution

• Low-level Shear Relative to Boundary

(Thorpe et al. 1982, Rotunno et al., 1988, Weisman and Klemp 1986)

The low-level shear is the vector difference, normal to the boundary,of the surface wind minus the 2.5 km wind. It can varyconsiderable along the boundary.

This parameter is indicative of How tilted the updrafts will be. Values < -8 m/s favor erect Updrafts and thus more intense and long lived storms.

Stability Influences on Thunderstorm EvolutionSoundings are of limited use for thunderstorm nowcasting because of small-scale variability in water vapor.

In this example threesimultaneous soundingsshow there are largevariations in the convective availablepotential energy (orange area) overshort distances in thevicinity of a convergence line.Wilson et al., 1992

CONVERGENCE LINE MODIFIESTHE WATER VAPOR FIELD

Nowcasting Methodology• Boundary layer structure

– convergence line position– colliding boundaries– strength of the

convergence– low-level shear– boundary-relative steering

flow– stability

• Cloud characteristics-cloud type-cloud growth-cloud top temperatures-new cloud motion

Cumulus Cloud Growth Above Boundary

2 June 2000

Visible Infrared

Infrared Satellite Cloud Top Temperatures Used as Predictor of Storm Growth

Satellite-based, Feature Detection Algorithms

• Cross-correlation tracker (CTREC) on GOES-IR

• Upstream thresholding of satellite data (satThresh)

• IR Temperature Change (RateOfChange)

• Shortwave Reflectance (satDerive)

• Cloud classification (CloudClass)• Atmospheric Stability CTREC

satThreshsatThreshsatThreshRateOfChange (ROC)satDerive (satellite reflectance)CloudClass

High CAPE

Low CAPE

Nowcasting Methodology• Boundary layer structure

– convergence line position– colliding boundaries– strength of the

convergence– low-level shear– boundary-relative steering

flow– stability

• Cloud characteristics-cloud type-cloud growth-cloud top temperatures-new cloud motion

• Storm Characteristics- position and motion- growth rate - storm structure

- storm merger - storm-boundary interaction- storm decay

(

Thunderstorm Characteristics:Radar can provide time trends of thunderstorm movement, size, height, intensity, etc.

TITAN

Radar ReflectivityStorm Area Growth RateMaximum ReflectivityExtrapolatedStorms

Nowcasting Methodology• Boundary layer structure

– convergence line position– colliding boundaries– strength of the

convergence– low-level shear– boundary-relative steering

flow– stability

• Cloud characteristics-cloud type-cloud growth-cloud top temperatures-new cloud motion

• Storm Characteristics- position and motion- growth rate - storm structure

- storm merger - storm-boundary interaction- storm decay

What do you do with all this information???

Produce 1-2 hr Nowcasts of Storm Initiation, Growth and Decay

Blue Regions - Little chance of storm developmentGreen Regions - Moderate likelihood for thunderstormsRed Regions - Areas of forecast storm initiation

60 min Storm InitiationLikelihood Field

• Environmental conditions (Numerical Model)

– Frontal forcing– CAPE/CIN – CAPE – Relative humidity

• Boundary-layer– Convergence/vertical shear along

boundary– Colliding boundaries– Vertical velocity along boundary– Boundary-relative steering flow– New storms along boundary

• Clouds– Clear sky or cumulus clouds– Cloud growth observed with cloud top

cooling rate

Process a lot of information quickly.Nowcasts produced every 6-10 min

Provides 1 hr Nowcasts of:Thunderstorm Initiation, Extrapolation, Growth and Decay

1 hour forecast Verification

Storm Initiationnowcasts

Extrapolated nowcasts

Auto-Nowcaster System

Forecast Length

Extrapolation

NWP

Fore

cast

Ski

ll

Future Nowcast Systems

Blended Nowcast Systems

6 hr Blended Forecasts

Issued at July 14, 2004 at 19:00, Valid July 15, 2004 01:00

Green Contour - 35 dBZ radar reflectivity at valid time

Observations6 hr Extrapolations

Numerical Model6 hr Forecast

What is needed to get started?

• Well-calibrated radar• High Speed communications (mosaic’ing

in real-time)• Data quality algorithms• Radar detection and forecast algorithms• Forecasters trained in very short period

forecasting

Build Infrastructure:

End