A Doppler Radar Emulator and its Application to the Detection of Tornadic Signatures Ryan M. May.

A Doppler Radar Emulator and its Application to the

Detection of Tornadic Signatures

Ryan M. MayRyan M. May

AcknowledgementsAcknowledgements

Reading CommitteeReading Committee Dr. Michael BiggerstaffDr. Michael Biggerstaff Dr. Ming XueDr. Ming Xue Dr. Robert PalmerDr. Robert Palmer

Dr. Tian-You YuDr. Tian-You Yu Curtis AlexanderCurtis Alexander Gordon CarrieGordon Carrie

MotivationMotivation

Create tool that generates radar moment Create tool that generates radar moment data for a given set of radar operating data for a given set of radar operating parametersparameters

Useful for:Useful for: Radar system designRadar system design Scanning strategy designScanning strategy design Algorithm developmentAlgorithm development Retrieval technique evaluationRetrieval technique evaluation

Previous WorkPrevious Work

Zrnic (1975) simulated time series radar data Zrnic (1975) simulated time series radar data using an assumed Gaussian distribution of using an assumed Gaussian distribution of velocities within a volumevelocities within a volume

Chandrasekar and Bringi (1987) simulated Chandrasekar and Bringi (1987) simulated reflectivity values as a function of raindrop size reflectivity values as a function of raindrop size distribution parametersdistribution parameters

Wood and Brown (1997) evaluated the effects of Wood and Brown (1997) evaluated the effects of WSR-88D scanning strategies on the sampling WSR-88D scanning strategies on the sampling of mesocyclones and tornadoesof mesocyclones and tornadoes

Capsoni and D’Amico (1998) simulated time Capsoni and D’Amico (1998) simulated time series radar data using returns from individual series radar data using returns from individual hydrometeors within a volumehydrometeors within a volume

Radar ConfigurationRadar Configuration

WavelengthWavelength LocationLocation Transmit PowerTransmit Power Antenna GainAntenna Gain Antenna BeamwidthAntenna Beamwidth Noise ThresholdNoise Threshold

Pulse LengthPulse Length PRFPRF Pulses per RadialPulses per Radial Rotation RateRotation Rate Gate LengthGate Length Scan AnglesScan Angles

CapabilitiesCapabilities

Azimuthal ResolutionAzimuthal Resolution Range ResolutionRange Resolution AttenuationAttenuation Range AliasingRange Aliasing Velocity AliasingVelocity Aliasing Anomalous PropagationAnomalous Propagation Antenna SidelobesAntenna Sidelobes

ScatteringScattering

Currently, the Rayleigh approximation is Currently, the Rayleigh approximation is used for scattering:used for scattering:

62

4

5

DKwb

Rain is assumed to have a Marshall-Palmer Rain is assumed to have a Marshall-Palmer distributiondistribution

Cloud droplets are assumed to be monodisperseCloud droplets are assumed to be monodisperse

)Im(32

wa KD

Emulator DesignEmulator Design

A “pulse” is propagated through the A “pulse” is propagated through the model’s numerical output grid along the model’s numerical output grid along the current pointing anglecurrent pointing angle

This pulse is subdivided into many small, This pulse is subdivided into many small, individual elementsindividual elements

Each element is assigned values for Each element is assigned values for reflectivity, radial velocity, and attenuation reflectivity, radial velocity, and attenuation factor from the model grid, using nearest factor from the model grid, using nearest neighbor samplingneighbor sampling

Emulator Design (cont.)Emulator Design (cont.)

Representation of segmented pulse being matched to model grid field


At a given instant, two pulses are being At a given instant, two pulses are being used, allowing for the simulation of 2used, allowing for the simulation of 2ndnd trip trip echoesechoes

For every range gate along the beam, the For every range gate along the beam, the pulses are sampled to produce a value of pulses are sampled to produce a value of returned power, Doppler velocity, and returned power, Doppler velocity, and velocity variancevelocity variance


Returned power is calculated as:Returned power is calculated as:

i ii

iiiit

rl

VWfgPrP

22

24

3

22

0 )4()(

Doppler velocity is the power-weighted average of all velocities for all pulse elementsDoppler velocity is the power-weighted average of all velocities for all pulse elements The velocity variance for the pulse is the power-weighted variance of velocities for all pulse elementsThe velocity variance for the pulse is the power-weighted variance of velocities for all pulse elements


When the returns for the specified number When the returns for the specified number of pulses for a radial have been of pulses for a radial have been calculated, a radial of data is generatedcalculated, a radial of data is generated

Returned power is the average returned Returned power is the average returned power for all pulsespower for all pulses

Doppler velocity is the power-weighted Doppler velocity is the power-weighted average velocity for all pulsesaverage velocity for all pulses

Spectrum width is the power-weighted Spectrum width is the power-weighted variance for all pulsesvariance for all pulses


At this point, the velocity is forced to a At this point, the velocity is forced to a value within the Nyquist co-interval, value within the Nyquist co-interval, simulating velocity aliasingsimulating velocity aliasing

Also, equivalent radar reflectivity factor is Also, equivalent radar reflectivity factor is calculated from the returned power as:calculated from the returned power as:

221

23

2210 2ln2

wt

re

KcgP

PrZ

Simulation CharacteristicsSimulation Characteristics

Simulation created using the Advanced Simulation created using the Advanced Regional Prediction System (ARPS)Regional Prediction System (ARPS)

Horizontal grid resolution: 50mHorizontal grid resolution: 50m Stretched vertical grid (~18m at surface)Stretched vertical grid (~18m at surface) Warm rain precipitation microphysicsWarm rain precipitation microphysics Produces a 200m diameter tornado with a Produces a 200m diameter tornado with a

160 m/s change in velocity across the 160 m/s change in velocity across the vortexvortex

ARPS Simulation Vector Velocity, Rain Water Mixing Ratio, and Total Buoyancy

Capabilities – Radar Capabilities – Radar CharacteristicsCharacteristics

Exp. λ (cm)

Beamwidth(deg)

PRF(Hz)

Pulse Length(μs)

Rot. Rate

(deg s-1)

Pulses Per Rad.

Gate Length(m)

control 10 1 1500 1.5 20 75 250

OS 10 1 1500 1.5 15 50 250

GS 10 1 1500 .75 20 75 125

NSL 10 1 1500 1.5 20 75 250

BW 10 2 1500 1.5 20 75 250

NY 10 1 1000 1.5 20 50 250

X 3 1 1500 1.5 20 75 250

ST 10 1 1500 1.5 20 75 250

Examples – 10cm, 1Examples – 10cm, 1o o BeamwidthBeamwidth

Returned Power Equivalent Reflectivity Factor

Doppler Velocity Spectrum Width

Examples – Azimuthal OversamplingExamples – Azimuthal Oversampling

CONTROL Equivalent

Reflectivity Factor

CONTROL Doppler Velocity

OVERSAMPLEDEquivalent

Reflectivity Factor

OVERSAMPLED Doppler Velocity

Examples – Azimuthal OversamplingExamples – Azimuthal Oversampling

Equivalent Reflectivity Factor Difference (Oversampled - Orginal)

Examples – 125m Gate SpacingExamples – 125m Gate Spacing

CONTROL Equivalent

Reflectivity Factor


125M GATE SPACING Equivalent

Reflectivity Factor

125M GATE SPACING

Doppler Velocity

Examples – No SidelobesExamples – No Sidelobes

NO SIDELOBES Equivalent

Reflectivity Factor

NO SIDELOBES Doppler Velocity

CONTROL Equivalent

Reflectivity Factor



Returned Power Difference (Original – No Sidelobes)

Examples – 2Examples – 2oo Beamwidth Beamwidth

Equivalent Reflectivity Factor

(original)Doppler Velocity

(original)


(2o Beamwidth)Doppler Velocity (2o Beamwidth)

Examples – Low PRFExamples – Low PRF



Examples – X-band (3cm)Examples – X-band (3cm)



Examples – X-band (3cm)Examples – X-band (3cm)

Returned Power Difference (Original – X-band)

Examples – 2Examples – 2ndnd Trip Echoes Trip Echoes



We’ve now seen examples of the We’ve now seen examples of the emulator’s capabilitiesemulator’s capabilities

Let’s move on to material that’s more…Let’s move on to material that’s more…practical: detecting tornadoespractical: detecting tornadoes

From Examples to ApplicationFrom Examples to Application

Example Application:Example Application:Tornado DetectionTornado Detection

Emulated data for prototype CASA radarsEmulated data for prototype CASA radars 4 Metrics for tornado intensity:4 Metrics for tornado intensity:

Maximum velocityMaximum velocity ΔΔVelocityVelocity DiameterDiameter Axisymmetric vorticity: 2Axisymmetric vorticity: 2ΔΔV / DV / D

4 Ranges : 3km, 10km, 30km, 50km4 Ranges : 3km, 10km, 30km, 50km Matched Sampling / OversamplingMatched Sampling / Oversampling

Tornado Detection – Radar Tornado Detection – Radar CharacteristicsCharacteristics

Parameter Matched Sampling Oversampled

λ (cm) 3 3

Beamwidth (deg) 2 2

PRF (Hz) 2000 2000

Rot. Rate (deg s-1) 40 40

Pulses Per Rad. 100 50

Pulse Length (μs) .5 .5

Gate Length (m) 100 100

Tornado – 3km, Matched SamplingTornado – 3km, Matched Sampling

Equivalent Reflectivity

Factor

Spectrum Width

Doppler Velocity Doppler Velocity

(no aliasing)

Tornado – 3km, OversamplingTornado – 3km, Oversampling


Factor

Spectrum Width


(no aliasing)



Factor

Spectrum Width


(no aliasing)



Factor

Spectrum Width


(no aliasing)



Factor

Spectrum Width


(no aliasing)



Factor

Spectrum Width


(no aliasing)



Factor

Spectrum Width


(no aliasing)



Factor

Spectrum Width


(no aliasing)

Tornado Detection ResultsTornado Detection Results

Experiment Vmax (m s-1) ΔV (m s-1) D (m) 2 ΔV/D (s-1)

3km, Matched 49.1 93.3 216 0.864

3km, Oversampled 55.7 110.6 216 1.024

10km, Matched 35.2 57.6 705 0.163

10km, Oversampled 36.3 62.7 529 0.237

30km, Matched 31.8 33.4 1047 0.064

30km, Oversampled 32.3 42.7 1047 0.082

50km, Matched 27.5 29.5 1749 0.034

50km, Oversampled 28.5 38.6 1749 0.044

ConclusionsConclusions

The large beamwidth of the CASA radars The large beamwidth of the CASA radars will be significant hurdle to the detection of will be significant hurdle to the detection of tornadoestornadoes Oversampling does help mitigate Oversampling does help mitigate somesome of this of this

problemproblem These sampling issues will compound the These sampling issues will compound the

dealiasing problems due to the low dealiasing problems due to the low Nyquist velocity at X-bandNyquist velocity at X-band The quality of the dealiasing procedure for the The quality of the dealiasing procedure for the

data will be data will be extremelyextremely important important

Future StudiesFuture Studies

Continue examining the detectability of Continue examining the detectability of tornadoestornadoes Test detection using objective algorithmsTest detection using objective algorithms Examine impacts of attenuationExamine impacts of attenuation Examine data for times when storm is not Examine data for times when storm is not

tornadictornadic Examine vertical continuityExamine vertical continuity

Evaluate scanning impacts on quality of Evaluate scanning impacts on quality of dual Doppler analysisdual Doppler analysis

Future DevelopmentFuture Development

Mie ScatteringMie Scattering Phased Array AntennaPhased Array Antenna Time Evolution of Model FieldTime Evolution of Model Field Polarimetric VariablesPolarimetric Variables Ground Clutter TargetsGround Clutter Targets

Questions?Questions?

Capsoni, C., and M. D'Amico, 1998: A physically based radar simulator. J. Atmos. Oceanic Technol., 15, 593-598.

Chandrasekar, V., and V. N. Bringi, 1987: Simulation of radar reflectivity and surface measurements of rainfall. J.

Atmos. Oceanic Technol., 4, 464-478.Wood, V. T., and R. A. Brown, 1997: Effects of radar sampling on single-Doppler velocity signatures of mesocyclones and tornadoes. Wea. Forecasting, 12, 928-938.Zrnic, D. S., 1975: Simulation of weatherlike Doppler spectra

and signals. J. App. Meteor., 14, 619-620.

Examples – 10cm, 1Examples – 10cm, 1o o BeamwidthBeamwidth



Examples – 125m Gate SpacingExamples – 125m Gate Spacing

Returned Power (control)






Examples – 2Examples – 2oo Beamwidth Beamwidth



A Doppler Radar Emulator and its Application to the Detection of Tornadic Signatures Ryan M. May.

Documents

radial velocity

velocity aliasingalso

pulsesdoppler velocity

average returned power

generatedreturned power

value of returned power

radar moment data

doppler radar emulator