Staggered PRT Update

Part I

Sebastián Torres and David Warde CIMMS/The University of Oklahoma

and National Severe Storms Laboratory/NOAA

NEXRAD TACNorman, OK

29 August, 2012

Why Staggered PRT?• Less “Purple haze”: larger ra

• Less Velocity aliasing: larger va

Doppler Velocity from KTLXBatch Mode

Doppler Velocity from KOUNStaggered PRT

2

Early SPRT Concerns• Extending the range coverage– Overlaid echoes

• Extending the Nyquist velocity– Velocity dealiasing algorithm– Catastrophic errors

• Defining scanning strategies– PRT selection– Dwell-time selection

• Mitigating ground-clutter contamination

3

SPRT Range Coverage

ra,2

ra,1

ra,2

ra,1

ra,2-ra,1

Batc

hSP

RT

ra,L 2ra,S

ra,S

2ra,S

ra,S

ra,2

ra,1

Reflectivity Doppler Var. Polarimetric Var.

4No “purple haze”

possibleSome “purple haze”

possiblekey: vs.

SPRT Velocity Dealiasing• Velocities are estimated for each PRT– v1 are estimated from the short-PRT pairs– v2 are estimated from the long-PRT pairs

• Maximum unambiguous velocities for the short and long PRT pairs are different

– Velocities v1 and v2 alias in different ways• The true (de-aliased) velocity can be obtained by

“analyzing” how v1 and v2 alias

T1 T2

timeT1 T2

…R1

R2 R1 R2

5

“Catastrophic” Velocity Errors• A “catastrophic error” occurs if errors of

estimates are so large that v1 and v2 cannot be properly dealiased

– These appear as speckles in the velocity fields

– Catastrophic errors are more likely for wider normalized spectrum widths

– The ORPG velocity dealiasing algorithm has been modified to mitigate these

6

ORPG Legacy Velocity Dealiasing

KCRI - 31 March 2008

3.1 degNo Dealiasing

KCRI - 31 March 2008

3.1 degBaseline

Dealiasing

KCRI - 31 March 2008

3.1 degModified

Dealiasing

(courtesy of David Zittel, ROC)

7

Milestones• 2003: 2/3 PRT ratio with DC filter

– NSSL Report 7• 2005: SACHI filter (standalone)

– NSSL Report 9• 2007: Informal presentation to ROC DQ Team• 2008: Any PRT ratio with DC filter

– Last NEXRAD TAC informational briefing– NSSL Report 12

• 2009: 2/3 PRT ratio with SACHI filter– Stand-alone algorithm description delivered on 03/09

• 2009: 2/3 PRT ratio with SACHI filter and overlaid echo recovery– Stand-alone algorithm description delivered on 07/09– NSSL Report 13

• 2010: Extension to dual polarization– Last NSSL/NCAR/ROC TIM briefing– NSSL Report 14

• 2012: CLEAN-AP– NSSL Report 16

8

2012 SPRT Algorithm Description• Dual-pol extension– H- and V-channel processing– Spectral moments and

polarimetric variables• CLEAN-AP integration– Detection and filtering functions– Modular description to facilitate

future uniform-PRT implementation• Recovery of overlaid echoes– Allows using shorter PRTs for better

performance

9

ROC’s 3-Phase Approach

SPRT (NSSL)

DPT2 (SIGMET)

Algorithm Selection

Engineering Implementation

Operational Implementation

DC Removal GCF

RPG Velocity

Dealiasing Algorithm

Spectral GCF

Scanning Strategies

10

Batch PRT vs. Staggered PRT Reflectivity

(courtesy of D. Saxion, ROC)11

BATCHSPRT

Batch PRT vs. Staggered PRT Doppler Velocity

(courtesy of D. Saxion, ROC)12

BATCHSPRT

VCPs for SPRT• Careful VCP design for SPRT is needed– SPRT algorithm performance is more intimately

tied to acquisition parameters• Ground clutter suppression• Catastrophic errors

– More trade-offs than with other techniques• We want everything we have and more!

• SPRT can complement SZ-2 in VCPsthat achieve a “complete solution” for the mitigation of R/V ambiguities

13

Legacy Mitigation Strategy

0.5°

1.5°

19.5°

Long-PRT scan followed by short-PRT scan.Long-PRT reflectivities are used to unfold short-PRT velocities (at most, strongest overlaid trip can be recovered)

Alternating batches of long- and short-PRT pulses. Long-PRT reflectivities are used to unfold short-PRT velocities(at most, strongest overlaid trip can be recovered)

Split Cuts2 scans at each elevation angle

Batch PRT

1 scan at each elevation angle

Uniform PRT

1 scan at each elevation angle7.0°

14

Evolutionary Mitigation Strategy

0.5°

1.5°

19.5°

Long-PRT scan followed by phase-coded short-PRT scan.Long-PRT reflectivities are used to unfold short-PRT velocities (two strongest overlaid trips can be recovered) (ORDA Build 9)

Phase coding (SZ-2)2 scans at each elevation angle

Staggered PRT

1 scan at each elevation angle

Uniform PRT (Legacy)

1 scan at each elevation angle7.0°

Alternating long- and short-PRT pulses.

Range-unfolded reflectivities and velocities with good maximum unambiguous velocity (ORDA Build 14+)

15

Staggered PRT

1 scan at each elevation angle

What does a good VCP look like?

16

T1 = 1.6 msT2 = 2.4 msM = 28

SPRT

T1 = 1.23 msT2 = 1.84 msM = 40

SPRTSPRT

T1 = 1.6 msT2 = 2.4 msM = 28

TL = 3.11 msTS = 0.98 msML = 6, MS = 41

Batch(~ same dwell times)

Choosing the PRTs• PRT ratio: 2/3– Not required but recommended for initial

implementation• Choose the PRTs as short as possible– The proposed SPRT

algorithm can handle overlaid echoes

– Maximum overlay condition result in shortest PRTs

• ra,2 = rmax T2 = 2rmax /c

17

Choosing the Dwell Times• Dwell times must be chosen to…– Meet meteorological-variable variance

requirements– Meet ground-clutter suppression requirements– Satisfy operational needs for short update

times

18

Standard vs. Proposed VCP 221

Same update time Larger va

Reduced “purple haze” for v and v

No “purple haze” for Z, ZDR, DP, and HV

Lower variance of Z compared to batchSlightly increased variance of v

19

Future Work• Document performance of CLEAN-AP for SPRT– Similar to what was done for SACHI

• Support implementation of SPRT on the ORDA• Support engineering evaluation of SPRT• Support ORPG 2D-VDA modifications for SPRT– Similar to what was done for legacy VDA

• Support VCP design for SPRT• Investigate advantages of

other PRT ratios– Algorithm already generalized to

work with any PRT ratio20

Summary• SPRT is a mature technique– dual-pol extension– CLEAN-AP integration

• Algorithm description delivered to the ROC in April ‘12

• SPRT improves– range coverage

• less purple haze for v and v , no purple haze for the polarimetric variables

– velocity measurements• significantly less velocity aliasing

– data quality• reflectivity and polarimetric variables with lower variance

• SPRT is expected to replace Batch and Doppler waveforms in R/V ambiguity mitigation VCPs

21

Backup Slides

Another great example

Batch Mode SPRTDoppler Velocity (04 Mar 2004)

23

Case I: April 22, 2004 – 2.5 deg

Batch ModeVCP 11

ra = 147 kmva = 28.8 m/s

Doppler VelocityStaggered PRT(2/3, same DT)

ra = 184 kmva = 45.1 m/s

Doppler Velocity

Reflectivity

ra = 276 km

24

Case II: June 30, 2004 – 1.5 deg

Staggered PRT(2/3, same DT)

ra = 240 kmva = 34.7 m/s

Doppler VelocityBatch Mode

VCP 11

ra = 147 kmva = 28.8 m/s

Doppler Velocity

Reflectivity

ra = 360 km

25

Case III: March 3, 2004 – 2.5 deg

Batch ModeVCP 11

ra = 147 kmva = 28.8 m/s

Doppler VelocityStaggered PRT(2/3, same DT)

ra = 184 kmva = 45.1 m/s

Doppler Velocity

Reflectivity

ra = 276 km

26

Staggered PRT Processing

• Reflectivity and polarimetric-variable estimation– Segment I: short PRT samples– Segment II: short and long PRT samples– Segment III: long PRT samples

• Velocity and spectrum width estimation– Segment I: overlaid echoes on one sample of every pair

• No bias for “dominant” echo– Segment II: clean pairs– Segment III: overlaid echoes on one sample of every pair

• Non-dominant echo must be censored!27

T1 T2

I II I II III I

= 2/3R1 R2R1

R2R1R2