2010 phased-array radar innovative sensing experiment
DESCRIPTION
2010 Phased-Array Radar Innovative Sensing Experiment. Photo by Adam Smith. Photo by James Murnan. Pam Heinselman NSSL. Sebastian Torres NSSL and CIMMS. Daphne LaDue OU/CAPS. Heather Lazrus SSWIM. Photo by M. Benner. Acknowledgements. National Weather Center Contributions - PowerPoint PPT PresentationTRANSCRIPT
2010 Phased-Array Radar Innovative 2010 Phased-Array Radar Innovative Sensing ExperimentSensing Experiment
Sebastian TorresNSSL and CIMMS
Photo by Adam Smith
Daphne LaDueOU/CAPS
Pam HeinselmanNSSL
Photo by James Murnan
Heather Lazrus SSWIM
2010 PARISE April 2010 Norman, OK
Acknowledgements
National Weather Center Contributions– Signal processing techniques and software
development• Ric Adams, Chris Curtis, Eddie Forren, Igor Ivic, Dave
Priegnitz, John Thomson, David Warde (NSSL/CIMMS)
– WARNGEN for WDSS-II• Charles Kerr & Kurt Hondl (NSSL/CIMMS)• David Andra and other fcsters at Norman WFO
– WES cases• Mark Sessing, Darrel Kingfield, & Ben Baranowski
(WDTB)
– Playback Simulations• Kevin Manross (NSSL/CIMMS)
– Experimental Warning Program Logistics• Greg Stumpf and Travis Smith (NSSL/CIMMS and MDL)
2010 PARISE April 2010 Norman, OK
What’s Unique to PAR?Parabolic Antenna– Single radiation
element• Single transmitter• Single receiver
– Non-conformal– Fixed beam pattern– Mechanical steering
Phased Array Antenna
– Multiple radiation elements
• Multiple transmitters• Multiple receivers
– Conformal– Variable beam pattern– Electronic steering
Unique Capabilities
2010 PARISE April 2010 Norman, OK
PAR Electronic Beam Steering
Adapted from Jeff Herd, MIT LL
Beam Perpendicular to the Array
Scan To 30 deg
Want fields to interfere constructively in desired directions, and interfere destructively in the remaining space
2010 PARISE April 2010 Norman, OK
Why a PAR at NSSL?
The WSR-88D (NEXRAD) is ~20 years old
– End of life around the year 2020– Investigating replacement technology
• Affordability of new technology vs. operating costs for obsolete technology
• Improved weather surveillance capabilities– e.g., faster updates
• Combine weather and aircraft surveillance networks
– Reduce number of radars– Reduce maintenance costsThe MPAR Concept
+
2010 PARISE April 2010 Norman, OK
Multi-function PAR Concept
Long-Range Surveillance
Severe Weather Non-Cooperative Targets Weather Fronts
Terminal Surveillance
WMD Cloud
2010 PARISE April 2010 Norman, OK
What is the NWRT PAR?
SPY-1A Antenna
U.S. Navy
Partnership+
GovernmentAcademia
Private Industry
National Weather Radar Testbed Phased-Array
Radar
=
NWRT PAR
Photo by A. Zahrai
2010 PARISE April 2010 Norman, OK
Purpose of the NWRT PARWhich type of scanning
improvement do forecasters consider most
important?
62%
Stakeholders’ needs:Faster Updates
Why are faster updates needed?
1) Tornado cyclone and mesocyclone evolution can occur in 10s of seconds
2) Significant storm evolution and transition between storm types can occur between WSR-88D scans
3) Mid and upper-level signatures indicative of downbursts are not reliably detected
Source: Radar Operations Center Source: LaDue et al. 2010, in press (BAMS)
Determine how to best capitalize on PAR capabilities to address 21st century forecast and warning needs
Determine how to best capitalize on PAR capabilities to address 21st century forecast and warning needs
2010 PARISE April 2010 Norman, OK
Future PAR Design?
Ultimate GoalWhat we have now
2010 PARISE April 2010 Norman, OK
NWRT PAR Characteristics
90
1.5
2.1 2.1
Characteristics Similar to WSR-88D
Wavelength: PAR = 9.4 cm / WSR-88D = ~10 cm (S-band)
Range Resolution: PAR = 240 km / WSR-88D = 250 km
Sector Scans Polarization
90
Beam Width
1.5
2.1 2.1
10 km
Characteristics Different from WSR-88D
2010 PARISE April 2010 Norman, OK
Signal Processing Upgrades
– Data Quality: brings performance closer to that of operational radars
• Artifact removal– Ground clutter, interference, DC bias, point targets
• Range and velocity ambiguity mitigation• Calibration
– Evolutionary: demonstrates PAR technology for weather applications
• Faster updates– Range oversampling– Adaptive scanning
2010 PARISE April 2010 Norman, OK
Interference FiltersFilters OFF Filters ON
NWRT1 May 200800:25 UTC
2010 PARISE April 2010 Norman, OK
CLEAN-APCLutter Environment ANalysis Using Adaptive Processing
CLEAN-AP is OFFAre these storms?
NWRT10 Feb 09 16:19
CLEAN-AP is ONAP contamination was removed!
NWRT10 Feb 09 16:19
KTLX ReflectivityAll-bins filtering
2010 PARISE April 2010 Norman, OK
Range Unfolding
14
2010 PARISE April 2010 Norman, OK
Dynamic PRT Adjustmentvia the Radar Control Interface
PRT Change
PRT Change
2010 PARISE April 2010 Norman, OK
Want Faster Updates?There’s no such thing as a free lunch
Full ScanFull Scan Full ScanFull ScanShort Obs. Short Obs.
Time Time
Partial ScanPartial Scan
Less coverage
Less accuracy
Faster updates
Faster updates
Adaptive Adaptive ScanningScanning
Range Range OversamplingOversampling
2010 PARISE April 2010 Norman, OK
Range OversamplingA Signal Processing Solution
• Range oversampling (RO) adds more information without increasing the observation time
– RO leads to overlapping radar volumes
– RO results in more accurate estimates and/or faster updates
c/2
c/2L
c/2
Traditional Sampling Oversampling
These are These are cleverly cleverly combined combined
for for improved improved accuracyaccuracy
2010 PARISE April 2010 Norman, OK
Range OversamplingPerformance Demonstration
Data collected using same observation
time
Standard Processing Range Oversampling Processing
A smoother field is an indication of more accurate data
2010 PARISE April 2010 Norman, OK
Adaptive scanning
Areas of interest onlyArbitrary
Adaptive scanning
Areas of interest onlyArbitrary
Conventional scanning
EverywhereSequential
Conventional scanning
EverywhereSequential
Electronic Adaptive Scanning
Courtesy of Chris Curtis
Goal: Faster UpdatesGoal: Faster Updates
2010 PARISE April 2010 Norman, OK
What is ADAPTS?• Adaptive DSP Algorithm for PAR Timely
Scans– Beam positions are classified as active or
inactive• Only active beam positions are scanned
– Active beam positions are updated after every scan– Full surveillance volume scans are scheduled periodically
AZ
EL
Radar scans full surveillance volume
ADAPTS determines
active beam
positions
ADAPTS redetermines active
beam positions
time
SURV ADAPTS ADAPTS ADAPTS ADAPTS SURV ADAPTS
tn
AZ
EL
Radar scans active beam positions
tn+1
AZ
EL
Radar scans active beam positions
tn+2
AZ
EL
Radar scans active beam positions
tn+3
AZ
EL
Radar scans active beam positions
AZ
EL
Radar scans active beam positions
Surveillance update time
Scanning strategy schedule:
tn tn+1 tn+2 tn+3
2010 PARISE April 2010 Norman, OK
1st Criterion1st Criterion 2nd Criterion2nd Criterion 3rd Criterion3rd Criterion
How does ADAPTS work?• Active beam positions
meet one or more criteria• Elevation angle• Continuity and coverage• Neighborhood
Zth
continuity
Always scan at the lower elevations
Range
Z
Determine significant weather
signals
coverage
Real-time display of active beam positions
Look around to respond to storm
evolution
EL
AZleft right
up
dow
n
2010 PARISE April 2010 Norman, OK
ADAPTS PerformanceQualitative Evaluation
ADAPTS is OFF
ADAPTS is ON
09 AUG 2008 – Reflectivity - 8.7 deg
09 AUG 2008 – Reflectivity - 8.7 deg
2010 PARISE April 2010 Norman, OK
ADAPTS PerformanceQuantitative Evaluation
0040:59 0104:35 0129:24 0154:27 0217:27Time (UTC)
Te
mp
ora
l R
es
olu
tio
n (
min
)
0.9
1.0
1.1
1.2
1.3
1.4
a)
0040:59 0129:24 0217:27b)
200 km 150 km 100 km 200 km 150 km 100 km 200 km 150 km 100 km
1 May 2009 Nominal Update Time
ADAPTS Update Times
2010 PARISE April 2010 Norman, OK
PARISE: MotivationNWRT PAR has unique capabilities
Research Questions:1) How can technology be used to produced more focused, efficient, and effective weather surveillance?
2) Can improved understanding of storm processes be attained from high-temporal resolution data?
3) What are potential operational impacts of higher temporal resolution data on the warning decision process?
4) What additional warning lead time might be gained? Photos by James
Murnan
2010 PARISE April 2010 Norman, OK
NWRT PAR: More focused, efficient, effective weather surveillance?
1) Oversampled VCP2) Weather-driven ADAPTIVE scanning
Automated and Manual
2010 PARISE April 2010 Norman, OK
What is an Oversampled VCP?
Dense Vertical Sampling
Most dense sampling near the ground
Tilts chosen to sample storm extending to 18 km AGL w/in 20 km of the PAR
Optimized when the max height uncertainty (%) is ~same at all ranges and heights of storm features (Brown et al. 2000)
Overlapped Azimuthal Sampling
50% overlapped azimuthal sampling at all tilts to improve the apparent resolution of azimuthal signatures
Data Quality
Number of pulses for reflectivity and velocity estimates meet or exceed VCP 12 standards
2010 PARISE April 2010 Norman, OK
Oversampled VCP
22 tilts; max height uncertainty 18%
Update Time: 2 min
Red lines = VCP 12
2010 PARISE April 2010 Norman, OK
Weather-driven Adaptive Scanning
Are storms located only w/in 120 km of NWRT PAR?
Yes No
Choose uniform-PRT version of Oversampled VCP and run ADAPTS
Max Update Time: 1.4 min
Choose split-cut-PRT version of Oversampled VCP and run ADAPTS
Max Update Time: 2 min
2010 PARISE April 2010 Norman, OK
Weather-driven Adaptive Scanning
Are storms located only w/in 120 km of NWRT PAR?
Are storms potentially tornadic?
Yes No
Are storms located within 120 km of PAR?
Choose 2-tilt, split-cut VCPUpdate Time: 22 s
Composite Update Time: 2.7 min
Yes No
Continue to sample as assessed above
If only located w/in 120 km: Choose 4-tilt, uniform PRT VCP
Update Time: 18 sComposite Update Time: 2.3 min
Else: Choose 4-tilt, split-cut PRT VCP
Update Time: 38 sComposite Update Time: 3.3 min
2010 PARISE April 2010 Norman, OK
Weather-driven Adaptive Scanning
VCP # Tilts Waveform Update Time (s)
Oversampled VCP (OVCP) 22 Split cut < 6 120
OVCP_within_120km 22 Uniform 82
Storm Surveillance 11 Uniform 55
Tornadic_outside_120km 2 Split cut 22
Composite_outside_120 2 + 22 Split cut 164
Tornadic_SplitCut 4 Split cut 38
Composite_SplitCut 4 + 22 Split cut 196
Tornadic_within_120km 4 Uniform 18
Composite_within_120km 4 + 22 Uniform 138
2010 PARISE April 2010 Norman, OK
PARISE: MotivationNWRT PAR has unique capabilities
Research Questions:1) How can technology be used to produced more focused, efficient, and effective weather surveillance?
2) Can improved understanding of storm processes be attained from high-temporal resolution data?
3) What are potential operational impacts of higher temporal resolution data on the warning decision process?
4) What additional warning lead time might be gained? Photos by James
Murnan
2010 PARISE April 2010 Norman, OK
1) How might higher temporal resolution data impact the warning decision process?
2) Will faster data updates increase warning lead time?
How will we answer these questions?
Tuesday Evening and WednesdayFor a variety of playback and real-time (hopefully!) cases, interrogate PAR data and issue warnings as you would in your office.
After each event, discuss your experience in making warning decisions (or not)
Thursday: Impact of Temporal Resolution ExperimentDirectly compare how forecasters (you!) issue warnings based on data provided at current radar update rates, with warnings issued based on faster data updates provided by Phased Array Radar (PAR).
James Murnan
Impacts on warning decision process?
2010 PARISE April 2010 Norman, OK
Become comfortable using WDSS-II to interrogate storms and issue warnings
Issue warnings for 5 playback and/or real-time events
Participate in discussions on warning decision experience
Photos by James Murnan
Your Goals!
2010 PARISE April 2010 Norman, OK
Impact of Temporal Resolution
Experiment PurposeDetermine potential operational impact of temporal resolution on warning decision process and warning lead time
34
MethodControl and experiment groups• Control: PAR data degraded to WSR-88D update time • Experiment: PAR data with full-temporal resolution
Matched groups: You help us determine which pairings results in equivalent radar-interpretation skills
2010 PARISE April 2010 Norman, OK
What you will do during experiment
• Form two teams• Work through two cases– Gain situational awareness (WES)– Write an Area Forecast Discussion– Work through the case in displaced real time– Issue warning products as needed– Discuss your experience within your team– Contrast experiences with other team
• Break for lunch / work through 2nd case / final debrief to end the day
35
2010 PARISE April 2010 Norman, OK
Consent• Read through consent forms
• Participation– Provides first rigorous test of impact of temporal
resolution on warning decision process– Lasts one 8-hr workday– Provides opportunity to learn from each other– Is not completely confidential – Is not reimbursed above normal compensation– Is voluntary
• Indicate your preference on quotation and recording
36
2010 PARISE April 2010 Norman, OK
2:45 Gain experience using Warning Decision Support System – Integrated Information (WDSS-II)
4:00 Tour National Weather Center
5:00 Group Dinner (wx dependent)
6:30 First playback or real-time event
9:00 End of day!
Photos by James Murnan
Today’s Schedule