NEXRAD TAC MeetingAugust 22-24, 2000

Norman, OK

AP Clutter Mitigation Scheme

Cathy Kessinger

Scott Ellis

Joseph VanAndel

http://www.atd.ucar.edu/rsf/NEXRAD/TAC_DQ_Aug00/index.html

August 23, 2000 Cathy Kessinger Slide 2

Ground clutter due to anomalouspropagation degradesthe performance of rainfall estimatesfrom radar

Currently, it must be detected by operators and clutterfilters turned onmanually

Automation!

Reflectivity RadialVelocity

ReflectivityAP Clutter

ReflectivityPrecipitation

August 23, 2000 Cathy Kessinger Slide 3

AP Clutter Mitigation Scheme

• Improve quality of rainfall estimates

• 4 CCR’s have been approved

• Preliminary implementation of the AP clutter detection algorithm is underway at OSF

• Planned implementation for Open Build 2

August 23, 2000 Cathy Kessinger Slide 4

AP Clutter Mitigation Scheme• Automatic clutter filter control

• Radar Echo Classifier– Uses fuzzy logic techniques

(for details, see http://www.atd.ucar.edu/rsf/NEXRAD/index.html)

– AP Detection Algorithm (APDA)– Precipitation Detection Algorithm (PDA)– Clear Air Detection Algorithm (CADA)– other algorithms, as needed

• Reflectivity compensation of clutter filter bias

• Tracking of clutter filtered regions

August 23, 2000 Cathy Kessinger Slide 5

Radar Echo Classifier

• Tim O’Bannon is implementing REC – AP clutter detection algorithm is first– Sharing NEXRAD data sets for testing– Working out methods of comparing results

between NCAR and OSF

August 23, 2000 Cathy Kessinger Slide 6

Evaluation of REC• Use statistical indices to measure

performance of algorithms against “truth”– CSI, POD, FAR computed from

2x2 contingency table

• For NEXRAD cases, truth defined by human experts (subjective)

• For S-Pol cases, truth defined by Particle Identification algorithm (objective)

August 23, 2000 Cathy Kessinger Slide 7

Radar Echo Classifier

• Objective truth fields with S-Pol added– Truth derived from Particle Identification algorithm

using multi-parameter fields as input – No human truthing– Improves algorithm optimization– Lower FAR realized for APDA for S-Pol cases

• Less improvement for NEXRAD cases (subjective truth)

– Allows for real-time statistical evaluation

August 23, 2000 Cathy Kessinger Slide 8

Example of S-Pol “truth”

• 11 February 1999• AP, clear air &

precipitation• Truth:

– green = AP

– gold = precipitation

– red = clear air

Reflectivity Radial Velocity

Spectrum Width Truth

August 23, 2000 Cathy Kessinger Slide 9

Radar Echo Classifier• Two reflectivity features added

– Both computed over a local area– Matthias Steiner “spin” variable

• Reflectivity difference from gate to gate > threshold

• Difference > 0, spin > 0; Difference <0, spin <0

• Percentage of maximum possible spin changes

• Sign =100 for “speckled” fields, 0 for pure gradients

– Tim O’Bannon “sign” variable• Reflectivity difference from gate to gate

• Accumulate + or -1 depending on sign of difference

• Sign=0 for “speckled” fields, +1 for pure gradients

August 23, 2000 Cathy Kessinger Slide 10

Radar Echo Classifier

• Evaluating APDA in non-Doppler region (430 km) – Using spin and sign with reflectivity “texture”

and vertical difference in reflectivity – Membership functions are not optimized– KNQA movie loop

August 23, 2000 Cathy Kessinger Slide 11

Radar Echo Classifier• New Python Environment for Radar

Processing (PERP)– Better software development environment– Debugging nearly done– Implemented Radar Echo Classifier on S-Pol

• Web-based display of REC output (http://www.atd.ucar.edu/rsf/STEPS/AP/index.html)

• Real-time operation during STEPS

• Continue development during IMPROV this winter

August 23, 2000 Cathy Kessinger Slide 12

PrecipitationDetection Algorithm

Clear AirDetection Algorithm

Particle Identification(truth)

APDetection Algorithm

Radial VelocityReflectivity

Movie loop of June 19

Movie loop of June 22

S-Pol real-time display of REC output

August 23, 2000 Cathy Kessinger Slide 13

Radar Echo Classifier

• Development continuing on – Precipitation Detection Algorithm– Clear Air Detection Algorithm

August 23, 2000 Cathy Kessinger Slide 14

Reflectivity

FY99 PDA

Truth

FY98 CPDA

Precipitation Detection Algorithm

S-Pol scan with convective and stratiform precipitation (gold), clutter (green) and clear air return (red)

Note improved detection of all precipitation regions with PDA vs CPDA

CPDA is very good at detecting noise

August 23, 2000 Cathy Kessinger Slide 15

CADA Truth

CADA Truth

Clear Air Detection Algorithm

Results shown from two of the S-Pol cases

CADA performs well at detecting the clear air and does not detect most of the clutter return

Edges of precipitation echoes are falsely detected

August 23, 2000 Cathy Kessinger Slide 16

Summary• Implementation of REC at OSF is primary

emphasis

• Continuing to develop REC algorithms– Sea clutter algorithm next

• Next fiscal year NCAR will: – Add reflectivity compensation to PERP– Start development of automatic clutter filter

control