1 clutter mitigation decision (cmd) theory and problem diagnosis radar monitoring workshop erad 2010...
TRANSCRIPT
11
CLUTTER MITIGATION DECISION (CMD)CLUTTER MITIGATION DECISION (CMD)THEORY AND PROBLEM DIAGNOSISTHEORY AND PROBLEM DIAGNOSIS
RADAR MONITORING WORKSHOP
ERAD 2010SIBIU ROMANIA
Mike DixonNational Center for Atmospheric Research
Boulder, Colorado
22
Why do we need to know where clutter is in order Why do we need to know where clutter is in order to filter it effectively?to filter it effectively?
We need to filter We need to filter normal-propagationnormal-propagation (NP) clutter and (NP) clutter and anomalous-anomalous-
propagationpropagation (AP) clutter (AP) clutter
The filters used remove weather power in some circumstances:The filters used remove weather power in some circumstances:
– Velocity close to 0Velocity close to 0
– Spectrum width close to 0Spectrum width close to 0
These conditions occur both in:These conditions occur both in:
– ClutterClutter
– Stratiform precipitationStratiform precipitation
We therefore need a technique which identifies likely locations of clutterWe therefore need a technique which identifies likely locations of clutter
The filter is applied only at those gates with a The filter is applied only at those gates with a high probability of clutterhigh probability of clutter..
33
Problem - weather and clutter combinedProblem - weather and clutter combinedReflectivity plotReflectivity plot
44
Problem - weather and clutter combinedProblem - weather and clutter combinedVelocity plotVelocity plot
55
What happens if we filter everywhere?What happens if we filter everywhere?
Filtered reflectivity – applying clutter filter everywhere
Weather power removed
77
Combined clutter and weather.Combined clutter and weather.Weather and clutter are distinct.Weather and clutter are distinct.
88
Combined clutter and weather spectrum.Combined clutter and weather spectrum. Weather and clutter overlap somewhat. Weather and clutter overlap somewhat.
99
Combined clutter/weather spectrum.Combined clutter/weather spectrum.Weather and clutter overlap completely. Weather and clutter overlap completely.
1212
Adaptive filter has a difficult timeAdaptive filter has a difficult timewhen clutter and weather overlap when clutter and weather overlap
1414
MotivationMotivation
Adaptive spectral clutter filters show great promise for Adaptive spectral clutter filters show great promise for intelligently filtering clutter power while leaving weather intelligently filtering clutter power while leaving weather power largely unaffected.power largely unaffected.
However, these filters still remove weather power under the However, these filters still remove weather power under the following circumstances:following circumstances:– the weather return has a the weather return has a velocity close to zerovelocity close to zero;;– the weather return has a the weather return has a narrow spectrum widthnarrow spectrum width..
This tends to occur with This tends to occur with stratiformstratiform weather in the region of weather in the region of the the zero isodopzero isodop..
In order to mitigate the problem, information other than In order to mitigate the problem, information other than that used by the filters must be used to determine whether that used by the filters must be used to determine whether clutter exists at a gate.clutter exists at a gate.
1515
Principal feature fields for CMDPrincipal feature fields for CMD
In order to identify gates with clutter, we use a number of In order to identify gates with clutter, we use a number of so-called feature fields. These contain information which is so-called feature fields. These contain information which is independent of that used by the clutter filter.independent of that used by the clutter filter.
The feature fields used in the latest CMD version are:The feature fields used in the latest CMD version are:
– The The TEXTURETEXTURE of the reflectivity field – of the reflectivity field – TDBZTDBZ..
– The The SPINSPIN of the reflectivity field. This is a measure of of the reflectivity field. This is a measure of how often the reflectivity gradient changes sign.how often the reflectivity gradient changes sign.
– The The Clutter Phase AlignmentClutter Phase Alignment or or CPACPA, which is a measure , which is a measure of the pulse-to-pulse stability of the returned signal.of the pulse-to-pulse stability of the returned signal.
1616
TEXTURE of reflectivity - TDBZTEXTURE of reflectivity - TDBZ
TDBZ is computed as the TDBZ is computed as the mean of the squared mean of the squared reflectivity differencereflectivity differencebetween adjacent gates.between adjacent gates.
TDBZ is computed at each TDBZ is computed at each gate along the radial, with the gate along the radial, with the computation centered on the computation centered on the gate of interest.gate of interest.
TDBZ at a gate is computed TDBZ at a gate is computed using the dBZ values for the 4 using the dBZ values for the 4 gates on either side of the gates on either side of the gate of interest.gate of interest.
Computed for this gate
Using data from these gates
1717
TDBZ feature fieldTDBZ feature field
DBZ TDBZ
Example of TDBZ – Denver Front Range NEXRAD - KFTG
1818
Reflectivity SPINReflectivity SPIN
For a point at which a gradient sign change occurs, let x be the reflectivitychange from the previous gate and y be the reflectivity change to the next gate.
Then
SPIN CHANGE = (|x| + |y|) / 2
xy
x y
x y
1919
SPIN feature fieldSPIN feature field
DBZ SPIN
Example of SPIN – Denver Front Range NEXRAD – KFTG(SPIN is noisy in low SNR regions)
2020
Clutter Phase Alignment - CPAClutter Phase Alignment - CPA
In In clutterclutter, the phase of each pulse in the time series for a , the phase of each pulse in the time series for a particular gate is particular gate is almost constantalmost constant since the clutter does not since the clutter does not move much and is at a constant distance from the radar.move much and is at a constant distance from the radar.
In In noisenoise, the phase from pulse to pulse is , the phase from pulse to pulse is randomrandom..
In In weatherweather, the phase from pulse to pulse , the phase from pulse to pulse will varywill vary depending on the velocity of the targets within the depending on the velocity of the targets within the illumination volume. illumination volume.
2323
CPA feature fieldCPA feature field
CPA is computed as the length of the cumulative phasor vector, CPA is computed as the length of the cumulative phasor vector, divided by the sum of the power for each pulse.divided by the sum of the power for each pulse.
CPA is computed at a single gate.CPA is computed at a single gate.
It is a normalized value, ranging from 0 to 1.It is a normalized value, ranging from 0 to 1.
In clutter, CPA is typically above 0.9In clutter, CPA is typically above 0.9..
In weather, CPA is often close to 0In weather, CPA is often close to 0, but increases in weather with a , but increases in weather with a velocity close to 0 and a narrow spectrum width.velocity close to 0 and a narrow spectrum width.
In noise, CPA is typically less than 0.05In noise, CPA is typically less than 0.05..
CPA was originally developed as a quality control field for clutter CPA was originally developed as a quality control field for clutter targets used for refractivity measurements.targets used for refractivity measurements.
2525
Combining TDBZ, SPIN and CPACombining TDBZ, SPIN and CPA
The individual The individual featurefeature fields, TDBZ, SPIN and CPA, are fields, TDBZ, SPIN and CPA, are combined into a single combined into a single interestinterest field using fuzzy logic. field using fuzzy logic.
First, each First, each featurefeature field is converted into an field is converted into an interestinterest field, field, using a membership transfer function.using a membership transfer function.
Interest fields have a range from 0.0 to 1.0.Interest fields have a range from 0.0 to 1.0.
The The interestinterest fields are assigned a fields are assigned a weightweight..
The The combined interestcombined interest field is computed as a field is computed as a weighted weighted meanmean of the individual interest fields. of the individual interest fields.
2626
Steps in computing the single-pol CMDSteps in computing the single-pol CMD
Step 1:Step 1:
– Compute TDBZ and TDBZ-interestCompute TDBZ and TDBZ-interest
– Compute SPIN and SPIN-interestCompute SPIN and SPIN-interest
– Compute CPA and CPA-interestCompute CPA and CPA-interest
Step 2:Step 2:
– Compute Texture-interest = maximum of (TDBZ-interest, SPIN-interest)Compute Texture-interest = maximum of (TDBZ-interest, SPIN-interest)
Step 3:Step 3:
– Compute CMD valueCompute CMD value
= fuzzy combination of CPA-interest and Texture-interest= fuzzy combination of CPA-interest and Texture-interest
Compute CMD flag: true if CMD >= 0.5, false if CMD < 0.5Compute CMD flag: true if CMD >= 0.5, false if CMD < 0.5
2727
Membership functions for single-pol CMDMembership functions for single-pol CMD
-> (15,0)
-> (30,1)
2828
Membership function combinationMembership function combination as used in single-pol CMD as used in single-pol CMD
-> (15,0)
-> (30,1)
Weight=1.01
Weight=1.0
3030
Logic for setting the clutter flagLogic for setting the clutter flag
1.SNR> 3dB?
3.CMD> 0.5?
3.SetFlag.
4040
THERE ARE 3 MAIN ERROR TYPESTHERE ARE 3 MAIN ERROR TYPES
1.1. FALSE DETECTIONSFALSE DETECTIONS: the algorithm detects clutter incorrectly, so : the algorithm detects clutter incorrectly, so
that the filter is applied excessively. This is particularly that the filter is applied excessively. This is particularly
problematic when it occurs in the region of 0 velocity, since the problematic when it occurs in the region of 0 velocity, since the
filter cannot distinguish between clutter and weather.filter cannot distinguish between clutter and weather.
2.2. MISSED DETECTIONSMISSED DETECTIONS: the algorithm fails to identify clutter, and : the algorithm fails to identify clutter, and
the filter is therefore not applied where it should be.the filter is therefore not applied where it should be.
3.3. FILTER FAILUREFILTER FAILURE:: the CMD algorithm works correctly, but the the CMD algorithm works correctly, but the
filter fails to work effectively.filter fails to work effectively.
4141
ERROR TYPE 1:ERROR TYPE 1:
FALSE DETECTIONSFALSE DETECTIONS
This is the most common error type.This is the most common error type.
4242
Example 1 - Example 1 - Filtered DBZFiltered DBZ
Filtered reflectivity using CMD Filtered everywhere
4444
Example 2 : Filtered DBZExample 2 : Filtered DBZ
Filtered using CMD Filtered everywhere
Weather powerremoved
4646
Example 3 : KFTG 2006/10/10, 1000 UTCExample 3 : KFTG 2006/10/10, 1000 UTC
Filtered using CMD Filtered everywhere
Weather powerremoved
5050
Example Example 44 : : Operational NEXRADOperational NEXRAD
Filtered reflectivity Filtered velocity
5151
ERROR TYPE 2:ERROR TYPE 2:
MISSED DETECTIONSMISSED DETECTIONS
It was found that in some clutter regions, CPA It was found that in some clutter regions, CPA values can be lower than expected.values can be lower than expected.
5252
KEMX reflectivity 2009/04/22 21:22 UTCKEMX reflectivity 2009/04/22 21:22 UTCAlso showing counties, interstates and US highwaysAlso showing counties, interstates and US highways
We will investigate the region indicated by the ellipse
5656
Region 1 – CPARegion 1 – CPA
This region has CPA values with considerable variability, with some values in the clutter region being as low as 0.2.
5757
Characteristics of the CPA fieldCharacteristics of the CPA field
It was noted that in some of the clutter regions, there are a It was noted that in some of the clutter regions, there are a
considerable number of gates with low CPA values. The CPA considerable number of gates with low CPA values. The CPA
values can vary from high to low in adjacent gates.values can vary from high to low in adjacent gates.
Examining the phase time series for these gates shows that Examining the phase time series for these gates shows that
the phase can change significantly over a small part of the the phase can change significantly over a small part of the
time series.time series.
It is probable that 2 separate targets are illuminated during It is probable that 2 separate targets are illuminated during
the dwell.the dwell.
This phase change reduces CPA.This phase change reduces CPA.
However, for the remainder of the time series the phase is However, for the remainder of the time series the phase is
stable.stable.
5858
Example - spectrum of clutter point with low CPAExample - spectrum of clutter point with low CPACPA = 0.19CPA = 0.19
A-scopeX-axis: range (km)
Red – unfiltered spectrumX-axis: samples
Pink – filtered spectrum
Power time seriesX-axis: time
Phase time seriesX-axis: time
Pulse-to-pulse phaseDifference time seriesX-axis: time
Magenta line shows range
Change in phase for partof the time series leads to low CPA value
5959
ERROR TYPE 3:ERROR TYPE 3:FILTER DOES NOT WORK EFFECTIVELYFILTER DOES NOT WORK EFFECTIVELY
The clutter filter can have problems dealing with The clutter filter can have problems dealing with multi-mode spectra.multi-mode spectra.
6060
Traffic clutter spectraTraffic clutter spectra
It appears that some of the missed detections and poor It appears that some of the missed detections and poor
clutter filter performance are caused by traffic echoes.clutter filter performance are caused by traffic echoes.
The following slides show spectra from normal clutter The following slides show spectra from normal clutter
echoes and suspected traffic echoes.echoes and suspected traffic echoes.
The traffic echoes exhibit multi-modal spectra. This makes The traffic echoes exhibit multi-modal spectra. This makes
them both difficult to detect as clutter, and difficult to filter them both difficult to detect as clutter, and difficult to filter
with the current adaptive filters.with the current adaptive filters.
6161
Unfiltered reflectivityUnfiltered reflectivity
Note interstate 10 – shown in
bold – traversing this
area,and smaller
roads
6262
FFiltered reflectivity showing gates atiltered reflectivity showing gates atwhich CMD and the clutter filter failedwhich CMD and the clutter filter failed
Ellipse high-lightsgates for whichCMD and/or theclutter filterfailed
6363
Normal-propagation clutter signatureNormal-propagation clutter signatureCPA = 0.88CPA = 0.88
A-scopeX-axis: range (km)
Red – unfiltered spectrumX-axis: samples
Pink – filtered spectrum
Power time seriesX-axis: time
Phase time seriesX-axis: time
Pulse-to-pulse phasedifference time seriesX-axis: time
Magenta line shows range
6464
Spectrum of suspected traffic targetsSpectrum of suspected traffic targetsCPA = 0.17CPA = 0.17
A-scopeX-axis: range (km)
Red – unfiltered spectrumX-axis: samples
Pink – filtered spectrum
Power time seriesX-axis: time
Phase time seriesX-axis: time
Pulse-to-pulse phasedifference time seriesX-axis: time
Magenta line shows range
Multi-modal spectrum
6565
Spectrum of suspected traffic targetsSpectrum of suspected traffic targetsCPA = 0.30CPA = 0.30
A-scopeX-axis: range (km)
Red – unfiltered spectrumX-axis: samples
Pink – filtered spectrum
Power time seriesX-axis: time
Phase time seriesX-axis: time
Pulse-to-pulse phasedifference time seriesX-axis: time
Magenta line shows range
Multi-modal spectrum