Operational use of dual-polarisation: lessons learned at M

1
Operational use of dual-polarisation lessons
learned at Météo France after 8 years of
experience at all wavelengths (S / C / X) P.
Tabary Météo France Head of Weather Radar
Centre pierre.tabary_at_meteo.fr TECO2012 18
October 2012 Brussels
2
Outline of the presentation

• The French metropolitan radar network
• Demonstrated benefits of polarimetry at X / C /
  S bands
• Challenges / Open issues

3
The French metropolitan radar network in 2012
DP
DP
DP
DP
DP
DP
In 1991 11 radars
DP
DP

• In 2012
• 26 radars
• All Doppler (Triple-PRT)
• 18 C band (13 DPOL)
• 6 S band (2 DPOL)
• 2 X band (2 DPOL)

DP
DP
DP
DP
DP
DP DPOL Radar Purple S Green C Brown X
4
2004 First polarimetric radar installed in
Trappes
5
Polarimetry Roadmap 2004 - 2014

• 2004 First C-band dual-pol radar installed in
  Trappes
• 2004 2008 Demonstration of the benefits for
• Non Precipitation Echo ID
• Attenuation Correction
• Self-consistency calibration
• Rainfall rate retrieval
• Hydrometeor Classification
• 2012 1ST version of DPOL processing chain
  operational
• Non Precipitation Echo ID
• Basic ?DP-based Attenuation Correction
• 2014 (plan) 2ND version of DPOL processing
  chain operational
• Hydrometeor ID (Rain, Hail, Wet Snow, Dry Snow,
  )
• Improved Rain Rate Estimation Hybrid Z-KDP
  estimator

• Data Quality
• Calibration
• Monitoring

6
Automatic Non Precipitation Echo ID
?HV
Texture of ZDR
Histograms of dual-polarisation variables (?HV
and texture of ZDR) in precipitation,
ground-clutter and clear-air.
no precipitation
Gourley, JJ, P. Tabary, J. Parent-du-Chatelet,
2007 A fuzzy logic algorithm for the separation
of precipitating from non-precipitating echoes
using polarimetric radar, J. Atmos. Oceanic
Technol. Vol. 24, No. 8, 14391451.
7
Automatic Non Precipitation Echo ID
200 km
Green clear air
Blue ground-clutter
Echo Type
Reflectivity (dBZ)
Yellow Precipitation
Gourley, JJ, P. Tabary, J. Parent-du-Chatelet,
2007 A fuzzy logic algorithm for the separation
of precipitating from non-precipitating echoes
using polarimetric radar, J. Atmos. Oceanic
Technol. Vol. 24, No. 8, 14391451.
8
Quantitative Precipitation Estimation

• Evaluation at hourly time step against rain
  gauges in rain
• Comparison restricted to within 60 km of the
  radar
• Evaluation at the 3 wavelengths X / C / S
• Comparison of 3 different rain rate estimators
• QPE algorithm is adapted from Tabary (2007) and
  includes VPR and beam blocking correction,
  advection correction, .
• No real-time gauge adjustment is applied ?
  radar only QPE

Tabary P. 2007. The New French Operational Radar
Rainfall Product. Part I Methodology. Wea.
Forecasting. 22 393-408.
9
Results at S-band - Summer 2010 - 1 radar - 4
EventsEvaluation at hourly time step against
rain gauges
RR NB corr 5.0 -0.27 0.82
RR NB corr 5.0 -0.18 0.84
RR NB corr 5.0 -0.09 0.88

• Z-KDP
• If KDP lt 1/km ? Use of Z-R (Marshall-Palmer)
  with attenuation correction
• If KDP gt 1/km ? Use of R(KDP)

Z-R (Marshall-Palmer) without attenuation
correction
Z-R (Marshall-Palmer) with attenuation
correction PIA (dB) 0.04 ?DP ()
RR Hourly Rain Gauge Accumulation (in mm) NB
Normalized Bias (Radar vs. Gauge) Corr
Correlation coefficient
10
Results at C-band - Summer 2010 - 4 radars - 26
EventsEvaluation at hourly time step against
rain gauges
DBP2
No RGAdj
RR NB corr 5.0 -0.47 0.54
RR NB corr 5.0 -0.34 0.70
RR NB corr 5.0 -0.19 0.79

• Z-KDP
• If KDP lt 1/km ? Use of Z-R (Marshall-Palmer)
  with attenuation correction
• If KDP gt 1/km ? Use of R(KDP)

Z-R (Marshall-Palmer) without attenuation
correction
Z-R (Marshall-Palmer) with attenuation
correction PIA (dB) 0.08 ?DP ()
RR Hourly Rain Gauge Accumulation (in mm) NB
Normalized Bias (Radar vs. Gauge) Corr
Correlation coefficient
11
Results at X-band 2011 - 1 radar - 4
EventsEvaluation at hourly time step against
rain gauges
RR NB corr 5.0 -0.74 0.52
RR NB corr 5.0 -0.51 0.63
RR NB corr 5.0 -0.28 0.70

• Z-KDP
• If KDP lt 0,5/km ? Use of Z-R (Marshall-Palmer)
  with attenuation correction
• If KDP gt 0,5/km ? Use of R(KDP)

Z-R (Marshall-Palmer) without attenuation
correction
Z-R (Marshall-Palmer) with attenuation
correction PIA (dB) 0.28 ?DP ()
RR Hourly Rain Gauge Accumulation (in mm) NB
Normalized Bias (Radar vs. Gauge) Corr
Correlation coefficient
12
Data Quality Polarimetric monitoring indicators
12

• If well calibrated / processed (?DP ZDR),
  polarimetric variables improve the quality of all
  conventional radar products
• If not well calibrated / processed, polarimetric
  variables may lower the quality of all
  conventional radar products
• Examples
• 1) Large biases on ZDR may strongly impact rain
  rate estimation (0.2 dB 15)
• 2) Remaining ground-clutter may corrupt entire
  range profiles because of errors in ?DP offset
  computation
• Need to have very robust calibration, monitoing
  correction procedures

13
Long-term monitoring of polarimetric
indicatorsBlaisy (C-band) August 2010 ? April
2011
ZDR for ZH20-22 dBZ
12 13-10-2010 Maintenance on the radar
Typical scatter ??0.3 dB(Required ?0.2 dB)
28-03 01-03-2011 Maintenance on the radar
?DP offset
Slight positive bias (0.2 dB)
?HV
9 months
ZDR at 90
14
Long-term monitoring of polarimetric
indicatorsBlaisy
Stability of ZDR is close to but still slightly
below - requirements (?0.3 dB vs. ?0.2 dB
required) Temperature electronic calibration
procedures are thought to be responsible for the
observed scatter Work under progress The
quantitative use of ZDR remains a challenge
ZDR for ZH20-22 dBZ
12 13-10-2010 Maintenance on the radar
Typical scatter ??0.3 dB(Required ?0.2 dB)
28-03 01-03-2011 Maintenance on the radar
?DP offset
Slight positive bias (0.2 dB)
?HV
9 months
ZDR at 90
15
Conclusions
15

• Polarimetry has become the new standard in
  operational radar networks
• Polarimetry improves the quality of all radar
  products (e.g. rain rate estimation) especially
  at high frequency (X)
• New products can be proposed with polarimetry
  (e.g. hydrometeor classification)
• Phase-based parameters (?DP and KDP) are very
  valuable for attenuation correction and rain rate
  estimation
• The quantitative use of ZDR is still a challenge
  (calibration / stability issues vs. 0.2 dB
  precision required)
• The benefits for Quantitative Precipitation
  Estimation have been demonstrated in rain. Solid
  precipitation estimation is still an open area of
  research
• Rain gauges are still needed !

16
Questions