Advances in polarimetric X-band weather radar

Remote Sensing of the Environment

AATTMMOOSS

AATTMMOOSS

DelftUniversity ofTechnology

Advances in polarimetric X-band weather radarTobias Otto


AATTMMOOSS

AATTMMOOSS


Contents• motivation

• weather radar polarimetry

• X-band challenge

• radar data processing• attenuation correction• differential phase processing

• raindrop-size distribution

• quantitative precipitation estimation (QPE)

• further applications

• limitations of X-band weather radar

• radar technologies for polarimetric X-band weather radar


AATTMMOOSS

AATTMMOOSS


Compact, easily deployable and cheaper than the usual S- or C-band weather radars.

Used for dedicated, short-range (< 60km) applications such as• gap-filling radars in complex terrain such as moutainous areas, e.g.

- RHyTMEE project of Météo France

• high-resolution precipitation measurement in densly populated areas in order toimprove urban water management and flood prediction, e.g. - polarimetric X-band radar network in Tokyo, Japan (http://www.bosai.go.jp/kiban/radar) - RAINGAIN project in Paris, Rotterdam, London and Leuven (http://www.raingain.eu) - CASA Dallas Fort Worth Urban Demonstration Network (http://www.casa.umass.edu/)

• improve the low-altitude radar coverage

They can provide a higher temporal and spatial resolution than standard operational weather radars due to the reduced range coverage and less stringent requirements on the scanning strategy due to their focused application.

But• attenuation due to rain is stronger than at S- or C-band, total signal extinction within

few kilometres is possible in a cloudburst (instantaneous rain rates >100 mmh -1)

• resonance scattering (Mie scattering) occurs in moderate to strong rain

Why X-band*?

*electromagnetic frequency band from 8 – 12 GHz


AATTMMOOSS

AATTMMOOSS


radar.dhigroup.com

metek.de

Marine radars turned into weather radars.

usually power measurement onlywith fan beam antenna coarse resolution in elevationgood for a spatial overview of precipitation but

not for quantitative precipitation estimation (QPE) cheap

gematronik.com

novimet.com

Dedicated polarimetric weather radars.

beside power also Doppler and polarimetricmeasurements

very good for quantitative precipitation estimationnot that cheap

The two X-band weather radar worlds


AATTMMOOSS

AATTMMOOSS


• motivation









Contents


AATTMMOOSS

AATTMMOOSS


Most hydrometeors are not spherical, andthey show distinct polarimetric signatures at microwave frequencies.

- ice particles

- hail

- raindrops

Beard, K.V. and C. Chuang: A New Model for the Equilibrium Shape of Raindrops, Journal of the Atmospheric Sciences, vol. 44, pp. 1509 – 1524, June 1987. http://commons.wikimedia.org/wiki/Category:Hail

Why polarimetry?

http://upload.wikimedia.org/wikipedia/commons/a/a5/Hail_-NOAA.jpg

http://upload.wikimedia.org/wikipedia/commons/a/a5/Hail_-NOAA.jpg


AATTMMOOSS

AATTMMOOSS


linear horizontal / vertical polarisations (H and V)

Motivation:- easier to understand especially for the weather radar user community

- close to the characteristic / principal polarisations for measurementsat low elevations, i.e. low depolarisation

- differential measurements (power, phase) between H and V are directlylinked to the anisotropy (oblateness) of the hydrometeors

What to measure?- ideally the complex polarisation scattering matrix which links the incident electric

field vector Ei with the backscattered electric field vector Es

re

EE

SSSS

EE jkr

iv

ih

vvvh

hvhhsv

sh

Which polarisations are used?


AATTMMOOSS

AATTMMOOSS


Zhh (dBZ)

Zvh (dBZ)

Zhv (dBZ)

Zvv (dBZ)

transmit

rece

ive

(alternate polarisation mode)

Measurement principle

Data: C- Band POLDIRAD (DLR, Oberpfaffenhofen, Germany), Prof. Madhu Chandra


AATTMMOOSS

AATTMMOOSS


Zhh (dBZ)

Zvh (dBZ)

Zhv (dBZ)

Zvv (dBZ)

transmit

rece

ive

-= Zdrdifferentialreflectivity


Differential reflectivity


AATTMMOOSS

AATTMMOOSS


dBZlog10 2hhhh PCRZ dBlog10

vv

hhdr P

PZ

Differential ReflectivityReflectivity

rainmelting layeraggregates (snow)ice crystals


Differential reflectivity


AATTMMOOSS

AATTMMOOSS


Zhh (dBZ)

Zvh (dBZ)

Zhv (dBZ)

Zvv (dBZ)

transmit

rece

ive -

= LDR (dB)linear depolar-

isation ratio


Linear depolarisation ratio


AATTMMOOSS

AATTMMOOSS


dBZlog10 2hhhh PCRZ dBlog10

vv

hv

PPLDR

Linear Depolarisation RatioReflectivity

melting layerground clutter


Linear depolarisation ratio


AATTMMOOSS

AATTMMOOSS


range-normalised microwave propagation through rain

range r

phase difference between H and Vdifferential phase Φdp (deg)

range

The slope of the differential phase is calledspecific differential phase:

12

121

2)()(

kmdegrr

rrK dpdp

dp

The measurement of the differential phase is crucial for polarimetric X-band weather radars because it is:

- independent from radar calibration - independent from partial beam blocking and attenuation as long as the signal is not totally extinct - almost linearly related to rain attenuation - very useful at X-band for rainfall rate estimation when R 3 mm h-1

Differential phase


AATTMMOOSS

AATTMMOOSS


• motivation









Contents


AATTMMOOSS

AATTMMOOSS


attenuation A

differential propagation phase Φdp

propagation(forward-scattering)

backward-scattering

differential backscatter phase δco

reflectivity Z

Power and differential phase measurements by X-band weather radars are always a combination of propagation and backward-scattering effects that need to be separated before analysing the weather radar data.

drrrZrZnr

rrnn

1

1

2' α

1

1

2 ( )nr

dp n co n dpr r

r r K r dr

X-band challenge


AATTMMOOSS

AATTMMOOSS


0.5°

reflectivity (dBZ)

differential phase (deg)

differential reflectivity (dB)

Data: TU Delft X-band IDRA, data freely available at http://data.3tu.nl/repository/collection:cabauw

differential backscatter phase(an indicator of resonance/Mie scattering)

differential attenuation

A clutter-filtered polarimetric X-band weather radar measurement.

X-band challenge


AATTMMOOSS

AATTMMOOSS


• motivation









Contens


AATTMMOOSS

AATTMMOOSS


• attenuation can be estimated via the specific differential phase Kdp:

X-band scattering computation using measured drop-size distributions(by 2D-video disdrometer) and several raindrop-shape models

• rule of thumb for S-, C- and X-band:whenever microwave attenuation due to rain is substantial, the differential phase accumulation is significant enough that Kdp can be estimated

αhh specific one-way attenuation at horizontal polarisation (dB km-1)

αh-v differential attenuation (dB km-1), i.e. αh-v=αhh- αvv

Estimation of attenuation


AATTMMOOSS

AATTMMOOSS


• a more complex attenuation correction method relies on the determination of the path-integrated attenuation (PIA), e.g. by

• differential phase (no estimation of Kdp required),• power measurement of a fixed target at far range (ground clutter), …

• the PIA is distributed over the range bins weighted by the reflectivity

bzaα

0.1

0.11

' 10 1

: 10 1 :

b b PIAn

n b PIAN n N

z rr

I r r I r r

: 0.46 'N

n

rb

n N nr r

I r r b z r dr

α specific one-way attenuation (dB km-1)

z reflectivity in linear units (mm6m-3)z′ attenuated reflectivity (mm6m-3)

PIA (dB)

Estimation of attenuation


AATTMMOOSS

AATTMMOOSS


• motivation









Contents


AATTMMOOSS

AATTMMOOSS


2011-09-10 19:45:19UTC, az. 324.4 deg

Goal is the estimation of the slope of the differential propagation phase Kdp.

1

1

2 ( )nr

dp n co n dpr r

r r K r dr

most likely differentialbackscatter phase


Differential phase processing


AATTMMOOSS

AATTMMOOSS


2011-09-10 19:45:19UTC, az. 324.4 deg


Most common method:Linear regression with a runningwindow length of about 1-3km.

Disadvantage:• leads to negative Kdp in the presence

of differential backscatter phase• reduced range resolution of the

resulting Kdp

• Kdp peaks are underestimated



AATTMMOOSS

AATTMMOOSS


2011-09-10 19:45:19UTC, az. 324.4 deg


• the difference of Ψdp between the ranges ra and rb can be distributed among the range bins including a weighting with the reflectivity zhh and the differential reflectivity zdr

with

• the differential reflectivity is closely related to the backscatter phase,

0.69 0.42

0.69 0.42

range

12dp n dp

hh n dr n

hh dr

K r wr

z r z rw

z z

(coefficients valid for rain, X-band, zhh and zdr in linear units)

brar

ΔΨdp = Ψdp(rb) – Ψdp(ra)

X-band scattering computations based onraindrop-size distributions measured by a disdrometer

Differential phase processingAATTMMOOSS


AATTMMOOSS

AATTMMOOSS


2011-09-10 19:45:19UTC, az. 324.4 deg


• the difference of Ψdp between the ranges ra and rb can be distributed among the range bins including a weighting with the reflectivity zhh and the differential reflectivity zdr

with

• the differential reflectivity is closely related to the backscatter phase,ra and rb can be chosen such thatZdr(rb) - Zdr(ra) 0, therefore δco(rb) - δco(ra) 0,

in this case, ΔΨdp is due to the differential propagation phase only.

0.69 0.42

0.69 0.42

range

12dp n dp

hh n dr n

hh dr

K r wr

z r z rw

z z

(coefficients valid for rain, X-band, zhh and zdr in linear units)

brar

ΔΨdp = Ψdp(rb) – Ψdp(ra)



AATTMMOOSS

AATTMMOOSS


attenuated reflectivity (dBZ)

differential phase (deg)

attenuated differential reflectivity (dB)corrected reflectivity (dBZ) corrected differential reflectivity (dB)

specific differential phase (deg km-1) differential backscatter phase (deg)

The separation of the forward- and backward-scattering components is crucial at X-band.

Only after a separation of both components, the data can be further processed and analysed (rainfall rate retrieval, hydrometeor classification).


X-band challenge


AATTMMOOSS

AATTMMOOSS


• motivation









Contents


AATTMMOOSS

AATTMMOOSS


The weather radar measurements are connected via the raindrop-size distribution (RDSD)to meteorological parameters such as liquid water content or rainfall rate.

• Raindrop-size distribution normalised with respect to the liquid water content:

Nw .. concentration parameterD0 .. median volume diameterµ .. shape parameter

• for simplicity, often µ = 0 is assumed such that the RDSD becomes a two-parameter exponential distribution

0-(3.67 )

0

( ) ( ) eD

µD

wDN D N f µD

4

4

6 (3.67 )( )3.67 ( 4)

µµf µµ

Raindrop-size distribution


AATTMMOOSS

AATTMMOOSS


4

18 6 1825

10 ( ) 10 ( )D D

z D N D dD D N D dDK

• reflectivity (mm6m-3)

9 3LWC 10 ( )6 D

D N D dD • liquid water content (mm3m-3)

6 33.6 10 ( ) ( )6 D

R D v D N D dD • rainfall rate (mm h-1)

terminal fall velocity (m s-1)

raindrop volume

• specific differential phase (deg km-1) 318010 ( ) ( ) ( )dp hh vvD

K f D f D N D dD

valid for Rayleigh scattering

wavelength

dielectric factorradar coss-section

forward-scattering amplitudes

Meteorological parameters:

Polarimetric weather radar measurements:

Raindrop-size distribution


AATTMMOOSS

AATTMMOOSS


• motivation









Contents


AATTMMOOSS

AATTMMOOSS


X-band scattering computations based onraindrop-size distribution measured by a disdrometer

reflectivity zhh

specific differentialphase Kdp

Variability due to:• raindrop-size distribution

numeric example assuming Rayleigh scattering

A fixed parameterisation of Z-R / Kdp-R relations leads to uncertainties due to the natural variability of rainfall.

• raindrop shape (Kdp)

Note:• Kdp can be estimated up to ~0.1 deg,

only useful for instantaneous rainfall rates larger than ~3 mmh-1 at X-band

• Kdp – based rainfall rate estimates tend to be more accurate also due to its independence from radar calibration and signal attenuation

raindropdiameter #/m3 Z water volume

per cubic meter1 mm 4096 36 dBZ 2144.6 mm3

4 mm 1 36 dBZ 33.5 mm3

Logarithmic scale.Z-R / Kdp-R relations are not linear!

Rainfall rate estimation


AATTMMOOSS

AATTMMOOSS


Data processing and rainfall rate estimation of the TU Delft polarimetric X-band radar IDRA:

• spectral clutter suppression [1]

• estimation of the specific differential phase Kdp [2](reflectivity-weighted to overcome the coarse range-resolution of conventional Kdp estimators, the estimated Kdp is unaffacted by signal attenuation and independent of the radar calibration)

• estimation of the one-way specific attenuation by αhh = 0.34∙Kdp with αhh (dB km-1) and Kdp (deg km-1) and attenuation correction of the reflectivity

• the parametrisations for the rainfall rate estimation are based on 41530 raindrop-size distributions measured by a 2D-video disdrometer data at Cabauw (Netherlands) in 2009:

• zhh = 243∙R1.24 with the rainfall rate R (mm h-1) and the reflectivity at horizontal polarisation zhh (mm6 m-3)

• R = 13∙Kdp0.75 with the rainfall rate R (mm h-1) and the one-way specific differential phase Kdp (deg km-1)

• for the final rainfall rate product, R(Kdp) is chosen if the reflectivity is above 30 dBZ, and the standard deviation of Kdp is below 2 deg km-1, else R(zhh) is used

[1] C. Unal, 2009: Spectral Polarimetric Radar Clutter Suppression to Enhance Atmospheric Echoes,J. Atmos. Oceanic Technol., 26, 1781–1797.

[2] T. Otto and H.W.J. Russchenberg, 2011: Estimation of Specific Differential Phase andDifferential Backscatter Phase from Polarimetric Weather Radar Measurements of Rain,IEEE Geosci. Remote Sens. Lett., 8, 988-992.



AATTMMOOSS

AATTMMOOSS


corrected differential reflectivity (dB)corrected reflectivity (dBZ)

specific differential phase (deg km-1) differential backscatter phase (deg)


rainfall rate estimate (mm h-1)



AATTMMOOSS

AATTMMOOSS


• motivation









Contents


AATTMMOOSS

AATTMMOOSS


Hydrometeor classification• the hydrometeors (snow, ice, rain, hail) show different polarimetric signatures a classification is possible and can improve rainfall rate estimation

Adaptive clutter suppression• robust suppression of clutter (ground targets, birds, planes) is possible taking

advantage of the different polarimetric signatures see next presentation by Christine Unal

Raindrop-size distribution retrieval• the polarimetric parameters can be combined to estimate the parameters of

the raindrop-size distribution and to improve the rainfall rate estimation

Further applications of radar polarimetry


AATTMMOOSS

AATTMMOOSS


• motivation









Contents


AATTMMOOSS

AATTMMOOSS


• major limitation of X-band weather radar systems is attenuation in heavy rain / wet hail:

• if the purpose of an X-band radar is the observation of heavy precipitation:• instead of using a single X-band radar, use a network of X-band radars, or• complement the X-band radar measurements with measurements of the

operational weather radar network (S- or C-band observations).

ΔΨ = 180 deg, that correspondsto ~60 dB round-trip attenuationover 8 km distance!


Limitations of X-band radar


AATTMMOOSS

AATTMMOOSS


• motivation









Contents


AATTMMOOSS

AATTMMOOSS


• most commercially available polarimetric weather radars do not employ the alternate polarisation mode, instead they use the “simultaneous H/V mode”: simultaneous transmission of a horizontally and a vertically polarised wave with equal amplitude they will combine dependening on their phase offset to an elliptically polarised wave the radar measures a combination of co- and cross-polarised scattering matrix components:

only in case of very low cross-polarisation!

Advantages• no need of a high-power ferrite switch• double unambiguous Doppler velocity interval

Disadvantages• very demanding requirements on the radar cross-polarisation isolation• depolarisation in the melting layer / ice clouds will deteriorate the measurements• reduced accuracy of polarimetric weather radar measurements due to cross-pol component• no measurement of the linear depolarisation ratio, instead cross-correlation coefficent • loss of 3dB in sensitivity compared to alternate mode because the transmit power is split

equally over the H and V transmit channel

s i i ih hh h hv v hh h

s i i iv vh h vv v vv v

E S E S E S E

E S E S E S E

Simultaneous H/V mode


AATTMMOOSS

AATTMMOOSS


• important antenna specifications for polarimetric weather radars are• high resolution in azimuth and elevation, i.e. pencil beam (large directional gain),• ideally equal specifications for horizontal and vertical polarisation

(e.g. matched co-polarised beam patterns, S-parameters),• low cross-polarisation levels.

• usually parabolic reflector antennas are employed by polarimetric weather radars

• there is some on-going research in order to use phased-array antennas,e.g. by the Engineering Research Center for Collaborative Adaptive of the Atmosphere (CASA, USA)[1]:

• 64 T/R modules with 1.25W transmit power each• electronic phase steering in azimuth (±45 deg) and mechanical steering in elevation• elevation beamwidth of 2.8 deg, azimuth beamwidth of 1.8 deg – 2.4 deg• alternate polarisation mode due to limited cross-polarisation isolation

[1] J.L. Salazar, E.J. Knapp and D.J. McLaughlin, 2010: Dual-polarization performance of the phase-tilt antenna array in a CASA dense network radar, Geoscience and Remote Sensing Symposium, IGARSS 2010, 3470-3473.

Phased-array antennas


AATTMMOOSS

AATTMMOOSS


• first commercial systems are on the market that use solid-state transmitter instead of the traditionally used high-power microwave sources:

• long lifetime• compact, no high-power microwave circuits (waveguides etc.)• combined with an arbitrary waveform generator (e.g. direct digital-synthesizer), high

flexibility of the transmitted waveform software-defined radar• to retain the sensitivity of such systems, pulse-compression is employed

• e.g. alternate transmission of a modulated long pulse (~50 µs) for far-range measurements and a short pulse (~1 µs) for close-range measurements

time

Txlong Rxlong Txshort Rxshort Txlong Rxlong

far-range measurement close-rangemeasurement

combination

Solid-state transmitter


AATTMMOOSS

AATTMMOOSS


IDRA is mounted on top of the 213 m high meteorological tower.

CESA

R – C

abau

w Ex

perim

enta

l Site

for A

tmos

pher

ic Re

sear

chSpecifications• 9.475 GHz central frequency• FMCW with sawtooth modulation• transmitting alternately horizontal and vertical

polarisation, receiving simultaneously the co- and the cross-polarised component

• 20 W transmission power• 102.4 µs – 3276.8 µs sweep time• 2.5 MHz – 50 MHz Tx bandwidth• 60 m – 3 m range resolution• 1.8° antenna half-power beamwidth

ReferenceJ. Figueras i Ventura: “Design of a High Resolution X-band Doppler Polarimetric Weather Radar”, PhD Thesis, TU Delft, 2009. (online available at http://repository.tudelft.nl)

Near real-time display:http://ftp.tudelft.nl/TUDelft/irctr-rse/idra

Processed and raw data available at:http://data.3tu.nl/repository/collection:cabauw

TU Delft X-band weather radar: IDRA


AATTMMOOSS

AATTMMOOSS


• motivation









Contents


AATTMMOOSS

AATTMMOOSS


Advances in polarimetric X-band weather radar

Tobias Otto

e-mail [email protected]

web http://atmos.weblog.tudelft.nl

radar data http://data.3tu.nl/repository/collection:cabauw

references R. E. Rinehart, “Radar for Meteorologists”,Rinehart Publications, 5th edition, 2010.

V. N. Bringi and V. Chandrasekar, “Polarimetric Doppler Weather Radar: Principles and Applications”, Cambridge University Press, 1st edition, 2001.

R. J. Doviak and D. S. Zrnić, “Doppler Radar and Weather Observations”, Academic Press, 2nd edition, 1993.