Download - Doppler Weather Radar Calibration Final

Doppler Weather Radar CalibrationG.VISWANATHAN

Director(Rtd)[email protected]

DWR CalibrationHow DWR works & what constitute the radar dataMeasurement Accuracy, Precision ,BiasRadar Hardware Calibration

Antenna,Rx Sensitivity, Tx Power,Signal ProcessingCalibration vs ValidationZ estimate ValidationDoppler Velocity ValidationInter comparison of data – Radar-Radar, Radar-

GaugeNetworked Radars – Cal/Val IssuesDual Pol Radar CalibrationWhat IMD needs to do

RADAR TARGETS

HARD TARGET SOFT TARGET

Distinct reflecting targets small enough not to fill the beam (Point target)

Distributed targets like hydro meteors or turbulent medium which fills the beam.(Volume reflectivity)

R

Ct/2

C /2

3.R

Radar Resolution Elements Range : C /2Angle : 3.R

Doppler: 1/ T, where T is Integration Time

1 2 3Dr=c/2fs IPP=T=1/prf

SPATIO TEMPORAL SAMPLING OF DYNAMIC WEATHER PHENOMENABY DWR IN A HEMISPHERICAL VOLUME OF SPACE AROUND RADAR

Dr 3q xR3q xR= Dv

Radar DataRaw Data @ I/Q level, or Signal power received

Base Data – or – Moment Data (Z,V, )s

Derived Products, like Rain Rate(R)

H Pol radars depend on emperical exponential, Z-R relationship, further compounded by DSD, Climatology etc.

BASE PRODUCTS OF THE DWR ARE THE THREE MOMENTS OF THE TIME SERIES THUS GENERATED- Z = Precipitation Intensity is the area under the curve of Doppler Velocity Spectrum for an incremental volume of space as seen by the

Radar resolution elements,D v=( q3 R)2(DR) & filled with Hydrometeors and

is estimated from the 0th moment of the Time Series generated by the radar.

VD= Mean Doppler Velocity of the Hydrometeors in the resolution volume &

is estimated from the First Moment of the Time Series.

= s Spectrum Width of Doppler Spectrum & is estimated from the Second Momentof the Time Series.

whereq3 = Beam widthDR = Range resolutionR = Range to the volume element

ISRO RADAR DEVELOPMENT UNIT, BANGALORE.

sZ

V

ISRAD

Accuracy, Precision, Resolution

*** *** ***

*** *** ***

* * *

* * ** * *

Neither precise nor accurate

Precise but not accurate

Accurate but not precise

Accurate & Precise

*** *** ***

*

***

Measurement AccuracyAccuracy

Frames of Reference for Radar

Tropo Centric Frame

Xg

Yg

Zg

XtYt

, , h

OriginEarth Centre

Geocentric Frame

r = Slant range or radial distance (Kms) = Elevation angle [measured from local horizon (mils)] = Azimuth Angle [measured clock wise from North (mils)]

Z

NX

Y

r

= El

= Az

Target positionP(r, , ) @ t

T

Radar Data Array

Elevation(-2 to+30deg)Azimuth for Each Elevation (0-

360deg)Range ( in terms of Range Bin) for

each AzimuthData values(say Z,V,) for each Range

Bin

DWR Hardware Calibration

Weather Radar Equation relates the signal power received in terms of the System level parameters of the radar and the backscattering cross section of the target, in this case the hydrometeors in a cloud.

It can be written in a simpler form as follows:Sr = KZ/(R*R)

Where K is known as the Radar Constant Z is precipitation intensity in sample volume

R is Slant range to Target in a given Az & El

IMD-ISRO CAL-VAL EXPERIMENT

Lat: 13.374

Lon: 80.258

Lat: 13.664

Lon: 80.227

Lat: 13.083

Lon: 80.289

13

13.1

13.2

13.3

13.4

13.5

13.6

13.7

80.00 80.10 80.20 80.30 80.40 80.50

Lat

itu

de(

deg

)

Longitude(deg)

Shar

Lat: 13.374

Lon: 80.258

Lat: 13.664

Lon: 80.227

Lat: 13.083

Lon: 80.289

CDR


: Two radar beams intersecting. Here red and blue lines mark the range bins of radar A and B respectively. Note that the difference in the sizes of range bins between two radars leads to different sampling volumes in the overlapping volume.


feed

DWR at SDSC SHAR

Az RJ, Cable, RF

Amplifier

Base band Down Conversion

Cable,

Az RJ

TRLDirectional.

Coupler

Down

Converter

LNAIFA & FILTER

10 MHz ref Pulse/CW

2 x ADC I

Q

I

Q

RF source

2.7-2.9 GHz

Modulator Variable

Attenuator

RF FRONT ENDIF RECEIVER

EXT.SIMULATOR

PIN

DSP dBm (POUT ) / dBZ

Figure 2. Experimental setup at DWR-SHAR for calibrating receive chain dynamic range.

IMD-ISRO CAL-VAL EXPERIMENTCDR, SHAR - LP

-10

0

10

20

30

40

50

60

70

80

90

100

-120 -110 -100 -90 -80 -70 -60 -50 -40 -30 -20 -10 0

Input power(dBm)

Re

fle

cti

vit

y(d

BZ

)

DWR, SHAR CDR, Chennai


The radius of the metal sphere considered in Spherecal is 15 cms. Radar cross section of this metal sphere falls close to optical region as shown in the calculations below:

2 r/ = 9.01 & / r2 = 0.99


. Launching of metal sphere from SHAR for spherecal experiment. Also shown Is the video output of metal sphere. Upper trace is log video and lower trace is linear output


Table 2. Summary of Spherecal experiment conducted on 4-4-2007 for Pulse width = 2 s S. No

Description DWR, SHAR CDR, Chennai

1 Radar Constant in use

-7.53 dB -7.0 dB

2 Radar Constant measured from Spherecal expt.

-6.66 dB -4.56 dB

3 Difference between measured and estimated

+0.87 dB +2.44 dB

Remarks

Average value from both the spheres

Value corresponding to 2nd sphere only


Table – 1. Comparison of antenna gain using Sun calibration

DWR System Location

Expected Antenna Gain

Measured Antenna Gain

Difference (measured – expected)

SHAR 45.0 dB 44.1 dB -0.91 dB

Chennai 44.23 dB 45.6 dB + 1.37 dB

Joint CAL-VAL Committee of IMD-ISRO

Prof.G.S,Bhat-CAOS/IISc ChairmanMr.SK.Banerji- IMD HQ MemberMr. B.Thampi – IMD© MemberMr.SK.Kundu - IMD HQ MemberMr.S.Venkateswarlu-IMD MemberDr.B.Manikiam-ISRO HQ MemberDr.GV.Rama- ISRO SHAR MemberMr.V.Rama Rao – ISRAD MemberMr.S.Rajendran - ISRAD MemberMr.R.Ranga Rao –ISRAD Member-SecretaryData Analysis ably assisted by Mrs.Anitha

Daniel,Subbu,&Vikas



CDR-2006_10_28_16_30_29_72

CDR-2006_10_28_16_30_29_72

Comparison of PPI-Z @ Elevation 1.0SHAR-2006_10_28_16_31_10

SHAR22

PPI (Z) comparison of SHAR and Chennai DWRs before applying the corrections.

IMD-ISRO CAL-VAL EXPERIMENTComparison of PPI-Z @ Ele 1

2006_10_28_19_36_28

CDR-2006_10_28_19_34_01

Figure 8. c) PPI (Z) comparison of SHAR and Chennai DWRs before applying the correction.


Figure 8. d) PPI (Z) comparison of SHAR and Chennai DWRs after applying the correction.

Comparison of PPI-Z @ Ele 1 after added bias +5dB to SHAR

2006_10_28_19_36_28

CDR-2006_10_28_19_34_01

SHAR25


10/2

8/06

14:

24

10/2

8/06

19:

12

10/2

9/06

0:0

0

10/2

9/06

4:4

8

10/2

9/06

9:3

6

10/2

9/06

14:

24

10/2

9/06

19:

12

10/3

0/06

0:0

0

-30

-20

-10

0

10

20

30

40

50

60

Intersection point @ 13.374,80.258 after correction

Time(UT)

Z(d

Bz)


10/28/06 14:24 10/28/06 19:12 10/29/06 0:00 10/29/06 4:48 10/29/06 9:36 10/29/06 14:24 10/29/06 19:12 10/30/06 0:00-20

-10

0

10

20

30

40

50

60

Middle [email protected],80.30 after correction

Time(UT)

Z(d

Bz)

Scatter Figure 11. a) Scatter plot of cloud reflectivity of SHAR and Chennai DWRs without correction.

Figure 11. b) Scatter plot of cloud reflectivity of SHAR and Chennai DWRs with correction (added +2dBz to SHAR and –3dBz to Chennai data).

Subsequently an inter comparison between SHAR DWR and TRMM – (Joint PR-TMI )revealed that SHAR – DWR makes an under estimate of ~5-6dB

Most likely cause is the difference in Antenna gain as measured in Test range(Without Bends& twists) & the effective gain after field installation, namely plumbing losses)

In the production version of the radar WG plumbing is optimized by welding /brazing WG components with out using standard flanges etc.



Figure 11. a) Scatter plot of cloud reflectivity of SHAR and Chennai DWRs without correction.

Figure 11. b) Scatter plot of cloud reflectivity of SHAR and Chennai DWRs with correction (added +2dBz to SHAR and –3dBz to Chennai data).

Scatter plot_without correction

y = 0.8802x + 5.0463

0

5

10

15

20

25

30

35

40

45

50

55

60

0 5 10 15 20 25 30 35 40 45 50 55 60

SHAR(Z)

CD

R(Z

)

Scatter plot_withcorrectionadded 2dBz to SHAR,added -3dBz to CDR

y = 0.8802x + 0.286

0

5

10

15

20

25

30

35

40

45

50

55

60

0 5 10 15 20 25 30 35 40 45 50 55 60

SHAR(Z)

CD

R(Z

)

IMD-ISRO CAL-VAL EXPERIMENTRain Acc. comparison between radar and radar

0.00

20.00

40.00

60.00

80.00

100.00

120.00

80.26 80.10 80.20 80.30 80.40 80.05 80.08 80.35 80.15

13.374 13.358 13.368 13.378 13.389 13.353 13.356 13.383 13.363

Middle points

Rain

Acc.(

mm

/day)

DWR-SHAR DWR-CDR

Figure 12. Histogram of rain accumulation of both radars at equidistant locations.

IMD-ISRO CAL-VAL EXPERIMENTRain Acc. comparison between gauge and radar

0

20

40

60

80

100

120

140

160

180

Sulurpet Gudur Nellore Nungambakkam

Gauge locations

Rain

Acc.(

mm

/day)

Raingauge DWR-SHAR DWR-CDR

Figure 13. Histogram of daily rainfall measured by rain gauges and DWRs.

4. Conclusions For the Committee Members, visit to the

radar sites and seeing DWRs in operation was an unforgettable experience. The radar calibration experiments could be successfully carried out because IMD and ISRAD teams worked together as one unit with lot of understanding and full cooperation. This spirit should continue. Some important conclusions from calibrations carried out are as follows.


IMD-ISRO CAL-VAL EXPERIMENTSHAR_2006_10_29_05_35_28 CDR_2006_10_29_05_33_55

Figure 14.a) Radar to radar PPI (V) image comparison.

(Note: The cyclone crossed Nellore. Since the cyclone is north of both the DWRs similar velocity patterns are seen by both the DWRs.

IMD-ISRO CAL-VAL EXPERIMENT The response of the receive chain sub-system of both the DWRs

(forward of the LNA input point) is linear over a dynamic range of ~100 dB. The repeatability is excellent. The performance of this sub-system is as per the design specifications.

The measured radar constant based on metal sphere calibration differs from the values being used at present by +1.57 dB and +0.87 dB respectively for Chennai and SHAR DWRs. These numbers are based on the maximum value of Z measured during Spherecal experiment at a given range of the sphere.

Suncal results show that the measured antenna gain differs from the values being used at present by +1.4 dB and -0.9 dB respectively for the Chennai and SHAR DWRs. The measurement uncertainty in Suncal experiment is ±1 dB, and the observed differences are within or close to this limi

Radar derived daily rain accumulations are consistently lower than that measured by the rain gauges. The Z-R relationships used were developed for the mid-latitude systems. Perhaps, these relations need re-evaluation for the Indian cloud systems.

The intercomparison of Z products in areas that are almost equi-distant from two radars and in volumes having some overlap show that there is a difference of about 4 to 6 dB between the two radars with Chennai values being higher. Even among radars of the same make, differences of 2-3 dB are common.

Essentiality of CAL-VAL for IMD All networked DWR’s to be Calibrated

periodically as per well laid out procedure.Sun Cal to be once in 3 months & sphere Cal

every 6 monthsEach DWR to be supported by a network of fast

response TM Rain GaugesAtleast Two Distrometers to be part of Rain

Gauge networkIMD to organize a DWR CAL-VAL workshop

once a year, with participation of Designers, IMD,ISRO, Mnfrs, Modelers & Researchers in Radar meteorology .

Database for each Network radar on CAL-VAL to be archived.

Seasonal Differences in DSD at Gadanki:

Rao et al. 2009, QJRMS

Radhakrishna et al. 2009, JGR

THANK YOU