GV for the Evaluation of High Resolution Precipitation Products using WPMM in Korea J.C. Nam 1, K.Y. Nam 1, G.H. Ryu 2, and B.J. Sohn 2 1 Korea Meteorological.

GV for the Evaluation of High Resolution Precipitation Products using WPMM

in Korea

GV for the Evaluation of High Resolution Precipitation Products using WPMM

in Korea

J.C. Nam1, K.Y. Nam1, G.H. Ryu2, and B.J. Sohn2

1 Korea Meteorological Administration(KMA)2 Seoul National University (SNU)

J.C. Nam1, K.Y. Nam1, G.H. Ryu2, and B.J. Sohn2

1 Korea Meteorological Administration(KMA)2 Seoul National University (SNU)

The 2nd GPM-GV Workshop, 27-30, September 2005, TaiwanThe 2nd GPM-GV Workshop, 27-30, September 2005, Taiwan

Remote Sensing Research Lab.

Ground Observation Networks of KMA - Automatic Weather Station(AWS) Network - Radar Network - Haenam Supersite

Window Probability Matching Method (WPMM)

Evaluation Precipitation Products using WPMM

Comparison precipitation products in space and time

Concluding Remarks

ContentsContents


ConfidenceGPM

Product

II. KEOP Supersites (Hae Nam) Micro rain radar Autosonde Wind profiler Optical rainguage Meteorological tower(30m)

I. Basic Rainfall Validation Raingauges(536 AWS) Radar(10) Radiometer(2)

RainfallRetrieva

lGPM

SatelliteData

Potential Applications

Weather prediction

Water management

Severe weather monitoring

Flood prediction

Cloud microphysics

Cloud-radiation modeling

Climate research

Understanding weather

phenomena

Agriculture III. Data Assimilation

GV

Basic Science Modeling

Physics, etc.

Calibration

Calibration

Improve Retrieval

Algorithms

Schematic Diagram of Korea GPM(K-GPM)Schematic Diagram of Korea GPM(K-GPM)


Observation NetworkConventional Station

Automatic Weather Station

No.

Obs.

No. of station

No. of daily observation

Surface 74 8-24

Upper-air 3 2-4

Aeronautical 9 24-48

No.

Obs.No. of station

No. of daily observation

Measured elements

AWS 536Every min.

(Continuous)Temp., Wind,

Preci., etc

Ocean Buoy

5 24Temp., Wind,

Wave, SST, etc.

Ground Observation NetworkGround Observation Network


Automatic Weather Station(AWS) Network

- ASOS(Automated Surface Observing System) : 42 sites

- Manned AWS(Automatic Weather System) : 35 sites - Unmanned AWS(Automatic Weather System) : 459 sites

ASOS AWS Mountain AWS



• Spatial resolution ASOS + AWS network : 13 km

Unmanned AWS network : 14 km

• Temporal resolution : 1 min.

•Data Collection - DSU Modem leased line(9,600 bps)

- DSU Modem + Microwave comm.

- ORBCOMM Satellite comm.



Real Time Data Collection NetworkReal Time Data Collection NetworkTelecommunication Network in KMA


Radar Network of KMARadar Network of KMA

No. of station Time resolution Observation eliments

Operational Weather Radar

10(1) Every 10 min.Reflectivity, Radial Velocity, Spectral Width

Research radar

Muan

Operational radar • 5 C-band radars

•Baekryungdo•Kunsan•Donghae•Cheju•Chungsong

• 4 S-band radars•Gwangduksan•Jindo•Gwannaksan•Pusan

• 1 C-band(Airport)•Incheon

Research radar • 1 X-band radar

•Muan


C-band radar(ROKAF)

S-band radar

Aerosonde(from Australia)

X-band radar

Haenam Special observation site• autosonde for continuous upper air obs.• boundary layer wind profiler• micro rain radar for vertical structure of rain• optical rain gauge for continuous accurate rain rate observation• conventional synoptic weather observation

Ground Observation SupersiteGround Observation Supersite


Heanam Super sites

Understanding of the land-surface hydrological and cloud-precipitation processes in cloud physics and numerical model.

IntensiveObservation

Micro Rain RadarProducing vertical profiles

of rain rate, LWC anddrop size distribution

Flux TowerProducing sensible, latent, and radiative

fluses over land surface

Optical Rain GaugeContinuous accurate rain rate observation.

AutosondeContinuous upper air

observation

Boundary Layer RadarProducing one-minute profile

of vertical and horizontal winds

Produce high resolution temporal and spatial data for the monitoring, analysis and prediction of severeweather phenomena(typhoon, fronts…)

Ground Observation SupersiteGround Observation Supersite



Facility Agency Equipment Status Additional Info

National KMA

536 Rain Gauges, 13 x 13 km 1 minute readout

Operational.1 and .5 mm tipping

bucket

10 Radars (C-, S-band) Operational

10 Wind ProfilersPlanned2004 (2),2005-’08 (8)

HaenamKEOP

Supersite

METRI(KMA)

X-band Doppler Radar OperationalOperating as CEOP Supersite

Autosonde Operational

Boundary Layer Wind Profiler

Operational

Micro Rain Radar Operational

Flux Tower (10 m) Operational

Optical Rain Gauge Operational

Radiometer Operational


Radar-Raingauge data processing - Special resolution : 1 km (Reflectivity, using Radar Software Library) - Temporal resolution: 1 min. (Rain rate, using TRMM/GSP algorithm)

Z-R Relationship - Set the minimum radar reflectivity corresponding with rain gauge (10 dBZ) - Estimation of Z-R relationship from Z-R pairs in real-time

Window Probability Matching Method (WPMM)Window Probability Matching Method (WPMM)

00

)()(RZ

dRRRPdZZZP

Space resol. = 1 kmTime resol. = 1 min.

1x1 km1x1 km


Radar DataRadar Data

NCAR/SPRINTNCAR/SPRINTNCAR/CEDRICNCAR/CEDRIC

2-D Reflectivity data2-D Reflectivity data

Raingauge DataRaingauge Data

Raw data checkRaw data checkTRMM/GSPTRMM/GSP

Rain rateRain rate((mm/h) Datamm/h) Data

Calculated the Z-R relationship using WPMM each radar

Convert Rain Intensities using the real-time Z-R relationship

Composite of all of Radar Intensity (Overlapping Maximum value select)

Data Procedure for WPMMData Procedure for WPMM


Radar Scan strategy and characteristics

Range (km)

0 20 40 60 80 100 120 140 160 180 200 220 240

He

igh

t (k

m)

0

1

2

3

4

5

6

7

8

9

10

0.0

1.0

2.03.04.05.06.07.0

Bright Band

Ground and sea clutter

Jindo, Gwangduksan, Kwanaksan, Pusan (4 S-band Radar)Jindo, Gwangduksan, Kwanaksan, Pusan (4 S-band Radar) Baekyeongdo, Donghae, Kunsan, Cheju, Myeonbongsan, Youngjongdo Baekyeongdo, Donghae, Kunsan, Cheju, Myeonbongsan, Youngjongdo (6 (6 C-band)C-band) 0.0 – 7.00.0 – 7.0° (C-band: 8 elevations ° (C-band: 8 elevations , , interval 10 minutesinterval 10 minutes)) 0.0 – 19.0 ° ( S-band: 10 elevations, interval 10 minutes)0.0 – 19.0 ° ( S-band: 10 elevations, interval 10 minutes) Melting layer level is about Melting layer level is about 3.5 – 5.5 3.5 – 5.5 km from June to Augustkm from June to August

Melting layer height Osan(47122) 00 UTC

Time (month)5 6 7 8 9

He

igh

t (k

m)

3.0

3.5

4.0

4.5

5.0

200020012002


Beam blocking areaBeam blocking area Range attenuation errorRange attenuation error Bright-band contaminationBright-band contamination C-band(5), S-band(4)C-band(5), S-band(4)

NCAR/SPRIINT -Resolution : 1x 1 x 0.5 km(Cartesian)-Height 1.5 – 4.0 km (interval: 0.5 km)

NCAR/CEDRIC-Standard deviation check-Ground clutter check-Beam filling

2-Dimensional Reflectivity Data-Effective reflectivity height eliminated the ground and bright band

Rdar DataRdar Data

1

2

3

4

Hei

ght (

km)

150 170 190 210100 240Range (km)

0 0

Radar Data ProcessingRadar Data Processing

Research

radar

Muan


Precipitation events check (B, rain event ) : if the time interval between Tips less than 30 minutes, B is rain event

Half tip (C) : if the time interval between tips is within 20-30 minutes, insert the half tip in the middle

Calculate the 1-minute rain rate(mm/h) using Cubic Spline Interpolation each rain eventCalculate the bias of measured rainfall and interpolated rainfall

Bias = rainfall(measured) / rainfall(interpolationed)

Rain-rate greater than 1000 mm/hr

single tip or isolated tips (A) -> Gaussian interpolation is applied( R=R0exp(-x2/100) )

Calculated rain-rate = bias * {rain(1st sec. of min.) – rain(2nd sec. of min.} *60TRM

M/G

SP

t

mm

AB

C TRMM/GSP algorithmTRMM/GSP algorithm

Raingauge Data ProcessingRaingauge Data Processing


0-15 15

100/0

2

)( xeRtR

Rai

n-r

ate

(mm

/hr)

Time (min.)

x j-1

Time (min.)

Acc

um

ula

ted

Rai

nfa

ll (

mm

)

x j x j+1

y j-1

y j

y j+1

x

y

21

3

21

3

1

1

''1

''1

))((6

1

))((6

1

1

jj

jj

jj

j

jjjj

xxBBD

xxAAC

ABxx

xxA

DyCyByAyy

Cubic Spline Interpolation - tip interval within 30 minutes, effective data > 3 point - the slope of accumulated precipitation convert to rain-rate

Gaussian Interpolation(Not TRMM/GSP algorithm) - tip interval greater than 30 minutes => Single tip - single tip consider as small convective precipitation

Raingauge Data ProcessingRaingauge Data Processing


Single tip does not accord to the rain-rate of ORG (Optical Raingauge) Rain event accords to the rain-rate of ORG One tip of AWS raingauge is 0.5 mm (rain-rate=30 mm/h)

Haenam(AWS id:261), Date : 8 Jul. 2002

Time (hour)

6 7 8 9 10 11 12 13 14 15 16 17 18

Rai

n-ra

te (

mm

/h)

0

1

2

3

4

5

6

7

8

ORGAWS 1-min.AWS Tip time

TRMM/GSP & ORGTRMM/GSP & ORG


Computation of Z-R pair every 10 minutes during 1 hour : (1) Rain rates of raingauges (- 9 ~ 0 minutes) (2) Reflectivites of radar grids(3×3) around raingauge (horizontal res.: 1 km) (3) Threshold : 10 ~ 60 dBZ, 0.5 mm/h (4) Rain-rate calculates using M – P relationship (if Reiteration Num. = 1)

Computation of rain rates applied Z-R relationship every 10 min. (No precipitation under threshold dBZ)

Z-R Fitting : - Median Fitting ( Under and Over the 30 dBZ )

Z-R pairs > Threshold Num.

Z-R Fitting : - Z-R relationship of former time

Yes No

Data reading : (1) 1-minute rain rate of each raingauge using TRMM-GSP (2) 10-minute reflectivity of each radar using RSL

Produce Rain Rate from each radar-raingauges rain rate

WPMM (Window Probability Matching Method)

Reiteration Num. > 1

YesNo

Rain Rate ProductRain Rate Product


Z-R relationship

Rain-Rate(mm/h)

0.01 0.1 1 10 100

Re

fle

cti

vit

y(d

BZ

)

-10

-5

0

5

10

15

20

25

30

35

40

45

50

55

Stratiform : 200R1.6 Convective : 300R1.4 Thunderstorm : 31R1.71 Oragraphic : 486R1.37

b

adBZb

R

aRbdBZ

aRZ b

log

10

1log

log10log10

Y X )(

6)(

)(

0

3

0

6

dDDVD

DNR

dDDDNZ

t

Reflectivity (dBZ)

5 10 15 20 25 30 35 40 45

Ra

in r

ate

(m

mh-1

)

0

5

10

15

20

Observed data

Z = 20.6R1.6 (median fitting under 30 dBZ)Z = 2.19R3.3 (median fitting over 30 dBZ)M - P relationship

Y=AX +B Linear equation Intersection of Y-axis : 10 loga Slope : 10 b Fitting Method : Median Fit

Fitting example

[Gwangduksan] 1720 LST July 7, 2004.

Z-R fittingZ-R fitting


0

2

4

6

8

10

12

14

16

J indo Gwangduksan Gwannaksan Baereongdo Kunsan Pusan Cheju

Mean E

rror(

mm

/h)

Z=200R1.6(1- hour)

Z=200R1.6(10- minute)

WPMM(1- hour)

WPMM(10- minute)

0

2

4

6

8

10

12

14

16

J indo Gwangduksan Gwanaksan Baereongdo Kunsan Pusan Cheju

Mean E

rror(

mm

/h)

Z=200R1.6(1- hour)

Z=200R1.6(10- minute)

WPMM(1- hour)

WPMM(10- minute)

7 July 2004

18 August 2004

S-band RadarOperation stop(reinstalling)

Sensitivity test for Z-R relationshipSensitivity test for Z-R relationship

S-Band C-Band


19 Jun. 20 Jun.

19 Jun. 20 Jun.

Nowon(407)Corr. = 0.94, RMSE=3.04 mm/h

Dongdaemun (408)Corr. = 0.95, RMSE=3.05 mm/h

Jungrang(409)Corr. = 0.90, RMSE=3.12 mm/h

Dongjak(410)Corr. = 0.97, RMSE=4.4 mm/h

Precipitation Product Comparison Precipitation Product Comparison 19-20 June 2004

TRMM/GSPWPMM


TOTAL RAINFALL [ 19-20 Jun. 2004]

Time(hour)

1 3 5 7 9 11 13 15 17 19 21 23 1 3 5 7 9 11 13 15 17 19 21 23

To

tal R

ain

fall

(mm

)

0

1000

2000

3000

4000

AWS WPMMZ=200R1.6Z=300R1.4TRMM/GSP

19 Jun. 20 Jun.

Comparison Total RainfallComparison Total Rainfall


TOTAL RAINFALL [ 19-20 Jun. 2004]

Time(hour)

1 3 5 7 9 11 13 15 17 19 21 23 1 3 5 7 9 11 13 15 17 19 21 23

BIA

S

0.0

0.5

1.0

1.5

2.0

WPMM Z=200R1.6Z=300R1.6

19 Jun. 20 Jun.

Comparison BAISComparison BAIS


AWS No. 140

Time (KST)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Ra

in R

ate

(mm

/h)

0

5

10

15

20

TRMM-GSPWPMM

CORR = 0.89, RMSE = 2.33, ME = 0.88,TRMM-GSP mean = 3.59, WPMM mean = 4.48

AWS No. 327

Time (KST)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Rai

n R

ate

(mm

/h)

0

2

4

6

8

10

12CORR = 0.85, RMSE = 1.45, ME = 0.20,TRMM-GSP mean = 5.19, WPMM mean = 5.39

18 August 2004[ Time: 1200 – 1400 KST ]

Precipitation Product Comparison Precipitation Product Comparison

TRMM/GSP

WPMM


Radar – AWS differenceRadar – AWS difference Radar rain intensity object analysis (AWS grid point)

AWS rain intensity object analysis (AWS grid poing)

Verification in SpaceVerification in Space


WPMM rain intensity

AWS rain intensity

Z=200R1.6 rain intensity

Verification AreaManned AWS site

-Seoul-Gyungki-Gwangwon-Chungcheong-Jeonla-Gyungsang-Cheju

Verification in TimeVerification in Time


Products• Rain intensity Composite• Each radar rain intensity• Each radar reflectivity• Verification in space• Verification in time series

Products• Rain intensity Composite• Each radar rain intensity• Each radar reflectivity• Verification in space• Verification in time series

http://wpmm.metri.re.kr

Real time Web ServiceReal time Web Service


KMA’s operational Automatic Weather Station Network (13km*13km, one minute) and Weather radar(10 stations) can be used for GPM calibration and validation.

High resolution(1km x 1km) precipitation intensity were estimated from radar reflectivity with the various Z-R relationship obtained by WPMM using raingauges data of AWS operated by Korea Meteorological Administration (KMA).

Rain intensity produced by WPMM has a good agreement with ground rainfall data measured by raingauge and Optical Rain Gauge (ORG).

Rain intensities of S-band and C-band radars obtained by WPMM were more accurate than Z-R relationship (Z=200R1.6) and S-band radars were more accurate than C-band radars.

Concluding RemarksConcluding Remarks

Thank you for your attention !

Thank you for your attention !

GV for the Evaluation of High Resolution Precipitation Products using WPMM in Korea J.C. Nam 1, K.Y. Nam 1, G.H. Ryu 2, and B.J. Sohn 2 1 Korea Meteorological.

Documents

unmanned aws network

temporal resolution

spatial data

xband radarmuanno

data collection dsu

cloudprecipitation processes

minute profileof vertical

landsurface hydrological