The development of GRAPES_RAFS and its applications Xu Zhifang Hao Min Zhu Lijuan Gong Jiangdong Chen Dehui National Meteorological Center, CMA Wan Qilin.

The development of GRAPES_RAFS and its applications

Xu Zhifang Hao Min Zhu Lijuan Gong Jiangdong Chen DehuiNational Meteorological Center, CMA

Wan Qilin

Guangzhou Institute of Tropical and Marine Meteorology, CMA

(24 October, 2011, for Workshop-NWP Nowcasting in Boulder-USA)

Outline• 1 Introduction• 2 RAFS Based on GRAPES_Meso• 3 Results of some experiments• 4 Plans for future work

1 Introduction

The requirements are increasing in short-term forecasts of the severe high impact weather which occurred more frequently in recent years in China.

1 IntroductionSever thunderstorm in He-nan ( 河南 ) on June, 2009

Mud-rock flow on August 07, 2010 in Zhou-qu ( 舟曲 )

1 Introduction (cont.)

• Many data sets from different obs. systems (AWS, aircraft, GPS, radars, satellites) have been frequently (hourly or shorter) are available in real time.

(Meng Zhaolin ， 2011)

2 RAFS Based on GRAPES_Meso

Flow Chart of Grapes_RAFS

Data processing

ObservationDATA A

ssimilation

C

ycle

FirstguessForecast

Digital Filter initialization

DFI

GRAPES_ MODELGRAPES-Model

background

GRAPES_ 3DVARGRAPES-3DVar

Cloud Analysis

nudging

Data Assimilation

– Radio sonde data (wind, temperature, humidity, pressure)

– AWS data (pressure)– VAD winds (Doppler Radar)– GPS/PW– Ship reports (pressure)– Aircraft (wind, temperature)– FY-2C/2D cloud-drift winds – Cloud Analysis (Radar , Satellite data,

surface observation et al.)

Data can be used in GRAPES_3DVAR :

3 Results of some experiments

3.1 Design of the experiments

– GRAPES_Meso: 15km L31– GRAPES_3DVAR (model grid space)– 1-hourly cycle rapid analysis– 12 hour forecast at

03,06,09,15,18,21UTC– 24 hour forecast at 00,12UTC– Data used: GTS, local radiosonde,

Doppler radar VAD, AWS, Aircraft

3.1 Design of the experiments

VADRadiosonde

SYNOPAIREP

Horizontal distribution of one case

3.2 Verification

left ： observation (radar reflectivity)

right ： model forecasting （ Initial time ： 0300GMT 3 June, 2009 ）

The position prediction of strong convective system is very close to the observation.

The moving direction of the system is close to observation

left ： observation (radar composite reflectivity)

right ： model forecasting （ Initial time ： 0600GMT 3 June, 2009 ）

2009 6- 8 00- 06 6 TS TS Score年 月 全国区降水 小时 评分

0

0. 05

0. 1

0. 15

0. 2

0. 25

0. 3

0. 35

0. 4

0. 45

6月 7月 8月months

RUC_小雨OPT_小雨RUC_中雨OPT_中雨RUC_大雨OPT_大雨RUC_暴雨OPT_暴雨

2009 6- 8 12- 18 6 TS TS Score年 月 全国区降水 小时 评分

0

0. 05

0. 1

0. 15

0. 2

0. 25

0. 3

0. 35

0. 4

0. 45

6月 7月 8月months

RUC_小雨OPT_小雨RUC_中雨OPT_中雨RUC_大雨OPT_大雨RUC_暴雨OPT_暴雨

Initial Time ： 0000GMT Initial Time ： 1200 GMT

(TS score) (TS score)

TS-Verification of 6h accumulated precipitation over whole Chinafor the period of June-August 2009

Inter-comparison between GRAPES_RAFS (RUC) with operational GRAPES_Meso (OPT): TS of RAFS is better than that of OPT for all thresholds: 0.5mm, 5mm, 10mm, above 15mm.

2009 6-8 24 TS年 月 小时降水 评分

0

0. 1

0. 2

0. 3

0. 4

0. 5

0. 6

0. 7

6月 7月 8月

RUC_小雨OPT_小雨RUC_中雨OPT_中雨RUC_大雨OPT_大雨RUC_暴雨OPT_暴雨

TS-Verification of 24h accumulated precipitation over whole China

for the period of June-August 2009

Inter-comparison between GRAPES_RAFS (RUC) with operational GRAPES_Meso (OPT): TS of RAFS is better than that of OPT for all thresholds: 1mm, 10mm, 25mm, above 50mm.

1mm

10mm

25mm

50mm

(TS score)

(Wang Yu ,2007)Sub-domains for verifications

1 North-Eastern

2 Xinjiang

3 East of West-Northern

4 North of China

7 M. & D. basins of Yangtz River

8 South of China

5 Tibetan Plateau

6 East o

f West-

So

uth

ern R

.

TS-Verification of 6h accumulated precipitation forecasts over whole China for the period of June-August 2011: RAFS vs OPT.

1 2 3 4 5 6 7 81 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8

Threat ScoreBias

Threat Score Bias

3.3 Recent updates

(1)

Old vertical correlation: too big from bottom to top

V. Potential

HumidityS. function

Pressure

new vertical correlation: reduced significantly

S. function

Humidity

Pressure

V. potential

Red: with old background covariance ; Grey: with new background covariance

0.1 10 25 50 100 (mm)

0.1 10 25 50 100 (mm)

0.1 10 25 50 100 (mm)

0.1 10 25 50 100 (mm)

TS-Verification of 6h accumulated precipitation over whole ChinaPeriod: 072009

I. Time: 06Z

Period: 072009

I. Time: 18Z

Period: 082009

I. Time: 06ZPeriod: 082009

I. Time: 18Z

Red: with old background covariance ; Grey: with new back-ground covariance

0.1 10 25 50 100 (mm)

0.1 10 25 50 100 (mm)

TS-Verification of 24h accumulated precipitation over whole China for August (left), July (right) 2009

Configurations• GRAPES_Meso 3.0

15km L31 with m-top at 10 hPa•Radiation: no-change•Cumulus: Betts-Miller-Janjic•Microphy.: WSM-6 (soon by 2-momment (Liu, 2010))•Cloud: no-change•Land surface: NOAH •PBL: no-change•q-adv.: PRM(Xiao, 2002)

• GRAPES_Meso 2.5

15km L31 with m-top at 10 hPa•Radiation: RRTM LW & Dudhia SW•Cumulus: SAS•Microphy.: WSM-6•Cloud: Xu & Randall diagnostic cloud•Land surface: SLAB•PBL: MRF PBL•q-adv.: QMSL

Grey: with GRAPES_V2.5; Pink: with GRAPES_V3.0

0.1 10 25 50 100 (mm) 0.1 10 25 50 100

(mm)

0.1 10 25 50 100 (mm)

0.1 10 25 50 100 (mm)

TS-Verification of 6h accumulated precipitation over whole ChinaPeriod: 072009

I. Time: 06Z

Period: 072009

I. Time: 18Z

Period: 082009

I. Time: 06Z

Period: 082009

I. Time: 18Z

Grey: with GRAPES_V2.5; Pink: with GRAPES _V3.0

0.1 10 25 50 100 (mm)

0.1 10 25 50 100 (mm)

TS-Verification of 24h accumulated precipitation over whole China for July (left), August (right) 2009

4 Plans for future work

4 Plans for future work

To Improve QC scheme and assimilation approach of local intensified observation.

To make better use of satellite data. To perform the cloud analysis. To improve boundary layer and cloud

microphysical process in GRAPES model. To add 1DVAR assimilation of precipitations

Thanks for your attention