A new operational medium-range numerical weather forecast system of CHINA

NWPD/NMC/CMA(Beijing,CHINA)

Organizational Chart of CMACMA

(China MeteorologicalAdministration)

NSMC(National Satellite

MeteorologicalCenter)

NMC(National

Meteorological Center)

CAMS(Chinese AcademyOf Meteorological

Science)

NCC(National Climate Center)

CMO(Central

MeteorologicalObservatory)

NIC(National

InformationCenter)

NWPD(NumericalWeather

Prediction)

WFOD(WeatherForecastOffice)

RSAAMD(Remote SensingApplication and

AgriculturalMeteorology)

Role of NWPD : Research, development and support of themodels meant for operational weather prediction

Operational Numerical Forecast SystemsMain Numerical Forecast Systems

Global Forecast System (T213)Regional Forecast System (HLAFS)Typhoon Track Forecast System (MTTP)Mesoscale Forecast System (MM5)Ensemble Forecast System (T106)

Environmental Meteorology Forecast Systems Based on Main Model Systems

Forest Fire Meteorological Condition Forecast SystemNuclear Contamination Dispersion and Transportation Forecast SystemUltraviolet Index Forecast SystemDust Devil Forecast SystemAir Quality Forecast System

Interactive

Structural Chart of the Operational NWP Systems

Globalassimila-tion

pre-processanalysismodelpost-process

Medium rangeforecasts

Regionalassimila-tion

Typhoonforecasts

Regionalforecasts

Mesoscaleforecasts

L.B.C&

First guess

Remote -files

Grid-pointproducts

Verification

TVservice

Archiving

Ele

men

t dat

a-ba

se

Fiel

d da

ta-b

ase Graphs

Interactive

Put into full operationin place of previous operational system (T106) on September 1st 2002.

Runs twice a dayfrom 00 UTC up to 72h range and from 12 UTC up to 10days range

New Medium-range Forecast System T213

New Medium-range Forecast System T213

Runs on the massive parallel computerthe first operational medium-range weather forecast systemOperational machine: IBM/SP with 88 processors and peak calculation speed up to 80 GFLOPs

New Medium-range Forecast System T213

Contituents of the New SystemT213L31 global spectral modelData preprocessingOptimum Interpolation (OI) analysisPost-processingJob running watchingField database, archiving , forecast verification , product making and distribution

Global data assimilation system flow chart21UTC 03UTC 09UTC 15UTC 21UTC

Element data

Analysis observation file

Analysis observation file

Analysis observation file

Analysis observation file

Analysis observation file

Analysis observation file

Analysis observation file

Analysis observation file

Box highresolution analysis

Box highresolution analysis

Box highresolution analysis

Box highresolution analysis

Box highresolution analysis

Box highresolution analysis

Box highresolution analysis

Box highresolution analysis

Diabaticnormal modeinitialization

Diabaticnormal modeinitialization

Diabaticnormal modeinitialization

Diabaticnormal modeinitialization

Diabaticnormal modeinitialization

Diabaticnormal modeinitialization

Diabaticnormal modeinitialization

Diabaticnormal modeinitialization

T213L31model6h forecast

T213L31model6h forecast

T213L31 model6h forecast

T213L31 model6h forecast

T213L31 model6h forecast

T213L31 model6h forecast

T213L31 model6h forecast

T213L31 model6h forecast

(00UTC) (06UTC) (12UTC) (18UTC)Analysis time

T213L31 model240h forecast

T213L31 model240h forecast

T213L31 model72h forecast

T213L31 model72h forecast

Data assimilationOperational: optimized interpolation upgrading the previous operational 3D-OI with a resolution of T106L19Optimized and Parallelized using MPIUsing 8 CPU of IBM sp10 minutes to complete each assimilationTest running: 3Dvar

Global spectral model T213L31Based on global model contituent of IFSMarch,1998 from ECMWFOptimizing subroutineswith large quantity of computation (FFT, Legendre transformation)Using asynchronous and nonblocking messagepassing increase the speed of communicationJoint application of distributed memory and shared memory parallelizationreduce communication and memory use

Thread 0

Thread 1Thread 0

Thread 1

Thread 0

Thread 1

Thread 0

Thread 1

task 0

task 1

task2

task3

2个线程

4MPI T

asks

Parallel mode of T213L31 on one node (8 CPUs)

Using 6 nodes (48 CPUs), 10 days forecast can be finished in 2.5h

Global spectral model T213L31Dynamical Featuretriangular truncation, resolving 213 waves around a great circle on the globe;31 levels (up to 10 hPa). Semi-lagrangian, semi-implicit temporalintegration, time step 15minutes.use of a reduced Gaussian grid.Hydrostatic. Forth order linear horizontal diffusion.

Global spectral model T213L31Physical processorography (terrain height, US Navy data-set, 10 minutes of arc resolution) three surface and sub-surface levels (allowing for vegetation cover, gravitational drainage, capillarity exchange, surface and sub-surface runoff, deep-layer soil temperature and moisture), clouds (high, medium, low, convective)stratiform and convective precipitation, carbon dioxide (345 ppmv fixed), aerosol, ozone

diffusionground and sea roughness, ground and sea-surface temperature, ground humidity, snow-fall, snow-cover & snow melt radiation (incoming short-wave and out-going long-wave)friction (at surface and in free atmosphere)gravity wave drag evaporation, sensible and latent heat flux

Verification of T213 in last year

---parallel running of T213 and T106

2002年6-8月T213/T106 500hPa高度场北半球月平均距平相关系数比较

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

6 7 8

T213_48

T106_48

T213_72

T106_72

T213_96

T106_96

T213_120

T106_120

2002年6-8月T213/T106 500hPa高度场东亚月平均距平相关系数比较

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

6 7 8

T213_48

T106_48

T213_72

T106_72

T213_96

T106_96

T213_120

T106_120

500 hPa height monthly mean anomaly correlation coefficient comparision of T213/T106 in June-August, 2002

North Hemisphere East AsiaJune July August

June July August

2002.4.7-9.20 T213 500hPa高度场预报北半球距平相关系数逐日变化

0

0.2

0.4

0.6

0.8

1

1 9 17 25 33 41 49 57 65 73 81 89 97 105 113 121 129 137 145 153 161

day1

day2

day3

day4

day5

day6

day7

2002.5.17-9.20 日本 500hPa高度场预报北半球距平相关系数逐日变化

0

0.2

0.4

0.6

0.8

1

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148 155

day1

day2

day3

2002.4.7-9.20 EC 500hPa高度场预报的北半球距平相关系数逐日变化

0

0.2

0.4

0.6

0.8

1

1 9 17 25 33 41 49 57 65 73 81 89 97 105 113 121 129 137 145 153 161

day1

day2

day3

day4

day5

day6

day7

500 hPa height anomaly correlation coefficient evolution with day in North Hemisphere (April 7-September 20, 2002)

ECMWF

JMA

CMA

图7:2002.4.7-7.20 T213 500hPa高度场预报的东亚距平相关系数逐日变化

0

0.2

0.4

0.6

0.8

1

1 5 9

13

17

21

25

29

33

37

41

45

49

53

57

61

65

69

73

77

81

85

89

93

97

101

105

day1

day2

day3

day4

day5

day6

day7

图6:2002.4.7-7.20 EC 500hPa高度场预报的东亚距平相关系数逐日变化

0

0.2

0.4

0.6

0.8

1

1 4 7

10

13

16

19

22

25

28

31

34

37

40

43

46

49

52

55

58

61

64

67

70

73

76

79

82

85

88

91

94

97

100

103

day1

day2

day3

day4

day5

day6

day7

500 hPa height anomaly correlation coefficient evolution with day in East Asia (April 7-July 20,2002)

ECMWF

CMA

Verification Statistic of Precipitation

Threat ScoreTS=NA/(NA+NB+NC)

Forecasting BiasB=(NA+NB)/(NA+NC)

NANCNB

forecasts

observation

Standard station：400 stations；

• grid standard station；

interpolation

rank：

• Light rain： 0.1 mm～9.9mm

• Moderate rain：10.0 mm～24.9mm

• Heavy rain：25.0 mm～49.9mm

• Storm rain: 50.0 mm～99.9mm

• Torrential rain : 100.0mm ～

Rainfall forecast Comparision of T213 and T106 (12-36h) with observation (18-19, august, 2001)

observation

2002年6月1日到9月27日T213及预报员24小时降水预报累加检验TS评分

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

预报员

T213

预报员

T213

预报员

T213

预报员

T213

预报员

T213

预报员T213

2002年6月1日到9月27日T213及预报员48小时降水预报累加检验TS评分

0

0.2

0.4

0.6

0.8

预报员

T213

预报员

T213

预报员

T213

预报员

T213

预报员

T213

预报员T213

forecasters (June 1-September 27, 2002)Forecaster

Threat Scores of 48h forecasting accumulative rainfall of T213 and forecasters (June 1-September 27, 2002) Forecaster

L M H S T

L M H S T

Forecaser

Forecaser

Forecaser

Forecaser

Forecaser

Forecaser

Forecaser

Forecaser

Forecaser

Forecaser

Real SST Experiment

Real SST ExperimentThe main problems exist in forecasting precipitation of T213 in CHINA in July to August in 2001.the typhoon rainfall often lies southerly to the main land of CHINA and lies on the sea.Case ExperimentFive typhoon synoptic procedures which affected CHINA in July to August of 2001

operational test

observation

operational test

ncep reanalysis

Ncep

reanalysisoperational

test

Rainfall Verification in Summer of 2003

T213DWDJMA

L M H S T L M H S T L M H S T L M H S TL M H S T

Comparision of rainfall verification for CHINA with different models (June,2003)

T213DWDJMA

L M H S T

Comparision of rainfall verification for CHINA with different models (July,2003)

T213DWD

JMA

L M H S T

L M H S T

36 60 84 108 132

36 60 84 108 132

Comparision of rainfall verification for CHINA with different models (August,2003)

L M H S T

L M H S T

36 60 84 108 132

36 60 84 108 132

T213DWDJMA

Prospects for development

Prospects for developmentimprovement to operational system

Realization of SETTLS(Stable Extrapolation Two-Time_level Scheme): with the kindly help of Dr. Hortal of ECMWF

Use of linear gaussian grid

raise of resolution to TL511L60(before 2008)

More physical components in the models

Improvement of assimilation algorithms, withmore efforts put on satellite and radar observation data

Development of surface analysis

Prospects for developmentDevelopment of a unified model

Unified grid model

Common dynamic frame used by multiply scale

global, mesoscale and climate model

4Dvar with stress on the use of satellite data

Physical processes suitable to different sacles

Suportive environment

program coding standards, graphic package, interface for research and development ,product interpretation

Atm

osphere comm

only used dynamic fram

e

MesoscaleAtmosphere

Physis

GlobalAtmospheric

Physics

Monthly and Seasonally

AtmosphericPhysics

MesoscaleAtmospheric

Model

RegionalOceanicModel

GlobalAtmospheric

model

GlobalOceanicModel

Coupled ensemble subsystem

EnsembleAtmospheric

Model

SimplifiedOceanic

Model

Monthly and Seasonally

AtmosphericModel

SimplifiedOceanic

Model

simplified

Coupled climate subsystem

Regional coupled subsystem

Global coupled subsystemGlobal

Atmospheric 4DVARGlobal

Oceanic 4DVAR

Mesoscale Atmospheric 4DVAR

RegionalOceanic 4DVAR

RegionalOceanicPhysics

GlobalOceanicPhysics

Oceanic com

monly used dynam

ic frame

The structure of the next generation numerical forecast system of CHINA

图 7 新一代多尺度通用模式体系结构图

动 力 框 架

（ 静 力 /非静 力 ）

中 尺 度

（ 物 理 ）

集 合 预 报

短 期 气 候 （ 物 理 化

学 ）

中 期 （ 物 理 ）

海 洋

模 式

Ensemble forecasts

Mesoscale(physics)

Dynamic Frame

(hydrostatic/nonhydrostatic)

Short term Climate

(physics and Chemistry)

Oceanic model

Mediumrange

(physics)

Illustration of multiple scale unified model

中国新一代模式与同化系统基本构想

大

平流层动力——辐射——简单化学

（多尺度物理过程）

气

层 对流层 动力——辐射——云——能量与水循环

（多尺度物理过程）

动 分力 析框 同架 化

海洋 陆地/水圈 陆地/生物圈

海浪——海冰

洋流

雪盖

土壤水冰冻

路面过程

土壤水、植被

动 分力 析框 同架 化

大

气

层

平流层

对流层

动力——辐射——简单化学

（多尺度物理过程）

动力——辐射——云——能量与水循环

（多尺度物理过程）

动力框

架 分

析同化

陆地/生物圈陆地/水圈海洋

路面过程

土壤水、植被

雪盖

土壤水冰冻

海浪——海冰

洋流

动 分

力 析

框 同

架 化

地

陆

海

洋

Atm

osphere

stratosphere

troposphere

land

ocean

ocean

Sea wave Sea ice

Ocean circulation

Dynamics-radiation-simplified chemistry(multiple scale physical process)

Dynamics-radiation-cloud-energy and water cycling(multiple scale physical process)

Land/hydrosphere

Snow coverSoil water

congelation

Land/ecosphere

Soil water,vegtation

Dynam

ic frame

Analysis and assim

ilation

Dynamic frameAnalysis and assimilat-ion

The unified model and assimilation system of CHINA

图 — 1 中 国 气 象 局 国 家 级 数 值 预 报 模 式 水 平 分 辨 率 逐 年 提 高

1 9 9 9 2 0 0 2 2 0 0 5 2 0 0 8 2 0 0 k m 1 6 0 k m 1 2 0 k m 6 0 k m 3 0 k m

表 — 1 中 国 气 象 局 国 家 级 数 值 预 报 模 式 水 平 分 辨 率 逐 年 提 高

年 份

模 式

1 9 9 9

2 0 0 1

2 0 0 5

2 0 0 8

B J 中 尺 度 1 5 K M 6 K M 3 K M 1 K M 全 国 区 域 5 0 K M 2 5 K M 1 5 K M 5 K M 全 球 中 期 1 2 0 K M 6 0 K M 4 0 K M 2 0 K M 中 期 集 合 2 0 0 K M 1 2 0 K M 8 0 K M 6 0 K M

中 期 集 合 全 球 中 期 区 域 短 期 B J 中 尺 度

The increase of resolution of numerical forecast models as time goes on

ensemble

global

regional

mesoscale

year

modelmesoscale

regionalglobal

ensemble