The Impact of the Reduced Radiosonde Observation in Russia on GRAPES Global Model Weihong Tian, Ruichun Wang, Shiwei Tao, Xiaomin Wan Numerical Prediction.

The Impact of the Reduced Radiosonde Observation in Russia on GRAPES Global Model

Weihong Tian, Ruichun Wang, Shiwei Tao, Xiaomin Wan

Numerical Prediction Center of CMA

Contents

• Background• Russia Radiosonde quality assessment• Impact study • Summary

Background• Russia reduced & adjusted radiosonde observation

from January 2015.• The impact caused by reduced data on GRAPES

global model need to assess.

RAOB data coverage in 1 Jan 2015

RAOB data coverage in 1 Jan 2014

00UTC. 12UTC.

00UTC. 12UTC.

The location of reduced radiosonde coverage

Blue dots for 12UTC(51) Red dots for 00UTC(28).

RS92/III Reduced stationHeight

Temperature

O-A (ERA) statistic: Reduced observations at 00UTC compared with Vaisala RS92/Digicora III (Finland)

RS92/III Reduced station

negative bias

Temperature observation bias (10hpa)

Vaisala RS92/Digicora III (Finland) AVK-MRZ (Russian Federation)

Summer Winter

Kelvin

RS92/III

U-Wind

Reduced station

RS92/III Reduced station

O-A (ERA) statistic: Reduced observations at 00UTC compared with Vaisala RS92/Digicora III (Finland)

V-Wind

Impact Study I

• Experiments: Use RAOB Pressure observation– GRAPES 3DVar global data assimilation system– Resolution :0.5*0.5 *62– Observations:

• GTS conventional data, AMV, NOAA-15 /16/18/19 AMSUA

• METOP-A AMSUA– No temperature bias correction for all RAOB– Experimental Period : 1-31 January 2014– Test 1 : use all the data (control)– Test 2 : reduced radiosonde observation

GRAPES Height analysis bias compare with FNL at 00UTC

Reduce RAOB at 00UTC decrease the height analysis bias

All data

Reduce RAOB

GRAPES Height analysis bias compare with FNL at 12UTC

Reduce RAOB at 12UTC has neutral impact on height

All data

Reduce RAOB

00UTCu-wind analysis bias compare with FNL (700hpa)

All data

Reduce RAOB

Reduce RAOB at 00UTC increase the wind analysis bias

12UTCu-wind analysis bias compare with FNL (700hpa)

All data

Reduce RAOB

Reduce RAOB at 12UTC has neutral impact on wind

GRAPES Height analysis Bias & RMSE (Unit:m)

East AsiaRussia

（ 40-90N, 0-180E）

Russia

East Asia

BiasBias

RMSE RMSE

GRAPES U-wind analysis RMSE (Unit:m/s)

Russia East Asia

（ 40-90N,0-180E）

00UTC

Anomaly correlation coefficient 00/12UTC

East Asia

12UTC

East Asia

NH

NH

Impact Study II

• Experiments: Use RAOB Temperature observation– GRAPES 3DVar global data assimilation system– Resolution :0.5*0.5 *60 (new model version)– Observations:

• No AMDAR Temp obs (Temp bias), use wind observation• GTS conventional data, AMV, NOAA-15 /16/18/19 AMSUA• METOP-A AMSUA, GPS/RO Refractivity

– No temperature bias correction for all RAOB– Experimental Period : 1-31 May 2013– Test 1 : use all the data (control)– Test 2 : reduced radiosonde observation

00UTCHeight analysis bias compare with ERA (700hpa)

Reduce RAOB at 00UTC has neutral impact

00UTCu-wind analysis bias compare with ERA (250hpa)

Reduce RAOB at 00UTC increase the wind analysis bias

GRAPES Height analysis Bias & RMSE (Unit:m)

East AsiaRussia

（ 40-90N, 0-180E）

Russia

East Asia

BiasBias

RMSE RMSE

Wind analysis error

Russia

（ 40-90N, 0-180E）

East Asia

Summary• RAOBs in Russia have large negative temperature

bias. bias correction is needed. • The reduced RAOB in Russia has clear negative

impact on wind analysis• but with neutral impact on geopotential height

analysis.

Thank you for attention!