Extreme Precipitation by High Resolution RegCM3Over East AsiaJing ZHENG , Zhenghui Xie Institute of Atmospheric Physics (IAP),CAS, ChinaXunqiang Bi the Abdus Salam International Centre for Theretical Physics (ICTP), Italy

OutlineIntroductionModel & Exp.Results Discussion

OutlineIntroductionModel & Exp.Results Discussion

Giorgi & Bi, 2000:

The internal model variability exhibits a pronounced summer maximum. It significantly influences the day-to-day model solution, especially for summer precipitation.Also affected by the model internal variability was the frequency of occurrence of heavy daily precipitation events.

A regional climate model simulation is characterized by an intrinsic level of internal variability which can be excited by any type of perturbation and is regulated by synoptic conditions, season, model domain, region of application, and specific simulation period.

Giorgi & Bi, 2000:

The internal model variability exhibits a pronounced summer maximum. It significantly influences the day-to-day model solution, especially for summer precipitation.Also affected by the model internal variability was the frequency of occurrence of heavy daily precipitation events.

A regional climate model simulation is characterized by an intrinsic level of internal variability which can be excited by any type of perturbation and is regulated by synoptic conditions, season, model domain, region of application, and specific simulation period.

Resolution?Model? Its internal model variability?the extreme precipitation,Yangzi and Huaihe Valley summer, 1998 compared to 60km resolution & observational data.the results indicated that high resolution can give more details about the region of the extreme precipitation. Additionally, the maximum of the extreme precipitation were different by the two simulations of different resolutions.Difficulties in simulating precipitation, esp. extreme prec.20km resolution of RegCM3

OutlineIntroductionModel & Exp.Results Discussion

Summary of RegCM3 CoreDynamics:MM5 Hydrostatic (Grell et al 1994)Radiation:CCM3 (Kiehl 1996)Large-Scale Clouds & Precipitaion:SUBEX (Pal et al 2000)Cumulus convection:Anthes-Kuo (1977)Grell (1993)Emanuel (1991)Boundary Layer:Holtslag (1990)

Tracers/Aerosols/dust:Qian et al (2001); Solmon et al (2005); Zakey et al. (2006)Land Surface:BATS (Dickinson et al 1993)SUB-BATS (Giorgi et al 2003)Ocean FluxesBATS (Dickinson et al 1993)Zeng et al (1998)ComputationsParallel Code (Bi, Gao, Yeh)Multiple PlatformsMore User-Friendly Code(Pal et al 2006; Since Giorgi et al 1993ab)ICTP

Experiment Design Region East ChinaPeriod 1998. Jun. ~ Aug. (June: Spinup time)Resolution 2.5o60km20km (80*100, (32N,112E))ICBC Experiment ERA40.vs.NNRP2Cumulus Scheme Experiment (ERA40)Domain Size experiment (NNRP2, icup=4)Buffer zone experiment (ERA40, icup=4)

60km20km

ICBC -- ERA40 .vs. NNRP2ERA40NNRP2?

OutlineIntroductionModel & Exp.ResultsDiscussion

Time Series for PrecipitationNNRP2 vs ERA40 (icup=4) 60km 20kmERA40(icup=1,2,4)60km20km4 observational stations: Wuhan, Hefei, Nanjing, Shanghai

ERA[i1]ERA[i2]ERA[i4]NNRP[i4]EnsembleObs.

ERA[i1]ERA[i2]ERA[i4]NNRP[i4]EnsembleObs.

ERA[i1]ERA[i2]ERA[i4]NNRP[i4]EnsembleObs.

ERA[i1]ERA[i2]ERA[i4]NNRP[i4]EnsembleObs.

Domain Size Experiment (NNRP2,i4)LargeSmallObs

Buffer Zone Experiment (ERA40,i4)BZ12BZ18 Obs

GPCPRegCM3CMAPCRUMet. stationsSpatial pattern

Spatial pattern ERA40_i4NNRP2_i4NNRP2_SERA40_BZtpr .vs. prcv July1998

Spatial pattern ERA40_i4NNRP2_i4NNRP2_SERA40_BZtpr .vs. prcv Aug1998

OutlineIntroductionModel & Exp.Results Discussion

Preliminary conclusion & Discussionsimulations were different by different resolutions/ domain sizes/ cumulus schemes/ forcing fields.Resolution: higher resolution can give more details about the region of the extreme precipitation. Domain: it depends, but smaller seems much higher peaks.Cumulus schemes: MIT-Emanuel seems better here.Forcing fields: ERA40 southward rain band [vs. NNRP2 ]Difficulties in simulating precipitation, esp. extreme prec.Resolution & Precipitation?Cumulus scheme? Domain size? Forcing fields?Internal model variability?Land Component of the Model? More accurate physical process?A CASE STUDY ----

Thanks for Your Attention!

Small Domain size seems much higher peaks than large domain.()ERA40NNRP