J. Meteor. Res.  2015, Vol. 29 Issue (1): 132-143   PDF    
http://dx.doi.org/10.1007/s13351-014-4934-1
The Chinese Meteorological Society
0

Article Information

Zhong Shuixin, Chen Zitong, 2015
Improved wind and precipitation forecasts over South China using a modified orographic drag parameterization scheme
J. Meteor. Res., 29(1): 132-143
http://dx.doi.org/10.1007/s13351-014-4934-1

Article History

Received February 28, 2014
in final form September 2, 2014
Improved Wind and Precipitation Forecasts over South China Using a Modified Orographic Drag Parameterization Scheme
ZHONG Shuixin1,2 , CHEN Zitong1,     
1 Institute of Tropical and Marine Meteorology/Guangdong Provincial Key Laboratory of Regional Numerical Weather Prediction, China Meteorological Administration, Guangzhou510080;
2 State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, Beijing100081 (Received February 28, 2014; in final form September 2, 2014)
Abstract:To improve the wind and precipitation forecasts over South China, a modified orographic drag parameterization (OP) scheme that considers both the gravity wave drag (GWD) and the mountain blockingdrag (MBD) effects was implemented in the Global/Regional Assimilation and Prediction System TropicalMesoscale Model (GRAPES−TMM). Simulations were performed over one month starting from 1200 UTC19 June 2013. The initial and lateral boundary conditions were obtained from the NCEP global forecastsystem output. The simulation results were compared among a control (CTL) experiment without the OPscheme, a GWDO experiment with the OP scheme that considers only the GWD effect, and an MBD experiment with the modified OP scheme (including both GWD and MBD). The simulation with the modifiedOP scheme successfully captured the main features of precipitation, including its distribution and intensity,and improved the wind circulation forecast in the lower troposphere. The modified OP scheme appearsto improve the wind forecast by accelerating the ascending air motion and reinforcing the convergence inthe rainfall area. Overall, the modified OP scheme exerts positive impacts on the forecast of large-scaleatmospheric fields in South China.
Key words: orographic drag parameterization (OP)     gravity wave drag (GWD)     wind circulation, South China    
1. Introduction

The dynamic and thermal effects of large-scaleorography can significantly impact the atmosphericcirculations and climate patterns. In large-scale atmospheric models,the gravity waves(GWs)inducedby subgrid-scale orography(SSO)largely influence thewind and temperature in the middle troposphere(e.g.,Palmer et al., 1986; Kim and Arakawa, 1995; McL and ress et al., 2012). The dissipation of GWs generatessynoptic-scale forces on the atmospheric flow,knownas gravity wave drag(GWD). Simulations of shortterm evolution of weather systems and long-term climate change are improved with parameterization ofthe GWD induced by SSO(Lilly,1972; Matsuno,1982; Boer et al., 1984). Without parameterization ofthe subgrid-scale GWD,excessively strong westerlies(easterlies)may appear in the simulations of the winter(summer)midlatitude Northern Hemisphere(Palmer and Mansfield, 1986; Kim et al., 1998; Bauer et al., 2000).

In the early atmospheric models,parameterization of the GWD induced by SSO was based on thetwo-dimensional linear wave theory that describes orographic GWs(e.g., Helfand et al., 1987; Miller et al., 1989; Broccoli and Manabe, 1992)with the “satura-tion hypothesis”(Lindzen,1981). Boer et al.(1984)first considered the influence of GWD on lowerlevelatmospheric circulations in large-scale models. Theeffect of wave breaking on GWD,first considered by Palmer and Mansfield(1986)in a saturation theoryframework,has been adopted in many later studies( Helfand andLabraga,1988; Miller et al., 1989; Qian,2000). Besides improving the low-level atmosphericcirculation simulations,the GWD parameterization alleviates the simulation system bias,such as the warmpool effect caused by southerly deviation,leading toimproved simulations of large-scale atmospheric fieldsat upper levels in the Northern Hemisphere(Hong et al., 2008; McL andress et al., 2013).

The GWD parameterization scheme of Kim and Arakawa(1995; hereafter KA95)has considered the enhancement of drag by low-levelwave breaking and the trapping of the hydrostatic and non-hydrostatic GWs. Zhong et al.(2014a)implemented the KA95 scheme in theGlobal/Regional Assimilation and Prediction SystemTropical Mesoscale Model(GRAPES−TMM), and demonstrated that the scheme improved the overall ability of GRAPES−TMM in forecasting typhoonpath and intensity,as well as the cold air outbreakover South China(Zhong et al., 2014b). However,theKA95 scheme excludes the mountain blocking drag(MBD)effect,which has been considered in severalother orographic drag parameterizations(Lott and Miller, 1997; hereafter LM97; Webster et al., 2003;Kim and Doyle, 2005; Wang et al., 2008). In additionto the GWD induced by SSO,the MBD induced byflows over/around complex orography also needs to beparameterized to reduce the mountain wave effect and to simulate realistic atmospheric flows.

Kim and Doyle(2005)extended the KA95 schemeby including the MBD effect in their drag parameterization. They adopted the bulk aerodynamic drag formulation,which physically resembles the parameterizations of LM97 and Scinocca and McFarlane(2000).As pointed out by LM97,the inclusion of the MBDeffect improves simulation of the low-level flow deflection and resembles the effect of the enhanced orography. A drawback of the enhanced orography is thatstagnant flow effects are treated as invariant,whereasthey are actually time dependent. Furthermore,envelope orography may artificially increase the modeledprecipitation(Lott and Miller, 1997). This problem isaverted by GRAPES−TMM,which has replaced theenvelope orography with a more realistic “mean” gridscale orography. In the present study,we modifiedthe KA95 scheme by including the MBD in a regionalmodel developed on the basis of GRAPES−TMM. Weaccount for the orographic angle,anisotropy, and slopein a three-dimensional(3D)framework.

Section 2 introduces the model we used and themodified KA scheme, and summarizes the experimental design. Section 3 experimentally evaluates the impact of the modified scheme on precipitation and windforecasts over South China in July 2013. The performance of the modified scheme is compared with that ofthe original KA95 scheme. A summary and discussionare presented in Section 4.2. Model description and experimental design 2.1 GRAPES−TMM and RMSC

Based on the GRAPES−TMM(Chen et al., 2008),a regional model for South China(RMSC)wasdeveloped. The RMSC adopts a semi-implicit Lagrangian temporal difference scheme on the ArakawaC grid with a Charney-Philip vertical layer-skippingsetup. It also employs a modified hydrostatic deduction technique,a relaxed correction method in theboundary zone, and effective orography construction.The RMSC,first implemented in 2009,focuses on forecasting the weather systems over South China. Itproduces operational forecasts of isobaric atmosphericfields twice a day(Chen et al., 2010). Since its deployment,the RMSC has provided important reference for weather forecasters and researchers. RealtimeRMSC forecasts and verification results are availableat: http://www.grapes-trams.org.cn.2.2 The orographic drag parameterization

In this study,the term orographic drag parameterization is abbreviated as OP. To improve the KA95GWD scheme in the RSMC,we combined the MBDparameterization with bluff-body dynamics(Miller,2004) and incorporated this design into the KA95scheme. The mesoscale flow dynamics are described bythe following two conceptual schemes,whose relativeimportance depends on the non-dimensional mountainheight. If the maximum height of a mountain range isH,its non-dimensional height(also called the inverseFroude number)is written as

whereUis the horizontal wind speed and Nis theBrunt–Vaisala frequency. IfHn is small,the atmospheric flow passes over the mountain and triggers orographic GWs. The corresponding stress at the reference level(the level of the generated GWD)is givenby(Kim and Arakawa, 1995)where ρ is the atmospheric density, and Δxis the horizontal scale of the model grid. The subscript 0 denotesthe reference level,i.e.,the level at which GWs aregenerated. E is the enhancement factor of the dragforce at the reference level that is used to enhance thereception of the drag force caused by lower-level wavebreaking or suppression effects,which represents thenumber of mountains within a grid specified by m', and G''is a progressive function for judging whetherthe atmospheric flow is free or blocked.

At largeHn,the vertical motion of the atmospheric flow is blocked and part of it flows aroundthe mountain,generating the MBD. The height of theblockage(block level)is estimated by the followingequation:

Unlike the GWD parameterization process,blocking drag in the MBD parameterization occurs mainlywithin the block level(Zblk); hence,the drag forceis integrated from z=0 to Zblk. The correspondingblocking drag is

where l(z)is the effective orographic height related tothe wind direction and blocked height(Alpert,2004),μis an altitude above the mountain,ψ denotes theangle between the directions of the low-level wind and the principal axis of the topography, and a≈μ/σ and in Eq.(4)relate to SSO parameters; namely,the anisotropy σ and mean slopeγ. To ensure thestability of MBD parameterization,the wind speedUin Eq.(3)is determined by a semi-implicit methodthat updates U at stept+dt and |U| at stept.The3D GWD is parameterized by adding Eqs.(2) and (3)in the physics equations of the GRAPES model,givenby 2.3 Experimental setup

The fundamental effects of MBD are investigatedin two sensitivity experiments: one accounting for the3D GWD parameterization effect(i.e.,the MBD experiment considering both GWD and MBD) and theother neglecting this effect(i.e.,the GWDO experiment with the original KA95 scheme considering onlythe GWD induced by SSO). In addition,a control(CTL)experiment without considering any orographicdrag effect was also conducted. All the three experiments employed the same boundary layer,l and surface, and convective parameterization schemes, and identical st and ard initialization(SI); namely,the Yonsei University(YSU)boundary layer scheme,the simplified Arakawa-Schubert(SAS)cumulus parameterization scheme,the SLAB l and surface scheme,theSWRAD short-wave radiation scheme, and the RapidRadiative Transfer Model(RRTM)long-wave radiation scheme. The simulation domain comprises 195×153 grid points with a horizontal resolution of 18 km,as shown in Fig. 1. There are 55 layers in the verticaldirection, and both the initial and lateral boundaryfields are obtained from the 0.5°×0.5°forecast fieldsof the NCEP Global Forecast System(GFS)with thelateral boundary fields updated every 6 h.

We conducted a one-month simulation for thethree experiments: GWDO,CTL, and MBD. TheMBD uses the modified OP scheme based on the KA95scheme. The experiment period is from 1200 UTC 19June to 1200 UTC 21 July 2013. The verification isperformed with a nationwide uniform st and ard verification program provided by the Center for NumericalPrediction Research,Chinese Academy of Meteorological Sciences. This system has been localized, and itsvalidity has been confirmed in a previous operationtest. We verified the 850-hPa wind,the temperatureat 2 m above ground level(T2m), and the 24- and 48-h forecast accumulative precipitation. The surfacevariables(T2m and precipitation)are compared withdata from 1298 automatic stations in the study domain(Fig. 1), and the isobaric variables(850-hPawinds)are examined against the GFS analysis dataover 11.20°–33.76°N,97.40◦–128.84°E(the statisticaldetails will be given in Section 3.4).

Especially in summer,the RMSC forecasts oflow-level strong southwest winds are widely variable.Therefore,we selected a 48-h simulation of the weatherpatterns from 30 June to 1 July 2013,when sustainedrainstorms with strong low-level southwest winds occurred in the Sichuan basin. We investigated the influence of MBD on the forecasting of the low-levelwind field and precipitation by the RMSC,throughthe GWDO experiment in parallel with an MBD experiment. As shown in Fig. 1,the RMSC domain embraces South and West China,characterized by complicated orography of plateaus and basins.

Fig. 1. Domain of the RMSC. Color shadings denote orography, and the black dots indicate locations of the 1298automatic stations used in the RMSC verification system.
3. Results3.1 Applicability of the MBD parameterization

As discussed in Section 2,the blocking drag isintegrated from the surface up to the blocking levelZblk. Palmer and Mansfield(1986)posed an upperlimit on the st and ard deviation of the orography forthe blocking level. To represent the blocking situation,Kim and Doyle(2005)exp and ed the definition ofthe Froude number to include the 3D flow over mountains. They computed the horizontal aspect ratio ofthe mountains, and calibrated by comparing the para-meterized vertical distribution of the momentum fluxwith that obtained from explicit mesoscale simulations. In the present paper,the one-month simulationresults are compared with the observation data. Acritical effective height exists such that

where KE and PE denote the kinetic and potential energy,respectively. According to the dividing streamline theory(Snyder et al., 1985),when KE >PE,theatmospheric flow passes over the mountain and triggers orographic GWs. If KE As the terrain height over the Sichuan basin is approximately 300 m(see Fig. 1),we analyzed the KE and PE at 417 m. Figure 2 shows the distributions ofKE and PE at 417 m after 10 h of model integration,as well as the related blocking area calculated in theMBD parameterization. PE exceeds KE everywhereexcept in the central Sichuan basin,a characteristically low-lying region. High PE flows are concentratedaround steep mountains(see topographical details inFig. 1),whereas their high KE counterparts dominate the northwest of Sichuan and may cause GWD asthey pass over mountainous regions. In other words,itis necessary to apply the MBD parameterization overthe complicated geography of the Sichuan basin. Theblocking drag at different levels and its effect on thewind field are discussed next.
Fig. 2.(a)KE,(b)PE, and (c)the corresponding blocking drag region(blue shading denotes the MBD parameterizationapplicable area),at 417 m above the surface after 10-h model integration.
3.2 The MBD effect

Figure 3 shows the MBD values at 417 and 4109m. The large value area covers the eastern low-lyingregion and the western steep mountains around theSichuan basin. The drag magnitude is much higher inthe eastern than in the western region; consequently,the wind disturbance(accelerated wind speed determined by MBD)is stronger in the east of Sichuanbasin. In other words,a large amount of surfacestress exists in the eastern region,triggering appreciable wind disturbance in the low troposphere. Figure 3 reveals a striking drag stress east of the Sichuanbasin,where the wind toward south meets the windward mountain. This drag may trigger GWs and transport the kinetic energy,thereby redistributingthe wind field and altering the convergence and moisture distribution. These conditions are ideal for precipitation. According to observations at the automaticweather stations over East Sichuan,several heavy rainfall events did occur in Guangyuan Municipality ofSichuan on 1,18, and 19 July 2013. Hence,we studythe effects of MBD parameterization on the wind field and precipitation over the above area.

Fig. 3. The stress of MBD(10−2Nm−2) and wind turbulence(10−5s−1)at 417 m(vertical rainbow scale) and 4109m(horizontal gray scale)after 10-h model integration.

Figure 4 presents the vertical velocity differencesbetween the MBD and GWDO experiments at 850 and 500 hPa. The MBD parameterization introduces significant changes to the middle and lower troposphere.Specifically,it generates disturbances in the wind field,especially over the east of Sichuan. The ascending motion is accelerated in the affected region, and the cyclonic circulation is also strengthened. We concludethat the wind circulation over complex topography isespecially sensitive to the MBD parameterization innumerical weather prediction models. The MBD parameterization also affects other simulated fields,suchas the convergence(divergence),vorticity, and watervapor transport in the atmosphere. This indicates thatgood simulations of an atmospheric model depend toa certain degree on successful parameterization of theMBD effect.

Fig. 4. Vertical wind speed(shaded) and horizontal wind difference(arrows)between the MBD and GWDO experiments(MBD–GWDO)at(a)850 and (b)500 hPa,valid at 0200 UTC 1 July 2013(14-h forecast time). The contours in(a)represent terrain height.
3.3 Short-range forecast

As discussed in Section 2,when the GWD inducedby SSO is included in the model physics,the resultingGWD significantly affects the kinetic energy dissipation. This dissipation is essential to the short-rangeforecasting of wind fields,accumulative precipitation, and other meteorological factors. In this subsection,we analyze the effect of MBD on the lower tropospherewind and accumulative precipitation forecasting. Figure 5 compares the moisture convergence and lowlevel jets simulated in the MBD and GWDO experiments. The MBD parameterization significantly altersthe maximum wind field at 850 hPa. The area of thewinds greater than 12 m s−1is closer to northern and northeastern Sichuan basin in the MBD experimentthan in the GWDO experiment, and the bulk wind direction changes from southwesterly in the GWDO experiment to southerly in the MBD experiment. Therefore,the MBD parameterization strengthens moistureconvergence in the heavy rainfall region east of Sichuan and enhances the ascending motion in that region.

Fig. 5. Simulated moisture convergence(shaded; 10−7gmkg−1s−1),vertical velocity(contours), and low-level wind(vectors;>12 m s−1)at 850 hPa in(a)MBD and (b)GWDO experiment,at 0200 UTC 1 July 2013.

These modifications significantly impact the precipitation forecasts.Figure 6 shows the 24-h precipitation forecast inMBD and GWDO experiments. The center of heavyrainfall and the position of the 24-h rain belt simulatedin the MBD experiment are in better agreement withthe observations at 1200 UTC 30 June 2013. Morestriking improvements on the precipitation forecast areobserved at 0000 UTC 1 July 2013. The magnitude and area of the precipitation are significantly improvedby the MBD parameterization,especially in forecasting of the 24-h rainfall exceeding 50 mm. By contrast,the GWDO experiment yields much weak precipitation.

Fig. 6. The 24-h accumulative precipitation from(a,c)GWDO and (b,d)MBD experiments at(a,b)1200 UTC 30June and (c,d)0000 UTC 1 July 2013. The red dots represent observed values.

The MBD affects the wind forecasts at almost allforecast times. The analysis field and 24-h simulationsof 850-hPa wind in MBD and GWDO experiments at1200 UTC 1 July 2013 are plotted in Fig. 7. It isfound that the excessively strong southerly wind oversoutheast Sichuan is substantially alleviated by theMBD parameterization. In the GWDO experiment,the strong southerly wind in that region exceeds 16ms−1(Fig. 7c),but reduces to approximately 12 ms−1in the MBD experiment(Fig. 7b),showing animprovement on the wind forecast in the lower troposphere,compared with the GFS analysis(Fig. 7a). Inthe simulated 700-hPa vorticity field(figure omitted),the intensity of the low-vorticity center is closer to thatof the GFS analysis in MBD than in CTL. Withoutthe blocking drag parameterization,the low-vorticitycenter is too strong over Sichuan Province.

Fig. 7. Total wind speed(shaded) and wind vectors at 850 hPa at 0000 UTC 1 July 2013 from the(a)NCEP GFSanalysis field,(b)MBD, and (c)GWDO experiment.
3.4 Monthly verification

To verify the effects of the modified OP scheme,we conducted a month-long verification starting from1200 UTC 19 June 2013 using the verification programprovided by the Center for Numerical Prediction Research,Chinese Academy of Meteorological Sciences.The verified variables include the wind field at 850 hPa and T2m. The root-mean-square error(RMSE)between the forecast and observation is calculated, and the 24-h precipitation(24P)threat scores(TSs)at 5levels,i.e.,LR(24P ≤9.9 mm),MR(10 mm≤24P≤24.9 mm),SR(25 mm≤24P≤49.9 mm),HR(50mm≤24P≤99.9 mm), and SHR(100 mm≤24P≤249.9 mm),are also computed. Here,the larger the TSvalues,the better the model forecast of precipitation.The RMSE is calculated as follows:

where F and Av are the forecasted and analysis values,respectively, and Nis the number of grid points inthe verification region. In verifying the 850-hPa windfield,the NCEP analysis is used as the analysis value,whereasinverifyingthetemperatureaT2maboveground level,the ground-level SYNOP data collectedby the South China Regional Meteorological Centerat 1298 observation stations are used as the observeddata. The stations are distributed as shown in Fig. 1.

Table 1 presents the verification scores of the 24- and 48-h forecasts from the MBD,GWDO, and CTLexperiments. Shown are RMSEs and TSs over SouthChina. The results demonstrate the overall improvements of the modified OP scheme. RMSEs of the 24- and 48-h forecasts of the 850-hPa wind and those ofT2m are reduced by the modified OP. In particular,the RMSE of the 24-h 850-hPa wind forecast reducesfrom 4.36 m s−1 in the GWDO experiment to 3.93 ms−1 in the MBD experiment,while that of the 48-hforecast decreases from 5.17 to 4.94 m s−1 . In otherwords,the modified OP reduces the RMSEs in the 24- and 48-h 850-hPa wind forecasts by 0.43 and 0.23 ms−1,respectively. Clearly,the RMSEs of the 24- and 48-h forecasts of the 850-hPa wind are much larger inthe CTL experiment than in MBD and GWDO experiments,indicating that parameterizing the GWDinduced by SSO alleviates biases in low-level wind forecasting. The RMSEs ofT2m(the TS values of precipitation)are slightly smaller(larger)in the MBD experiment than in GWDO and CTL experiments. TheMBD parameterization has mainly improved the 24- and 48-h forecasting of LR,MR,HR, and SHR.

Table 1. Skill scores of RMSE and TS for the meteorological variables at 24- and 48-h forecast time in MBD,GWDO, and CTL experiments over 11.2°–33.76°N,97.4°–128.84°E for the extremely heavy rainfall process overthe Sichuan basin from 1200 UTC 19 June to 1200 UTC 21 July 2013
4. Summary and discussion

To improve the forecasting of wind circulation and precipitation over South China,we implementeda modified OP scheme in the RMSC developed fromthe GRAPES−TMM. The simulation using the modified OP scheme that accounts for the MBD effectsuccessfully captured the main features of the precipitation,including its distribution and intensity, and also improved forecast of the lower tropospheric windcirculation,demonstrating an overall improvement inmodel performance.

This study did not investigate the effect of MBDparameterization beyond 48 h. It did not comparethe gravity wave breaking criteria with observationdata either. Rather,we compared explicitly simulatedwind fields and 24-h precipitation forecasts generated by the OP scheme and by the modified OP. Onthe other h and ,Kim and Doyle(2005)evaluated theoragraphic drag parameterization by calculating thesurface pressure drag,which is essential to verify theGWD parameterization induced by SSO in the physical model. Kim and Hong(2009)suggested thatGWD is sensitive to the boundary layer scheme and shortwave radiation scheme(Shin et al., 2010). Thesensitivity of MBD to the l and surface scheme is alsoworthy of investigation. Furthermore,the boundaryfluxes could be verified by calculating the verticalReynolds stress profiles(Laprise and Peltier, 1989),as well as the self-acceleration in the parameterization of GWD(Scinocca and Sutherl and ,2010),waveresonance( Grubiˇsic and Stiperski, 2009; Teixeira et al., 2012) and wave breaking in directionally shearedflows(Teixeira,2014),which were not considered inthe present paper.

In summary,parameterization of the MBD effectin different orographical settings is necessary for better simulating the atmospheric processes,especiallyover large-scale orography such as the southwesternregions of China. The modified OP scheme might alsoinfluence the cold air outbreak into southern China,particularly in winter and spring. This influence hasimportant implications for weather forecast and related application, and will be investigated in futurework.Acknowledgments.

The authors are thankfulto Dr. Meng Weiguang for his enthusiastic and rigorous comments to this article. The comments from theanonymous reviewers are also appreciated.

References
[1] Alpert, J. C., 2004: Sub-grid scale mountain blocking at NCEP. Proceedings of the 16th conference on NWP, Seattle, Washington, Amer. Meteor. Soc., 2-4.
[2] Bauer, M. H., G. J. Mayr, I. Vergeiner, et al., 2000: Strongly nonlinear flow over and around a three-dimensional mountain as a function of the horizontal aspect ratio. J. Atmos. Sci., 57, 3971-3991.
[3] Boer, G. J., N. A. Mcfarlane, R. Laprise, et al., 1984: The Canadian Climate Centre spectral atmospheric general circulation model. Atmos. Ocean, 22, 397-429.
[4] Broccoli, A. J., and S. Manabe, 1992: The effects of orography on midlatitude Northern Hemisphere dry climates. J. Climate, 5, 1181-1201.
[5] Chen Dehui, Xue Jishan, Yang Xuesheng, et al., 2008: New generation of a multi-scale NWP system (GRAPES): General scientific design. Chin. Sci. Bull., 53, 3433-3445.
[6] Chen Zitong, Huang Yanyan, Wan Qilin, et al., 2010: Rapid updating cycle assimilation forecasting system and its experimental analysis in flood season. J. Trop. Meteor., 26, 49-54. (in Chinese)
[7] Grubišic, V., and I. Stiperski, 2009: Lee-wave resonances over double bell-shaped obstacles. J. Atmos. Sci.,66, 1205-1228.
[8] Helfand, H. M., J. C. Jusem, J. Pfaendtner, et al., 1987: The effect of a gravity wave drag parame-terization scheme on GLA fourth order GCM fore-cast. J. Meteor. Soc. Japan, Special volume for the WMO/IUGG NWP Symposium, Tokyo, Japan,729-742.
[9] Helfand, H. M., and J. C. Labraga, 1988: Design of a nonsingular level 2. 5 second-order closure model for prediction of the atmospheric turbulence. J. Atmos. Sci., 45, 113-132.
[10] Hong, S. Y., C. Jung, E. C. Chang, et al., 2008: Lower-tropospheric enhancement of gravity wave drag in a global spectral atmospheric forecast model. Wea. Forecasting, 23, 523-531.
[11] Kim, Y.-J., and A. Arakawa, 1995: Improvement of orographic gravity-wave parameterization using a mesoscale gravity-wave model. J. Atmos. Sci., 52,1875-1902.
[12] Kim, Y.-J., J. D. Farrara, and C. R. Mechoso, 1998: Sensitivity of AGCM simulations to modifications in the ozone distribution and refinements in selected physical parameterizations. J. Meteor. Soc. Japan,76, 695-709.
[13] Kim, Y.-J., and J. D. Doyle, 2005: Extension of an orographic-drag parametrization scheme to incorpo-rate orographic anisotropy and flow blocking. Quart. J. Roy. Meteor. Soc., 131, 1893-1921.
[14] Kim, Y.-J., and S. Y. Hong, 2009: Interaction be-tween the orography-induced gravity wave drag and boundary layer processes in a global atmo-spheric model. J. Geophys. Res., 36, L12809, doi:10.1029/2008GL037146.
[15] Laprise, R., and W. R. Peltier, 1989: The structure and energetics of transient eddies in a numerical simu-lation of breaking mountain waves. J. Atmos. Sci.,46, 565-585.
[16] Lilly, D. K., 1972: Wave momentum flux——GRAP problem. Bull. Amer. Meteor. Soc., 53, 17-23.
[17] Lott, F., and M. J. Miller, 1997: A new subgrid-scale orographic parameterization: Its formulation and testing. Quart. J. Roy. Meteor. Soc., 123, 101-127.
[18] Matsuno, T., 1982: A quasi one-dimensional model of the middle atmosphere circulation interacting with internal gravity waves. J. Meteor. Soc. Japan, 60,215-226.
[19] McLandress, C., T. G. Shepherd, S. Polavarapu, et al., 2012: Is missing orographic gravity wave drag near 60°S the cause of the stratospheric zonal wind biases in chemistry-climate models? J. Atmos. Sci., 69,802-818.
[20] McLandress, C., J. F. Scinocca, T. G. Shepherd, et al., 2013: Dynamical control of the mesosphere by oro-graphic and nonorographic gravity wave drag during the extended northern winters of 2006 and 2009. J. Atmos. Sci., 70, 2152-2169.
[21] Miller, M. J., 2004: Subgrid-scale orographic drag. Meteorological Training Course Lecture Series for ECMWF, 1-9.
[22] Miller, M. J., T. N. Palmer, and R. Swinkbank, 1989: Pa-rameterization and influence of subgrid-scale orog-raphy in general circulation and numerical weather prediction models. Meteor. Atmos. Phys., 40, 84-109.
[23] Palmer, T. N., and D. A. Mansfield, 1986: A study of win-tertime circulation anomalies during past El Niño events, using a high resolution general circulation model. I: Influence of model climatology. Quart. J. Roy. Meteor. Soc., 112, 613-638.
[24] Palmer, T. N., G. J. Shutts, and R. Swinbank, 1986: Alleviation of a systematic westerly bias in general circulation and numerical weather prediction models through an orographic gravity wave drag parame-terization. Quart. J. Roy. Meteor. Soc., 112,1001-1039.
[25] Qian, Y. F., 2000: Effects of envelope orography and gravity wave drag on performance of climate mod-eling. J. Appl. Meteor. Sci., 11, 13-20. (in Chinese)
[26] Scinocca, J. F., and N. A. McFarlane, 2000: The param-eterization of drag induced by stratified flow over anisotropic topography. Quart. J. Roy. Meteor. Soc., 126, 2353-2393.
[27] Scinocca, J. F., and B. R. Sutherland, 2010: Self-acceleration in the parameterization of orographic gravity wave drag. J. Atmos. Sci., 67, 2537-2546.
[28] Shin, H. H., S. Y. Hong, J. Dudhia, et al., 2010: Orography-induced gravity wave drag parameter-ization in the global WRF: Implementation and sensitivity to shortwave radiation schemes. Adv. Meteor., doi: 10.1155/2010/959014.
[29] Snyder, W. H., R. S. Thompson, R. E. Eskridge, et al., 1985: The structure of strongly stratified flow over hills, dividing-streamline concept. J. Fluid Mech.,152, 249-288.
[30] Teixeira, M. A. C., 2014: The physics of orographic gravity wave drag. Front Phys., 2, 43, doi:10.3389/fphy.2014.00043.
[31] Teixeira, M. A. C., J. L. Argaínband, and P. M. A.Miranda, 2012: The importance of friction in moun-tain wave drag amplification by Scorer parameter resonance. Quart. J. Roy. Meteor. Soc., 138,1325-1337.
[32] Wang Yuan, Cai Ninghao, and Tang Jinyun, 2008: A Two-wave scheme for orographic gravity wave drag parameterization. J. Meteor. Res., 22, 152-161.
[33] Webster, S., A. R. Brown, D. R. Cameron, et al., 2003: Improvements to the representation of orography in the Met Office Unified Model. Quart. J. Roy. Meteor. Soc., 129, 1989-2010.
[34] Zhong Shuixin, Chen Zitong, Dai Guangfeng, et al., 2014a: Impacts of orographic gravity wave drag parameterization on typhoon intensity and path. Chinese J. Atmos. Sci., 38, 273-284. (in Chinese)
[35] Zhong Shuixin, Chen Zitong, Huang Yanyan, et al., 2014b: Applying orography-induced gravity wave drag parameterization to GRAPES model. J. Trop. Meteor., 30, 413-422. (in Chinese)
[35] Lindzen, R. S., 1981: Turbulence and stress owing to gravity wave and tidal breakdown. J. Geophys. Res., 86, 9707-9714.