Meteorol Atmos Phys 100, 53–72 (2008) DOI 10.1007/s00703-008-0295-6 Printed in The Netherlands
1 2
Atmospheric Sciences Program, School of Earth and Environmental Sciences, Seoul National University, Seoul, Korea Department of Atmospheric Sciences, Global Environment Laboratory, Yonsei University, Seoul, Korea
Impact of boundary layer processes on seasonal simulation of the East Asian summer monsoon using a Regional Climate Model D.-H. Cha1 , D.-K. Lee1 , S.-Y. Hong2 With 15 Figures Received 31 October 2007; Accepted 6 January 2008 Published online 14 August 2008 # Springer-Verlag 2008
Summary In this study, an improved planetary boundary layer (PBL) scheme (YSU scheme) is implemented in a regional climate model (SNURCM) to rectify the systematic bias of the model with the MRF PBL scheme, and the impact of new PBL processes on the simulation of precipitation in the East Asian summer monsoon (EASM) is investigated through regional climate simulations with the MRF and the YSU schemes for the summer of 1998 when extreme floods occurred over East Asia. Compared to the experiment with the MRF scheme that shows excessive monsoon precipitation, particularly over the ocean, the experiment with the YSU scheme improves the seasonal mean precipitation as well as associated large-scale circulations. The temporal progression of the monsoon precipitation and 500 hPa geopotential height, and vertical structure are also improved by the revised scheme. The MRF scheme simulates more convective precipitation over the ocean than the YSU scheme, since excessive PBL mixing results in the positive feedback between convective precipitation and latent heat flux at the sea surface. The MRF scheme also simulates more non-convective precipitation over the ocean due to distorted large-scale circulations and excessive PBL mixing. In the experiment with the YSU scheme, the simulated precipitation over the ocean is in good agreement with the observation, since the positive feedback
Correspondence: Dong-Kyou Lee, Atmospheric Sciences Program, School of Earth and Environmental Sciences, Seoul National University, Seoul 151-747, Korea (E-mail:
[email protected])
is relatively reduced and large-scale features are reasonably reproduced due to decreased PBL mixing. Excessive PBL mixing in the MRF scheme can also amplify the biases of precipitation over the ocean due to uncoupled air–sea interaction that can result in the imbalance of the energy budget at the sea surface. This implies that new YSU PBL processes can have crucial influences on the regional climate simulation without an ocean model. The results of additional experiments without the spectral nudging technique also reaffirm the impact of the YSU scheme, and further indicate that the development of a coupled regional climate model is required for more reasonable simulation of the EASM.
1. Introduction The East Asian summer monsoon (EASM) plays an important role in the weather and climate over the surrounding areas as well as on a global scale (Lau et al. 2000). Since the impacts of floods and droughts occurring in the EASM on human lives and economics are significant, a number of researches on the EASM have been conducted using observations and numerical models. One of methods to study the EASM is regional climate models (RCMs), which can reproduce regional or local details embedded within a low-resolution large-scale forcing data (e.g.,
54
D.-H. Cha et al.
general circulation model and global reanalysis data). In a number of studies, RCMs have been used for the simulation of the EASM and the investigation of regional climate processes in the EASM. Wang et al. (2003) and Lee et al. (2004) reproduced the extreme floods over East Asia during the summer of 1998 using RCMs. Through regional climate simulations, Kang et al. (2005) showed that the precipitation variability in the EASM is significantly linked with the largescale moisture convergence. Fu et al. (2005) also inter-compared the results of 18-month simulation over Asia using 8 RCMs through the regional climate model inter-comparison project (RMIP). In regional climate simulation of the EASM, strong internal forcing can be generated by natural characteristics in East Asia, such as complicated topography and land-use, complex physical processes among land, ocean, and atmosphere, and various weather and climate phenomena (e.g., typhoon, flood, and drought). In general, most RCMs have systematic errors in regional climate simulation over East Asia. The systematic errors of RCMs are usually larger in summer than other seasons, since the dynamics and physical processes of the regional climate over East Asia are more complex in summer. Wang et al. (2003) showed that their RCM had less predictability for the EASM in the tropics (e.g., southern China) than in the mid-latitudes (e.g., central China and northern China), and Lee et al. (2004) and Chow et al. (2006) showed that precipitation over the ocean were simulated with prominent biases in their regional climate simulation of the EASM. Fu et al. (2005) also showed that all RCMs participating in the RMIP had systematic biases in the simulation of severe precipitation events related with the EASM. One of the reasons that systematic errors arise in RCMs is the uncertainty of physical parameterizations such as cumulus convective parameterization scheme (CPS), explicit moisture scheme (EMS), radiative transfer package, and planetary boundary layer (PBL) scheme. In many studies, the issues with regard to the uncertainties in physical parameterizations have focused on CPS and EMS in regional climate modeling (e.g., Giorgi and Shlieds 1999; Gochis et al. 2002; Lee et al. 2005). In addition to the CPS and the
EMS, the boundary layer scheme also causes large systematic biases of RCMs, because it plays an important role to determine the interactions and exchanges of moisture, momentum, and energy between land, ocean, and atmosphere in regional climate simulation. There have been some impact studies of the boundary layer scheme in both short-term weather prediction modeling and general circulation modeling. For example, Hong and Pan (1996) showed that a slight change in parameters in the boundary layer formulism could significantly affect the simulation of precipitation in the National Centers for Environmental Prediction (NCEP) Medium-Range Forecast (MRF) Model. Mass et al. (2002) demonstrated that the MRF PBL scheme simulated too much mixing and resulted in excessive winds near the surface at night. Braun and Tao (2000) showed that in the simulation of Hurricane BOB (in 1991), simulated horizontal precipitation structures varied substantially between the different PBL schemes. Martin et al. (2000) showed that the change in the PBL scheme in the general circulation model (GCM) resulted in the apparent improvement of PBL and the cloud structure over the eastern subtropical oceans. In the regional climate modeling, however, there have been only a few studies on the impact of the boundary layer scheme, such as Giorgi et al. (1993). In particular, the impact of the boundary layer scheme on the regional climate simulation of the EASM has not been investigated. The objective of this study is to investigate the impact of the boundary layer scheme on the simulation of the EASM using an RCM. To achieve this objective, we implement the Yonsei University (YSU) boundary layer scheme (Hong et al. 2006) to the Seoul National University Regional Climate Model (SNURCM), instead of the Medium-Range Forecast (MRF) scheme (Hong and Pan 1996), and analyze the impact of the boundary layer scheme on the simulation of the EASM in terms of precipitation and circulations. We describe the regional climate model used in this study in Sect. 2. The results of the simulations for the EASM using different boundary layer schemes are discussed in Sect. 3, and the differences in physical processes of precipitation between two simulations and the impact of uncoupled air–sea interaction are explained in
Impact of boundary layer processes on seasonal simulation of the ESAM
Sects. 4 and 5, respectively. Finally, the Summary and Conclusion are provided in Sect. 6. 2. Model description and experiments 2.1 Seoul National University Regional Climate Model The SNURCM (Lee et al. 2004) was developed based on the Penn State University=National Center for Atmospheric Research Mesoscale Model version 5 (MM5) (Grell et al. 1994) in the late 1990s. An advanced and comprehensive land surface parameterization scheme, the community land model version 3 (CLM3) (Bonan et al. 2002), was coupled to the SNURCM for land surface and soil physical processes. To reduce systematic bias, a spectral nudging technique (von Storch et al. 2000) was implemented for lateral boundary handling by a modified relaxation method (Liang et al. 2001). Driving forcings in the spectral nudging technique are stipulated not only at the lateral boundaries but also in the model interior. The SNURCM has been applied to several studies on regional climate simulation of the EASM. Using the SNURCM, Lee et al. (2004) reasonably regenerated the record-breaking floods over East Asia during the summer of 1998, and Kang et al. (2005) investigated the causes of extreme climate events; the floods in 1991 and the drought in 1994. The performance of the SNURCM was also evaluated by participating in the International Regional Climate Model Inter-comparison Project (RMIP) (Fu et al. 2005). Lee et al. (2006) performed a 20-year (1980–1999) continuous regional climate simulation over East Asia using the SNURCM forced by the reanalysis, and they showed that the SNURCM can reproduce the EASM with a level comparable to that in the observations. In all studies using the SNURCM, however, a significant bias was also generated in that the precipitation over the ocean tended to be overestimated as compared with the observation. In particular, the overestimation of oceanic precipitation was pronounced, if cumulus convection parameterization schemes with downdraft processes were used (Lee et al. 2005). As one of the ways to resolve this problem, in this study, we improve the parameterization
55
of physical processes in the boundary layer over the ocean. A new PBL scheme, the Yonsei University (YSU) boundary layer scheme, is implemented in the SNURCM. Recently, Hong et al. (2006) developed the YSU scheme by improving the MRF scheme. The turbulence diffusion equation of the YSU scheme is described by 3 @C @ @C z 0 0 ¼ Kc c ðw c Þh ð1Þ @t @z @z h where C is a prognostic variable such as u, v, t, and q; Kc , the eddy diffusivity coefficient; c , a correction to the local gradient that incorporates the contribution of large-scale eddies to the total flux; ðw0 c0 Þh , the flux in the mixed layer; h, the boundary layer height; and z the vertical level. Major modifications of the YSU scheme were (1) the explicit treatment of the entrainment process of heat and momentum fluxes between surface and the PBL top by adding the second term on the right hand side of Eq. (1); (2) the use of vertically varying parameters in the PBL, such as the Prandtl number and mixed-layer velocity scale; and (3) the inclusion of nonlocal-K mixing for momentum. Hong et al. (2006) indicated that the first factor was the most critical to the improvement. The factor resolved the problems of too much mixing with strong wind shear and too little mixing in the convection-dominated boundary layer. Therefore, in the short-term simulations using the Weather and Research Forecast (WRF) model, the YSU scheme reduced some systematic biases of large-scale features, such as an afternoon cold bias at the low levels in the MRF scheme, thereby improving simulated precipitation over the Unite States. Byun and Hong (2004) also showed that the YSU scheme of the National Centers for Environmental Prediction (NCEP) global spectral model improved the simulated tropical precipitation by a direct effect of the new scheme over the eastern Pacific and by an indirect effect over the western Pacific. However, the impact of the YSU scheme on regional climate simulation of the EASM has not been investigated yet. 2.2 Experiments and data To investigate the impact of the PBL process on the regional climate simulation of the EASM, we
56
D.-H. Cha et al.
Fig. 1. Model domain with topography and sub-region A for the area mean analysis
perform two regional climate simulations using the SNURCM. Hereafter, the two experiments are referred to as the MRF run and the YSU run, in which the MRF scheme and the YSU scheme are used for vertical diffusion parameterization, respectively. We also perform two additional experiments, the MRF_NOSP run and the YSU_NOSP run, that are the same as the MRF run and the YSU run, respectively, except that they have no application of the spectral nudging technique. The model domain consists of 140 by 110 grid points along the zonal and meridional directions (8400 km 6600 km), respectively, with a horizontal grid spacing of 60 km, including nearly all parts in East Asia and its nearby the Western North Pacific ocean (Fig. 1). Along the vertical, there are 23 layers between the 70 hPa model top and the surface, and the time step of model integration is 180 s. The physical parameterization schemes used in this study are the Kain-Fritsch cumulus convective parameterization scheme (Kain and Fritsch 1990), the Reisner II explicit moisture scheme (Reisner et al. 1998), the CCM2 radiative transfer scheme (Briegleb 1992), and the CLM3 land surface model (Bonan et al. 2002). To improve the simulation of the EASM, the traditional relaxation method and spectral nudging technique are simultaneously applied for lat-
eral boundary condition. As shown in Kang et al. (2005) and Miguez-Macho et al. (2004), the systematic biases of the seasonal regional climate simulation can be significantly reduced by the spectral nudging technique. The spectral nudging technique is applied to the horizontal wind components, and the large-scale spectral regimes are assumed to have wave numbers up to eight and six in zonal and meridional directions, respectively, that corresponded to a wavelength of approximately 1000–1100 km. In order to provide initial and boundary data to drive the regional climate model, upper-air variables (zonal and meridional winds, temperature, and relative humidity) on a 2.5 2.5 horizontal grid mesh and at 13 pressure levels from 1000 to 70 hPa along with surface variables on a T63 Gaussian grid (approximately a 1.875 1.904 grid mesh) are obtained from 6-hourly data from the National Centers for Environmental Prediction=Department of Energy (NCEP=DOE) R-2 reanalysis (Kanamitsu et al. 2002). These data are bi-linearly interpolated to the model grids. The observed sea surface temperature (SST) is updated every 24 h from the 1 1 weekly data (Reynolds et al. 2002). The SNURCM also requires initial soil moisture and temperature fields, that are obtained from an off-line simulation of the CLM3, after it attains an equilibrium
Impact of boundary layer processes on seasonal simulation of the ESAM
57
under the forcing of 6-hourly atmospheric values and surface radiation from the reanalysis, as well as daily observed precipitation from the NCEP= Climate Prediction Center (CPC) archives for the year 1998. The model is integrated for four months (MJJA), starting from May 1 in 1998 when extreme floods occurred over East Asia. Simulated upper-air variables are compared with the reanalysis (R-2) driving fields in order to evaluate the performance of the model. In addition, daily Global Precipitation Climatology Project (GPCP) data with 1 1 horizontal grid spacing are used for precipitation evaluation. Sounding data from the South China Sea Monsoon Experiment (SCSMEX) are also compared with simulated results. 3. Simulation results of the EASM in 1998 Record-breaking abnormal floods over East Asia occurred during the summer of 1998. Ding and Liu (2001) demonstrated that this severe flood over East Asia was related to the prolonged impact of the strongest El-Ni~ no (1997=98) activity of the last century. More than 150% of the normal rainfall that was observed over most of East Asia resulted in a number of casualties and tremendous damage in East Asian countries. In this section, the simulated results of EASM in 1998 from the YSU run and the MRF run are compared with observations and the reanalysis in terms of seasonal mean synoptic features, monsoon evolution, and vertical structure. Simulated seasonal mean (JJA) precipitation in the MRF and the YSU runs is shown in Fig. 2. Both experiments reproduce precipitation over land, properly capturing the extreme flooding in the Yangtze River basin and heavy rainfall in Korea and Japan during the summer of 1998. In other words, both runs tend to simulate similar precipitation patterns over land, although the simulated precipitation over the Yangtze River basin from the MRF run is more than that from the YSU run. The reason for the analogous simulations of precipitation over land is likely to be related to the application of the spectral nudging technique, which can reduce the deviation of monsoon circulation between the model solution and the large-scale forcing (i.e., the reanalysis). However, the difference in precipitation over the ocean between the two is much larger than
Fig. 2. Seasonal mean precipitation over East Asia during the summer of 1998 (JJA) in (a) GPCP observation, (b) MRF run, and (c) YSU run. The contour intervals are 2 mm day1
that over land. The MRF run significantly overestimates precipitation over the ocean, in particular, the Bay of Bengal (BOB) and the South China Sea (SCS), as compared to the observation.
58
D.-H. Cha et al.
Table 1. Seasonal mean precipitation of observation (GPCP) and simulations (mm day1 ) in the summer of 1998 (JJA) averaged over land and the ocean of the whole domain except for the buffer zone, and statistics of the simulated precipitation. Bias, SC, and RMSE indicate the seasonal mean bias, spatial correlation coefficient, and root mean square error between simulation and observation, respectively Areas
Experiments
Obs.
Land
MRF YSU
4.2
Ocean
MRF YSU
5.7
Simulation Total
Convective
Non-convective
4.5 4.1 8.7 5.5
2.8 2.7 6.2 4.2
1.7 1.4 2.5 1.4
The monsoon rain-band in the MRF run, extending from the SCS to the ocean area south of Japan, is southward shifted as compared with the observation. Precipitation over the subtropical ocean south of 20 N is also overestimated in the MRF run, while the observed is relatively less due to the abnormally expanded western Pacific subtropical high (WPSH) during the summer of 1998. Simulated precipitation over the ocean is decreased in the YSU run, thus, the bias of the oceanic precipitation is significantly reduced. In addition, the spatial distribution of the EASM rain-band is reasonably simulated, and the light precipitation region over the subtropical ocean is properly captured in the YSU run. Therefore, the YSU scheme brings about considerable improvement in the simulation of precipitation over East Asia, in particular, over the ocean. For further quantitative analysis, the seasonal mean precipitation was averaged over land and over the ocean, and the statistics are shown in Table 1. Over land, the simulated precipitation from the YSU run is slightly less than that from the MRF run, and the BIAS and RMSE are improved to some extent. However, over the ocean, the simulated precipitation from the YSU run is decreased to great extent as compared that from the MRF run. Therefore, all statistics (i.e., BIAS, SC, and RMSE) of the oceanic precipitation are significantly improved in the YSU run. The decrease in the oceanic precipitation in the YSU run results from the reductions in convective as well as non-convective precipitation, which are about 70% and 60% of those in the MRF run, respectively. The reasons for these reductions will be further explained in Sect. 4. The YSU run also improves atmospheric circulations over East Asia, which is considerably associated with precipitation in the EASM.
BIAS
SC
RMSE
0.3 0.1 3.0 0.2
0.82 0.82 0.58 0.70
2.7 2.2 5.0 2.1
Figure 3 shows the seasonal mean upper-level and low-level circulations and the seasonal mean geopotential height with 5880 gpm in the reanalysis and simulations. The spatial patterns of the simulated upper-level jet at 200 hPa from both runs are close to that from the reanalysis (R-2), which extends from northwestern China to northern China, Korea, and Japan. However, the intensities of the upper-level jet are slightly underestimated as compared to the reanalysis. Over land, the low-level southwesterly wind resulting in the extreme floods over southern and central China is properly reproduced in the MRF run as well as in the YSU run. However, the difference of low-level wind between the two is nontrivial over the ocean. In the MRF run, the low-level wind over the SCS is stronger as compared with that in the reanalysis, and the anticyclonic low-level wind near the eastern model boundary associated with the WPSH is not well captured. Meanwhile, the bias of the low-level wind is prominently reduced in the YSU run, where the intensity and spatial pattern of lowlevel wind over the ocean are comparably simulated with respect to the reanalysis. In the MRF run, the large bias of the low-level circulation over the ocean is associated with the suppressed westward expansion of the WPSH. As is well known, the westward expansion of the WPSH is significantly related to the atmospheric circulations of the EASM (Chang et al. 2000; Huang et al. 2003). In the reanalysis, the 5880 gpm contour reaches the SCS (110 E) and southern China (Fig. 3a). However, the WPSH in the MRF run expands to just about 135 E, rather than the SCS, due to the underestimation of its intensity (Fig. 3b). Since low-level circulation of the EASM tends to be along the periphery of the WPSH, the expansion of the
Impact of boundary layer processes on seasonal simulation of the ESAM
Fig. 3. Seasonal mean 850 hPa wind vector, 200 hPa wind speed (m s1 ; contour and shading), and 5880 geopotential height contour (solid line) over East Asia during the summer of 1998 (JJA) in (a) R-2 reanalysis, (b) MRF run, and (c) YSU run
WPSH has a significant influence on the EASM in terms of the temporal variability and the spatial pattern. Therefore, the reduced westward
59
expansion of the WPSH in the MRF run unreasonably intensifies the anticyclonic low-level circulation over the subtropical western Pacific; this circulation can transport the excessive moisture to East Asia. The bias in the low-level wind in the YSU run over the SCS and the subtropical western Pacific is significantly reduced, since the contour with 5880 gpm expands to the Philippines and Taiwan, which is in agreement with the reanalysis (Fig. 3c). Figure 4 shows the observed and simulated time-latitude distributions of the daily total precipitation averaged in the area between 110 E–130 E during the EASM. In the observation, five abnormal flooding events occur over the SCS (15 N) in the second half of May, over southern China (22 N) in early June, over central China (28 N) in the middle of June, over central China (28 N) in the second half of July, and over Korea (37 N) in the first half of August. The temporal evolution of the simulated precipitation from the MRF run is similar to that from the observation. For example, the northward transition of the rainband related with the Meiyu front from early June to mid July is captured, and the five flooding events are simulated in the MRF run. However, precipitation in the five events is significantly overestimated as compared with the observation. In addition, precipitation over the subtropical ocean south of 35 N is overestimated during the entire summer in 1998. In the YSU run, the simulated precipitation associated with the five flooding events and over the subtropical ocean is in good agreement with the observation, and the overestimated precipitation over the subtropical ocean in the MRF run is significantly reduced. Since the temporal evolution of the precipitation in the EASM is related to that of the WPSH, the time-latitude distributions of the daily mean geopotential height averaged between 110 E and 130 E during the EASM in the reanalysis and simulations are analyzed in Fig. 5. The MRF run tends to underestimate the geopotential height at 500 hPa as compared to the reanalysis during the entire summer in 1998, especially during the periods associated with the five flooding events (Fig. 5b). In other words, the westward expansion of the WPSH is inhibited in the MRF run, as shown in Fig. 3. In the YSU run, the temporal evolution of the geopotential height at
60
D.-H. Cha et al.
Fig. 4. Time-latitude distribution of daily precipitation (mm day1 ) averaged between 110 E and 130 E from 1 May to 31 Aug. in 1998 of (a) GPCP observation, (b) MRF run, and (c) YSU run. The contour intervals are 5 mm day1
500 hPa is reasonably simulated in terms of the intensity and distribution (Fig. 5c). The inhibited expansion of the WPSH in the MRF run also results in the pronounced bias in the temporal evolution of the low-level wind (not shown). Figure 6 shows the vertical profiles of 20-day (6 May–25 May) mean biases in wind speed, potential temperature, and water vapor mixing ratio between the simulations and observation averaged over 3 stations in central China and 2 stations in the South China Sea (SCS). This period is an intensive observing period (IOP) of SCSMEX. Simulated results are compared with sounding data from SCSMEX (Johnson and Ciesielski 2002) after interpolation to observed locations. Over the stations in central China, the biases of all variables are prominently reduced in the YSU run except for potential temperature
above 600 mb. In particular, the YSU run tends to simulate enhanced mid-level wind, decreased upper-level wind, and colder and wetter midand low-level air compared to the MRF run. The biases over the stations in the SCS from the YSU run are also smaller than those from the MRF run. The YSU run reduces the biases over the SCS simulating weaker low-level wind, and colder and drier air in the entire troposphere compared to the MRF run. 4. Differences in simulated precipitation between MRF and YSU runs In this section, the reasons why the large bias of precipitation over the ocean in the MRF run is apparently reduced in the YSU run are further investigated by analyzing the differences in the
Impact of boundary layer processes on seasonal simulation of the ESAM
61
Fig. 5. The same as in Fig. 4 but for 500 hPa geopotential height (gpm) with contour intervals of 20 gpm
physical processes of precipitation between the two. Since RCMs with about 60 km horizontal grid spacing use the explicit moisture scheme as well as the cumulus convective parameterization scheme for precipitation physical processes, the simulated total precipitation consists of convective precipitation (i.e., sub-grid scale precipitation) and non-convective precipitation (i.e., grid-scale precipitation). Figure 7 shows the seasonal mean differences of convective precipitation and non-convective precipitation between the MRF and YSU runs. The differences of convective precipitation and non-convective precipitation over land are relatively small as compared with those over the ocean. However, it is obvious that the MRF run simulates more convective precipitation over the ocean as compared to the YSU run, in particular, over
the BOB, the SCS, and the ocean around the Philippines (Fig. 7a). Furthermore, the simulated non-convective precipitation over the ocean from the MRF run is larger than that from the YSU run, in particular, over the SCS, the ocean south of Taiwan, and the ocean south of Japan (Fig. 7b). This means that the YSU scheme can improve the simulation of precipitation over the ocean, since it plays a role to reduce convective precipitation as well as non-convective precipitation over the ocean, both of which are prominently overestimated by the MRF scheme. Therefore, in this section, we examine reasons for different simulations of both convective precipitation and nonconvective precipitation between the two runs. Figure 8 shows the vertical profiles of the 24-h mean potential temperature and water vapor mixing ratio during the first day (May 1) from
62
D.-H. Cha et al.
Fig. 6. Vertical profiles of 20-day mean (6 May–25 May) bias in (a, d) wind speed (m s1 ), (b, e) potential temperature (K), and (c, f) water vapor mixing ratio (g kg1 ) averaged over 3 stations in central China (left panels) and 2 stations in the SCS (right panels). 3 stations in central China are Shao-Wu (27.3 N, 117.5 E), Gan-Zhou (25.9 N, 114.9 E), and Nanchang (28.6 N, 115.9 E), and 2 stations in the SCS are Dongsha Island (20.7 N, 116.7 E) and Shiyan 3 (20.4 N, 116.9 E)
the model initial time. The values are averaged over land and over ocean. Despite the 24-h simulation from the model initial time, there are large differences in the PBL vertical structures over the ocean between the MRF and YSU runs, while the differences over land are relatively small. The variations of the potential temperature and moisture over the ocean with height from the MRF run are smaller than those from the YSU run. Similar difference in seasonal mean vertical structure between two runs is also simulated (now shown). This means that the MRF scheme simulates stronger PBL mixing over the ocean as
compared to the YSU scheme. Excessive PBL mixing over the ocean in the MRF scheme can be attributed to strong low-level wind over the ocean. As shown in Byun and Hong (2004), Braun and Tao (2000), and Hong et al. (2006), the MRF scheme tends to exaggerate PBL mixing when the low-level wind is strong. The YSU scheme, however, reduces excessive PBL mixing by the explicit treatment of the entrainment of fluxes at the mixed layer top (Hong et al. 2006). The YSU scheme can weaken the turbulent mixing compared to the MRF scheme during the summer monsoon season over East Asia,
Impact of boundary layer processes on seasonal simulation of the ESAM
Fig. 7. Seasonal mean differences (JJA) of (a) convective precipitation (mm day1 ) and (b) non-convective precipitation (mm day1 ) between the MRF and the YSU runs (MRF – YSU)
when winds are strong. The weakened oceanic convection by the YSU than by the MRF scheme is similarly found in Byun and Hong (2004), for a simulated precipitation climatology over the equatorial central Pacific oceans. The MRF scheme overestimates the PBL mixing over the ocean consecutively during the entire summer of 1998. In the MRF run, warmer and drier air tends to be simulated in the lowest layer over the ocean below 950 hPa, while colder and wetter air tends to be simulated in the layer over the ocean between 850 and 950 hPa (Fig. 9a and b). Moreover, the equivalent potential temperature in the lower troposphere over the ocean is increased in the MRF run; this also results in an increased convective available potential energy (CAPE) and enhanced convection over the ocean (Fig. 9c and d). Therefore, convective
63
precipitation over the ocean in the MRF run is overestimated by excessive PBL mixing, as shown in Fig. 6a, while that in the YSU run is reduced by decreased PBL mixing. Over land, too, PBL mixing is exaggerated by the MRF scheme when the simulated low-level wind is strong, but it is not as excessive as that over the ocean (not shown). Then, what is the reason for persistent and excessive PBL mixing, and consecutive and enhanced convection over the ocean resulting in overestimated convective precipitation in the MRF run during the entire summer? Figure 10 shows the seasonal mean differences in the frictional velocity, latent heat flux at the surface, vertical motion at 500 hPa, and wind speed at 850 hPa between the MRF and YSU runs. A stronger frictional velocity (i.e., intensified PBL mixing) over the ocean such as the BOB and the SCS is simulated in the MRF run while the difference in the frictional velocity over land is not large except for that over southern China and the northwestern region of the Indochina Peninsula (Fig. 10a). Stronger PBL mixing by the MRF scheme can increase the latent heat flux at the sea surface, which is controlled by the sea surface temperature (SST) and the moisture content at the lowest level (Fig. 10b). This is because the moisture at the lowest level is excessively transported to free atmosphere by stronger PBL mixing of the MRF scheme, as shown in Fig. 9b. Therefore, the MRF scheme simulates less moisture content at the lowest level over the ocean and reproduces more latent heat flux at the sea surface as compared to the YSU scheme. As explained in Fig. 9c and d, more latent heat flux at the sea surface in the MRF run leads to an increase in the equivalent potential temperature in the lower troposphere (i.e., above PBL); this can increase the triggering of convection over the ocean by enhanced CAPE. Therefore, convection over the ocean, in particular, over the SCS and the ocean around the Philippines is prominently enhanced by the MRF scheme (Fig. 10c). Enhanced convection over the ocean around the Philippines in the MRF run can also inhibit the westward expansion of the WPSH due to the enhanced low-level low over these regions. Therefore, the 5880 gpm contour in the MRF run expands to just about 135 E, as shown in Fig. 3b. The shrunken WPSH in the MRF run
64
D.-H. Cha et al.
Fig. 8. Vertical profiles of 24-h mean (00 UTC 1 May–00 UTC 2 May) (a) potential temperature (K), and (b) water vapor mixing ratio (g kg1 ) averaged over land and the ocean in the whole domain except for buffer zone
Fig. 9. Time-height graph of difference (MRF – YSU) of (a) potential temperature (K), (b) water vapor mixing ratio (g kg1 ), (c) equivalent potential temperature (K), and (d) vertical velocity (cm s1 ) averaged over the ocean in the whole domain except for buffer zone. Contour intervals are 0.2 K, 0.3 g kg1 , 1 K, and 0.1 cm s1 , respectively
strengthens the anticyclonic low-level circulation over the subtropical ocean and weakens that over land such as China, Korea, and Japan (Fig. 10d). This indicates that the intensity of the low-level wind over the subtropical ocean (the mid-latitude land) is increased (decreased) in the MRF run
as compared with that in the YSU run. The enhanced low-level wind over the subtropical ocean in the MRF run can further increase excessive PBL mechanical mixing, which in turn, can increase the latent heat flux from the sea surface to the atmosphere again. In other words, a posi-
Impact of boundary layer processes on seasonal simulation of the ESAM
65
Fig. 10. Seasonal mean differences (JJA) in (a) frictional velocity (cm s1 ; contour and shading), (b) latent heat flux at the sea surface (W m2 ), (c) 500 hPa vertical motion (cm s1 ), and (d) 850 hPa wind speed (m s1 , contour and shading) and wind vector between the MRF and the YSU runs (MRF – YSU)
tive feedback over the ocean between convective precipitation and latent heat flux at the sea surface is likely to exist in the MRF run, thereby inducing persistent overestimation of convective precipitation over the ocean. In contrast, this positive feedback over the ocean does not almost appear in the YSU run, since excessive PBL mixing is relatively reduced and large-scale circulation is realistically reproduced. To clarify the positive feedback over the ocean in the MRF run, the time series of the daily mean differences (MRF – YSU) in frictional velocity, surface latent heat flux, vertical motion at 500 hPa, wind speed at 850 hPa, and convective precipitation averaged over land and the ocean in region A are analyzed in Fig. 11. The MRF scheme tends to simulate stronger frictional velocity (i.e., stronger PBL mixing) over the ocean than the YSU scheme during the entire summer. In addition, more surface latent heat flux, stronger convection, stronger low-level wind, and
more convective precipitation are simulated in the MRF run when the MRF scheme simulates stronger PBL mixing as compared to the YSU scheme. It should be noted that over the ocean, the time series of the surface latent heat flux, mid-level vertical motion, low-level wind speed, and convective precipitation between the two runs are consistent with that of the frictional velocity. In other words, the temporal variations of the differences in all variables over the ocean are highly correlated with each other. Over land, however, those variations are not consistent with each other, except for those between the frictional velocity and the low-level wind speed and between mid-level vertical motion and the convective precipitation. In the analysis of the cross correlations among the temporal variations of these variables (Table 2), there are strong positive correlations among these variables with above 0.5 only over the ocean. This feature indicates the positive feedback process over the
66
D.-H. Cha et al.
Fig. 11. Time series of daily mean differences (MRF – YSU) of (a) frictional velocity (cm s1 ), (b) latent heat flux at the sea surface (W m2 ), (c) 500 hPa vertical motion (cm s1 ), (d) 850 hPa wind speed (m s1 ), and (e) convective precipitation (mm) averaged over land and the ocean in the region A (see Fig. 1) Table 2. Cross correlation coefficients of differences of frictional velocity (UST), latent heat flux at the surface (LHF), 500 hPa vertical motion (W500), 850 hPa wind speed (WS850), and convective precipitation (CRAIN) averaged over land and the ocean in the region A. Correlation exceeding 99% (95%) significance is boldfaced (in parentheses) UST
LHF
W500
WS850
CRAIN
1
0.05 1
0.02 0.11 1
0.79 0.06 0.24 1
0.01 0.31 0.68 (0.18) 1
1
0.93 1
0.67 0.67 1
0.86 0.80 0.58 1
Land UST LHF
W500 WS850 CRAIN
OCEAN UST LHF
W500 WS850 CRAIN
0.50 0.59 0.63 0.58 1
ocean between convective precipitation and surface latent heat flux in the MRF run. As previously explained, in the MRF run, enhanced convection over the ocean around the
Philippines resulting in the shrunken WPSH can also distort large-scale monsoon circulations, which determine the moisture transport and moisture convergence in the EASM. Therefore, non-convective precipitation (i.e., large-scale precipitation) can also be unreasonably simulated in the MRF run. In Fig. 12a, the regions where the MRF run simulates more moisture convergence than the YSU run correspond with the regions with overestimated non-convective precipitation in the MRF run (see Fig. 7b). Since the MRF run simulates more moisture transport from the tropics to the mid-latitudes as compared to the YSU run, an unreasonable band with enhanced moisture convergence extending from the SCS to the ocean south of Japan (i.e., the periphery of the WPSH) is reproduced in the MRF run. This induces the overestimation of non-convective precipitation. Overestimated non-convective precipitation in the MRF run is also caused by an enhanced downdraft process of the CPS and excessive PBL mixing, which can increase the relative humidity at the low level. The regions with overestimated non-convective precipitation in the MRF run are consistent with those with enhanced
Impact of boundary layer processes on seasonal simulation of the ESAM
67
Fig. 12. The same as in Fig. 7 except for (a) vertically integrated moisture convergence (mm day1, contour and shading) and moisture transport (kg m1 s1 ; vector), and (b) relative humidity at 850 hPa (%)
convection at the mid-level (see Fig. 10c). The lower troposphere over the regions with enhanced convection is likely to be cooled and moistened due to the downdraft process of the CPS, so that relative humidity in the lower troposphere is increased in the MRF run (Fig. 12b). Increased relative humidity at the low level in the MRF run can also result from excessive PBL mixing, which makes the low-level air between 850 and 950 hPa cooler and wetter, as shown in Fig. 9. However, in the YSU run, non-convective precipitation is not significantly overestimated, since large-scale atmospheric circulations and convection are reasonably simulated, and PBL mixing is reduced. 5. Impact of uncoupled air–sea interaction It should be noted that the differences in all variables over the ocean are much larger than those over land, indicating that the impact of the PBL scheme over the ocean is relatively larger than that over land. This is consistent with the results of the seasonal mean precipitation (see Fig. 2) and the seasonal mean atmospheric circulation (see Fig. 3). The large impact of the PBL scheme over the ocean can be related to the absence of the ocean model in the SNURCM. Over land with the land surface model, there is coupled interaction between the land and atmosphere, so that the transportation of energy, moisture, and momentum are determined not only by the land surface model but also by the atmosphere model. Over the ocean, however, there is just
uncoupled interaction between the ocean and the atmosphere, so that the transportations are only determined by the atmosphere model, since the prescribed SST is used due to the absence of an ocean model. Wang et al. (2005) showed that GCMs uncoupled with ocean models cannot simulate a realistic SST-rainfall relationship over the western Pacific, while coupled GCMs can simulate it. Similarly, in the SNURCM, which is not coupled with an ocean model, uncoupled air–sea interaction can amplify the discrepancy in the simulated precipitation over the ocean. To clarify the discrepancy resulting from uncoupled air–sea interaction, we analyzed the simulated relations between atmospheric hydrometeors and the surface energy budget. Figure 13 shows the vertical profiles of the seasonal mean (JJA) atmospheric hydrometeors (i.e., cloud water, rain water, ice, and snow) averaged over land and the ocean. The differences in the atmospheric hydrometeors over the ocean between the MRF and YSU runs are more prominent as compared with those over land. It is clear that the MRF run simulates more atmospheric hydrometeors over the ocean, except for cloud water in PBL, than the YSU run. These different vertical distributions of atmospheric hydrometeors over the ocean can lead to the differences in the energy budget at the sea surface (Table 3). The differences in the surface energy budget at the sea surface between the two runs are also larger than those on the land surface. The simulated incoming solar radiation at the sea surface from the MRF run is reduced by about 90% of that from
68
D.-H. Cha et al.
Fig. 13. Vertical profiles of seasonal mean (JJA) hydrometeors averaged over land (upper panels) and the ocean (lower panels) in the whole domain except for buffer zone, which were simulated in the MRF run (left panels) and the YSU run (right panels). All units are 105 kg kg1 Table 3. Surface energy budgets during the summer of 1998 (JJA) averaged over land and the ocean in the whole domain except for buffer zone. Sdown, Ldown, Sup, Lup, Sh, and Lh, indicate incoming solar radiation, incoming longwave radiation, outgoing solar radiation, outgoing longwave radiation, sensible heat flux, and latent heat flux, respectively. Units are W m2 Experiment
Sdown
Ldown
Sup
Lup
Lh
Sh
Land
MRF YSU
Ocean
MRF YSU
239.3 247.8 228.2 252.3
358.1 359.3 411.2 405.4
53.7 55.1 18.3 20.2
427.6 429.7 453.3 453.3
38.7 37.2 130.2 96.4
70.3 78.0 7.9 9.8
the YSU run due to more atmospheric hydrometeors. Despite the reduced incoming solar radiation, however, the simulated latent heat flux at
the sea surface from the MRF run is increased by about 140% of that from the YSU run. Therefore, the energy budget at the sea surface
Impact of boundary layer processes on seasonal simulation of the ESAM
69
Fig. 14. The same as in Fig. 7 except for (a) incoming solar radiation (W m2 ), and (b) ground temperature ( C)
in the MRF run is likely to be unbalanced as compared with that in the YSU run. This imbalance of the energy budget at the sea surface in the MRF run is associated with the uncoupled air–sea interaction resulting from the prescribed SST. Figure 14 shows the differences in the incoming solar radiation at the surface and the surface skin temperature (including SST) between the MRF and YSU runs. Over most of the ocean, the MRF run simulates less incoming solar radiation at the surface due to more atmospheric hydrometeors (Fig. 14a). However, there is no difference in SST due to the prescribed SST, while there is a difference in the surface skin temperature over land despite the slight difference in the incoming solar radiation at the land surface (Fig. 14b). In the MRF run, SST should be reduced by the decreased incoming solar radiation at the sea surface, but it is not changed according to the atmospheric conditions due to the absence of an ocean model. Since the prescribed SST in the MRF run plays the role of an increased SST, which can increase the latent heat flux at the sea surface, the imbalance of the energy budget at the sea surface can be easily yielded by the uncoupled air–sea interaction. In the MRF and YSU runs, the spectral nudging technique (SNT) has an important role to reduce the systematic biases induced by uncoupled air–sea interaction. To clarify the impact of uncoupled air–sea interaction on the regional climate simulation of the EASM, we also performed two additional experiments, the MRF_NOSP and the YSU_NOSP runs, which are the same as the MRF and YSU runs, respectively, except that the SNT is not applied in them.
It is apparent that the biases of precipitation and atmospheric circulations not only over the ocean but also over land in the MRF_NOSP and YSU_NOSP runs are increased as compared with those in the MRF and YSU runs, respectively (Fig. 15). In the MRF_NOSP run, extreme floods over the East Asian continent during the summer of 1998 are not captured and precipitation over the ocean is unrealistically overestimated (Fig. 15a). As explained previously, the uncoupled air–sea interaction can result in the unreasonable overestimation of precipitation over the ocean. Further, the overestimated precipitation over the ocean can induce distorted atmospheric circulations and thus inhibits the expansion of the WPSH, since the biases of monsoon circulations are not reduced by the SNT (Fig. 15b). Therefore, the increased biases of precipitation over the ocean are likely to be expanded to the entire domain, when the SNT is not applied to the SNURCM. Similarly, the biases of precipitation and atmospheric circulations from the YSU_NOSP run are also increased as compared with those from the YSU run (Fig. 15c and d). This indicates that the YSU scheme cannot perfectly reduce the systematic biases in regional climate simulation of the EASM caused by uncoupled air–sea interaction if the SNT is not applied to the RCM. However, it is obvious that the YSU scheme improves the simulated precipitation and atmosphere circulations as compared to the MRF scheme in the experiments without the SNT. The imbalances in the surface energy budget over the ocean induced by uncoupled air–sea interaction are further increased in the experiments
70
D.-H. Cha et al.
Fig. 15. Seasonal mean precipitation (mm day1 ) (left panels), and seasonal mean 850 hPa wind vector, 200 hPa wind speed (m s1 ; contour and shading), and 5880 geopotential height contour (solid line) (right panels) over East Asia during the summer of 1998 (JJA). Upper and lower panels indicate MRF_NOSP run and YSU_NOSP run, respectively Table 4. The same as in Table 3 except for the MRF_NOSP and the YSU_NOSP runs Experiment
Sdown
Ldown
Sup
Lup
Lh
Sh
Land
MRF_NOSP YSU_NOSP
Ocean
MRF_NOSP YSU_NOSP
247.6 252.0 204.8 242.6
349.7 357.7 424.5 419.7
55.8 56.0 16.4 19.4
425.1 430.7 453.3 453.3
38.0 36.7 223.3 114.7
72.3 79.2 11.4 11.8
without the SNT as compared with those in the experiments with the SNT (Table 4). It should be noted that the increased imbalance of the surface energy budget over the ocean is much larger in the MRF_NOSP run as compared with that in the YSU_NOSP run. Despite the reduced incoming solar radiation at the sea surface, the latent heat flux at the sea surface from the MRF_NOSP run is increased by about 170% of that from the MRF run, while that from the YSU_NOSP run is increased by about 120% of that from the YSU run. This indicates that the YSU scheme plays a role to reduce the imbalance of the surface energy bud-
get over the ocean resulting from uncoupled air– sea interaction as compared to the MRF scheme, even if the SNT is not applied to the RCM. 6. Conclusions In this study, the YSU scheme is implemented into the SNURCM, and two regional climate simulations for the summer of 1998 using the MRF scheme and the YSU scheme are performed to investigate the impact of the boundary layer processes of the YSU scheme on the simulation of the EASM.
Impact of boundary layer processes on seasonal simulation of the ESAM
The YSU scheme simulates the precipitation in the EASM more reasonably as compared to the MRF scheme that significantly tends to overestimate the precipitation over the ocean. The simulated atmospheric circulations by the YSU scheme are also closer to the observation than those by the MRF scheme. The temporal evolutions of the precipitation and 500 hPa geopotential height are more reasonably simulated in the YSU run. The YSU scheme also improves vertical structure in the troposphere compared to the MRF scheme. The difference in the simulated results (e.g., precipitation and atmospheric circulations) over the ocean between the MRF and YSU runs is more apparent than that over land, since the YSU scheme does not show such large biases of precipitation and atmospheric circulations over the ocean as simulated in the MRF run. Therefore, the YSU scheme can have considerable impact on the regional climate simulation over East Asia, in particular, over the ocean. The causes of different simulations of convective precipitation and non-convective precipitation between the MRF and YSU runs were also examined. In the MRF run, the overestimation of convective precipitation is induced by the positive feedback between the convective precipitation and the latent heat flux at the sea surface that results from excessive PBL mixing. In addition, unreasonable synoptic conditions over the ocean, such as enhanced low-level moisture convergence and increased low-level relative humidity, result in the overestimation of non-convective precipitation. In contrast, the YSU scheme reasonably simulates precipitation over the ocean, since the positive feedback is relatively reduced and synoptic conditions are appropriately reproduced. Excessive PBL mixing of the MRF scheme can yield the biases of precipitation over the ocean due to uncoupled air–sea interaction. In the MRF run, uncoupled air–sea interaction results in more simulated atmospheric hydrometeors which in turn lead to imbalance of the energy budget at the sea surface, including enhanced latent heat flux at the sea surface. In the experiments without the SNT, the biases of precipitation and atmospheric circulations are increased and expanded to the entire domain as compared with those in the experiments with the SNT. However, the YSU scheme improves the simula-
71
tion of precipitation and atmospheric circulations and reduces the imbalance of the surface energy budget over the ocean caused by uncoupled air– sea interaction as compared to the MRF scheme, even if the spectral nudging technique is not applied to the RCM. Consequently, this study shows that the new PBL processes in the YSU scheme can improve the simulation of the EASM in a RCM, in particular, over the ocean. The results of this study, however, may not be general, since simulated results are substantially dependent on the physical and dynamical systems of a model such as the parameterization of precipitation physical process, model resolution, and domain size and location. The impact of the YSU scheme on the simulation of the EASM can be changed by different models. Additional sensitivity studies to model systems are needed to clarify the effect of the YSU scheme. Nevertheless, the results of this study obviously indicate that the PBL processes can have significant impact on the regional climate simulation of the EASM using the RCM. Also, it should be noted that the YSU scheme cannot perfectly reduce the systematic biases in regional climate simulation of the EASM caused by uncoupled air–sea interaction in the experiment without the SNT. Furthermore, the development of the coupled RCM with an ocean model is required for more reasonable regional climate simulations of the EASM, in which the uncertainties resulting from uncoupled air–sea interaction can be overcome. Acknowledgement This study was supported by the Korea Meteorological Administration Research and Development Program under Grant CATER 2006–2303. Also, this study was partially supported by the BK21 program of the Korean Government Ministry of Education. The authors also appreciate to Drs. Hans von Storch and Frauke Feser at the GKSS Research Centre for providing the source codes of spectral nudging and to the National Center for Atmospheric Research for access to surface observation data.
References Bonan GB, Oleson KW, Vertenstein M, Levis S, Zeng X, Dai Y, Dickinson RE, Yang Z-L (2002) The land surface climatology of the community land model coupled to the NCAR community climate model. J Climate 15: 3123–49
72
D.-H. Cha et al.: Impact of boundary layer processes on seasonal simulation of the ESAM
Braun SA, Tao W-K (2000) Sensitivity of high-resolution simulations of Hurricane Bob (1991) to planetary boundary layer parameterizations. Mon Wea Rev 128: 3941–61 Briegleb BP (1992) Delta-eddington approximation for solar radiation in the NCAR community climate model. J Geophys Res 97: 7603–12 Byun Y-H, Hong S-Y (2004) Impact of boundary layer processes on simulated tropical rainfall. J Climate 17: 4032–44 Chang C-P, Zhang Y, Li T (2000) Interannual and interdecadal variations of the East Asian summer monsoon and tropical Pacific SSTs. Part I: Roles of the subtropical ridge. J Climate 13: 4310–25 Chow KC, Chan JCL, Pal JS, Giorgi F (2006) Convection suppression criteria applied to the MIT cumulus parameterization scheme for simulating the Asian summer monsoon. Geophys Res Lett 33: L24709; DOI: 10.1029= 2006GL028026 Ding Y, Liu Y (2001) Onset and the evolution of the summer monsoon over the South China Sea during SCSMEX field experiment in 1998. J Meteor Soc Japan 79: 255–76 Fu C, Wang S, Xiong Z, Gutowski WJ, Lee D-K, McGregor JL, Sato Y, Kato H, Kim J-W, Suh M-S (2005) Regional climate model intercomparison project for Asia RMIP. Bull Amer Meteor Soc 86: 257–66 Giorgi F, Marinucci MR, Bates GT (1993) Development of a second-generation regional climate model (RegCM2). Part I: Boundary-layer and radiative transfer processes. Mon Wea Rev 121: 2794–813 Giorgi F, Shields C (1999) Tests of precipitation parameterizations available in the latest version of the NCAR regional climate model (RegCM) over the continental U.S. J Geophys Res 104: 6353–76 Gochis DJ, Shuttleworth WJ, Yang ZL (2002) Sensitivity of the modeled North American monsoon regional climate to convective parameterization. Mon Wea Rev 130: 1282–98 Grell GA, Dudhia J, Stauffer DR (1994) A description of the fifth-generation Penn State=NCAR mesoscale model (MM5). NCAR=TN-398 þ STR, NCAR Technical Note, NCAR, Boulder, CO. 122 pp Hong S-Y, Pan H-L (1996) Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon Wea Rev 124: 2322–39 Hong S-Y, Noh Y-G, Dudhia J (2006) A new vertical diffusion package with an explicit treatment of entrainment processes. Mon Wea Rev 134: 2318–41 Huang R-H, Zhou LT, Chen W (2003) The progress of recent studies on the variabilities of the East Asian monsoon and their causes. Adv Atmos Sci 20: 55–69 Johnson RH, Ciesielski PE (2002) Characteristics of the 1998 summer monsoon onset over the northern South China Sea. J Meteor Soc Japan 80: 561–78 Kain JS, Fritsch JM (1990) A one-dimensional entraining=detraining plume model and its application in convective parameterization. J Atmos Sci 47: 2784–802
Kanamitsu M, Ebisuzaki W, Woollen J, Yang S-K, Hnilo JJ, Fiorino M, Potter GL (2002) NCEP-DOE AMIP-II reanalysis (R-2). Bull Amer Meteor Soc 83: 1631–43 Kang H-S, Cha D-H, Lee D-K (2005) Evaluation of the MM5=LSM coupled model for East Asian summer monsoon simulations. J Geophys Res 110: D10105; DOI: 1029=2004JD0005266 Lau K-M, Kim K-M, Yang S (2000) Dynamical and boundary forcing characteristics of regional components of the Asian summer monsoon. J Climate 13: 2461–82 Lee D-K, Cha D-H, Kang H-S (2004) Regional climate simulation for the 1998 summer flood over East Asia. J Meteor Soc Japan 82: 1735–53 Lee D-K, Cha D-H, Choi S-J (2005) A sensitivity study of regional climate simulation to convective parameterization schemes for 1998 East Asian summer monsoon. J Terres Atmos Ocea Sci 16: 989–1015 Lee D-K, Cha D-H, Choi S-J, Jin C-S, Kim Y-O, Kang D-H, Jung D-I (2006) Impact assessment of climate changes on regional climate and water cycle in the Korean Peninsula, 2006 Western Pacific Geophysics Meeting, July 24–27, 2006, Beijing, China Liang X-Z, Kunkel KE, Samuel AN (2001) Development of a regional climate model for U.S. midwest applications. Part I: Sensitivity to buffer zone treatment. J Climate 14: 4363–78 Martin GM, Bush MR, Brouwn AR, Lock AP, Smith RNB (2000) A new boundary layer mixing scheme. Part II: Tests in climate and mesoscale models. Mon Wea Rev 128: 3200–17 Mass CF, Ovens D, Westrick KW, Colle BA (2002) Does increasing horizontal resolution produce better forecasts? Bull Amer Meteor Soc 83: 407–30 Miguez-Macho G, Stenchikov GL, Robock A (2004) Spectral nudging to eliminate the effects of domain position and geometry in regional climate model simulations. J Geophys Res 109: D13104; DOI: 10.1029=2003JD004495 Reisner J, Rasmussen RM, Bruintjes RT (1998) Explicit forecasting of supercooled liquid water in winter storms using the MM5 mesoscale model. Quart J Roy Meteor Soc 124: 1071–107 Reynolds RW, Rayner NA, Smith TM, Stokes DC, Wang W (2002) An improved in situ and satellite SST analysis for climate. J Climate 15: 1609–25 Von Storch H, Langerberg H, Feser F (2000) A spectral nudging technique for dynamical downscaling purposes. Mon Wea Rev 128: 3664–73 Wang B, Ding Q, Fu X, Kang I-S, Jin K, Shukla J, DoglasReyes F (2005) Fundamental challenge in simulation and prediction of summer monsoon rainfall. Geophys Res Lett 32: L15711; DOI: 10.1029=2005GL022734 Wang Y, Sen OL, Wang B (2003) A highly resolved regional climate model (IPRC-RegCM) and its simulation of the 1998 severe precipitation event over China. Part I: Model description and verification of simulation. J Climate 16: 1721–38