SCIENCE CHINA Earth Sciences - LASGstaff.lasg.ac.cn/ueditor/php/upload/file/20180927/... · 2018. 9. 27. · showed that if AGCM utilized different cumulus parame-terization schemes,

SCIENCE CHINA Earth Sciences

© Science China Press and Springer-Verlag Berlin Heidelberg 2011 earth.scichina.com www.springerlink.com

*Corresponding author (email: [email protected])

• RESEARCH PAPER • November 2011 Vol.54 No.11: 1761–1771

doi: 10.1007/s11430-011-4215-0

Sensitivity of simulated tropical intraseasonal oscillations to cumulus schemes

HU WenTing1,2, DUAN AnMin1* & WU GuoXiong1

1 LASG, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China; 2 Graduate University of Chinese Academy of Sciences, Beijing 100049, China

Received June 3, 2010; accepted October 12, 2010; published online July 23, 2011

The sensitivity of simulated tropical intraseasonal oscillations (ISO) to different cumulus parameterization schemes was ana-lyzed using an atmospheric general circulation model (latest version-SAMIL2.2.3) developed at the Laboratory for Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics (LASG) at the Institute of Atmospheric Physics (IAP) of the Chinese Academy of Sciences. Results show that the basic features of tropical climatological intraseasonal oscillations (CISO) can be captured using all three cumulus schemes. The CISO simulated by the Tiedtke scheme was found to be more realistic than that of the Manabe and Zhang-McFarlane schemes. The results of simulated transient intraseasonal oscillations (TISO) indicate that although the Tiedtke and the Zhang-McFarlane schemes in the new version SAMIL2.2.3 have been ad-justed according to different problems, only the latter can simulate the eastward propagation of the 27–50-day TISO mode. It may be associated with the more realistic diabatic heating profile simulated by the Zhang-McFarlane scheme. In addition, the Manabe scheme in SAMIL2.2.3 is the same as that in the prior version SAMIL2.08. However, some aspects of the physical process, such as the radiation scheme and aerosol condition, have been changed. Conversely the eastward propagation from 100°E to the west of the tropical 27–50-day TISO mode only can be simulated using the Manabe scheme of SAMIL 2.08. Consequently, not all the improvements of physical parameterization schemes work well in every respect. The coordinated de-velopments between dynamic frame and physical processes, and among different physical processes, are important methods that may be used to improve the model.

cumulus parameterization scheme, intraseasonal oscillation, atmospheric general circulation model, diabatic heating profile

Citation: Hu W T, Duan A M, Wu G X. Sensitivity of simulated tropical intraseasonal oscillations to cumulus schemes. Sci China Earth Sci, 2011, 54: 1761–1771, doi: 10.1007/s11430-011-4215-0

The tropical intraseasonal oscillation (ISO) is one of the most significant signals in the atmosphere, and an important phenomenon that influences global weather and climate. Madden and Julian [1–3] analyzed and detected ISO signals, with a period of 41–53 days, in the pressure and zonal wind fields over Canton Island, and then recognized that this os-cillation exits globally as well. Subsequent meteorological studies have revealed that both the 30–60-day oscillations

[4–6] and 10–20-day quasi-two weeks oscillations [7, 8] exist in the tropical atmosphere. Various studies have shown that the tropical ISO has an impact on the onset and retreat of the Asian monsoons [9], tropical storm activities [10] and ENSO [11, 12], thus connecting short-term weath-er changes to long-term climate change. The purpose of investigating tropical ISO is not only to describe its charac-teristics, but more importantly to provide theoretical infor-mation to improve dynamic seasonal forecast capacity.

The forecasting error of the ISO plays an important role

1762 Hu W T, et al. Sci China Earth Sci November (2011) Vol.54 No.11

in operational forecasting, such as 3-day and 10-day fore-casting [13]. Unfortunately, today’s model capacity of sim-ulating ISO is not satisfactory. Using the Atmosphere Mod-el Intercomparison Program (AMIP), Slingo et al. [14] compared simulation results of the ISO in 15 GCMs, and found that ISO signals were obtained in most GCMs, but strictly speaking, the major feature of the observed tropical ISO cannot be reproduced in any GCM.

Different models use different dynamic cores and physi-cal processes and possess different behavior in ISO simula-tion. This is because the factors which influence the simula-tion of the ISO in AGCMs are different. These differences can be attributed to three main points: model resolution, air-sea interaction and cumulus parameterization scheme. Jia et al. [15] used SAMIL-R42L9 to analyze the sensitivity of simulated tropical ISO to cumulus schemes, and found that the performance of the model in simulating the MJO changed apparently when using two different cumulus pa-rameterization schemes: the Manabe scheme and the Zhang-McFarlane scheme. In addition, other studies [16–18] showed that if AGCM utilized different cumulus parame-terization schemes, the results of simulating tropical ISO would be very different. Such studies revealed that cumulus parameterization schemes have key effects on tropical ISO simulation.

Wang et al. [19] found that the Northern Hemisphere summer monsoons display statistically significant climato-logical intraseasonal oscillations (CISO). The CISO results from a phase-locking of intraseasonal oscillation to annual cycle. The extreme phases of CISO correspond to charac-teristic development of the summer monsoons. The TISO [20] is defined as the remaining part after removing CISO from the total ISO, which represents the year-to-year vary-ing portion of the ISO. This part is closely related to the occurrence of intraseasonal extreme events. Yang [20] evaluated the performance of GAMIL1.1.1 simulations of summer CISO and TISO over East Asia (EA)-Western North Pacific (WNP). They indicated that although GAMIL1.1.1 could simulate some of the major characteris-tics of CISO and TISO, it failed to capture the observed eastward propagation of the 27–50-day TISO mode.

Based on previous studies of ISO numerical simulation and the definition of CISO and TISO, we systematically analyzed the CISO and TISO simulation capacity of SAMIL2.2.3. The effect of cumulus parameterization schemes on ISO simulation and the possible causes are dis-cussed through comparisons of results of different cumulus schemes.

1 Data, model and method

1.1 Data

(1) Daily outgoing longwave radiation (OLR) from the National Oceanic and Atmospheric Administration (NOAA)

for 1980–2006 [21]. (2) National Centers for Environmental Prediction (NCEP)/

Department of Energy (DOE) Reanalysis 2 (NCEP- 2), for 1980–2006 [22].

1.2 Model

The AGCM used in this paper is a global spectral model with high-resolution developed in the State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Ge-ophysical Fluid Dynamics (LASG). The model is abbrevi-ated SAMIL (Spectral Atmospheric Model of IAP LASG). The horizontal direction of SAMIL2.08 is rhomboidally truncated at a zonal wave number (R42), roughly equal to a grid of 2.8125° longitude and 1.67° latitude. The comple-mentary external forcing data were developed accordingly. In the vertical direction, a 26-layer in hybrid-coordinate was adopted. The dynamical framework used a “standard at-mosphere reduction” scheme [23], and the semi-implicit time integration program is introduced here. The radiation scheme comes from Edwards and Slingo [24]. The land surface process implemented here is the SSiB model [25, 26]. A more detailed description and recent improvements of this model are given by Zhou et al. [27].

Three cumulus parameterization schemes in SAMIL2.2.3 included the Manabe [28], Zhang-McFarlane and Tiedtke scheme. The Zhang-McFarlane [29, 30] and Tiedtke [31] schemes have been improved.

1.3 Experiment description and method

Three model experiments were carried out by utilizing three cumulus parameterization schemes in SAMIL2.2.3, respectively. These experiments were the AGCM control runs, in which the 20-year averaged monthly SST and sea-ice data, as required by the Atmospheric Model Intercomparison Project (AMIP) II (see http://www-pcmdi.llnl.gov/projects/amip/ AMIP2EXPDSN/BCS_OBS/amip2_bcs.htm for details), were prescribed and the initial conditions were based on the NCEP/DOE reanalysis of January 1, 1979 for SAMIL. Each experiment integrated 28 years. The first year of the inte-gration was ignored during the diagnostics. To compare with model data, the period of the observed data (daily OLR and NCEP/DOE reanalysis) was chosen as 1980–2006. We evaluated the performance of ISO simulation using different cumulus parameterization schemes with the two indicators CISO and TISO.

We partitioned climatological daily mean series yc(i), i=1365 into three components by Fourier analysis: yciso (i)=yc(i)yac(i)R(i), where yac denotes the sum of the first three Fourier harmonics (annual cycle), R represents synop-tic disturbance (removed by 5-day running mean). CISO was derived by removing the annual cycle and synoptic noise from the climatological daily mean field. Then, the time series of the TISO was obtained by subtracting both

Hu W T, et al. Sci China Earth Sci November (2011) Vol.54 No.11 1763

climatological parts and synoptic noise from the raw daily time series of a particular year. CISO and TISO discussed in this paper were calculated by observed OLR daily data and model daily precipitation.

In this study, we focused on the ISO characteristics of three typical tropical monsoon regions in Asia. These three regions are the tropical Indian Ocean region (10°–20°N, 50°–75°E), the Bay of Bengal region (10°–20°N, 80°– 100°E) and the South China Sea region (10°–20°N, 110°– 120°E). For simplicity, the following sections all use IDO, BOB and SCS. Major statistical diagnostic methods include point-based lead-lag correlation analysis, band-pass filtering and Fourier harmonic analysis.

2 Simulation of CISO in SAMIL2.2.3

In 1979, Yasunari [32] first discovered the significant northward propagation, from the equator to 30°N, of cloud and cumulus convection related to ISO over the South Asian monsoon region. The characteristics are reflected well by CISO. In this section, we use three outputs of cu-mulus schemes to compare and analyze meridional propa-gations of the CISO over the longitudes of the three regions

(BOB, IDO and SCS). The effects of the cumulus parame-terization schemes on CISO simulation capacity were eval-uated as follows.

Figure 1(a) displays meridional propagation of boreal tropical CISO along the longitudes between 50°E and 75°E (IDO). The CISO signals of IDO can propagate from the equator to 30°N. The model outputs of the Manabe scheme (Figure 1(d)) and the Zhang-McFarlane scheme (Figure 1 (g)) revealed that CISO almost disappears to the north of 10°N, and that the CISO signal of the Manabe scheme is weaker than that of the Zhang-McFarlane scheme from the equator to 10°N. Only in the Tiedtke scheme output can CISO be found from 20°N to the pole, and the dry/wet phases are very clear.

Observations (Figure 1(b)) show that the dry/wet phases of boreal tropical CISO along the longitudes between 80°E and 100°E (BOB) spread northward very regularly. The onset of the first wet phase occurs in mid-May and the ob-served speed of northward migration is 0.4 latitude degrees per day. Compared to observations, the Manabe scheme (Figure 1(e)) and the Tiedtke scheme (Figure 1(k)) can successfully reproduce the main features of CISO, but there are still some problems. For example, the onsets of the wet phase in the simulation results of the Manabe and the Tiedtke schemes were earlier than observed onsets, and the

Figure 1 Meridional propagations of CISO averaged along longitudes between 50°E and 75°E (first column, IDO), between 80°E and 100°E (second col-umn, BOB), and between 100°E and 120°E (third column, SCS) derived from observed daily OLR (first line), the Manabe daily precipitation (second line), the Zhang-McFarlane daily precipitation (third line) and the Tiedtke daily precipitation (fourth line). Yellow shading represents dry anomalies and green shading represent wet anomalies. Units: W m2 for OLR and mm d1 for precipitation.


speeds of the wet phase’s northward march (0.36 latitude degrees per day for the Manabe scheme and 0.3 latitude degrees per day for the Tiedtke scheme) also were slower. The two simulated northward propagating dry spells respec-tively occurring prior to and after the rainfall season also indicate the same problem. The model output of the Zhang- McFarlane scheme failed to simulate the northward propagation characteristics of CISO with orderless dry/wet phases, which greatly differed from the observations.

Over the SCS longitudes (100°E–120°E), observations (Figure 1(c)) clearly demonstrate that the onset of the wet phase occurs in mid-May and the corresponding dry phases propagate northward, then the CISO signal becomes much weaker after August. However, model simulations of the three cumulus parameterization schemes failed to reproduce the dry/wet phases. In contrast to the Zhang-McFarlane scheme, CISO simulated by the Manabe and the Tiedtke schemes were much improved. The northward march of the rain belt could be seen, but the spread speed was faster than the observed speed, and the CISO signal almost disappeared south of 10°N.

The propagation of dry/wet phases was connected to the advance and the retreat of monsoons. In addition, the northward spread of CISO was related to the onset of mon-soons.

Figure 2 exhibits the daily evolution of the climatological (1980–2006) OLR anomaly and the model precipitation anomaly over the Indian South region (7.5°–12.5°N, 75°–80°E), with respect to the onset date of the Indian summer monsoon. The date of the onset was defined as the first day when the sustained 850 hPa zonal wind averaged over the south Arabian Sea (5°–15°N, 40°–80°E) exceeded 6.2 m s1, with the provision that the area-averaged 850 hPa wind in the ensuing consecutive 6 days also exceeded 6.2 m s1 [33]. In Figure 2, the observations show that after onset of the Indian summer monsoon (June 3), the climatological OLR anomaly suddenly declined. This means that there tended to be a rainy season over the south Indian region and, at the same time, the maximum wet phase of the CISO de-veloped. In the simulation results of three cumulus parame-terization schemes, the onset date was earlier than the ob-served date. Only the daily precipitation and rapid increase of monsoon rainfall after the onset, as simulated by the Tiedtke scheme, were close to the observations.

Generally, CISO simulations along BOB longitudes were more reliable than in the other two regions. From Figures 1 and 2, it is possible to see that although the CISO simulation output of the Tiedtke scheme had many differences from the observed results, Tiedtke scheme is still the best choice for simulating CISO.

3 Simulation of TISO in SAMIL2.2.3

Two major TISO modes (12–25 days and 27–50 days) over

Figure 2 The antecedent and following development of the Indian sum-mer monsoon over the southern tip of India (7.5°–12.5°N, 75°–80°E), shown by climatological daily NOAA OLR anomalies and daily model precipitation simulated by three cumulus parameterization schemes. Dash- ed lines indicate the climatological onset date. Red solid lines represent the OLR or model precipitation anomalies, and blue solid lines represent CISO calculated by OLR and model precipitation.

the three regions (BOB, IDO and SCS) were identified through multi-year mean spectra. In this section, the zonal propagation characteristics of two TISO modes are analyzed. A series of time-longitude cross-sections (see Figures 3–5) are shown by point-based lead-lag correlation analysis with reference to the BOB, IDO, and SCS.

In Figures 3–5, the observations demonstrate the fol-


Figure 3 Zonal propagations of two boreal summer TISO modes aver-aged along latitudes between 10°N and 20°N, where (a), (b), (e), and (g) represent the 12–25-day TISO mode calculated by OLR and model precip-itation, and (b), (d), (f), and (h) delineate the 27–50-day TISO mode. The calculation method is point-based lead-lag correlation analysis with refer-ence to the BOB (10°–20°N, 80°–100°E). The gridded areas are regions above 99% significance tests.

lowing points: (1) the 12–25-day TISO mode propagated eastward from 60°E to 70°E, and turned westward to the east of 100°E; (2) the 27–50-day TISO mode spread east-ward from 70°E to 120°E, and moved westward to the east of 120°E. The simulation results show that when the refer-ence region was chosen as BOB (Figure 3), the Zhang- McFarlane scheme could reproduce well the eastward propagation of two TISO modes from 60°E to 90°E. The Tiedtke scheme could simulate the eastward march of the

Figure 4 Same as in Figure 3, except that IDO (10°–20°N, 50°–75°E) is used as a reference region.

27–50-day TISO mode from 60°E to 75°E. However, the Manabe scheme failed to simulate the eastward propagation of two modes. When IDO was the reference region (Figure 4), all three cumulus schemes could simulate the eastward moving features, but the features simulated by the Zhang- McFarlane scheme were closer to the observed data. Since SCS was the reference region (Figure 5), only the Zhang- McFarlane scheme successfully simulated the eastward propagation of the 27–50-day TISO mode from 60°E to 80°E. In other words, the Zhang-McFarlane scheme was more suitable to simulate the TISO mode eastward propaga-tion than the other two schemes.

The old Zhang-McFarlane scheme could not reproduce the eastward spread of TISO modes Furthermore, the modi-


Figure 5 Same as in Figure 3, except that SCS (10°–20°N, 110°–120°E) is used as a reference region.

fied Zhang-McFarlane scheme, which was used in SAMIL2.2.3, was more conducive to simulate the eastward propagation features of the 27–50-day TISO mode.

4 Diabatic heating vertical distribution and deep-shallow convection interaction

To some extent, the simulated ISO was related to the heat-ing vertical structure, whereas the atmospheric heating pro-cess in the model was largely determined by the cumulus parameterization schemes.

The diabatic heating profile of the observations, the three cumulus schemes in SAMIL2.2.3, and the old Zhang-

Figure 6 Boreal summer mean diabatic heating profiles averaged over BOB, IDO and SCS. The datasets used were NcepR2 reanalysis, and the model data were simulated by three cumulus parameterization schemes in SAMIL2.2.3 and the Zhang-McFarlane scheme in SAMIL2.08.


MacFarlane scheme are shown in Figure 6. In BOB, IDO and SCS, the heating maxima simulated by the three cumulus schemes in SAMIL2.2.3 existed in the middle of the troposphere, where the maximum diabatic heating intensity of the Tiedtke scheme, which is stronger than the others, reached 8 K/day in the SCS region. The old Zhang- McFar-lane scheme generated three heating centers at 900, 600 and 200 hPa, respectively, and the intensity in BOB was weaker than the new one. Clearly, the heating intensity of the modi-fied Zhang-McFarlane scheme was more reliable, with two heating centers at 900 and 500 hPa.

Li et al. [34] modified the diabatic heating profile of the Manabe scheme in SAMIL_R42L9, and found that when diabatic heating was bottom heavy (maximum in the lower troposphere), planetary-scale, intraseasonal, eastward propagating perturbations were reproduced with a phase speed similar to that of the MJO.

To further illustrate the effect of different diabatic heat-ing profiles from the modified Zhang-McFarlane scheme on the simulation of eastward propagation of the 27–50-day TISO mode, three sensitivity tests (CUP, CDN and CLH0) were carried out.

The purpose of the first two sensitivity tests (CUP and CDN) was to analyze the sensitivity of the simulated 27–50-day TISO mode to different diabatic heating vertical structures. The CUP sensitivity test was conducted in the following manner. The first step was to modify the top heavy diabatic heating profile between 20°S and 20°N. The amplitude of the heating profile was reduced by 90% at all levels, except at levels 11 and 12. These two layers were located in the upper troposphere, around 250–300 hPa. In order to maintain energy conservation, the total reduced heating needed to add uniformly to levels 11 and 12, and then we obtained the modified diabatic heating for these levels. The second step was to gradually reduce the ampli-tude of the original heating profile from 90% to zero be-tween 20° and 30° latitude. No heating profile was modified beyond 30° latitude. The same method was followed to cre-ate bottom-heavy heating profiles, except the peak levels were six and seven (600 and 700 hPa). Each application of the cumulus parameterization process in the model changed the atmospheric diabatic heating profile, and then the modi-fied profile was interated into the model to calculate the next step. The third sensitivity test (CLH0), of which the method cancelled the latent heating and only kept sensible heating and radiation effects, was to further confirm the effect of latent heating on the simulation of the 27–50-day TISO mode. Figure 7 shows the vertical profile of diabatic heating with three sensitivity tests, which correspond to and represent three different heating vertical structures.

Figure 8 shows the zonal propagation features of the 27–50-day TISO mode in three sensitivity tests. Both CUP and CDN could reproduce the eastward propagation from 80°E to 100°E. The results revealed that in the modified

Figure 7 Boreal summer mean diabatic heating profiles (10°–20°N, 50°–120°E) using three tests (CUP, CDN, CLH0).

Zhang-McFarlane scheme, in whichever level the changed heating center was located, there was little effect on simulating the 27–50-day TISO mode. However, if the impact of latent heating was cancelled, the eastward movement of the 27–50-day TISO mode could not be simulated, which indi-cated that the latent heating had a key effect on the repro-duction of the TISO eastward propagation.

In SAMIL2.2.3, deep convection was parameterized by the Zhang-McFarlane scheme and shallow convection by the Hack scheme. One sensitivity test (NS), which did not use the calculation of Hack scheme, was designed to demonstrate the effects of deep-shallow convection interactions on the eastward propagating TISO mode. The test revealed that the deep convection was suppressed after the shallow convection was removed, and SAMIL2.2.3 failed to reproduce the eastward propagation (figure omitted). For exam-ple, Figure 9 shows the BOB diabatic profile which is pro-duced by the NS test. The heating vertical distribution of the NS test was very similar to that of the old ZhangMcFarlane scheme. Both the NS test and the old ZhangMcFarlane scheme could not simulate the eastward propagation char-acteristics. As we can see, shallow convection has a key effect on the generation of deep convection, and the deep- shallow convection interaction also is essential for simulation of the eastward propagating TISO modes.

Thus, several conclusions were reached in the analyses, as follows: (1) Both heating amplitudes and center locations of the modified Zhang-McFarlane scheme were very dif-ferent from those of the old Zhang-McFarlane scheme and NS test, which failed to simulate the eastward propagating TISO mode. Simulation of the eastward propagation of the 27–50-day TISO mode was closely related to heating verti-cal structure and amplitude. (2) Simulation of the eastward propagation of the 27–50-day TISO mode did not depend on the location of the heating centers in the modified


Figure 8 Zonal propagations of the 27–50-day TISO mode along latitudes between 10°N and 20°N: detected from daily model precipitation of tests CUP, CDN and CLH0. The calculation method is point-based lead-lag correlation analysis with reference to the IDO, BOB and SCS. The gridded areas are the regions above 99% significance tests.

Zhang-McFarlane scheme. (3) Latent heating was essential for simulation of the eastward propagating 27–50-day TISO mode. (4) In the NS test, deep convection was constrained as the shallow convection was cancelled, which represented the key effects of the deep-shallow convection interaction on the reproduction of the eastward propagation.

5 Comparison of the Manabe scheme simula-tions in two model versions (SAMIL2.2.3 and SAMIL2.08)

The Manabe scheme in SAMIL2.2.3 was the same as the one obtained with the prior version SAMIL 2.08. However,

some parts of the physical process, such as radiation scheme and aerosol condition, changed. The eastward propagating TISO mode could be simulated by the Manabe scheme in SAMIL2.08, but the Manabe scheme in SAMIL2.2.3 could not replicate this phenomenon (Figures 10 and 11(a)–(d)). Furthermore, one sensitivity test (M_na) was conducted to understand the reason behind this difference. The method was to exclude the aerosol direct effect in SAMIL2.2.3 and compare it with that of SAMIL2.08.

As the reference region was BOB (Figure 10), the Ma- nabe scheme in SAMIL2.08 could reproduce reliable east-ward propagation of two TISO modes from 60° to 90°E. However, two TISO modes simulated by the Manabe scheme in SAMIL2.2.3 appeared to move westward from


Figure 9 Boreal summer mean diabatic heating profiles over BOB. The red line represents a modified Zhang-McFarlane scheme and the blue line represents an NS test.

Figure 10 Zonal propagations of two boreal summer TISO modes aver-aged along latitudes between 10°N and 20°N, where (a), (c), and (e) repre-sent the 12–25-day TISO mode calculated by OLR and model precipitation, and (b), (d), and (f) delineate the 27–50-day TISO mode. The calculation method is point-based lead-lag correlation analysis with reference to the BOB (10°–20°N, 80°–100°E). The gridded areas are the regions above 99% significance tests.

Figure 11 Same as in Figure 10, except that IDO (10°–20°N, 50°–75°E) is the reference region.

60° to 160°E. When the reference region was chosen to be IDO (Figure 11), the TISO component simulated by the Manabe scheme in SAMIL2.08 was more similar to obser-vations, although the Manabe scheme in the two versions succeeded in simulating the eastward propagation. With SCS as the reference region (similar to the results over BOB, figure omitted), the TISO mode simulated by the Manabe scheme in SAMIL2.2.3 moved westward, while the Manabe scheme in SAMIL2.08 could reproduce the reliable east-ward propagating TISO component. Eastward propagation characteristics of the TISO mode shown by M_na tests were close to those simulated by SAMIL2.08.

In other words, the Manabe scheme in SAMIL2.2.3 did not change, but the aerosol direct effects added to the model still influenced the TISO simulation by the same cumulus parameterization scheme.

6 Discussion and conclusion

In this paper, an atmospheric circulation spectral model (SAMIL2.2.3) developed in LASG/IAP was utilized to an-


alyze the significant differences of tropical ISO simulation results obtained by different cumulus parameterization schemes. The differences in the results simulated by the same cumulus parameterization scheme in two versions (SAMIL2.08 and SAMIL2.2.3) represent change of other physical processes and still induced the obvious differences of ISO simulations. The main conclusions are:

(1) CISO simulated by the Tiedtke scheme was found to be more realistic than that of the Manabe scheme and the Zhang-McFarlane scheme.

(2) Zonal propagation characteristics of different TISO modes clearly varied when the 27–50-day component had a significant eastward propagation. The capacity of the modi-fied Zhang-McFarlane scheme to simulate the eastward propagating TISO mode also was further strengthened.

(3) Simulation of the eastward propagation of the 27–50-day TISO mode was closely related to heating verti-cal structure and amplitude. Simulation of the eastward propagation of the 27–50-day TISO mode did not depend on location of the heating centers in the modified Zhang-McFarlane scheme. Latent heating was essential for simulation of the eastward propagating of the 27–50-day TISO mode. In the NS test, deep convection was con-strained as shallow convection was cancelled, which repre-sented the key effects of the deep-shallow convection inter-action on the reproduction of eastward propagation.

(4) The Manabe scheme in SAMIL2.2.3 was the same as that produced in the prior model version, SAMIL 2.08, but other parts of the physical process, such as radiation scheme and aerosol condition, were changed. However, the east-ward propagation from 100°E to the west of the tropical 27–50-day TISO mode only could be simulated using the Manabe scheme of SAMIL2.08. Thus, not all the changes of physical parameterization schemes work well in every re-spect. The coordinated developments between dynamic frames and physical processes, and among different physi-cal processes, are important features that would improve the model further.

A series of sensitivity tests was used to understand the factors which influence simulation of the eastward propa-gating 27–50-day TISO mode. However, we did not study the factors which have impacts on CISO northward propa-gation. Previous work has shown that the ISO northward propagation in air-sea coupled models is more accurate and realistic than the AGCM. It is possible that the air-sea in-teraction is one of the reasons for the tropical ISO genera-tion [35]. The effects of air-sea interaction on ISO simula-tion will be investigated in our future work.

The authors appreciate the comments of anonymous reviewers whose crit-ical reviews and valuable suggestions were essential for improvement of the manuscript. We thank Zhang Guangjun, Ling Jian and He Bian for their assistance. This work was supported by National Basic Research

Program of China (Grant Nos. 2010CB951703 and 2009CB421403), Knowledge Innovation Program of the Chinese Academy of Sciences (Grant Nos. KZCX2-YW-Q11-01 and KZCX2-YW-BR-14) and “Strategic Priority Research Program—Climate Change: Carbon Budget and Re- lated Issue” of the Chinese Academy of Sciences (Grant No. XDA- 05110303).

1 Madden R A, Julian P R. Detection of a 40–50-day oscillation in the zonal wind in the tropical Pacific. J Atmos Sci, 1971, 28: 702–708

2 Madden R A, Julian P R. Description of global-scale circulation cells in the tropics with a 40–50-day period. J Atmos Sci, 1972, 29: 1109–1123

3 Madden R A, Julian P R. Observations of the 40–50-day tropical os-cillation—A review. Mon Weather Rev, 1994, 122: 814–837

4 Lorenc A C. The evolution of planetary scale 200 mb divergence flow during the FGGE year. Q J R Meteorol Soc, 1984, 110: 427–441

5 Weickmann K M, Lussky G R, Kutzbach J E. Intraseasonal (30–60 days) fluctuations of outgoing long-wave radiation and 250 mb stream function during northern winter. Mon Weather Rev, 1985, 113: 941–961

6 Murakami T, Chen L X, Xie A, et al. Eastward propagation of 30–60-day perturbations as revealed from outgoing long-wave radia-tion data. J Atmos Sci, 1986, 43: 961–971

7 Krishnamurti T N, Bhalme H N. Oscillation of a monsoon system, Part I: Observational aspects. J Atmos Sci, 1976, 33: 1937–1954

8 Li C Y, Zhou Y P. Tropical atmospheric quasi-two weeks (10–20 days) oscillation (in Chinese). Chin J Atmos Sci, 1995, 19: 435–444

9 Yasunari T. A quasi-stationary appearance of 30 to 40 day period in the cloudiness fluctuations during the summer monsoon over India. J Meteorol Soc Jpn, 1980, 58: 225–229

10 Liebmann B, Hendon H H, Glick J D. The relationship between trop-ical cyclones of the western Pacific and Indian Oceans and the Mad-den-Julian Oscillation. J Meteorol Soc Jpn, 1994, 72: 401–411

11 Li C Y, Zhou Y P. Relationship between intraseasonal oscillation in the tropical atmosphere and ENSO (in Chinese). Chin J Geophys. 1994, 37: 17–26

12 Li G L, Li C Y. Activity of atmospheric intraseasonal oscillation and El Niño (in Chinese). J Tropical Meteorol, 1998, 14: 54–62

13 Hendon H H, Liebmann B, Newman M, et al. Medium-range fore-casts errors associated with active episodes of the Madden-Julian Os-cillation. Mon Weather Rev, 2000, 128: 69–86

14 Slingo J M, Coauthors. Intraseasonal oscillation in 15 atmospheric general circulation models: Results from an AMIP diagnostic sub-project. Clim Dyn, 1996, 12: 325–357

15 Jia X L, Li C Y. Sensitivity of numerically simulated tropical in-traseasonal oscillations to cumulus schemes (in Chinese). Acta Mete-orol Sin, 2007, 65: 837–855

16 Chao W C, Deng L. Tropical intraseasonal oscillation, supper cloud clusters, and cumulus convection schemes. Part II: 3D aquaplanet simulations. J Atmos Sci, 1998, 55: 690–709

17 Wang W, Schlesinger M E. The dependence on convection parame-terization of the tropical intraseasonal oscillation simulated in the UIUC 11-layer atmospheric GCM. J Clim, 1999, 12: 1423–1457

18 Maloney E D, Hartmann D L. The sensitive of intraseasonal variabil-ity in the NCAR CCM3 to changes in convection parameterization. J Clim, 2001, 14: 2015–2034

19 Wang B, Xu X H. Northern hemisphere summer monsoon singulari-ties and climatological intraseasonal oscillation. J Clim, 1997, 10: 1071–1085

20 Yang J. Climatological and transient intraseasonal oscillations in the East Asia-Western North Pacific summer monsoon (in Chinese). Dissertation for the Doctoral Degreee. Beijing: Institute of Atmos-pheric Physics, Chinese Academy of Sciences, 2008. 15

21 Liebmann B, Smith C A. Description of a complete (interpolated) outgoing long-wave radiation dataset. Bull Amer Meteorol Soc, 1996, 77: 1275–1277

22 Kanamitsu M, Ebisuzaki W, Woollen J. NCEP/DOE AMIP-II rea-


nalysis (R-2). Bull Amer Meteorol Soc, 2002, 83: 1631–1643 23 Zeng Q C. Characteristic parameter and dynamical equation of at-

mospheric motions (in Chinese). Acta Meteorol Sin, 1963, 33: 472– 483

24 Edwards J M, Slingo A A. Studies wit h a flexible new radiation code. I: Choosing a configuration for a large-scale model, Q J R Meteorol Soc, 1996, 122: 689–720

25 Xue Y, Sellers P J, Kinter J J, et al. A simplified biosphere model for global climate studies. J Clim, 1991, 4: 345–364

26 Liu H, Wu G X. Impacts of land surface on climate of July and onset of summer monsoon: A study with an AGCM plus SSib. Adv Atmos Sci, 1997, 15: 410–423

27 Zhou T, Yu R, Wang Z. The atmospheric general circulation model SAMIL and its associated coupled climate system model FGOALS-s (in Chinese). In: Impacts of the Ocean-Land-Atmosphere Interaction Over the Asian Monsoon Domain on the Climate Change Over China. Beijing: China Meteorological Press, 2005. 288

28 Manabe S, Smagorinsky J, Strickler R F. Simulated climatology of general circulation model with a hydrological cycle. Mon Weather Rev, 1965, 93: 769–798

29 Zhang G J. Convective quasi-equilibrium in mid-latitude continental environment and its effect on convective parameterization. J Geophys Res, 2002, 107: 4220, doi: 10.1029/2001JD001005

30 Zhang G J, Mu M Q. Simulation of the Madden-Julian Oscillation in the NCAR CCM3 using a revised Zhang-McFarlane Convection Pa-rameterization Scheme. J Clim, 2005, 18: 4046–4064

31 Liu Y M, Liu K, Wu G X. The impacts of the cumulus convective parameterization on the atmospheric water-content and rainfall simu-lation in SAMIL (in Chinese). Chin J Atmos Sci, 2007, 31: 1201– 1211

32 Yasunari T. Cloudiness fluctuations associated with the Northern Hemisphere summer monsoon. J Meteorol Soc Jpn, 1979, 57: 227 –242

33 Wang B, Ding Q H, Joseph P V. Objective definition of the Indian Summer Monsoon onset. J Clim, 2009, 22: 3303–3316

34 Li C Y, Jia X L, Ling J, et al. Sensitivity of MJO simulations to dia-batic heating profiles. Clim Dyn, 2009, 32: 167–187

35 Yang X Q, Huang S S. Intraseasonal Oscillations in air-sea coupled system (in Chinese). J Tropical Meteorol, CNKI: SUN: RDQX.0. 1993-03-001

SCIENCE CHINA Earth Sciences - LASGstaff.lasg.ac.cn/ueditor/php/upload/file/20180927/... · 2018. 9. 27. · showed that if AGCM utilized different cumulus parame-terization schemes,

Documents