Robustness of Precipitation Projections in China: Comparison between CMIP5 and CMIP3 Models

CHEN Huo-Po and SUN Jian-Qi

Nansen-Zhu International Research Centre, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China

Robustness of Precipitation Projections in China: Comparison between CMIP5 and CMIP3 Models

CHEN Huo-Po and SUN Jian-Qi

Nansen-Zhu International Research Centre, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China

Three sources of uncertainty in model projections of precipitation change in China for the 21st century were separated and quantified: internal variability, inter-model variability, and scenario uncertainty. Simulations from models involved in the third phase and the fifth phase of the Coupled Model Intercomparison Project (CMIP3 and CMIP5) were compared to identify improvements in the robustness of projections from the latest generation of models. No significant differences were found between CMIP3 and CMIP5 in terms of future precipitation projections over China, with the two datasets both showing future increases. The uncertainty can be attributed firstly to internal variability, and then to both inter-model and internal variability. Quantification analysis revealed that the uncertainty in CMIP5 models has increased by about 10%-60% with respect to CMIP3, despite significant improvements in the latest generation of models. The increase is mainly due to the increase of internal variability in the initial decades, and then mainly due to the increase of inter-model variability thereafter, especially by the end of this century. The change in scenario uncertainty shows no major role, but makes a negative contribution to begin with, and then an increase later.

precipitation, projection, uncertainty, CMIP3, CMIP5

1 Introduction

Future climate changes will have significant impacts on society, particularly in relation to impacts on the economy, ecosystems, and health. In order to adapt to a changing climate, decision-makers require quantitative projections of climate on regional scales. Such projections are available from coupled global climate models (CGCMs; Sun et al., 2007, 2010; Gao et al., 2008; Chen and Sun, 2009, 2013; Jiang et al., 2009; Lu and Fu, 2010; Li et al., 2011; Ma et al., 2012; Zhang and Sun, 2012; Chen, 2013; Chen et al., 2013a, b; Jiang and Tian, 2013; Lang and Sui, 2013), but they possess large uncertainties (Tebaldi et al., 2006). These uncertainties have been identified to mainly come from three sources; namely, inter-model variability, internal climate fluctuation, and scenario uncertainty. More details about the separation of these three sourcescan be found in the following sections in the present study or the study by Hawkins and Sutton (2009).

Numerous studies have been performed on future precipitation change projections for China through CGCMs (e.g., Li, 2008; Xu et al., 2010; Chen et al., 2012; Xu and Xu, 2012a, b) and regional models (e.g., Shi et al., 2009; Gao et al., 2012). For example, Jiang and Fu (2012) showed that annual mean precipitation is expected to increase on average by 3.4%-4.4% over China with a 2°C increase in global temperature. Sun et al. (2010) indicated that days of intense snowfall events over southern China are expected to decrease, while over northern China they will initially increase and then decrease. Such results have been well reviewed by Ding et al. (2007) and Wang et al. (2012). However, the uncertainties associated with these projections have not been examined in depth. Additionally, new-generation models have been substantially improved compared to those included in the third phase of the Coupled Model Intercomparison Project (CMIP3), and therefore whether or not improvements in the reliability of future climate projections by those models included in the fifth phase of CMIP (CMIP5) have taken place should be investigated. In this context, the present paper attempts to quantify the uncertainties associated with future precipitation change in China and identify its reliability in both CMIP3 and CMIP5 simulations. These results will not only provide decision-makers with more reliable information for climate change adaptation measures, but will also help identify the greatest sources of uncertainties in future projections, thus providing a basis for trying to reduce them.

2 Data and methodology

2.1 Data

Uncertainties were investigated in future precipitation projections from models included in CMIP3 and CMIP5. The simulations used in the case of the CMIP3 models (15 models) were the 20th Century Climate in Coupled Models (20C3M) for the present and SRES (Special Report on Emission Scenarios) B1, A1B, and A2 for the future. For the CMIP5 models (20 models), historical experiments were used for the present, and RCP2.6, RCP4.5, RCP6.0, and RCP8.5 (Representative Concentrations Pathways) for the future. To examine the simulations by the Chinese models involved in CMIP3 and CMIP5, the simulations for one model (Flexible Global Ocean-Atmosphere-Land System Model, Grid Version 1(FGOALS1.0g)) in CMIP3 and six models (BCC-CSM1.1, BCC-CSM1.1-m, BNU-ESM (Earth System Model, Beijing Normal University), FGOALS-g2 (Flexible Global Ocean-Atmosphere-Land System Model, Grid Version 2), FGOALS-s2, and FIO-ESM) in CMIP5 were compared with models from other nations. However the simulations in SRES A2 scenario by FGOALS1.0g and RCP6.0 scenario by BNU-ESM and FGOALS-g2 are not available here, they are thus not included for the comparison analysis on the projection uncertainty between CMIP3 and CMIP5. The simple introduction about the other referred 20 CMIP5 and 15 CMIP3 models can found in Table 1. For convenience of analysis, all the simulations were bilinearly interpolated onto a common 2.5°×2.5° (latitude × longitude) grid.

Table 1 Simple introduction to the models used in this study.

2.2 Separating the sources of uncertainty

Uncertainty in climate projections mainly arises from three distinct sources (Hawkins and Sutton, 2009, 2011). The first is internal variability of the climate system, which is the wiggle superimposed on the long-term trend in each projection. The second is inter-model variability; that is, different models produce somewhat different changes in climate in response to the same radiative forcings (shown by the spread between similarly colored lines in Fig. 1). The third is scenario uncertainty, caused by the future emissions of greenhouse gases (indicated by the spread in the thick colored lines in Fig. 1). The fractions of total variance from these three sources were calculated in the present study to investigate their contributions to the total uncertainty. The separation method for the sources of uncertainty into individual components was introduced in previous studies (e.g., Hawkins and Sutton, 2009), but its application in the context of the present study is summarized as follows.

Firstly, the simulation for each model and each scenario were fitted, using ordinary least squares, with a fourth-order polynomial over the years 1950-2099,

where raw data are represented byX, the smooth fit is represented byχ, the residual isε, and the reference denoted bycis calculated from the mean of the years 1980-99 for CMIP3 and 1986-2005 for CMIP5. The subscripts“m”, “s”, and “t” denote models, scenarios, and years, respectively.

Then, the internal variability for each model was defined as the variance of the residuals from the fits, calculated independently of scenario and time. The multimodel mean of these variances is considered as the internal variability component,

where vartmeans the variance across time andViis thus constant in time.NmandNsare the number of models and scenarios, respectively.

The inter-model variability for each scenario was calculated from the variance in the different model projection (varm) fits (χm,s,t), and finally the multi-scenario mean was taken as the model uncertainty component:

The scenario uncertainty was estimated from the variance of the multi-model means for the scenarios (vars):

It was assumed that there was no interaction between the three uncertainty sources. Thus, the total variability is then

Figure 1 Coupled global climate model (CGCM) projections of changes in Chinese annual mean precipitation, relative to the mean of 1980-99 in the third phase of the Coupled Model Intercomparison Project (CMIP3) and 1986-2005 in the fifth phase of the Coupled Model Intercomparison Project (CMIP5), for historical forcings and different future emission scenarios. Each line represents a different CGCM for SRES B1, A1B, A2 in CMIP3, and Representative Concentrations Pathways (RCP) RCP2.6, RCP4.5, RCP6.0, RCP8.5 in CMIP5, with historical simulations shown in gray. The thick lines show the multi-model means for each scenario. The bars in each panel represent the range of projected changes for 2080-99 for different scenarios and the dots are the ensemble results. (a) 15 CGCMs in CMIP3; (b) FGOALS1.0g in CMIP3; (c) 20 CGCMs in CMIP5; and (d) six Chinese models in CMIP5. Units: %.

Finally, we also estimated the reliability of projections for the future precipitation changes. The reliability was calculated as the signal to noise ratio (S/N),

where Δχ(t) is the signal calculated as the difference relative to the present simulation from projection fits. More details about this method can be found in Hawkins and Sutton (2009).

3 Results

3.1 Changes in precipitation

The CGCM is one of the main tools available at present for projecting future climate change. These models generally show higher performance in simulating surface air temperature than precipitation with respect to observations. The uncertainty is obvious and it must be considered in future climate change projections in order to understand the robustness of these models.

Future changes in precipitation projected by CMIP3 and CMIP5 models were analyzed before investigating their uncertainties. Figure 1 exhibits such changes on the annual scale. In general, the models project an increase in area-averaged annual mean precipitation over China, particularly in multiple model ensemble (MME) results. However, there are some differences between these two datasets. For example, although both groups of models show a positive change in the future, both inter-model and scenario spreads for the CMIP5 models are much larger than for the CMIP3 models, especially at the end of the 21st century. This positive change happens over all seasons, but with larger inter-model and scenario spreads. Additionally, this projected increase in magnitude generally increases with the increase in greenhouse gas emissions both in the CMIP3 and CMIP5 models.

We also compared the simulations of Chinese-developed climate models involved in CMIP3 and CMIP5 with those developed in other nations. In the case of CMIP3, only one Chinese model (FGOALS1.0g) was included and it shows a future increase under both the B1 and A1B scenarios, but with weaker linear trends compared to the MMEs of 15 models. However, the projected changes under different scenarios by the MME of the six Chinese models involved in CMIP5 are much more comparable with the ensemble results of 20 CGCMs. Both the inter-model variability and the scenario spread are much smaller than that from the 20 models. Additionally, it is interesting to note that there is almost no difference in the projected changes of precipitation between scenarios before 2040, but afterward, the spread is shown to increase with time. A similar change happens in the Chinese models. However, it should be pointed out that among the six Chinese models, the model FIO-ESM shows a negative change in future annual mean precipitation, which is opposite to the other five models.

Both datasets show large regionality in their future precipitation projections over China (not shown). For example, the annual precipitation over some parts of southern China is predicted to initially decrease and then increase, while it mostly consistently and significantly increases over other regions. The increase in magnitude is predicted to be larger over northern China, especially over Northeast China, North China, and eastern Northwest China, but smaller over southern China and southern Northwest China. By the end of this century, the areaaveraged annual precipitation in China is expected to increase by 6.48%-9.76% under different emission scenarios according to CMIP3 models, and by 5.89%-12.73% according to CMIP5 models, relative to present simulations. Similar increases in magnitude can be obtained by the Chinese models. There is a 3.33% increase under B1 and 4.32% under A1B from the simulations of FGOALS 1.0g in CMIP3, and a 2.35%-8.99% increase from the MME of the six Chinese models in CMIP5.

3.2 Uncertainty and reliability

In this section we examine the sources of uncertainty in the future precipitation projections for China. Figure 2 shows the fraction of total variance in Chinese mean precipitation projections due to each source of uncertainty in the 21st century for annual and seasonal scales. It is clear that internal variability is the dominant source of uncertainty at annual and seasonal scales, which can explain more than 90% of the total uncertainty for the first decade ahead. The model uncertainty shows less contribution to the total uncertainty in the first decade, but it increases rapidly with time and can explain more than 50% by the end of the 21st century, except for the winter season. With the increase of inter-model variability, the fraction of internal climate fluctuation is observed to rapidly decrease. Scenario uncertainty is rarely a significant source of uncertainty, especially before the year 2040, in which there is almost no contribution to the total uncertainty. However, it also increases with time in the last several decades. These are the common features in the changes of the three components of uncertainty in the CMIP3 and CMIP5 simulations for Chinese area-averaged precipitation. However, there are also some differences between these two datasets. For example, compared to CMIP3, model and scenario uncertainties increase much faster in CMIP5 simulations, which is true for both annual and seasonal scales. Thus, model and scenario spreads can explain much more of the total uncertainty than that in CMIP3. This is also clear from Fig. 1, in which model and scenario uncertainties are indicated by the colored bars.

Maps of the fraction of variance explained by each source of uncertainty also show large regionality over China, both from the simulations of CMIP3 and CMIP5 (not shown). The results consistently indicate that internal variability explains more than 60% of the total uncertainty over virtually all regions in China in the initial decades. It then decreases, but can still explain more than 50% over most regions, apart from the region of Tibet. In contrast to internal variability, model uncertainty shows a significant increase with time over all regions, especially in the Tibetan Plateau, in which the model component explains more than 50% of the total uncertainty by the end of thiscentury. This large inter-model variability is closely associated with the difficulty of the coarse resolution of CGCMs in depicting the complex topography of this region for different models. Additionally, the scenario uncertainty is also much larger in this region by the end of this century, while only less than 10% of the total uncertainty can be explained over the other regions in China. These changes in spatial patterns of the three sources of uncertainty are not only obvious at the annual scale, but are also true for seasonal scales.

Figure 2 The fraction of total variance in future precipitation projections explained by internal variability (orange), inter-model variability (blue), and scenario uncertainty (green), for Chinese annual (top panels), summer (middle panels), and winter (bottom panels) means in CMIP3 (left column) and CMIP5 (right column). Units: %.

Compared with CMIP3, the CMIP5 models have been significantly improved, including (i) the addition of interactive ocean and land carbon cycles of varying degrees of complexity; (ii) more comprehensive modeling of the indirect effect of aerosols; (iii) the use of time-evolving volcanic and solar forcing in most models; and (iv) higher horizontal and vertical resolutions (Taylor et al., 2012). Thus, an important question arises: has the uncertainty of future climate projections been significantly reduced in the CMIP5 models? To address this, we further calculated the total uncertainties of Chinese area-averaged precipitation for both the CMIP3 and CMIP5 simulations. Unexpectedly, the uncertainty has not been reduced, and in fact has significantly increased. Furthermore, this increase in magnitude increases with time from about 10% for the first decade to more than 60% by the end of the century with respect to CMIP3 simulations (Fig. 3c). This is also true at seasonal scales, but with larger uncertainty than at the annual scales both for CMIP3 and CMIP5 simulations (not shown). Further estimations of the contribution of each source of uncertainty to the change of total uncertainty revealed that the increase of total uncertainty in the first three decades is mainly attributable to the increase in internal variability, while the changes in model and scenario uncertainty exhibit negative contributions, relative to CMIP3 simulations. However, with the increase of inter-model variability in the CMIP5 simulations, the contribution by the change in internal climate fluctuation decreases rapidly, but the contribution by the change in model uncertainty shows an obvious increase. Additionally, the change in scenario uncertainty also shows a positive contribution after around 2030, but the change in magnitude exhibits a much smaller increase than model uncertainty. By the end of this century, the increase in model uncertainty explains about 65% of the change of total uncertainty, internal variability contributes about 12%, and scenario uncertainty explains about 23%. This implies that model uncertainty is the dominant factor forthe increase in total uncertainty of the projected precipitation changes in China in CMIP5 simulations relative to CMIP3, although both inter-model and internal variability are the main sources of total uncertainty by the end of this century.

Figure 3 (a) The signal to noise ratio (S/N) in future precipitation projections in China for annual and seasonal means simulated by CMIP3 models, (b) is the same as (a), but simulated by CMIP5 models, (c) change in total uncertainty of CMIP5 models relative to CMIP3 models for annual mean precipitation in China and the contributions by the changes in internal variability, inter-model variability, and scenario uncertainty. Units: %.

The reliability of projections represented by theS/Nratio was also explored in both models’ datasets. The magnitude ofS/N, in general, is inversely proportional to the noise under similar signals. Equivalently, largerS/Ndenotes less uncertainty if the change signal is specified. Figures 3a and 3b showS/Nplots that vary with time for future Chinese area-averaged precipitation according to CMIP3 and CMIP5 simulations, respectively. As expected,S/Nfor future years is generally low at the regional scale, especially in the first several decades, in which the values are even less than 0.5. This is mainly due to the large internal variability and often small change signal of precipitation for the first decades. At the annual scale, the magnitude ofS/Nis predicted to exceed 1.0 at around 2060 in CMIP3 simulations, but at around 2050 in CMIP5 simulations. Furthermore, the magnitude ofS/Nis much smaller in CMIP5 than in CMIP3 because of the increased total uncertainty for the precipitation change in China. At seasonal scales, the magnitude ofS/Nis much smaller than at the annual scale and the values are almost less than 1.0, except for spring and summer, by the end of this century based on the CMIP3 simulations. Plots ofS/Nfurther show a smallerS/Nin autumn and winter than in spring and summer.

2.4 各组间生命体征比较T0时的RR、HR、MAP、SpO2比较，差异均无统计学意义（P＞0.05）。 各组患者T1-T3时的RR、MAP、HR、 均低于同组T0时，且T1、T2时 A 组患者 SpO2低于 B、C、D、E 组， 差异均有统计学意义 （P＜0.05）；T1-T3其余指标各组间比较，差异均无统计学意义（P＞0.05）。见表4。

Maps ofS/Nfor annual precipitation by the end of this century (not shown) indicate that there are few regions where the magnitude ofS/Nis above 1.0 in China, and even show some areas where the magnitude ofS/Nis less than 0.5 for both CMIP3 and CMIP5 simulations. Spatially, the regions of Tibet and Northeast China are highlighted as having a relatively highS/Nby the end of this century, while the regions of Southeast China and some areas of Xinjiang are reported as having relatively lowS/N, with less than 0.3 for both CMIP3 and CMIP5 simulations. It is worth noting that theS/Nin Tibet is much larger in CMIP5 simulations than in CMIP3 by the end of this century, while much less in Northeast China. This is mainly associated with the magnitudes of change signals in precipitation over these regions. These results are also true for seasonal scales.

4 Discussion and conclusion

Sources of uncertainty in future precipitation projections for China from CMIP3 and CMIP5 simulations were investigated. Both datasets were compared to identify improvements in the latest generation of models. The sources of uncertainty due to internal, inter-model, and scenario variability were examined for future precipitation changes in China.

The separation and quantification analyses revealed that, to begin with, both sets of simulations show internal variability as providing the greatest contribution (90%) toward total uncertainty. However, with the increase of inter-model variability, model and internal uncertainty become the dominant sources of the total uncertainty, while scenario uncertainty plays no major role, although it also increases with time. Further analysis showed no improvements in the robustness of projections from CMIP5 models compared to CMIP3, but with increased uncertainty of about 10%-60%. This increase of total uncertainty in CMIP5 is mainly attributed to the initial increased internal variability, and then due to the increased inter-model variability. Scenario uncertainty also shows an increase in CMIP5, and the contribution of its increase is even larger than internal variability by the end of this century. Further analysis of theS/Nratio revealed that the projected precipitation change at the annual scale is generally more reliable than at any seasonal scale, which is also true for the uncertainty analysis in both CMIP3 and CMIP5 simulations.

Compared to CMIP3, the latest generation models have been significantly improved, including the inclusion of new feedback processes (e.g., carbon cycle) and new parameterizations (e.g., cumulus and radiation schemes). However, perhaps these changes lead to new sources of model uncertainty, as identified in the present study. In addition, the increased uncertainty in the CMIP5 projec-tions may be partly associated with the larger sample size compared to CMIP3. Further investigations are clearly needed into the uncertainty changes among CMIP3 models and their updated versions in CMIP5.

Acknowledgements. This research was jointly supported by the National Basic Research Program of China (2012CB955401), the“Strategic Priority Research Program—Climate Change: Carbon Budget and Relevant Issues” of the Chinese Academy of Sciences (XDA05090306), and the Chinese Academy of Sciences-the Commonwealth Scientific and Industrial Research Organisation (CAS-CSIRO) Cooperative Research Program (GJHZ1223).

Chen, H. P., 2013: Projected change in extreme rainfall events in China by the end of the 21st century using CMIP5 models,Chinese Sci. Bull., 58(12), 1462-1472.

Chen, H. P., and J. Q. Sun, 2009: How the “best” models project the future precipitation change in China,Adv. Atmos. Sci., 26, 773-782.

Chen, H. P., J. Q. Sun, X. L. Chen, et al., 2012: CGCM projections of heavy rainfall events in China,Int. J. Climatol., 32, 441-450.

Chen, H. P., J. Q. Sun, and X. L. Chen, 2013a: Future changes of drought and flood events in China under global warming scenario,Atmos. Oceanic Sci. Lett., 6, 8-13.

Chen, H. P., J. Q. Sun, and X. L. Chen, 2013b: Projection and uncertainty analysis of global precipitation-related extremes using CMIP5 models,Int. J. Climatol., doi:10.1002/joc.3871.

Ding, Y. H., G. Y. Ren, Z. C. Zhao, et al., 2007: Detection, causes and projection of climate change over China: An overview of recent progress,Adv. Atmos. Sci., 24, 954-971.

Gao, X. J., Y. Shi, R. Song, et al., 2008: Reduction of future monsoon precipitation over China: Comparison between a high resolution RCM simulation and the driving GCM,Meteor. Atmos. Phys., 100, 73-86.

Gao, X. J., Y. Shi, D. F. Zhang, et al., 2012: Uncertainties of monsoon precipitation projection over China: Results from two high resolution RCM simulations,Climate Res., 52, 213-226.

Hawkins, E., and R. Sutton, 2009: The potential to narrow uncertainty in regional climate predictions,Bull. Amer. Meteor. Soc., 90, 1095-1107.

Hawkins, E., and R. Sutton, 2011: The potential to narrow uncertainty in projections of regional precipitation change,Climate Dyn., 37, 407-418.

Jiang, D. B., Y. Zhang, and J. Q. Sun, 2009: Ensemble projection of 1-3°C warming in China,Chinese Sci. Bull., 54, 3326-3334.

Jiang, D. B., and Y. H. Fu, 2012: Climate change over China with a 2°C global warming,Chinese J. Atmos. Sci.(in Chinese), 36, 234-246.

Jiang, D. B., and Z. P. Tian, 2013: East Asian monsoon change for the 21st century: Results of CMIP3 and CMIP5 models,Chinese Sci. Bull., 58(12), 1427-1435.

Lang, X., and Y. Sui, 2013: Changes in mean and extreme climates over China with a 2°C global warming,Chinese Sci. Bull., 58(12), 1453-1461.

Li, H. M., L. Feng, and T. J. Zhou, 2011: Multi-model projection of July-August climate extreme changes over China under CO2doubling. Part I: Precipitation,Adv. Atmos. Sci., 28, 433-447.

Li, S. L., 2008: Projecting summer climate of mainland China in the middle 21st century: Will the droughts in North China persist?Atmos. Oceanic Sci. Lett., 1, 12-17.

Lu, R. Y., and Y. H. Fu, 2010: Intensification of East Asian summer rainfall interannual variability in the twenty-first century simulated by 12 CMIP3 coupled models,J. Climate, 23, 3316-3331.

Ma, J. H., H. J. Wang, and Y. Zhang, 2012: Will boreal winter precipitation over China increase in the future? The AGCM simulation under summer ‘ice-free Arctic’ conditions,Chinese Sci. Bull., 57, 921-926.

Shi, Y., X. J. Gao, Y. G. Wang, et al., 2009: Simulation and projection of monsoon rainfall and rain patterns over eastern China under global warming by RegCM3,Atmos. Oceanic Sci. Lett., 2, 308-313.

Sun, J. Q., H. J. Wang, W. Yuan, et al., 2010: Spatial-temporal features of intense snowfall events in China and their possible change,J. Geophys. Res., 115, D16110, doi:10.1029/2009JD 013541.

Sun, Y., S. Solomon, A. Dai, et al., 2007: How often will it rain?J. Climate, 20, 4801-4818.

Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design,Bull. Amer. Meteor. Soc., 93, 485-498.

Tebaldi, C., K. Hayhoe, J. M. Arblaster, et al., 2006: Going to the extremes-an intercomparison of model-simulated historical and future changes in extreme events,Climatic Change, 79, 185-211.

Wang, H. J., J. Q. Sun, H. P. Chen, et al., 2012: Extreme climate in China: Facts, simulation and projection,Meteor. Z., 21, 279-304.

Xu, C. H., and Y. Xu, 2012a: The projection of temperature and precipitation over China under RCP scenarios using a CMIP5 multi-model ensemble,Atmos. Oceanic Sci. Lett., 5, 527-533.

Xu, Y., X. J. Gao, and F. Giorgi, 2010: Upgrades to the reliability ensemble averaging method for producing probabilistic climate-change projections,Climate Res., 41, 61-81.

Xu, Y., and C. H. Xu, 2012b: Preliminary assessment of simulations of climate changes over China by CMIP5 multi-models,Atmos. Oceanic Sci. Lett., 5, 489-494.

Zhang, Y., and J. Q. Sun, 2012: Model projection of precipitation minus evaporation over China,Acta Meteor. Sinica, 26, 376-388.

:Chen, H.-P., and J.-Q. Sun, 2014: Robustness of precipitation projections in China: Comparison between CMIP5 and CMIP3 models,Atmos. Oceanic Sci. Lett., 7, 67-73,

10.3878/j.issn.1674-2834.13.0071.

Received 26 July 2013; revised 17 September 2013; accepted 18 September 2013; published 16 January 2014

CHEN Huo-Po, chenhuopo@mail.iap.ac.cn