QIN Danyu,LI Bo?,and HUANG YongKey Laboratory of Radiometric Calibration and Validation for Environmental Satellites,China Meteorological Administration,Beijing 100081

2Institute of Satellite Meteorology,National Satellite Meteorological Center,China Meteorological Administration,Beijing 100081

3School of Atmospheric Physics,Nanjing University of Information Science and Technology,Nanjing 210044

4Anhui Institute of Meteorology Science,Hefei 230031

1.Introduction

The East Asian monsoon region is surrounded by the high Tibetan Plateau to the west and the western Pacif i c Ocean to the east,as well as the complex coastlines along its southeastern edge.The unique orographic forcing and strong thermal contrasts between the ocean and the continent produce the notable East Asian monsoon,distinctive from other parts of the global monsoon system(Chen and Chang,1980;Tao and Chen,1987;Lau et al.,1988;Ding,1992).Mei-yu,which often occurs during late May and early July,is a typical weather and climate phenomena caused by the East Asian summer monsoon and has always been an important object of meteorological research.So far,studies on mei-yu have achieved fruitful results in terms of mesoscale meteorology,synoptic meteorology and climatology(Si,1989;Wang and Lin,2002;Zheng et al.,2008;Liang et al.,2010;Li and Zhou,2011),as well as its reproducibility and predictability in numerical simulations(Wang et al.,2004a,2005,Kitoh and Kusunoki,2008;Kusunoki and Mizuta,2008;Wang et al.,2009a,2009b;Fang et al.,2010,2013).

Studies on mei-yu can be conducted from different perspectives.On large scales,studies tend to concern the relationship between mei-yu and the East Asian summer monsoon(Tao and Chen,1987).In addition,research focused on the structure of the meso-γscale system in the mei-yu front has also been conducted(Yoshiaki et al.,2000).More studies are associated with synoptic-scale(Si,1989;Jiang and Ni,2004),meso-α-and meso-β-scale(Li et al.,1993;Ma and Zhao,1993;Xiang et al.,1993;Ninomiya,2000;Zhang et al.,2004)research of mei-yu and its numerical simulation(Wang et al.,2004a;Sun et al.,2007).In recent years,there have also been several studies emphasizing the multi-scale characteristics of the mei-yu front and its precipitation system(Ninomiya and Shibagaki,2007;Liu and Tan,2009).

Research on mei-yu can be conducted from multiple different views,among which the mei-yu front cloud system(MFCS)is often taken as a focus.As one of the important members of the mei-yu system,the scienti fi c signi fi cance of the MFCS is that: fi rst,storm clouds usually originate in the MFCS,and studying the characteristics is helpful for a better understanding of the generation,development and evolution of the storm clouds;and second,the formation of the MFCS is not only associated with the rainy weather system,but also involvescomplexphysicalprocesses,andsoresearchinsights can reveal the main physical processes of the MFCS.

Geostationary meteorological satellites have a unique advantage in terms of atmospheric observations due to their high spatial and temporal resolutions,as well as their wide fi eld of view.So far,observations from geostationary satellites have revealed many previously unknown facts and confi rmed many fi ndings.Most directly,black body temperature(TBB)is the manifestation of convective activities.Usually,areas with TBBs lower than ?32°C(about 241 K)correspond to active convection zones,while TBBs lower than?52°C(about 221 K)indicate strong active convection(Yao et al.,2005).Satellite observations,taking cloud images as representative,not only play an important role in day-to-day weather forecasting,but also re fl ect cloud distributions and the textural features,as well as the extinction and evolution of cloud,helpful in understanding the atmospheric state and ongoing physical processes such as dynamic and thermodynamic changes.

Studies in China of the mei-yu system using satellite observations have mainly concentrated on the mei-yu front rainstorm cloud cluster and the associated favorable environmental conditions.For example,much research has been devoted to the rainstorm and large-scale cloud features during the Yangtze River fl oods in 1991 and 1998(Li et al.,1993;Ye et al.,1993;Zheng et al.,1998;Shi et al.,2000;Yi and Xu,2001).Moreover,research on water vapor transport and wind fi elds in the rainy period have also been conducted(Lau and Chen,2004;Qin et al.,2004).So far,however,there have been fewer results reported from the perspective of satellite synoptic research associated with the MFCS,which is a mismatch considering the powerful observational capabilities of satellites today.

The active rainy period during a mei-yu process usually lasts a long time,undergoing constant changes in both intensity and morphology,accompanied by the establishment and reconstruction of the mei-yu front.Evolution of the MFCS is actually the result of interactions among the upper trough,the subtropical high,the quasi-stationary ridge and the rain band in speci fi c conditions(Ninomiya,2001;Qin et al.,2006).In fact,each dominant mode of mei-yu corresponds to a pattern of con fi guration and the in fl uences of different circulation systems.For example,interactions between the southerly on the west side of the western North Paci fi c subtropical high(WNPSH)and the northerly fl ow is the direct reason for the fi rst dominant mode of mei-yu(Qin et al.,2012).So,the purpose of the current study was to extract the leading modes during the active phase of the MFCS through EOF method,and to understand the transition processes from the typical fi rst mode to other leading modes,especially the accompanying circulation characteristics.In doing so,we hope to provide some reference for the predictability of mei-yu front activities.

The remainder of the paper is organized as follows.Section 2 describes the dataset and analysis procedures.Section 3 gives a description of the climatological and leading modes of the MFCS during the active phase of mei-yu days based on satellite observations.The transition processes from the typical mode to other modes are detailed in section 4.And fi nally,a summary and discussion are provided in section 5.

2.Data and method

2.1.Observational and reanalysis data

In this work,TBBs derived from three geostationary meteorological satellites,GMS-5(1998–2002),GOES-9(2003–2004)and FY-2C(2005–2008),were used to characterize the MFCS.The nadirs for GMS-5,GOES-9 and FY-2C are 140°E,155°E and 105°E,respectively.Diverse nadir positions lead to different observational angles for the same object.Meanwhile,different instrument performances,observational paths and observation times all result in differences among more than one satellite.As a result,data normalization is necessary when using observations from three satellites.

The normalization method used in this study was the cumulative distribution function(CDF).Without loss of generality,taking GMS-5 for example,the matching method adopted by CDF can be expressed by the following formula:

where TBBis the reference black brightness temperature,andis the adjusted correspondence of GMS-5.CGMS?5is the CDF of the TBBs for GMS-5,while Crefis the reference CDF from all GMS-5,GOES-9 and FY-2C TBB samples.By CDF-matching processing,the TBBs from GMS-5 can be adjusted to be consistent with the probability distribution of the reference.The same processes are then performed on GOES-9 and FY-2C TBB to complete the data normalization.More information is detailed in Qin and Li(2012).

Usually,a lower TBB indicates a higher altitude where the convection develops,especially in summer and over low latitudes.Before analysis,the normalized reconstruction was performed for the TBB data from the three different satellites,and then the latitude and longitude projection was carried out ontothedomainof(20°–45°N,100°–135°E).Ultimately,sixhourly data with spatial resolution of 0.1°×0.1°were formed and used for analysis.

In order to characterize the circulation in the active phase of the mei-yu period,NCEP II Reanalysis daily data(Kanamitsu et al.,2002)were adopted.Variables involved were the geopotential height,air temperature,relative humidity,and horizontal winds.

2.2.Analysis procedure

Taking the active period of the East Asian summer monsoon as the study focus,a total of 262 days were selected from the 18 mei-yu rainy active phases during 1998–2008 to be analyzed(Luo et al.,2012).Dates of the 18 rainy periods are shown in Table 1.

EOF decomposition is often applied to diagnostic analyses of meteorological series.For example,Zhou and Yu(2005)and Zhang et al.(2009)used the EOF method to isolate the leading modes of precipitation in China and the related patterns of water vapor transport in summer and spring,respectively.In the past,EOF was rarely used for weather research due to a lack of data;however,today,with increasingly rich model data and observations,the EOF method can also be applied to synoptic meteorological studies.For example,EOF analysis was performed to study a persistent heavy rainfall event that occurred from 21 to 22 July 1998 in Wuhan by use of high-resolution model outputs(Zhang et al.,2007).

At present,the EOF method can be applied to the longterm accumulation of satellite observations from an overall perspective.This isimportant forlearning about both the spatial and temporal characteristics of the MFCS,as well as the differences between the typical and atypical leading modes.In the current study,EOF analysis was performed using the six-hourly TBB anomalies over the active periods of meiyu during 1998–2008 to extract the principal modes of the MFCS.In the following discussion,the f i rst six EOF modes formei-yudaysarereferredtoasEOF1,EOF2,EOF3,EOF4,EOF5 and EOF6,respectively.

3.The climatology and leading modes of the MFCS

3.1.Climatology

From the cloud images provided by the geostationary meteorological satellites,the rainy region of the mei-yu front is a belt extending from South China and the Yangtze River Basin to the Japanese archipelago with low values of TBB(Nagata andOgura,1991).Thecloudbandofthemei-yufrontisoften stable and less dynamic.Many meso-scale convective systems(MCSs)are distributed within the cloud belt,and produce uneven precipitation both on the temporal and on the spatial scale(Yao et al.,2005).In terms of the temporal distribution,the cloud characteristics are different at each stage of the rainy period,mainly expressed by the direction and width of the cloud band as well as the changes in the convective activities(Ninomiya,2001).In terms of the spatial distribution,strong precipitation usually appears on the southern side of the MFCS belt,close to the transition zone with low values of precipitation echoes.Farther north,there is the wide stratiform precipitation zone and non-precipitating cloud anvil(Wang et al.,2009c;Fu and Qian,2011).

After the process of data quality control,there were 989 pieces of six-hourly TBB data during the active mei-yu period during 1998–2008.Before the leading modes are studied for the MFCS,we f i rstfocus on its climatological characteristics.As shown in Fig.1,during the active days of mei-yu,the high values of TBB are mainly located over the east of Taiwan and parts of western Inner Mongolia,while a TBB belt withlowvaluesstretchesfromthewesternpartoftheSichuan Basin to the southern regions of Japan.Five centers with TBBs lower than 253 K are located in the western Sichuan Basin,south of Yunan Province,the northwestern part of Hunan Province,southeastern Anhui,and South Japan,indicating more cloud and stronger convection activities over these areas.Previous studies considered the climatological distribution of the MFCS to be the result of the interaction between the cold and warm airf l ow over the Yangtze and Huai River Basin(Yao et al.,2005).

The climatological distribution of TBB over East Asia and the adjacent oceans is consistent with the rain belt of the East Asian summer monsoon(Zhou and Li,2002),especially with respect to June and July precipitation(Wang et al.,2009b;Li and Zhou,2011),for most of the active rainy mei-yu days occur in June and July,as indicated in Table 1.

Table 1.Onset and end dates of the mei-yu periods over the Yangtze and Huai River Basin in each year of the period 1998–2008.Numbers in the parentheses are the number of rainy days during each active period.

Fig.1.The climatological distribution of TBBs(units:K)on the active mei-yu days during 1998–2008,as shown in Table 1.

3.2.Leading modes

After gaining an overall understanding of the TBB climatology,EOF analysis was applied to the 989 pieces of TBB anomalydata,andthedominantmodesoftheMFCSwereextracted.These dominant modes interact with the mean state of the mei-yu cloud system,ultimately constituting different conf i gurations and distributions of the MFCS.The north criteria calculation revealed that the f i rst six modes are independent of each other,while the variances explained by these modes are 9.76%,5.89%,4.85%,4.49%,3.93%,and 3.57%,respectively.The spatial distributions of the f i rst six independent modes are shown in Fig.2.Note that the distribution of the leading modes should be combined with the principal components(PCs)(not shown)for analysis.

As shown in Fig.2a,most of the area for EOF1 is dominated by negative TBB anomalies.The central belt of the negative values elongates from central China to South Japan,which is the so-called MFCS,while the positive anomalies only appear in a small region in the north of Korea.Since the negative belt is located on the southeast side of its climatological position,this mode can be concluded as being the southern mode.

The EOF2 mode is characterized by a dipole structure.An almost east–west distribution of negative TBB anomalies is located over the central and eastern part of the Sichuan Basin,Hubei,Anhui and Jiangsu provinces,and then on to Japan.The strongest negative TBB anomalies are located in Anhui and Jiangsu,consistent with the location of the Changjiang-Huaihe rainy area.On the other hand,the whole of southern China is occupied by positive TBB anomalies,so we can take EOF2 as an east–west mode.

The EOF3 mode has a triple structure.Central and North China are occupied by positive TBB anomalies,indicating suppressed convection and less precipitation as well as seriousdrought.ThecenterofnegativeTBBanomaliesislocated over the Yellow Sea and the Korean Peninsula,corresponding to the location of the Korean rainy season.This mode can be named the northeastern mode.

The EOF4 mode is characterized by a northwest and southeast dipole structure.The center of negative TBB anomalies is located in Shandong,Henan,Anhui,and the northern part of Jiangsu,with strong developing convection.For the EOF5 mode,central and northern China are dominated by unanimous positive anomalous TBBs,and a center of negative values is located over Jiangsu,Anhui,Shandong and Henan.In the following discussion,EOF4 and EOF5 will be referred to as the northern mode and the dual centers mode,respectively.EOF6 is characterized by positive TBB anomalies over northern,eastern and southern China,while Southwest China and the adjacent western North Pacif i c is occupied by negative TBB anomalies with more precipitation.Typical EOF6 is a tropical-lowinf l uenced mode(HUANG Yong,2013,personal communication).

In a future paper,the background circulation features corresponding to these six leading modes and the mostlikelihoodmodeswillbediscussedindetail;thecurrentpaper does not cover these aspects.

Fig.2.The spatial distributions of the f i rst six principal modes of the TBB anomalies(units:K)on the active rainy mei-yu period from 1998–2008.

4.Transition from the southern mode to other modes

4.1.Determination of mode transition

From the previous section,we note that the f i rst principal mode of TBB anomalies explains 9.76%of the total variance,and this is a typical mode of the MFCS.Taking two standard deviations of PCs as a standard,we calculated the occurrence frequencies for the f i rst six principal modes of TBB.For determining the remarkable occurrence time for the leading modes of EOF analysis,the conventional method is to take a certain threshold of PC according to the purpose of study.For example,two standard deviations of PC were taken by Cao et al.(2004)as the threshold for determining the anomalous years of annual mean air temperature in Jiangsu Province.In another study,Xiao and Liu(2004)chose the raw PC to decide drought and f l ood years.The raw PC greater(less)than 300(?300)indicated a f l ood(drought)year.In the present study,we took two standard deviations of PC as the threshold for judging the presences of positive and negative leading modes.At the moment that the standardized PC for a certain mode became greater than two,we def i ned that the mode had appeared.This threshold(two standard deviations of PC)is helpful for selecting the time intervals when the leading modes are most signif i cant.In addition,the samples obtained met the basic needs of analysis.There were a total of 142 time intervals accompanied by the above six modes.In the 18 rainy phases during 1998–2008,these six leading modes occurred a total of 39 times,and the duration times differed from each other.

The transition process of the MFCS dominant modes is associated with the decay and reconstruction of the mei-yu front as well as the evolution of different background circulationconf i gurations.Therefore,thestudyoftransitionsamong different MFCS leading modes has important signif i cance.Currently,if the standardized PC for mode A is greater than two and lasts for several time intervals,and at later moments the standardized PC for mode A is less than two while the standardized PC for mode B is greater than two,then we take this process as a transition process from mode A to mode B.Based on this approach,in the 18 mei-yu active periods we studied,there were four transitions from the southern mode to other leading modes,among which,two conversion processes were to the northeastern mode,one was to the f i fth mode,and one was to the sixth principal mode.

The standard deviations of PCs for the rainy active periods in which the transition from the southern mode to other leading modes occurred are shown in Fig.3.The two transitions from the southern mode to the northeastern mode occurred in the f i rst active rainy period of 1998 and the second rainy season of 1999 respectively,while the transition from the southern mode converted to the dual centers mode occurred in the second rainy active period of 1998,and the transition from the southern mode to the tropical-low-inf l uenced mode occurred during the f i rst active rainy period of 2008.The circulation evolutions accompanied by these mode transitions are detailed in the next section.

4.2.Transition from the southern mode to the northeastern mode

The two transitions from the southern mode to the northeastern mode that occurred in 1998 and 1999 are denoted as“case 1998”and “case 1999”respectively in the following analysis.

4.2.1.Case 1998

As shown in Fig.3a,in case 1998,the southern mode persisted between 0000 UTC 23 June and 1800 UTC 23 June 1998,lasting for four time intervals.The third leading mode of TBB anomalies occurred at 1200 UTC 25 June and persisted until 0600 UTC 26 June.The six-hourly infrared(IR)TBB and the central location of the WNPSH are shown in Fig.4.The center of the WNPSH is represented by the location of 5880 gpm.

At the time the southern mode dominated the MFCS,the main cloud belt stretched from South China to Japan on the northwestern f l ank of the WNPSH(Fig.4).Several small convective cloud cells with signif i cantly low TBBs were located within the cloud band.The western ridge of 5880 gpm was to the east of 128°E.The area in the control of the WNPSH was increasing with time,especially in the north–south direction,and the WNPSH had a signif i cant southwestward extension.At 1200 UTC 25 June,the western ridge of 5880 gpm was at 116°E and continued to move westward later.Along with the evolution of the WNPSH,convective activities in the western part of the cloud belt were weakening and moving southward,while the convective center over southern Japan developed northward gradually.At the moment the northeastern mode dominated,the northern and southern parts of the cloud band merged into a uniform whole.A large hooked cloud anvil appeared over Northeast China,suggesting the impact of the cold low pressure system on its west side.

To describe the guiding circulation features in conversion from the southern mode to the northeastern mode,the air temperature,geopotential height and horizontal wind at 500 hPa and 850 hPa are shown in Figs.5 and 6.In the mid-troposphere(500 hPa),at the moment the southern mode(Fig.5)dominated,there was a large cold low over the Mongolia region with a central temperature of about 250 K.Note that this trough of low pressure was a deep system,which was very strong throughout the whole troposphere.Central and eastern China were under the control of a shallow trough,which was not signif i cant in the upper troposphere.This low trough strengthened with time and had a lean structure.For the northeastern mode,the main anvil over northeastern China just appeared between the front f l ank of the trough and the northern f l ank of the WNPSH.

In the lower troposphere(850 hPa,Fig.6),at the early time of 23 June(Fig.6a),the circulation pattern over central and southern China was characterized by two highs and one low.One of the highs was located over the central and northern parts of the Tibetan Plateau,and the other was the huge and strong WNPSH.Between the two highs,there was a low located over the Sichuan Basin corresponding to the trough in the mid-troposphere(shown in Fig.5a).The main convective cloud cells accompanying the southern mode arose over central and southern China.The low over southern China developed and moved northeastward,then merged with the trough over the Mongolia region with time.

At 0000 UTC 26 June(Fig.6g),the northeastern mode had been fully established.The main circulation pattern in the lower troposphere was also one of two highs and one low.However,at this moment,the lower troposphere over most of western and central China was controlled by a high with the center stronger than 1500 gpm,which was slightly weaker than that under the region of the WNPSH.In contrast,there was a low trough over northern China.The strongest convective clouds occurred on the northeastern end of the cloud band(Fig.4).As mentioned above,the low trough moving southeastward and developing over the Mongolia region was a deep system,signif i cant and strong through the upper,midand lower troposphere.At 850 hPa,the cold lows from North China and the Sichuan Basin,respectively,merged and combined to eventually form a new low trough.In the lower troposphere(Fig.6),this low tilted and diverged forward during its eastward movement.This structure is usually conductive to convection.Precipitation over northeastern China and the adjacent areas only occurred during the southeastward development of the low trough.The southwesterlies sandwiched between the font f l ank of the low trough and the northwestern f l ank of the WNPSH delivered rich warm water vapor to northeasternChina.Coldairmovedsouthwardalongwiththe northerlies behind the low trough and met the warm wet f l ow coming from the southwest,creating favorable conditions for precipitation.In addition,interaction between the low at 850 hPa and the shallow low at 500 hPa caused the major precipitation systems and corresponding cloud systems to move eastward,leading to the huge cloud anvil over the Yellow Sea and the East China Sea.

The westerly jet in the upper troposphere is also a key factor of the mei-yu circulation system.At 200 hPa(Fig.7),positive curvature of the eastern section of the subtropical westerly jet increased and the jet moved northward gradually,enhancing the instability of the atmosphere.Before the appearance of the northeastern mode,warm and humid air continuously injected into the eastern part of the westerly jet.This would have strengthened the moist baroclinicity and brought about a favorable background for convection,ultimately leading to the huge cloud anvil of the EOF3 mode.

4.2.2.Case 1999

In case 1999,shown by Fig.3b,the southern mode dominated the MFCS at 0000 UTC 22 June 1999 and lasted for only one period of time.The northeastern mode occurred at 1800 UTC 22 June and persisted for four time intervals until 1200 UTC 23 June.The distributions of the six-hourly TBBs and the central positions of the WNPSH(denoted by the 5880 gpm contour)are shown in Fig.8.

When the southern mode dominated the TBB anomalies,the western ridge of the WNPSH was at 130°E(Fig.8),similar to in case 1998.There was a cloud band along the northwestern f l ank of the WNPSH,extending from South China to the Japan Sea.Later,the WNPSH strengthened and moved westward.After the northeastern mode presented,several small convective cloud cells increased and gradually converged to form a belt of convection with strong lower TBBsstretchingfromthewestoftheWNPSHtonortheastern China.In case 1999,the westward movement of the WNPSH was not as signif i cant as that in case 1998,and there was no such huge hooked cloud anvil over North China.

Fig.3.The standardized PCs in the active mei-yu periods showing the transition from the southern mode(EOF1)to other leading modes of TBB anomalies on the rainy active mei-yu periods from 1998–2008:(a)PC1 and PC3 in the f i rst active mei-yu period in 1998;(b)PC1 and PC3 in the second active mei-yu period in 1999;(c)PC1 and PC6 in the second active mei-yu period in 1998;and(d)PC1 and PC5 in the f i rst active mei-yu period in 2008.The horizontal axis represents the number of time intervals in each of the active mei-yu periods.The vertical axis represents the standardized PCs.

Fig.4.TBBs from the geostationary satellites(shaded)and the locations of 5880 gpm(black solid line)from 0000 UTC 23 June to 1200 UTC 26 June 1998.

Fig.6.The same as Fig.5,except for 850 hPa.

Fig.7.The horizontal wind(vectors,units:m s?1)and the relative humidity(shaded,units:%)at 200 hPa from 0000 UTC 23 June to 0000 UTC 26 June 1998.

In the mid-troposphere(500 hPa,Fig.9),there was no closed low pressure moving southeastward as in case 1998.Over North China,a truncated low developed and moved eastward like the situation in case 1998,except for its northerly position.The cloud anvil in the northeast of the northeastern mode only occurred in the southwesterlies between the east f l ank of the low and the front f l ank of the WNPSH.In addition,the location of 5880 gpm indicates that the WNPSH underwent a north–south extension,similar to case 1998.In contrast to case 1998,before the advent of the northeastern mode,there was no apparent strengthening low over North China and Mongolia in the upper troposphere.At 200 and 300 hPa(not shown),the geopotential height was both smooth and straight,relatively.

Whenthesouthernmodedominated(0000UTC22June),the circulation pattern in the lower troposphere(700 and 850 hPa)(not shown)was characterized by two highs and one low.One of the highs was the weak high over the Tibetan Plateau with a central strength of about 1500 gpm,and the other was the powerful WNPSH.Between the two highs was a belt with low pressure with two low centers located over the southern part of the plateau and southeastern China,respectively,consistent with the strip of convection caused by the southern mode.This was similar to the situation in case 1998.After the establishment of the northeastern mode(Figs.9d andf),thecirculationoverEastChinacould alsobedescribed as being characterized by two highs and one low.At that moment,the western high extended from the Mongolian region tocentralChina,butitsintensitywasrelativelyweaker,witha central strength of 1460 gpm.The low sandwiched between the high zone and the WNPSH was located in northeastern China,northern China and the Bohai Sea,and the central strength was about 1400 gpm.Furthermore,there was also a closed low pressure system over southeastern China with a central strength of 1380 gpm.Therefore,there were two convective centers when the northeastern mode dominated,which were respectively located over the northeastern and southwestern part of the cloud band.For the diverse locations and intensities of the low pressure centers between case 1999 and case 1998,the corresponding northeastern mode was slightly different.

Fig.8.TBBs(shaded,units:K)and the location of 5880 gpm(black solid line)from 0000 UTC 22 June to 1200 UTC 23 June 1999.

We also examined the development and evolution of the westerly jet at 200 hPa.As shown in Fig.10,the positive curvature over the eastern section of the westerly jet strengthened with time,although weaker than that in case 1998.Before the appearance of the northeastern mode,a warm air current with high humidity f l owed into the eastern part of the westerly jet and further enhanced the baroclinicity in the exit area.This was conducive to the formation of precipitation,and f i nally led to the existence of the cloud anvil in the northeastern mode over northeastern China and adjacent regions.Because of the relatively weaker airf l ow humidity and the weaker positive curvature over the eastern section of the upper jet than in case 1998,the area of the cloud anvil in the northeasternpartofthecloudbandwasrelativelysmaller,and the strength of convection a little weaker than in case 1998.

4.3.Transition from the southern mode to EOF5 and EOF6

4.3.1.Transitionfromthesouthernmodetothedualcenters mode

A typical dual centers mode is characterized by aλshaped cloud distribution.The left branch of the cloud band extends from the Bay of Bengal to Northeast China,while the right branch stretches from Northeast China to the northwestern Pacif i c.In the current study,there are two approximatelyparallelcloudbandsatthemomentswhenEOF5dominates(not shown),so it was not a typical dual centers mode.Thus,the transition from the southern mode to the dual centers mode is not discussed in any further detail in this paper.

Fig.9.Distribution of air temperature(shaded,units:K),geopotential height(red solid line,units:gpm)and horizontal wind(vectors,units:m s?1)at 500 hPa from 0000 UTC 22 June to 0600 23 June 1999.

Fig.10.The horizontal wind(vectors,units:m s?1)and relative humidity(shaded,units:%)at 200 hPa from 0000 UTC 22 June to 0600 UTC 23 June 1999.

4.3.2.Transition from the southern mode to the tropicallow-inf l uenced mode

During the 18 active mei-yu rainy periods from 1998 to 2008,the southern mode converted to the tropical-lowinf l uenced mode only once.Thus,the process is not universal.AsshowninFig.3c,itwas1200UTC21July1998when the f i rst PC was the strongest,and the tropical-low-inf l uenced mode dominated from 1800 UTC 21 July until 0000 UTC 22 July,lasting for two time intervals(about 12 h).From the time series evolution of Fig.3c,we can see that the southern mode was mixed with the tropical-low-inf l uenced mode,and the separation between the two modes was not significant.So,in this case,the cloud distribution was not exactly the same as the typical EOF6 mode,which occurred in the early morning of 12 July 2007(HUANG Yong,2013,personal communication).Therefore,the transition from the southern mode to the tropical-low-inf l uenced mode is neither universal nor representative.

5.Summary and discussion

5.1.Summary

Based on the 18 active mei-yu rainy seasons during 1998 and 2008,the leading modes of the MFCS were extracted from the TBB datasets from three geostationary meteorological satellites by the EOF method.After calculating the statistics for the occurrences of each independent mode,the transition processes from the typical southern mode(EOF1)to other principal modes were discussed.The main fi ndings can be summarized as follows:

(1)The climatological distribution of the MFCS is characterized by a band with low TBBs extending from the western part of the Sichuan Basin to southern Japan with fi ve centers of low TBB values.

(2)The fi rst six modes of the MFCS are independent modes.According to the spatial distributions of TBB anomalies,the fi rst six leading modes can be named as the southern mode,the east–west mode,the northeastern mode,the northern mode,the dual centers mode,and the tropical-lowin fl uenced mode.

(3)The fi rst six leading modes occurred 39 times in total in the 18 active mei-yu periods.During 1998 to 2008,two transitions occurred from the southern mode to the northeastern mode,one transition between the southern mode and the dual centers mode,and one between the southern mode and the tropical-low-in fl uenced mode.

(4)During the transition processes from the southern mode to the northeastern mode,a low developed and moved eastward over central and eastern China in the midtroposphere(500 hPa).The low troposphere was occupied by two highs and one low over central and southern China.Cold air fl owed southward with the northerlies on the hindering fl ank of the low trough,and met the warm moisture sandwiched between the front fl ank of the trough and the WNPSH.Positive curvature in the eastern section of the upper westerly jet increased with time,and the injection of airfl ow with high humidity enhanced the baroclinicity of the jet.All of the above factors together producd a favorable environment for continuous convection and precipitation.The central area of the WNPSH increases while the western ridge extends westward and the northern boundary lifts northward.

(5)The transition processes from the southern mode to the dual centers mode and the tropical-low-in fl uenced mode are neither representative nor universal because of the lack of suf fi cient samples.

5.2.Discussion

It is worth noting that,in our analysis,the two cases of transition from the southern mode to the northeastern mode that were discussed were typical cases.That is,the cloud images were similar to the most typical cloud images and the two modes persisted for several time intervals and were not mixed with each other.This was the basis of the comparative analysis.Meanwhile,there were also some differences between case 1998 and case 1999.For example,case 1998 was accompanied by a deep cold low pressure moving eastward and shifting southward over the Mongolia region,while the cold low over the Mongolia region in case 1999 was shallow and only moved eastward.In the lower troposphere,although the circulation patterns in both cases were characterized by two highs and one low,the locations and strengths of the lows were not the same.In addition,in case 1999,a small closed low pressure system developed over time near the Tibetan Plateau,so the distributions of convective centers were broader compared to the situation in case 1998,and there were also signif i cant convective cloud clusters over the southwestern section of the main cloud band.

We also found that,in the analysis of mode transition,if at the moment the standard deviation of PC was larger than two,the standardized PC of another mode was also larger thantwo,meaningthatthetwomodesweremixed.Forexample,mixing of the southern mode and tropical-low-inf l uenced mode appeared in the second active mei-yu period in 1998,and the negative southern mode and negative northern mode were mixed in the f i rst mei-yu active period in 2004(not shown).In these cases,we cannot discuss the development and circulation evolutions in the transition processes.

Besides,for the determination of the dominant mode,two standard deviations of PC was taken as the threshold value.This led to the relatively small samples of conversion between two modes.In the 18 active mei-yu rainy periods during 1998 to 2008,there were four samples of transition from the southern mode to other modes,as mentioned in section 4.1.If the threshold value used is reduced appropriately,the results are different.For example,taking 1.8 as the threshold value,we f i nd that in the f i rst mei-yu period of 1998,EOF1 transits to EOF6 two times and to EOF3 one time;and in the f i rst mei-yu period of 1999,EOF1 transits to EOF3 three times.Thus,the numbers of transition samples should increase.However,for these transitions,the leading modes could not last long-term,and thus these short transition processes could not provide enough information for the circulation and convection changes in as detailed a manner as those outlined in the current study.Nevertheless,for transition processes among other leading modes,more appropriate threshold values should make the outcome of discussion on the evolution of circulation systems accompanied by mode transitions more robust and convincing.

Acknowledgements.This work was jointly supported by the National Natural Science Foundation of China(Grant No.40975023),the Special Promotion Program for Meteorology(Grant No.GYHY201406011 and No.GYHY201106044),and the NationalHighTechnologyResearchandDevelopmentProjectofChina(Grant No.2012AA120903).

Cao,N.H.,Z.W.Wu,X.Y.Chen,and G.P.Qian,2004:Variation of annual temperature anomaly in Jiangsu and its air-sea background features,1961–2000.Meteorological Monthly,30,31–34.(in Chinese)

Chen,G.T.J.,and G.P.Chang,1980:The structure and vortic-ity budget of an early summer monsoon trough(Mei-Yu)over southeasternChinaandJapan.Mon.Wea.Rev.,108,942–953.

Ding,Y.H.,1992:Summer monsoon rainfalls in China.J.Meteor.Soc.Japan.,70,373–396.

Fang,Y.J.,Y.C.Zhang,J.P.Tang,and X.J.Ren,2010:A regional air-sea coupled model and its application over East Asia in the summer of 2000.Adv.Atmos.Sci.,27,583–593,doi:10.1007/s00376-009-8203-7.

Fang,Y.J.,Y.C.Zhang,A.N.Huang,and B.Li,2013:Seasonal and intraseasonal variations of East Asian summer monsoon precipitation simulated by a regional air-sea coupled model.Adv.Atmos.Sci.,30,315–329,doi:10.1007/s00376-012-1241-6.

Fu,J.L.,and W.H.Qian,2011:The structure of a typical Mei-yu front identif i ed by the equivalent temperature.Atmos.Oceanic Sci.Lett.,4,109–113.

Jiang,J.Y.,and Y.Q.Ni,2004:Diagnostic study on the structural characteristics of a typical Mei-yu front system and its maintenance mechanism.Adv.Atmos.Sci.,21,802–813.

Kanamitsu,M.,W.Ebisuzaki,J.Woollewn,S.K.Yang,J.J.Hnilo,M.Fiorino,and G.L.Potter,2002:NCEP-DOE AMIP-II reanalysis(R-2).Bull.Amer.Meteor.Soc.,83,1631–1634.

Kitoh,A.,and S.Kusunoki,2008:East Asian summer monsoon simulation by a 20-km mesh AGCM.Climate Dyn.,31,389–401.

Kusunoki,S.,andR.Mizuta,2008:FuturechangesintheBaiurain band projected by a 20-km mesh global atmospheric model:Sea surface temperature dependence.SOLA,4,85–88.

Lau,K.H.,and S.J.Chen,2004:Water vapor winds in the vicinity of cloud bands along the Meiyu front.Acta Meteorologica Sinica,62,237–242.(in Chinese)

Lau,K.-M.,G.J.Yang,and S.H.Shen,1988:Seasonal and intraseasonal climatology of summer monsoon rainfall over East Asia.Mon.Wea.Rev.,116,18–37.

Li,B.,and T.J.Zhou,2011:ENSO-related principal interannual variability modes of early and late summer rainfall over East Asia in SST-driven AGCM simulations.J.Geophys.Res.,116,D14118,doi:10.1029/2011JD015691.

Li,Y.L.,S.Y.Tao,and C.X.Du,1993:An analysis of the mesoconvective cloud clusters in Mei-yu front.Quarterly Journal of Applied Meteorology,4,278–285.(in Chinese)

Liang,P.,Y.H.Ding,J.H.He,B.M.Chen,and X.T.Lei,2010:A study of determination index of regional Mei-yu over the Yangtze-Huaihe basin.Chinese J.Atmos.Sci.,32,418–428.(in Chinese)

Liu,J.Y.,and Z.M.Tan,2009:Mesoscale predictability of Mei-yu heavy rainfall.Adv.Atmos.Sci.,26,438–450,doi:10.1007/s00376-009-0438-9.

Luo,Y.L.,H.Wang,R.H.Zhang,W.R.Qian,and Z.Z.Luo,2012:Comparison of rainfall characteristics and convective properties of monsoon precipitation systems over South China and Yangtze-and-Huai River Basin.J.Climate,26,110–132.

Ma,Q.F.,and S.X.Zhao,1993:A study of Meiyu(Baiu)front and convective cloud clusters during TAMEX.Scientia Atmospheric Sinica,17,173–184.(in Chinese)

Nagata,M.,and Y.Ogura,1991:A modeling case study of interaction between heavy precipitation and a low-level jet over Japan in the Baiu season.Mon.Wea.Rev.,119,1309–1336.

Ninomiya,K.,2000:Large-and meso-α-scale characteristics of Meiyu/Baiu front associated with intense rainfall in 1–10 July 1991.J.Meteor.Soc.Japan,78,141–157.

Ninomiya,K.,2001:Largeλ-shaped cloud zone formed around July 6,1991 with pole-ward moisture transport from intense rainfall area in Meiyu-Baiu front.J.Meteor.Soc.Japan,79,805–813.

Ninomiya,K.,and Y.Shibagaki,2007:Multi-scale features of the Meiyu-Baiufrontandassociatedprecipitationsystems.J.Meteor.Soc.Japan,85,103–122.

Qin,D.Y.,J.X.Jiang,and Z.Y.Fang,2004:The characteristics of water vapor plume in the heavy rain events during 21–24 June 2002.Acta Meteorologica Sinica,62,329–337.(in Chinese)

Qin,D.Y.,and B.Li,2012:Matching FY-2 dual-Satellite infrared cloud top temperatures in typhoon monitoring application.Meteorological Monthly,38(12),1449–1455.(in Chinese)

Qin,D.Y.,Z.Y.Fang,and J.X.Jiang,2006:The cloud systems of heavy rainfall in the typical Meiyu period and their interactions.Chinese J.Atmos.Sci.,30,578–586.(in Chinese)

Qin,D.Y.,Y.Huang,and B.Li,2012:A typical cloud system of Mei-yu front.2012 Satellite Remote Sensing Application Technology Seminar Proceedings,105-112.(in Chinese)

Shi,C.X.,J.X.Jiang,and Z.Y.Fang,2000:A study on the features of severe convective cloud clusters causing serious fl ooding over Changjiang River Basin in 1998.Climatic and Environmental Research,5,279–286.(in Chinese)

Si,G.W.,1989:On the large-scale circulation of Mei-yu system over East Asia.Acta Meteorologica Sinica,47,312–323.

Sun,J.,X.F.Lou,Z.J.Hu,and S.X.Zhao,2007:A numerical simulation on torrential rain during the Meiyu period and analysis of mesoscale and microscale structure of convective systems.Chinese J.Atmos.Sci.,31,1–18.(in Chinese)

Tao,S.Y.,and L.X.Chen,1987:A review of recent research on the East Asian summer monsoon in China.Monsoon Meteorology,C.P.Chang and T.N.Krishnamuri,Eds.,Oxford University Press,New York,60–92.

Wang,B.,and H.Lin,2002:Rainy season of the Asian-Pacif i c summer monsoon.J.Climate,15,386–396.

Wang,B.,I.S.Kang,and J.Y.Lee,2004a:Ensemble simulations of Asian–Australian monsoon variability by 11 AGCMs.J.Climate,17,803–818.

Wang,B.,Q.H.Ding,X.H.Fu,I.-S.Kang,K.Jin,J.Shukla,and F.Doblas-Reyes,2005:Fundamental challenge in simulation and prediction of summer monsoon rainfall.Geophys.Res.Lett.,32,L15711,doi:10.1029/2005GL022734.

Wang,B.,and Coauthors,2009a:Advance and prospectus of seasonal prediction:Assessment of the APCC/CliPAS 14-model ensemble retrospective seasonal prediction(1980–2004).Climate Dyn.,33,93–117.

Wang,B.,J.Liu,J.Yang.T.J.Zhou,and Z.W.Wu,2009b:Distinctprincipalmodesofearlyandlatesummerrainfallanomalies in East Asia.J.Climate,22,3864–3875.

Wang,Y.,H.M.Xu,and Y.L.Luo,2009c:Structural characteristics of the cloud and heavy precipitation in a Mei-yu process.Science and Technology Information,12,13–14.(in Chinese)

Xiang,X.K,L.Ma,and D.C.Wang,1993:The mesoscale analysis of the Mei-Yu front cloud system in 1991.Quarterly Journal of Applied Meteorology,4,286–292.(in Chinese)

Xiao,X.Z.,and J.Liu,2004:The Climatic characteristics of summer precipitation of Longyan City.China Meteorological Society Annual Conference,Beijing,1–5.(in Chinese)

Yao,X.P.,Y.B.Yu,and B.K.Zhao,2005:Structural characteristic of Mei-yu frontal cloud system and its probable causes.Plateau Meteorology,24,1002–1011.(in Chinese)

Ye,H.M.,F.Y.Zhang,M.N.Ran,and X.X.Zhu,1993:The char-acteristics of the cloud system for Jiang-Huai severe rainfall in Jun of 1991.Quarterly Journal of Applied Meteorology,4,293–300.(in Chinese)

Yi,Q.J.,and X.D.Xu,2001:The propagation and development of cloud cluster systems and severe precipitation event in 1998.Climatic and Environmental Research,6,139–145.(in Chinese)

Yoshiaki,S.,D.Manabu,and S.S.Yamanaka,2000:Mesoβ to-γ-scale wind circulations associated with precipitation clouds near Baiu font observed by the MU and meteorological radars.J.Meteor.Soc.Japan,78,69–91.

Zhang,M.,J.An,and M.Zhu,2007:EOF expansion in one rainstorm.Chinese J.Atmos.Sci.,31,321–328.(in Chinese)

Zhang,J.,T.J.Zhou,R.C.Yu,and X.G.Xin,2009:Atmospheric water vapor transport and corresponding typical anomalous spring rainfall patterns in China.Chinese J.Atmos.Sci.,33,121–134.(in Chinese)

Zhang,X.L.,S.Y.Tao,and S.L.Zhang,2004:Three types of heavy rainstorms associated with the Meiyu front.Chinese J.Atmos.Sci.,28,187–205.(in Chinese)

Zheng,X.J.,Y.L.Li,and C.X.Du,1998:Analysis of GMS-5 water vapor images during torrential rain of Mei-yu in June 1995.Quarterly Journal of Applied Meteorology,9,246–250.(in Chinese)

Zheng,Y.G.,J.Chen,G.Q.Ge,Y.F.Huang,and C.X.Zhang,2008:Review on the synoptic scale Mei-yu front system and its synoptics’def i nition.Acta Scientiarum Naturalium Universitatis Pekinensis,44,157–164.(in Chinese)

Zhou,T.J.,and Z.X.Li,2002:Simulation of the East Asian summer monsoon using a variable resolution atmospheric GCM.Climate Dyn.,19,167–180.

Zhou,T.J.,and R.-C.Yu,2005:Atmospheric water vapor transport associated with typical anomalous summer rainfall patterns in China.J.Geophys.Res.,110,D08104,doi:10.1029/2004JD005413.