15OCTOBER2017 MURATA ET AL. 8237 Dominant Synoptic Disturbance in the Extreme Rainfall at Cherrapunji, Northeast India, Based on 104Years of Rainfall Data (1902–2005) FUMIEMURATA,aTORUTERAO,bHATSUKIFUJINAMI,cTAIICHIHAYASHI,d HARUHISAASADA,eJUNMATSUMOTO,fANDHIAMBOKJ.SYIEMLIEHg aFacultyofScienceandTechnology,KochiUniversity,Kochi,Japan bFacultyofEducation,KagawaUniversity,Takamatsu,Japan cInstituteforSpace-EarthEnvironmentalResearch,NagoyaUniversity,Nagoya,Japan dCenterforSoutheastAsianStudies,KyotoUniversity,Kyoto,Japan eNaraWomen’sUniversity,Nara,Japan fTokyoMetropolitanUniversity,Hachioji,andJapanAgencyforMarine-EarthScienceand Technology,Yokosuka,Japan gDepartmentofGeography,North-EasternHillUniversity,Shillong,India (Manuscriptreceived5June2016,infinalform18July2017) ABSTRACT Thecharacteristicsofactiverainfallspells(ARSs)atCherrapunji,northeastIndia,whereextremehigh rainfallisexperienced,andtheirrelationshipswithlarge-scaledynamicswerestudiedusingdailyrainfalldata from1902to2005andJapanese55-YearReanalysisfrom1958to2005.Extremehighdailyrainfallsoccurin associationwithARSs.Theextremelylargeamountsofrainfallinthemonsoonseasonaredeterminedbythe cumulativerainfallduringARSs.ARSsstartwhenanomalousanticycloniccirculation(AAC)at850hPa propagateswestwardfromtheSouthChinaSeaandwesternNorthPacific,andcoversthenorthernBayof Bengal.TheAACpropagatesfartherwestwardandsuppressesconvectionovercentralIndiaduringARSsat Cherrapunji,andcontinuesfor3to14days.Consequently,anorthwardshiftofthemonsoontroughduring the‘‘break’’intheIndiancoreregionoccurs.Thewesterlywind,whichprevailsinthenorthernportionofthe AAC,transportsmoisturetowardnortheastIndiaandenhancesmoistureconvergenceovernortheastIndia withsoutherlymoisturetransportfromtheBayofBengal,andgreatlyintensifiestheorographicrainfall.In theuppertroposphere,theTibetanhightendstoextendsouthwardwiththeonsetofARSs.Alinearre- lationshipcanbeseenbetweenthelengthandtotalrainfallofanARS.LongerARSstendtoresultingreater totalrainfall.AACswithagreaterzonalscaletendtoproducelongerandmoreintenseARSs.Thisstudy providesevidencefortheeffectofwesternNorthPacificAACsontheIndiansummermonsoon. 1. Introduction The synoptic condition during break spells has been described as a northward shift of the trough of low The Indian summer monsoon has remarkable intra- pressuretowardthefoothillsoftheHimalayas. seasonalvariation.Rainfallperiodswithaboveandbe- Ohsawaetal.(2000)studiedtheintraseasonalrainfall lownormalrainfallamountshavebeencalled‘‘active’’ variationduringthe1995summermonsoonseason,and and ‘‘break’’ spells, respectively. A detailed review of found that rainfall in Bangladesh increased when the studies of active and break spells was conducted by monsoontroughwaslocatedatthefootoftheHimalayas. Rajeevan et al. (2010). Traditionally, the active and Shrestha et al. (2012) analyzed rainfall over the central breakspellsoftheIndiansummermonsoonhavebeen Himalayanregionusing11years’worthofdatafrom based on the convective activity over northwest and the Tropical Rainfall Measuring Mission (TRMM) central India [referred to as the ‘‘Indian core region’’ PrecipitationRadar,andshowedthatrainfallwasheavy following Rajeevan et al. (2010)]. Negative (positive) overthesouthernmosthillsoftheHimalayas[calledthe rainfallanomaliesovernortheastIndiacorrespondedto Siwalik Range or sub-Himalayas; approximately 500– active (break) spells over northwest and central India. 700mabovemeansealevel(MSL)]whenthemonsoon trough was lying in the foothills of the Himalayas. Correspondingauthor:FumieMurata,[email protected] Goswamietal.(2010)analyzedtheextremerainfallover DOI:10.1175/JCLI-D-16-0435.1 (cid:1)2017AmericanMeteorologicalSociety.Forinformationregardingreuseofthiscontentandgeneralcopyrightinformation,consulttheAMSCopyright Policy(www.ametsoc.org/PUBSReuseLicenses). 8238 JOURNAL OF CLIMATE VOLUME30 northeastIndia, and showedthat these extreme events fromtheHimalayanrangesbytheBrahmaputraValley. occurred in association with monsoon synoptic events As the plateau is the first elevated terrain that warm, rather than isolated thunderstorms. In general, studies moistairfromtheBayofBengalencounters,maximum that have focused on intraseasonal variation over precipitation tends to occur (Houze 2012). It is con- northeastIndiahavebeenlimited. nected to the Indo-Burma Range in the east. The Hartmann and Michelsen (1989) performed spectral southernmarginoftheMeghalayaPlateauissteepand analysis of rain gauge data for 70 years from 3700 sta- ruggedduetothetectonicmotionalongtheDaukifault tionsinIndia,andnotedthatseveralstationsalongthe at the southern boundary of the plateau. There are southern flank of the Himalayas had spectral peaks of severaldeepvalleysthatopentowardthesouthwithsteep nearly 15 days. The 40–50-day spectral peak was not cliffsonbothsides(Fig.1b):thetownofCherrapunjiis significant along the southern flank of the Himalayas, situatednearonesuchvalley.Orographicamplification whereas it was dominant at most stations elsewhere in isthemajorplausiblecausefortheextremeprecipitation India. Ohsawa et al. (2000) detected a periodicity of at thislocation (e.g., Romatschke et al. 2010), although around 20 days in an analysis of the 1995 summer otherfactors,suchasalocal frontbetweenthesouth- monsoon season. Fujinami et al. (2011) analyzed rain westmonsoonflowandeast–northeastmountainwinds gaugedatafor20yearsinBangladesh,andshowedthata intheBrahmaputraValley,mayalsohaveaninfluence 7–25-dayspectralpeakwasdominant,whilea30–60-day (Yoshino1975). spectralpeakwasnotprominent.Fujinamietal.(2014) NortheastIndiareceivestheheaviestrainfallinIndia. further investigated the spatiotemporal structure of a Astheregionincludeshillyareasandadjoinsthegreat 7–25-daymodeovertheMeghalaya–Bangladesh–western Himalayas, orographic effects are important in rainfall Myanmarregion. innortheastIndia(Goswamietal.2010).Ourstudyre- A 10–20-daymode has beendetected in various ele- vealstherelationshipbetweenthelocalorographicrain- ments of the Indian summer monsoon system (e.g., fall and large-scale circulation patterns at intraseasonal Krishnamurti and Bhalme 1976). Past studies suggest time scales. Understanding the rainfall patterns over thatthe10–20-daymodeintheAsiansummermonsoon northeast India might be crucial in understanding the region originates over the equatorial west Pacific hydrological processes over the Ganges–Brahmaputra– (Annamalai and Slingo 2001), propagates westward MeghnaRiverbasin(HoferandMesserli2006).Forex- (Krishnamurti and Ardanuy 1980) with a double-cell ample,heavyrainfallduringARSsatCherrapunjidirectly structure(ChenandChen1993),andisamanifestation runs off toward Bangladesh and triggers flash floods. of the moist equatorial Rossby waves modified by the Also, the high monsoon rainfall years at Cherrapunji climatological mean flow (Kikuchi and Wang 2009). tendedtocorrespondwithseverefloodyearsinBangla- Fujinami et al. (2014) revealed that a westward- desh(Murataetal.2007). propagating anticyclonic anomaly related to the equa- Causes of the negative correlation between rainfall torial Rossby wave from the western North Pacific anomaliesintheIndiancoreregionandnortheastIndia promotes rainfall over the Meghalaya–Bangladesh– intheactive–breakcyclehavenotbeendiscussed.The westernMyanmarregion. ARSsatCherrapunjicanberepresentativeinnortheast Thisstudyanalyzesactiverainfallspells(ARSs)inthe India because the extreme rainfall amount provides monsoonseasonatCherrapunji,whichisatownlocated noiseless signalfor favorable synoptic-scaleevents. The in northeast India. Cherrapunji is a unique station be- objectiveofthisstudyistoanswerthefollowingquestions cause it holds the world record for the highest rainfall by understanding the characteristics of the Cherrapunji in a calendar month and for longer time scales than a ARSs(CARSs),anddiscusstherelationshipwithactive– calendarmonth(Jennings1950;KiguchiandOki2010), breakcycleintheIndiancoreregion: as well as the highest daily rainfall records in India 1) WhatistherelationshipbetweenCARSsandlarge- (Guhathakurta2007).RainfallatCherrapunjihasclear scaledynamics? ARSs, and daily rainfalls with more than several hun- 2) What is the contribution of CARSs to the extreme dredmillimetersarecommonduringARSs(Fig.3).The rainfall? large daily rainfall at Cherrapunji during the ARSs is suitable to detect the synoptic-scale events with high Section 2 describes the datasets used in this study. sensitivity becauseitcanminimizeothernoisescaused The definition and characteristics of CARSs are de- byfewlocalcumulonimbusclouds. scribedinsection3.Section4givesthespatiotemporal Cherrapunji is located on the southern slope of the structure of CARSs. The synoptic condition during MeghalayaPlateau(Fig.1a)withamaximumelevation CARSs,therelationshipwiththeIndiancoreregion,and ofaround2000mabovesealevel.Theplateauisdivided thecontributiontotheextremerainfallarediscussed 15OCTOBER2017 MURATA ET AL. 8239 FIG.1.(a)TopographicalmapoftheregionaroundCherrapunji(shownwiththestar). (b)LandscapeofthesoutherncliffoftheMeghalayaPlateaunearbyCherrapunjitown. in section 5. Finally, the results are summarized in from 1902–70, and the Guwahati regional office of the section6. IMD for daily rainfall data from 1970 to 2003. Daily rainfalldatafor2003–05wereobtaineddirectlyfromthe CherrapunjiObservatoryoftheIMD.Fiveyearsofdaily 2. Data rainfall data at Cherrapunji are missing completely Data on the daily rainfall observed at Cherrapunji (1965,1982–84and1987),and7years(1946,1960,1961, observatory of the India Meteorological Department 1963,1964,1972,and1986)havemorethan10%missing (IMD) were obtained from the following different data for June–September. To study the relationship sources: the Research Data Archive at the National between rainfall at Cherrapunji and that at other CenterforAtmosphericResearch(Computationaland regions, the IMD rainfall dataset (Pai et al.2014) ona Information Systems Laboratory; ds.480.0) for data 0.258 latitude–longitude grid was used for the period 8240 JOURNAL OF CLIMATE VOLUME30 Fujinamietal.(2011,2014)examinedthecharacteristics ofthe10–20-daymodeovertheMeghalaya–Bangladesh– westernMyanmarregion.Thisstudyfocusedonthetiming ofonset,length,andtotalrainfallamountofCARSsbe- causetheyaretheimportanthydrologicalcharacteris- ticsofthisregion. Figure 3 shows examples of the daily rainfall from 21 May to 10 October in the years 1915 and 1974. The annual rainfall in 1915 was almost mean value and in- cludedfouractivespells.Theannualrainfallin1974was the highest between 1902 and 2005. The solid curve il- lustrates the daily climatological rainfall: it shows the averagedailyrainfallineachyear,andsmoothingbythe 15-day running mean twice. The daily climatological rainfall reaches a maximum in late June, which was more than 1 month earlier than the rainfall over the FIG.2.The92-summerensemblespectrumoftheCherrapunji dailyrainfall(mm).Arednoisespectrum(dashedcurve)and95% Indiancoreregion(Rajeevanetal.2010;Paietal.2016). levelofsignificance(solidcurve)basedonalag-1autocorrelation Heavyrainfalldays(HRDs)aredayswithmorethan arealsoshown. 1.5timesthedailyclimatologicalrainfallbetweenJune and September. CARSs are defined as consecutive 1902–2005,andtheAPHRODITEV1101rainfalldata- HRDsonwhichthetotalrainfallexceeds 500mm. For set (Yatagai et al. 2009, 2012) on a 0.258 latitude– HRDperiodsthatinclude1Juneor30September,the longitude grid was used for the period 1958–2005. The number of consecutive HRDs extended into May or streamfunction,horizontalwindcomponentsandverti- October. There were only three (two) cases of 1-day cally integrated water vapor flux components of the (2-day) intervals between the two adjacent CARSs. Japanese 55-Year Reanalysis (JRA-55) (Ebita et al. Definitions of active–break spells can vary (Rajeevan 2011;Kobayashietal.2015)ona1.258latitude–longitude et al. 2010). Fujinami et al. (2014) defined active and grid were used to analyze the features of circulation breakmonsoonspellsaccordingtofiltereddata.Inthis patterns and moisture transport during CARSs. Daily study unfiltered data are used to define ARSs, with no interpolated outgoing longwave radiation (OLR) data assumptionregardingfrequency.Theshadedperiodsin ona2.58latitude–longitudegridwereusedtoshowlarge- Fig. 3 represent ARSs in this study. There were nine scale convective activity, including ocean areas from CARSs in 1974, which was the maximum number re- 1979 to 2005 (Liebmann and Smith 1996). In the com- cordedduringtheanalysisperiod.TheCARSsrepeated posite analysis, the statistical significance of the differ- periodically at intervals of 10–20 days, which is consis- encesfromclimatologicalvaluesateachgridpointwas tentwiththeresultofFig.2. estimatedusingtheStudent’sttest. Table 1 shows the relationship between the total rainfallonconsecutiveHRDsandthedurationofcon- secutiveHRDs.ThenumbersinthecolumninTable1 3. FeaturesofARSsatCherrapunji indicatetheperiodsofconsecutiveHRDsbetween1902 In this section CARSs are defined and described. and2005.Thenumberofperiodsdecreasesasthetotal Figure2showsthe92-summerensemblespectrumofthe rainfall increases. Regarding the length of a period, Cherrapunji daily rainfall. The years for which more there was a greater percentage of consecutive 3-day than10%oftheJune–Septemberdailydataaremissing HRDs. A linear relationship is observed between the were excluded; in total, 92 years were used, the first number of consecutive HRDs and the total rainfall. threeharmonicsoftheannualcyclewereremoved,and Greater rainfall is observed in the periods with more the fast Fourier transform technique was used to cal- consecutiveHRDs.Therearesomeextremecaseswhen culate the spectra. A 5-point running mean in the fre- thetotalrainfallwasbetween3500and5000mmduring quencydomainwasappliedtotherawspectratoreduce 7-or9-dayconsecutiveHRDs.Inthisstudy,CARSsare estimate errors. Pronounced peaks appear around a definedaseventswithmorethan500mmtotalrainfall, 10–30-day period and a peak of about 11 days exceeds andtheminimumlengthofaCARSis3days. the95%confidencelevel.Thespectrumissimilartothat Variation between intra- and interannual scales has of all Bangladesh daily rainfall (Fujinami et al. 2011) been considered. For example, Gadgil and Joseph and represents the dominance of the 10–20-day mode. (2003)showedthattheinterannualvariationinall-India 15OCTOBER2017 MURATA ET AL. 8241 FIG.3.Dailyrainfall(histogram;mm)for21May–10Octoberin(a)1915and(b)1974,andtheclimatologicaldaily rainfall(solidcurve;mm).Thegray-shadedareasrepresentCherrapunjiactiverainfallspells(CARSs). summer monsoon rainfall is related to the number of Figure5showstherelationshipbetweenthecumulative daysofrainbreaksandactivespells.Fujinamietal.(2011) number of CARSs and the cumulative rainfall amounts showedthatmorefrequentandstrongsubmonthly-scale duringCARSsoveroneyear.AsintheanalysisofFig.2, oscillations, and a greater number of rainy days, were data for 92 years were used. The bottom and top of relatedtothewetmonsoonyearsinBangladesh.Figure4 theboxarethefirstandthirdquartiles,respectively.The shows that the cumulative rainfall of CARSs is pro- thickbandinsidetheboxisthemedian.Theendsofthe portionaltototalmonsoonrainfallamountfromJuneto whiskersshowtheminimumandmaximum.Thenumber September. The cumulative rainfall during the period of CARSs ranges from one to nine per year. The years withoutCARSsrangesfrom2000to6000mmandhasa with four to nine CARSs over one year compose ap- negative tendency with monsoon total rainfall because proximatelyhalfofthetotalsampleyears.Intherangeof the number of days without CARSs decreases when onetofour spellsperyear, there isa tendencyfortotal CARSsincreases. rainfalltoincreasewiththenumberofCARSs.However, TABLE1.Relationshipbetweenthetotalrainfallofconsecutiveheavyrainfalldays(HRDs;horizontalline)andthelengthofconsecutive HRDs(verticalline).Asterisksdenotecellswithmorethan10periods. 14-day 1 1 13-day 0 12-day 3 1 2 11-day 0 10-day 5 1 3 1 9-day 15 5 3 1 5 1 8-day 13 1 3 5 3 1 7-day 28 1 3 *13 4 3 2 1 1 6-day 57 1 *10 *30 *14 2 5-day 90 2 *46 *28 *11 3 4-day 103 *15 *70 *18 3-day 167 *103 *64 2-day 82 *82 1-day 68 *68 Total 632 272 194 98 38 15 11 1 1 1 1 Total ,500 ,1000 ,1500 ,2000 ,2500 ,3000 ,3500 ,4000 ,4500 ,5000 Totalrainfall(mm) 8242 JOURNAL OF CLIMATE VOLUME30 FIG.5.Thedependenceofthecumulativetotalrainfallamounts (mm) during CARSs over 1yr on the number of CARSs. The FIG.4. Scatterplotsof monsoonrainfall(totalrainfallamount bottomandtopoftheboxarethefirstandthirdquartiles.Thethick fromJunetoSeptember;mm)andthecumulativerainfallwithand bandinsidetheboxisthemedian.Theendsofthewhiskersshow without CARSs (solid and open circles, respectively; mm). The theminimumandmaximumvalues.Thedatainparenthesesnext regressionlineisshownforwithinCARSs. tothenumberofCARSsonthehorizontalaxisshowthenumberof periods. thereisnocleartrendtowardanincreaseinrainfallwhen therearemorethanfourCARSsperyear.Itisnotalways andnorthoftherelativelylowrainfallarea.Theareaof thecasethatyearswithagreaternumberofCARSsare relativelylowrainfallisconsistentwiththelocationofthe wet monsoon years.Itimplies thatCARSswith greater monsoontroughorcontinentaltropicalconvergencezone totalrainfallscontributetowetmonsoonyearsinaddition inpaststudies(Gadgil2003).Usually,typhoonsandother toanumberofCARSs. cyclonicdisturbances,suchasdepressionsorlows,move Figure6showsacompositeoftheanomalousrainfall westward along the monsoon trough zone and produce from daily climatological rainfall over India during rainfalloverthemonsoontrougharea. CARSs.ThismapwasproducedbyusingtheIMDgrid Toexplaintheanomalousrainfalldistributionduring rainfall data for the period 1902 to 2005. Positive and CARSs(Fig.7),datafromcompositeanomalousstream- negativevaluesshowregionsofexcessandlessrainfall functionandcompositeanomalouswind(Figs.8a,b)and during CARSs, respectively. The hatched regions are significant at the 95% confidence level. While excess rainfallisobservedovernortheastIndiaduringCARSs, less rainfall is observed over central India and the Western Ghats. Usually, the active and break spells of the Indian monsoon are defined based on the rainfall over central India. Thus, it is important to note that northeast India generally has the opposite tendency to that of central India: the spatial distribution is very similartothatofthebreakmonsooncompositionofthe Indiancoreregion(e.g.,Paietal.2016). 4. Relationshipwithlarge-scalecirculationfield Figure7isthesameasFig.6,buttheAPHRODITE datasetisusedtoprovideanextendedviewovertheAsian monsoon region. Although the analysis period (1958– 2005)isshorterthanthatinFig.6,thedistributionofhigh andlowrainfallareasoverIndiaissimilar.Theregionwith significantly lower rainfall during CARSs extends longi- FIG. 6. Composite of anomalous rainfall (mm) from daily cli- tudinally over Pakistan, central India, the Indochina matologicalrainfallduringCARSsoverIndia.TheIndiaMeteo- Peninsula,andthenorthernPhilippines.Theregionwith rologicalDepartment(IMD)griddatasetwasused(1902–2005). significantly excessive rainfall during CARSs lies south Hatchedareasarestatisticallysignificantatthe95%level. 15OCTOBER2017 MURATA ET AL. 8243 analysisperiod,andthecompositethatexcludesCARSs is almost same as the climatological mean. There is a largedifferenceinthemoisturefluxfieldover208–308N, whilethereisalmostnodifferencesouthof208N,except that the moisture flux is somewhat smaller during CARSs. Climatologically, cyclonic circulation due to a monsoon trough appears over central India, and mois- ture transport is limited over this area (Fig. 9a). In contrast, a strong westerly moisture transport appears over central India, which intensifies moisture conver- genceovernortheastIndiaduringCARSs(Fig.9b).The location of moisture convergence over the foot of the central Himalayas is shifted slightly northward and in- tensifiesduringCARSs. Figure 10 is the same as Fig. 8, but shows the com- posites of the CARS onset days. In the lower tropo- sphere(Fig.10b),AACisdistributedoverthenorthern FIG. 7. Composite of anomalous rainfall (mm) from daily cli- matological rainfall during CARSs over the Asian monsoon re- Bay of Bengal, the Indochina Peninsula, the South gion.TheAPHRODITEdatasetwasused(1958–2005). China Sea, and the western North Pacific along 108– 208N.ComparedwithFig.8b,theAACdoesnotcover the Indian subcontinent. The location of the AAC is composite vertically integrated anomalous moisture flux almost consistent with the convective suppression rep- anditsdivergence(Fig.8d)wereproducedbyusingJRA- resentedbypositiveOLRanomalies(Fig.10c)andthe 55reanalysisdata.Theanalysisperiodwas1958–2005.The divergence of the anomalous moisture flux (Fig. 10d). compositeofanomalousOLRisalsoshown(Fig.8c)for The anomalous negative streamfunction is located to comparison with the convective activity, including ocean the south of the AAC over the eastern Indian Ocean, areas.Theanalysisperiodis1979–2005,duetothelimited andshowsthedoublecellstructure.Anotheranomalous satellitedata.Anomalousanticycloniccirculation(AAC) negative streamfunction is located around Japan. In appearsovercentralIndiaandthenorthernBayofBengal, the upper troposphere (Fig. 10a), the AAC is located atboth200-and850-hPalevels.Anomalouscycloniccir- just over northeast India where convection is active culationsappearoverthewesternequatorialIndianOcean (Fig. 10c). There are two other anomalous negative andtheTibetanPlateauatthe200-hPalevel.Thepositive streamfunctions situated in the northeast and south of OLR anomaly (Fig. 8c) and anomalous moisture di- theAAC.ThenegativeOLRanomaly(Fig.10c)andthe vergence (Fig. 8d) show intense convective suppression convergenceoftheanomalousmoistureflux(Fig.10d) over the western Indian subcontinent. In the lower tro- are distributed not only in northeast India but also ex- posphere(Fig.8b),theAACisextendedfarthereastover tend to southern China along the 208–308N zone; this the South China Sea; this location is consistent with the locationisconsistentwiththenorthernedgeoftheAAC area of relatively low rainfall in Fig. 7 and the positive atthe850-hPalevel(Fig.10b). OLRanomalyovertheSouthChinaSeaandthePhilip- Table 1 shows that greater total rainfall tends to be pines in Fig. 8c. The convergence of the anomalous observed during longer CARSs. The data were then moisture flux over northeast India (Fig. 8d) corresponds divided by the length of CARSs, and compared with well with active convection over northeast India during thoseinFig.11.Thecompositenumbersofperiodswere CARSs(Figs.6and7),althoughnegativeOLR(Fig.8c) 21,40,and9for3-,5-,and7-dayCARSs,respectively. does not represent well the convective activity over Figure 11 shows the composite of anomalous stream- northeastIndia.Theanomalousnegativestreamfunctionis function and anomalous wind data at 850hPa for located over the central equatorial Indian Ocean at the 3- (Figs. 11a,d,g), 5- (Figs. 11b,e,h), and 7-day CARSs 850-hPalevel,andrepresentsadouble-cellstructure. (Figs. 11c,f,i), respectively. The time evolution is also To further investigate how synoptic conditions pro- shownfromday22today12.Onday22(Figs.11a–c), duce moisture convergence over northeast India, a an AAC forms over the South China Sea. This AAC compositeofverticallyintegratedmoisturefluxanddi- propagates westward, and the onset of CARSs vergence during CARSs (Fig. 9b) is compared with a (Figs.11d–f)occurswhentheAACcoversthenorthern composite of climatological means throughout June to Bay of Bengal. The AAC moves farther westward on September(Fig.9a).CARSsoccupyonly8.5%oftotal day12(Figs.11g–i).Thewestwardpropagationspeedis 8244 JOURNAL OF CLIMATE VOLUME30 FIG.8.Compositeofanomalousstreamfunction(shading;106m2s21)andanomalouswind(vector;ms21)at (a)200-and(b)850-hPalevelsfromdailyclimatologicalvaluesduringCARSs.TheJRA-55datasetwasused (1958–2005).(c)Asin(a)and(b),butforoutgoinglongwaveradiation(OLR)(shading;Wm22).(d)Asin(a)and (b),butforverticallyintegratedmoistureflux(vector;kgm21s21)anditsdivergence(shading;1024kgm22s21). Hatchedareasarestatisticallysignificantatthe95%confidencelevel.Onlyanomalouswindvectorsthatshowed a95%statisticallysignificantdifferenceareplotted. intherangeof5–10ms21.Onday22,theAACextends A significant AAC over the Tibetan Plateau tends to morezonallytowardthewesternNorthPacificin7-day movesouthwardwiththeonsetofCARSs,andliesover periodsthanin 3-dayperiods.Furthermore,theAACs the convectively active area. The comparison between aremoreintensein7-dayperiodsthanin3-dayperiods. theonsetdayoftheCARSsandday12showsawest- Onday12,thewesterlyanomaliesstillremainoverthe ward propagation. The composite of 3- and 7-day western North Pacific along 208N in the 5- and 7-day CARSs showed similar AAC movement toward the periods.Inthe7-dayperiods,thecenteroftheAACis south, but they did not reach significance (figures not stilllocatedinthewesternNorthPacific. shown), possibly due to the relatively weak signal for Thetimeevolutionoftheverticalstructurewasana- 3-day CARSs and insufficient number of samples for lyzed.Thecompositesoftheanomalousstreamfunction 7-dayCARSs. and anomalous wind at 500hPa were almost the same as in Fig. 11 (figures not shown). Figure 12 shows the composite of the anomalous streamfunction and 5. Discussion anomalouswindat200hPain5-dayperiodsonday22 a. SynopticconditionsduringCARSs (Fig. 12a), the onset of CARSs (Fig. 12c), and day 12 (Fig.12e).TheyarecomparedwithanomalousOLRon The composite of the CARS onset days implies that day 22 (Fig. 12b), the onset of CARSs (Fig. 12d), and the CARSs begin when the AAC in the lower tropo- day 12 (Fig. 12f) to examine the effect of diabatic sphereisovertheBayofBengal(Fig.10b).ThisAAC heatingonthecirculationfieldintheuppertroposphere. signal formed in the South China Sea and the western 15OCTOBER2017 MURATA ET AL. 8245 FIG.9.Compositeofverticallyintegratedmoistureflux(vector;kgm21s21)anditsdivergence(shading; 1024kgm22s21)for(a)June–Septemberclimatologyand(b)duringCARSs. North Pacific propagates westward along 108–208N propagatesfartherwestwardovercentralIndia(Fig.8b), (Fig. 11). An anomalous negative streamfunction is and causes convective suppression (Fig. 8c). The AAC situatedtothesouthoftheAAC,andmakesadouble- over central India has a barotropic structure (Fig. 8a). cell structure (Fig. 10b). During the CARSs, the AAC ThesefeaturesareconsistentwiththeresultsofFujinami FIG.10.AsinFig.8,butfortheonsetdaycompositesofeachCARS. 8246 JOURNAL OF CLIMATE VOLUME30 FIG.11.AsinFig.8b,butforthecompositesfor2daysbeforetheonsetof(a)3-,(b)5-,and(c)7-dayCARSs,respectively.(d)–(f)Asin (a)–(c),butforthecompositesoftheonsetdayofCARSs.(g)–(i)Asin(a)–(c),butforthecompositesofthe2daysaftertheonset ofCARSs. et al. (2014) and other studies in the 10–20-day mode the model developed by Wang and Li (1994). It is an analysis. The AAC signal in the upper troposphere equatorial beta-plane channel model with a model movedsouthwardfromtheTibetanPlateauandcovered atmosphereconsistingofatwo-layerfreetroposphere overcentralandnortheastIndia(Fig.12). and a well-mixed boundary layer, and includes sur- DuringCARSs,suppressedconvectiveanomaliesare faceevaporationinthemoisturebudget.Theinstabil- distributedoverazonallyelongatedbandfromsouthern ity generated by convection–frictional convergence Pakistan,viacentralIndia,thenorthernBayofBengal, produceda low-frequency multiscale convectivecom- the Indochina Peninsula, and the Philippines, to the plexinwhichthecondensationheatingcoupledmoist westernNorthPacific(Figs.7and10c).Asimilarelon- Kelvin wave and moist Rossby wave with the gravest gatedband,butwithoppositesign,wasshownbyMoon meridional structure (n 5 1 mode). Wang and Xie etal.(2013)duringtheactivephaseoftheIndiansum- (1997)introducedtheclimatologicalJulymeanflowat mer monsoon, and by Takahashi et al. (2015) during 200-and850-hPalevelsandJulymeansurfacespecific above-normal precipitation over the Indochina Penin- humidity over Asian monsoon region as the model’s sula. The zonally elongated band is formed due to the basicstate. westwardpropagationofAAC. Thepath,propagatingspeed,andhorizontalscaleof WangandXie(1997)simulatedthesimilarelongated theobservedwestwardpropagatingAACsinthisstudy precipitation band over northwest India toward the (Fig.11)aresimilartothatofthesimulatednortherncell equatorial central Pacific. They used the extension of ofthen51equatorialRossbywavesinWangandXie
Description: