title stability for summertime afternoon precipitation...

Title

A Regional-Scale Evaluation of Changes in EnvironmentalStability for Summertime Afternoon Precipitation under GlobalWarming from Super-High-Resolution GCM Simulations: AStudy for the Case in the Kanto Plain

Author(s) TAKEMI, Tetsuya; NOMURA, Syohei; OKU, Yuichiro;ISHIKAWA, Hirohiko

Citation Journal of the Meteorological Society of Japan. Ser. II (2012),90A: 189-212

Issue Date 2012

URL http://hdl.handle.net/2433/166082

Right © 2012 by Meteorological Society of Japan

Type Journal Article

Textversion publisher

Kyoto University

Journal of the Meteorological Society of Japan, Vol. 90A, pp. 189--212, 2012. 189

DOI:10.2151/jmsj.2012-A10

A Regional-Scale Evaluation of Changes in Environmental Stability for Summertime

Afternoon Precipitation under Global Warming from Super-High-Resolution GCM

Simulations: A Study for the Case in the Kanto Plain

Tetsuya TAKEMI, Syohei NOMURA1, Yuichiro OKU2, and Hirohiko ISHIKAWA

Disaster Prevention Research Institute, Kyoto University, Uji, Japan

(Manuscript received 8 March 2011, in final form 22 September 2011)

Abstract

Understanding and forecasting of summertime afternoon precipitation due to rapidly developing cumulonim-bus clouds without any significant synoptic-scale influences are important to prevent and mitigate the induceddisasters. Future changes in the behavior of such precipitation events have recently gained scientific and societalinterests. This study investigates the environmental stability for afternoon precipitation that develops under syn-optically undisturbed conditions in summer by using the outputs of 20-km-mesh, super-high-resolution atmo-spheric general circulation model (GCM) simulations for a present, a near-future, and a future climate underglobal warming with the Intergovernmental Panel on Climate Change A1B emission scenario. The Kanto Plainwas chosen as the analysis area. After verifying the usefulness of the GCM present-climate outputs with observa-tions and gridded mesoscale analyses, we examine the future changes in the environmental stability for the after-noon precipitation by conducting statistical analyses. In the future climates, temperature lapse rate decreased inthe lower troposphere, while water vapor mixing ratio increased throughout the deep troposphere. The changes inthe temperature and moisture profiles resulted in the increase in both precipitable water vapor and convectiveavailable potential energy. These projected changes will be enhanced with the future period. Furthermore, the sta-tistical analyses for the di¤erences of the stability parameters between no-rain and rain days under the synopti-cally undisturbed condition in each simulated climate period indicated that the representations of the stabilityparameters that distinguish no-rain and rain events are basically unchanged between the present and the futureclimates. This result suggests that the environmental characteristics favorable for afternoon precipitation in thesynoptically undisturbed environments will not change under global warming.

1. Introduction

Severe local rainfalls have potentially high soci-etal impacts in many regions of the world. Signifi-cant e¤orts from both research and operationalsector have been made on the understanding and

forecasting those events from planetary-scale,synoptic-scale, and mesoscale perspectives. Rapiddevelopments in observational and computationalresources have nowadays enabled us to focus moreon convective-scales that are on the order of 1 km.Specifically, local-scale convective rainfalls due torapidly developing cumulonimbus clouds withoutany significant synoptic influences have been oneof the main research issues. It is anticipated thatsuch rapidly developing, convective rainfalls undersynoptically undisturbed conditions would becomemore prevalent and more intensified in a futurewarming climate. In the Fourth Assessment Report(AR4) by the Intergovermental Panel on Climate

Corresponding author: Tetsuya Takemi, Disaster Pre-vention Research Institute, Kyoto University, Goka-sho, Uji, Kyoto 611-0011, Japan.E-mail: [email protected]

1 Present a‰liation: Narita International Airport Cor-poration.

2 Present a‰liation: Osaka City Institute of PublicHealth and Environmental Sciences.6 2012, Meteorological Society of Japan

Change (IPCC) (2007), it is stated that the intensityof precipitation events is projected to increase par-ticularly in tropical and high latitude areas, andthat precipitation extremes are more enhancedthan the precipitation means in most tropical andmid- and high-latitude areas.

Future projections of precipitation characteristicsnot only in global-scales but also in regional-scaleswere investigated through numerical simulationswith the use of general circulation models (GCMs)and regional climate models (RCMs) nested inGCMs. For cases in Japan, as an example, the gen-eral characteristics of the changes in precipitationover Japan were investigated by GCM studies (e.g.,Kimoto et al. 2005; Kamiguchi et al. 2006), and thechanges in precipitation behavior due specifically toBaiu frontal activities were also focused on, owingto the significance and frequency of heavy rainfallsduring the Baiu periods in Japan, by many studieswith both GCMs (Kusunoki et al. 2006; Kitoh andUchiyama 2006; Ninomiya 2009) and RCMs(Yoshizaki et al. 2005; Kanada et al. 2005; Waka-zuki et al. 2005; Yasunaga et al. 2006). Althoughthe quantitative analyses on the characteristics ofBaiu precipitation under global warming were ex-tensively investigated, there have been few studiesthat investigated the convective precipitation insynoptically undisturbed environments under globalwarming, owing to fine-scales of such convectiveprecipitation for numerical representations.

The future projections will benefit much from theobservational evidences revealed by the past re-cords. The long-term variability of the generalcharacteristics of precipitation for cases in Japanhas been investigated from statistical viewpoints bymany studies such as Yonetani (1982), Iwashimaand Yamamoto (1993), Fujibe (1998), Sato and Ta-kahashi (2000), Kanae et al. (2004), Fujibe et al.(2005, 2006, 2009), Kamiguchi et al. (2010), andIwasaki (2010). There have been arguments on thereason for the changes in the frequency and inten-sity of precipitation: some studies attributed toglobal climate change (i.e., warming); and othersimplied the e¤ects of urbanization. However, thestability conditions of the environmental atmo-sphere should also play a critical role in controllingthe behavior of convective rainfalls.

There are studies that emphasized the impor-tance of environmental conditions for convectivestorms and precipitation. For the change in tropicalcyclone climatology due to global warming, Sugiet al. (2002) considered that the decrease in the

number of tropical cyclones is due to more stabi-lized atmosphere and hence more weakened tropi-cal circulation. From the modeling studies it is wellknown that the tropical troposphere will be morestabilized in a warming climate; this modeled trendhas been shown to be consistent with the observa-tions (Santer et al. 2005, 2008; Thorne et al. 2010),but with some exaggeration in the GCM simula-tions (Fu et al. 2011). On the other hand, somestudies indicated from the observations and re-analysis data that the increase in the free tropo-spheric temperature in the Tropics is smaller thanthe temperature increase at the surface (Christyet al. 2007; Bengston and Hodges 2011), whichmeans that the stability of the tropical tropospheredecreases in a warming climate. In addition, thedi¤erences in the temperature trends in the lowerand upper troposphere widely vary over the globe(Bengston and Hodges 2011) and in the latitude(Thorne et al. 2010). Therefore, it is not obviouswhether the tropospheric stability increases also inextra-tropical regions in any seasons and any mete-orological settings. Considering this uncertainty,the changes in environmental stability conditionsfor the convective rainfalls in the undisturbed set-tings need to be investigated from a regional-scalepoint of view.

For the case in Japan in summer, Kanada et al.(2010a) examined the daily precipitation and theatmospheric state in July (which corresponds tothe Baiu period) by conducting the RCM simula-tions and found that static stability, moisture con-tent, and convective available potential energy(CAPE) are more increased in the future climatesthan in the present. They further noted that theincreased CAPE would lead to stronger updraftsand therefore more intense precipitation. Del Genioet al. (2007) also pointed out that due to increasedCAPE updrafts become stronger in global warmingclimate in their investigations for continental deepconvection within a GCM simulation. From cloud-resolving simulations in idealized settings that rep-resent horizontally homogeneous, undisturbed con-ditions, Takemi (2007, 2010) investigated the e¤ectsof static stability on the intensities of convectionand precipitation and suggested that stability diag-nosis for the climate simulation data should beuseful for examining the variation of precipitationcharacteristics in the simulated climate.

Nomura and Takemi (2011, hereafter NT11)have recently investigated the environmental stabil-ity for summertime afternoon precipitation under

190 Journal of the Meteorological Society of Japan Vol. 90A

synoptically undisturbed environments over theKanto Plain, a region that includes the Tokyometropolitan area in Japan, in terms of commonlyused stability indices and parameters. They usedthe gridded mesoscale analysis data at the re-gional-scale after validating the data against radio-sonde observations. Their analyses on the di¤eren-ces of temperature and humidity at each heightbetween no-rain and rainy cases indicated that thetemperatures and moistures at low to middle levelsclearly distinguish the stability conditions for theafternoon rain events, which confirmed previousobservational studies by Yonetani (1975), Taguchiet al. (2002), and Kawano et al. (2004) who exam-ined the environmental conditions by using radio-sonde data at a single site in the northern part ofthe Kanto Plain. The idea of these studies can beapplied to climate-simulation data to investigatethe long-term trend of the changes in the environ-mental conditions for convective precipitation.

In this study, we investigate the environmentalstability for convective rain events in synopticallyundisturbed conditions from climate-simulationdata and evaluate the changes in the environmen-tal stability at a regional-scale under global warm-ing. According to Chuda and Niino (2005), stabilityconditions widely change over Japan dependingon locations and seasons; therefore, we focus onthe summertime cases in the Kanto Plain where anumber of studies have been previously done forenvironmental conditions. Moreover, the study ofChuda and Niino strongly suggests that the envi-ronmental conditions that control the developmentof cumulus convection will not be unchanged if theclimate is changed from a region to a region, froma season to a season, and furthermore from thepresent to the future. Thus, the evaluation of con-vective potential for the future periods is of scien-tific interest. Since the focus of this study is to eval-uate the environmental conditions, high-resolutionGCM data are considered to su‰ciently resolveand represent the environments and therefore tobe useful. The outputs of super-high-resolutionglobal climate simulations with a warming-climatescenario are used to examine stability parametersfor rain events over the Kanto Plain. Although itis well known from GCM simulations that the sta-bility increases in the tropical atmosphere underglobal warming, the environmental changes inextra-tropical regions in undisturbed meteorologi-cal conditions have not been explicitly investigated.The purpose of this study is to provide a regional-

scale evaluation of the environmental changes forconvective precipitation under synoptically un-disturbed conditions in a warming climate.

2. GCM data and analysis procedure

2.1 GCM and data

The data used in this study are the gridded out-puts at a 20-km horizontal spacing of the super-high-resolution atmospheric GCM (AGCM) simu-lations at a resolution of T959L60 (i.e., triangulartruncation of 959 with 60 vertical layers) for pres-ent, near-future, and century-end future climates,conducted by Meteorological Research Institute(MRI), Japan Meteorological Agency (JMA)under the Innovative Program of Climate ChangeProjection for the 21st Century (the KAKUSHINprogram). The code name of this AGCM is MRI-AGCM3.2S. The outline of the relevant KA-KUSHIN program and the details of the AGCMwere given by Kitoh et al. (2009) and Murakamiand Wang (2010). This super-high-resolutionAGCM was originally developed by Mizuta et al.(2006), and the present-climate simulations werewell validated against observations in their study.This AGCM was extensively used for the futureprojections of the Baiu rainfalls (Kusunoki et al.2006), typhoons (Murakami et al. 2011), and otherclimate applications.

The GCM simulations are intended to representthe present climate (during 1979–2008), the near-future climate (during 2015–2044), and the lastquarter of the twenty-first century (during 2074–2104), each for 30 years with a global warmingscenario, that is, the IPCC A1B emission scenario(IPCC 2007). The preliminary experiments were in-tended for 25-year periods (1979–2003, 2015–2039,and 2075–2099), and Kitoh et al. (2009) confirmedthat the preliminary results were almost the sameas those in Mizuta et al. (2006). Therefore, it is ex-pected that the GCM results used in this study havegood performance in representing the present cli-mate. In this study, we used the outputs simulatedfor 25-year periods that correspond to 1980–2004for the present climate, 2020–2044 for the near-future climate, and 2075–2099 for the century-endfuture climate. Hereafter, these periods are referredto simply as the present, the near-future, and the fu-ture, respectively.

The present focus is on the regional-scale atmo-spheric states that are the environments for sum-mertime afternoon precipitation events undersynoptically undisturbed conditions. In order to

February 2012 T. TAKEMI et al. 191

examine the performance of the GCM present-climate simulation in representing the existing realclimate for the undisturbed environments intendedin this study, we use the upper-air radiosonde ob-servations at the Tateno site in Tsukuba, Ibarakiand the numerical analysis data from the JMA’soperational Mesoscale Model (MSM). The periodsof these data used are the same as in NT11: during1976–2010 for the radiosonde data; and during2002–2010 for the MSM data. We also use precipi-tation data which are obtained from the Auto-mated Meteorological Data Acquisition System(AMeDAS) observations.

The analysis area is a mesoscale region of about100 km by 100 km centered at around centralTokyo (i.e., 35.14�N–36.26�N and 138.94�E–142.25�E), which covers the most part of the KantoPlain. This area is almost the same as examined inNT11. Figure 1 shows the area of interest and thedata points of the gridded GCM outputs. The timeintervals of surface and upper-air variables fromthe GCM outputs are 1 hour and 6 hours, respec-tively. As described shortly in the next subsection,the data of August are used to better focus on syn-optically undisturbed conditions in summer.

2.2 Analysis procedure

In NT11, synoptically undisturbed conditions insummer were chosen with the following procedure.At first, days with daily maximum temperature of

30�C or greater at any AMeDAS sites in the analy-sis area were chosen. The second step was to choosedays when precipitation in the morning hours wasnot observed at all the AMeDAS sites in the analy-sis area. The third step was to exclude days thatwere considered to be a¤ected by fronts and ty-phoons by examining weather charts, typhoonbest-track data, and Baiu-period definitions ofJMA. The last step was to choose days when theMSM predicted successfully afternoon rain/no-rain events. All these steps assumed the use of theobservation data. However, weather charts andBaiu-period definitions are not available in thefuture-climate simulations. Therefore, an objectiveapproach is desired in order to extract days withno significant synoptic-scale forcings from theGCM simulation outputs.

For the purpose of setting an objective approach,the vertical profiles of wind speeds are examinedfrom radiosonde observations. Figure 2 indicatesthe composited profiles of wind speeds for no-rain(referred to as N in NT11), rain (referred to asR in NT11), and strong rain cases (referred to asS in NT11) from the 35-year radiosonde data atTateno at 0900 JST1. Wind shears and upper-levelwind speeds are generally quite small, comparedwith typical continental environments for severestorms and squall lines (Bluestein and Jain 1985).Based on these wind profiles, an objective approachto extract the summertime undisturbed conditionsfrom the GCM outputs is to set the criteria as fol-lows:

1) Use the data only for August to avoid includingthe Baiu periods;

2) Exclude days under conditions with 500-hPawind speed of equal to or greater than 10 m s�1

and wind-speed di¤erence between the 500- and975-hPa levels of equal to or greater than 8 m s�1;

3) Choose days having rainfall of less than 1 mmduring the morning period (i.e., from 0000 to1200 JST) at all the grids in the analysis area.

Hereafter we refer to the states that pass these threecriteria as synoptically undisturbed condition. Toapply these criteria, the GCM data are averagedover the analysis area at each vertical level, andthen the shear and velocity are calculated from thearea-averaged values.

Fig. 1. The map of the Kanto Plain andthe surroundings and the locations of theGCM grid points (red points) representingthe analysis area of the present study.

1 Japan Standard Time, which is 9 hours plus UTC.


Figure 3 compares the two approaches, that is,the NT11 method and the present objective ap-proach, in terms of the averaged vertical profiles oftemperature and relative humidity from the 35-yearradiosonde data at 0900 JST under the synopticallyundisturbed conditions. The number of days chosenby the NT11 method was 699, while that by thepresent approach was 344. Although the numberof days chosen by the present approach is signifi-cantly smaller than that by NT11, 300 cases out ofthe selected 344 cases satisfy the condition of NT11.The averaged profiles based on the present ap-proach are closely coincident with the averages bythe NT11 method. The present approach was alsocompared with the NT11 method against the 9-year MSM data (not shown); this comparison indi-cated that 69 cases out of the total 76 cases ex-tracted by the present procedure satisfied the NT11condition. Note here that the number of the caseschosen in NT11 from the MSM data was 124.Therefore, the present objective approach based onthe wind shear and the upper-level wind speed per-forms reasonably well in choosing the synopticallyundisturbed condition.

In the following sections, we describe the charac-

teristics of environmental stability under synopti-cally undisturbed conditions represented in theGCM simulations for the present, the near-future,and the future climates. The stability indices andparameters examined in this study are the sameas in NT11; namely, CAPE, convective inhibition(CIN), lifting condensation level (LCL), level offree convection (LFC), level of neutral buoyancy(LNB), Showalter stability index (SSI), lifted index(LI), K-index (KI), total-totals index (TT), precip-itable water (PW), and temperature lapse rate from950 hPa to 500 hPa (TLR). The definitions of theseparameters except TLR can be found in standardtextbooks such as Bluestein (1993). Temperaturelapse rate in the lower half of the troposphereis chosen since it was found that this lapse ratesignificantly regulates the intensity of convectiveprecipitation under idealized modeling settings(Takemi 2007, 2010). Note that the present studyexamines the parameters specifically related to thetemperature and moisture profiles in weak-shearenvironments and therefore shear-related parame-ters such as bulk Richardson number and storm-relative helicity are beyond the scope of the presentstudy.

Fig. 2. The vertical profiles of (a) the east-west and (b) the north-south components of wind speed averagedfor each rain category under the synoptically undisturbed condition from the 35-year radiosonde data atthe Tateno site. The solid lines denote the no-rain case, the dotted lines the rain case, and the dash-dottedlines the strong-rain case. See Nomura and Takemi (2011) for the definitions of these cases.


3. Results

3.1 Representation of the undisturbed condition

in the present-climate simulation

As described in Sections 1 and 2.1, the perfor-mance of the present GCM simulations has beenwell tested and evaluated in terms of seasonal and/or monthly climatologies, and thus the outputs ofthe present GCM have been used for various appli-cations. To our knowledge, however, there havebeen little studies that investigated the environ-mental conditions for convection and precipitationunder synoptically undisturbed conditions by usingthe GCM outputs. Therefore, the representationof the synoptically undisturbed condition in thepresent-climate simulation is evaluated here.

By applying the criteria described in Section 2.2against the present-climate simulation, we foundthat there were 182 days that passed the criteriaamong the total number (i.e., 775) of days in Au-gust during the 25 years. In this subsection, the gen-eral characteristics such as temperature and mois-ture profiles as well as stability parameters for the182 days are examined and are compared with thedata used in NT11.

Figure 4 compares the vertical profiles of temper-ature, relative humidity, and water vapor mixingratio averaged for the synoptically undisturbedcases obtained from the GCM present-climate sim-ulation, the 35-year radiosonde observations at Ta-teno, and the 9-year MSM analyses over the analy-sis area used in NT11. Note that the GCM datawere extracted by the present objective procedurewhile the radiosonde and MSM data were extractedby the method used in NT11; the idea is to examinehow the undisturbed conditions in the real cases arerepresented in the GCM present-climate simula-tion. The figure indicates that the temperatureand mixing ratio profiles obtained from the GCMoutputs seem to in general reproduce the existingclimates obtained by the observations and the nu-merical analyses, although there are some notice-able di¤erences between the GCM climate and theobservations/analyses. The temperature profiles in-dicate that there is a colder bias below about the800-hPa level in the GCM than in the radiosondeobservations, while the mixing ratio profiles showthat the amounts at the middle levels in the GCMare smaller than those in the observations. The dif-ference between the GCM results and the radio-

Fig. 3. The vertical profiles of (a) temperature and (b) relative humidity at the Tateno site averaged for thesynoptically undisturbed conditions which are determined by the method used by Nomura and Takemi(2011) (solid line) and the present objective procedure (dashed line).


sonde observations in the profile of relative humid-ity appears to be more pronounced; this relative hu-midity di¤erence seems to be closely linked to thelow-level colder bias and the middle-level drier biasin the GCM simulations.

Figure 4 also shows that the MSM temperature(mixing ratio) profile has a warmer (moister) biasbelow (above) the 800-hPa level, which is oppositeto the GCM bias. Because the GCM and MSMdata both have some biases against the radiosondeobservations, it is not appropriate to think thatonly the GCM simulation fails to accurately repre-sent the observed climate. It is fair to say that bothGCM and MSM cannot be free from a certainamount of biases, although the biases seen in theGCM data are somewhat larger than those in theMSM analyses.

The comparison between the GCM data and theobservations/analyses is further extended to the sta-bility parameters examined in this study. The fre-quency distributions of the stability parameters forthe undisturbed conditions among the GCM data,the observations, and the MSM analyses are com-pared in Fig. 5. Among the parameters, the distri-butions for SSI, LI, CAPE, LCL, and LNB ob-tained by the GCM outputs agree quite well withthose obtained by the radiosonde observations aswell as the MSM analyses. The distributions formost of the other parameters obtained from theobservations/analyses are also represented favor-ably in the GCM data, although the di¤erences be-tween the GCM data and the observations are seen

to be large in the distributions of TLR and CIN.These di¤erences in TLR and CIN are consideredto be due to the representation of the temperatureprofile in the lowest layer in GCM (Fig. 4a). Thereis a colder bias below the 800-hPa level in GCM,which leads to the decrease in the magnitude ofTLR. On the other hand, the lapse rate belowabout the 900-hPa level in GCM is close to the dryadiabat, which tends to reduce the magnitude ofCIN.

From the comparisons indicated in Figs. 4 and 5it is said that the present climate in the GCM datacaptures the overall features obtained from theobservations and analyses considerably well. Thereare of course noticeable di¤erences between theGCM data and the observations/analyses. Amongthe di¤erences, the bias of the temperature profilein GCM from the observations is more pronounced(Figs. 4a, 5f ) and may not be negligible. In the fol-lowings, we are mostly interested in the anomaliesof the environmental conditions in the future andnear-future climate from those in the present cli-mate. It may be safely to assume that, as a first ap-proximation, the biases of the GCM simulationswill appear in the future climates in the same wayas in the present climate. Therefore, the biases suchas in TLR and CIN will not be significant in discus-sing the di¤erences among the simulated climateperiods.

We are confident with the fair agreement of the20-km GCM data with the observations/analysesexhibited in Figs. 4 and 5, in spite of the general

Fig. 4. The vertical profiles of (a) temperature, (b) relative humidity, and (c) water vapor mixing ratio aver-aged for the synoptically undisturbed condition, obtained from the GCM present simulation (solid lines),the Tateno radiosonde observations (dotted lines), and the MSM analyses (dashed lines).


Fig. 5. The frequency distribution of the stability parameters examined in this study under the synopticallyundisturbed condition, obtained from the GCM present simulation (solid lines), the Tateno radiosonde ob-servations (dotted lines), and the MSM analyses (dashed lines).


thinking that the GCM simulations at the resolu-tion of the order of 10 km have inevitably deficien-cies in exactly reproducing convective-scale pro-cesses and the GCM simulations, even for thepresent climate, are not able to exactly mimic thereal climates. One reason for this favorable agree-ment is that we are focusing on the conditionsthat are the environments for convective-scale pro-cesses. Therefore, it is demonstrated that the GCMoutputs are useful in investigating the environ-mental characteristics under the synoptically un-disturbed conditions. We will further examine thecharacteristics of environmental stability and pre-cipitation in the GCM present climate in the fol-lowing.

3.2 Characteristics of precipitation and

environmental stability in the present-climate

simulation

One-hundred eighty two days that are extractedas the synoptically undisturbed condition will bedivided into two categories, that is, no-rain andrain days, depending on rainfalls in the afternoon(i.e., from 1200 to 0000 JST). For dividing thesecategories, hourly precipitation outputs between1300 and 0000 JST are used. In NT11, the rain

days were defined as having a precipitation inten-sity of equal to or greater than 1 mm h�1 at anygrid points within the analysis area during the after-noon hours. However, it was indicated by Kanadaet al. (2010b) that the amount of precipitation inthe GCM outputs is in general underestimatedagainst the observed rainfalls. In addition, it is rea-sonable to think that the GCM simulations cannotreproduce quantitatively the convective precipita-tion with an intensity of a few mm h�1 at an hourlytimescale owing to a coarse resolution for resolvingsuch convective-scales. Therefore, the direct appli-cation of the definition of NT11 to the GCM datamay not be appropriate. In order to distinguish no-rain and rain days among the extracted 182 casesfrom the GCM outputs, the precipitation character-istics from the GCM outputs are examined first.

Figure 6 shows the frequency distributions ofthe maximum hourly precipitation and the totalamount of precipitation during the afternoon hourson the undisturbed days extracted from the GCMpresent simulation, the AMeDAS observations,and the MSM 3-hour forecasts. A sharp decreasein the frequency with the increase in the precipita-tion is overall reproduced in the GCM outputs forboth the maximum hourly precipitation and the

Fig. 6. Frequency distributions of (a) the maximum hourly precipitation during the afternoon and (b) theaccumulated precipitation in the afternoon for the undisturbed condition from the GCM present simulation(green), the AMeDAS observations (red), and the MSM forecasts (blue).


total afternoon precipitation. However, the GCMsimulation poorly represents the hourly precipita-tion of greater than about 5 mm h�1 (Fig. 6a).

On the other hand, the distribution of total pre-cipitation amounts up to about 25 mm in the AMe-DAS and MSM data seems to be better representedin the GCM simulation (Fig. 6b) than the distribu-tion of the maximum hourly precipitation. It issurprising that the frequency distribution of theGCM total precipitation closely captures that ofthe MSM precipitation. However, it is found thatthe GCM simulation fails to represent cases witha larger amount of precipitation of greater than25 mm. In addition, the GCM reproduces morefrequencies in representing total precipitation ofsmaller than 15 mm than do the AMeDAS obser-vations. As shown in Kanada et al. (2010b) in theirstudies on precipitation characteristics during warmseason (from June to October), the GCM present-climate simulation significantly underestimates thefrequency of stronger rain events (i.e., greater than26 mm day�1), while it reproduces weaker rainevents relatively well. Because this study focuseson precipitation events in the undisturbed environ-ments, it is expected that the daily precipitationamounts are not so extreme; namely, the total after-noon precipitation for the extracted days is shownto be mostly below 25 mm (Fig. 6b). By our defini-tion the total precipitation during the morninghours is less than 1 mm, and therefore the dailyprecipitation in the extracted cases is generally lessthan 25 mm. In this way, the result indicated inFig. 6b is consistent with the findings of Kanadaet al. (2010b).

Consequently, there are both agreements and dis-agreements in precipitation amounts between theGCM simulation and the observations/analyses.Although there is inevitably a deficiency of theGCM simulation in explicitly resolving convective-scale precipitation, the ability of the GCM in repre-senting the synoptically undisturbed conditions (seeFigs. 4 and 5) indicates that the agreement of thetotal afternoon precipitation between the GCMoutputs and the observed climate is evaluated to befair and is considered to be su‰cient for the presentanalyses on the environmental conditions. Theseconsiderations lead us to employ the total after-noon precipitation as a measure to distinguish no-rain and rain cases among the synoptically un-disturbed days extracted from the GCM outputs.It should be noted that the precipitation outputs ofthe present GCM have been used for hydrologic

applications and favorable results have been ob-tained (e.g., Kim et al. 2010).

In order to distinguish no-rain and rain cases, acertain threshold is used here. The present purposeis to clearly separate the environmental conditionsfor no-rain and rain events and to evaluate the dif-ferences in the mean atmospheric states from theselected cases. Therefore, we apply clearly di¤eren-tiating thresholds that will at the same time allowsu‰cient numbers of cases for statistical analyses.No-rain days (denoted as N days) are defined ashaving the total afternoon precipitation of lessthan 1 mm at every grid point within the analysisarea. On the other hand, rain days (denoted as Rdays) are defined as having the total precipitationof greater than or equal to 5 mm at any grid pointsover the analysis area. This setting of the thresholdsclassifies 84 cases as the N days and 58 cases as theR days, indicating that the both categories includesu‰cient numbers of cases. This rain categorizationis used to identify the di¤erences in the environ-mental conditions between the N and R days notonly in the present but also in the near-future andthe future climates represented in the GCM simula-tions.

To confirm that there are no significant synoptic-scale disturbances on the N and R days, we exam-ine large-scale pressure patterns averaged for eachcategory. The spatial distributions of geopotentialheight at the 500-hPa level at 0900 JST averagedfor the N and R days are shown in Fig. 7 obtainedfrom the MSM analyses and in Fig. 8 obtainedfrom the GCM present-climate simulation. TheMSM composites in Fig. 7 clearly demonstrate thatthe most part of Japan including the Kanto Plain iscovered by a high pressure system centered on thePacific just south of the Japanese Islands on boththe N and R days. This pressure pattern indicatedby the MSM analyses as the real climate is favor-ably represented in the GCM simulation (Fig. 8).It is also confirmed in Fig. 8 that no significantsynoptic-scale low-pressure system appears in theGCM composites for both the N and R days. Inaddition, the overall agreement shown in Figs. 7and 8 further supports that the GCM data underthe synoptically undisturbed condition capture theenvironmental characteristics that are seen in theexisting climate.

The di¤erence in the environmental characteris-tics between the N and R days from the GCM out-puts is examined here. Figure 9 indicates the envi-ronmental atmospheric states at 0900 JST for the


N and R days in terms of the vertical profiles oftemperature, relative humidity, and water vapormixing ratio averaged for the chosen days from thepresent-climate simulation. In the R cases, temper-ature increases in the lower layer, and both mois-ture variables, not only mixing ratio but also rela-tive humidity, increase at most of the vertical

levels. As will be shown later in this subsection,TLR and PW are more increased on the R daysthan on the N days. This point is consistently rep-resented in the vertical profiles demonstrated inFig. 9.

In order to statistically confirm the di¤erence be-tween the R and N days, we use t-test statistic to

Fig. 7. Geopotential height at the 500-hPa level at 0900 JST averaged for (a) the N days and (b) the R daysfrom the MSM analyses. The contour interval is 10 m.

Fig. 8. The same as Fig. 7, except from the GCM present-climate simulation.


verify the statistical significance, as was done inNT11. Test statistic, T , is defined as:

T ¼ jxA � xBj �s2A

nAþ s2

B

nB

� ��1=2

; ð1Þ

where xA and xB are the means for category A andB, respectively, sA and sB the standard deviations,nA and nB denote the number of cases in each cate-gory. When T is larger than a certain threshold, asignificant di¤erence between the categories is sta-tistically indicated. A larger value of T means thatthe di¤erence between each category is more signif-icant. In this study, a significance level is evaluatedat the 5-% level.

Figures 10a and 10b show the vertical distribu-tion of the di¤erences of temperature and relativehumidity at 0900 JST between the N and R daysrepresented in the present-climate simulation. Thetemperature in the lower layer (in the middle layer)is higher (lower) in the R days than in the N days.On the other hand, the relative humidity through-out the deep troposphere is higher in the R daysthan in the N days; the largest di¤erence is seen ataround the 500-hPa level. The characteristics ofthese di¤erences are quite similar to those foundby NT11 from the MSM analyses and the radio-sonde observations.

The statistical significances of these di¤erencesare evaluated at each level. In Figs. 10c and 10d,we examine the vertical profile of the T values forthe di¤erences shown in Figs. 10a and 10b. For thetemperature di¤erences, the T values exceeding the

significance level are found below the 800-hPa leveland at the 500-hPa level. This indicates that thelow-level higher temperature and the middle-levellower temperature in the R days than in the Ndays are statistically significant. For the relative hu-midity di¤erence, the T values exceed the signifi-cance level in the middle layer, with the maximumfound at the 500-hPa level. Thus, the humidity atthe middle levels is statistically higher in the Rdays than in the N days. Although the vertical dis-tributions of the statistical significance are a littledi¤erent from those obtained from the MSM anal-yses by NT11, the middle-level and low-level sig-nificance for the temperature di¤erence and themiddle-level significance for the relative humiditydi¤erence, which are identified by the MSM analy-ses, are favorably reproduced in the GCM present-climate simulation.

The results given in this subsection as well asin Section 3.1 strongly indicate that the GCMpresent-climate simulation well reproduces thecharacteristics of the environmental stability andits di¤erence between the N and R days under thesynoptically undisturbed condition in summer. Fur-thermore, it is indicated that the evaluation ofthe environmental characteristics at regional-scalesis quite reasonable even with the GCM outputs.Based on the analyses on the GCM present-climatesimulation, therefore, we will provide future projec-tions of the changes in the environmental condi-tions under the synoptically undisturbed conditionsin the next subsection.

Fig. 9. The vertical profiles of (a) temperature, (b) relative humidity, and (c) water vapor mixing ratio at0900 JST averaged for the N (thin solid line) and the R (dotted line) days under the synoptically un-disturbed condition in the present climate of the GCM simulation.


3.3 Future projections of environmental stability

The criteria given in Section 2.2 are applied tothe GCM simulation outputs for the present, thenear-future, and the future climates in order to ex-tract the synoptically undisturbed conditions ineach climate period. In addition, the extracted daysare divided into the N and R days by the rain cate-gorization defined in Section 3.2. Table 1 summa-rizes the numbers of the synoptically undisturbed

days and the N and R days in each simulatedperiod. It is seen that the total number of days ex-tracted as satisfying a synoptically undisturbedcondition as well as the number of the N days de-creases from the present to the future climate.

The characteristics of the frequency of precipita-tion, that is, the hourly maximum and the totalamount during the afternoon hours, under the syn-optically undisturbed conditions in the present, the

Fig. 10. The vertical profiles of the mean di¤erence of (a) temperature and (b) relative humidity at 0900 JSTbetween the categories of the N and the R days, and the T values for (c) the temperature di¤erence and (d)the relative humidity di¤erence between the two categories, obtained from the GCM present simulation. In(a) and (b), the positive di¤erence means that the R-day case is larger than the N-day case. In (c) and (d),the threshold for the statistical significance (T ¼ 1.99) is indicated by dotted lines.


near-future, and the future climates are comparedin Fig. 11. For both hourly maxima and totalamounts of afternoon precipitation, it is di‰cult tofind any pronounced di¤erences among the simu-lated periods in the frequency distribution. Somedi¤erences are seen for the total amount of after-noon precipitation; the number of cases with thetotal amount of greater than 5 mm appears toslightly increase with the future period. However,the tendency with time is not clearly identified.

As mentioned in Section 3.2 in relation to Fig. 6,it is reasonable to assume that the weaker rainevents (with the amount of less than 26 mm day�1)

are represented in the GCM simulations relativelywell. Figure 11b confirms that almost all the ex-tracted days have total precipitation of below25 mm day�1. In addition, the environmental con-ditions should be better represented than quantita-tive precipitation amounts, considering that thedownscaling experiments with a nonhydrostaticRCM nested in the present GCM using a nudgingmethod (which means that the background statesin RCM are essentially the same as in GCM) aresuccessful in quantitatively representing rainfall(Kanada et al. 2008, 2010b). Therefore, the envi-ronmental atmospheric conditions at regional-scales are appropriately evaluated with the GCMclimate simulations. In the followings, we focus onthe changes of the environmental stability in futureclimates.

The mean vertical profiles of temperature, rela-tive humidity, and water vapor mixing ratio underthe synoptically undisturbed conditions in the simu-lated present, near-future, and future climates areexhibited in Fig. 12. Temperature and mixing ratioat all the levels increase with the future period. Onthe other hand, relative humidity, especially belowthe 600-hPa level, does not largely vary among thesimulated period.

Fig. 11. The frequency distributions of (a) the maximum hourly precipitation during the afternoon and (b)the accumulated precipitation in the afternoon for the undisturbed condition from the present (blue), thenear-future (green), and the future (red) climate simulations of the GCM.

Table 1. The numbers of the synoptically undisturbeddays, the no-rain (N) days, and the rain (R) days inthe present, the near-future, and the future climatesrepresented in GCM.

Present Near future Future

Total 182 160 128

N days 84 57 45

R days 58 54 51


Figure 13 demonstrates the vertical profiles ofthe di¤erences of temperature, relative humidity,and water vapor mixing ratio averaged for the un-disturbed cases over each simulated period, andalso the profiles of the T values that indicate thestatistical significance for these di¤erences. Al-though there are di¤erences in the relative humidity(Fig. 13b) between the near-future (the future) andthe present climate, the existence of the statisticalsignificance for the di¤erences is not identified(Fig. 13e). On the other hand, the temperature andmixing ratio increase with the future period (Figs.13a and 13c). The T values for these di¤erences intemperature and mixing ratio shown in Figs. 13dand 13f exceed the significance level, which meansthat those di¤erences are statistically significant.Considering that the relative humidity change is notsignificant (Fig. 13e) and is basically none (Fig.12b), the increase in mixing ratio occurs in responseto the increase in temperature under the conditionregulated by the Clapeyron-Clausius equation.

The vertical profiles of temperature and moisturefor all the August days at 0900 JST in each climateperiod were also examined. The di¤erences in theAugust means of the temperature and moistureprofiles between the present and the future climateswere found to be significantly smaller than thoseshown in Fig. 13. Therefore, environmental changesin summer are more enhanced under the synopti-cally undisturbed conditions than under the meanconditions throughout the total August days. Fur-ther discussion on this topic is given in Section 4.

The changes in the temperature and moistureprofiles will lead to the changes in stability indicesand parameters. Temperature lapse rate, totalmoisture, one of stability indices, and an unstableenergy parameter are examined here. Figure 14shows the frequency distributions of TLR, PW, LI,and CAPE for the periods of the present, the near-future, and the future climate. The temperaturelapse rate becomes more stable in the future thanin the present climate, while the precipitable watervapor content increases in the future than in thepresent. LI and CAPE seem to indicate more un-stable values with the future period. The changesin TLR and PW seem to be consistent with thechanges indicated in Fig. 12.

The statistical significance for the di¤erences inthe means of the stability parameters between thepresent and the near-future (the future) climate isexamined by t-test statistic. The T value indicatingthe significance level for this analysis is 1.96. Theparameters that exceed the significance level forthe di¤erences between the present and the near-future are TLR (T ¼ 2.73), PW (T ¼ 3.62), andTT (T ¼ 1.97): TLR and TT become smaller withthe future period; while PW larger. For the di¤er-ences between the present and the future, the pa-rameters exceeding the significance level are TLR(T ¼ 3.53), PW (T ¼ 7.11), CAPE (T ¼ 4.27), CIN(T ¼ 2.73), LNB (T ¼ 2.23), and LI (T ¼ 2.14):CAPE, CIN, and PW become larger with the futureperiod; TLR and LI become smaller; and LNB be-comes higher. This result indicates that the atmo-

Fig. 12. The vertical profiles of (a) temperature, (b) relative humidity, and (c) water vapor mixing ratio at0900 JST averaged for the synoptically undisturbed condition in the present (blue), the near-future (green),and the future (red) climate simulations.


sphere becomes more unstable in terms of CAPE,PW, and LNB with the future while becomes morestable in terms of TLR and CIN. Coupled e¤ects inthe lapse rate decrease and the moisture increase re-sult in shifting LI towards more unstable. In thisway, this statistical analysis clearly demonstratesthat the di¤erences found in the frequency distribu-tions in Fig. 14 are significant in a statistical sense.

Furthermore, the di¤erences of the stability pa-rameters between the N and R days in each simu-lated climate are examined. Figure 15 comparesthe frequency distribution of the parameters, hav-ing distinct features identified in Fig. 14 and t-teststatistics, among the three simulated climate peri-

ods by classifying the N- and R-day cases. Com-pared with those shown in Fig. 14, the di¤erencesin the representation of the stability parametersamong the simulated climate periods are more en-hanced through distinguishing whether or not totalafternoon precipitation satisfies the defined thresh-old. With the future period, CAPE becomes larger,TLR becomes smaller, and PW becomes larger notonly on the N days but also on the R days. The in-creases in CAPE and PW with time seem to bemore distinct in the R days than in the N days. Aninteresting point is that the magnitude of TLR isfound to decrease even in the R days with the simu-lated period, despite that on rainy days convection

Fig. 13. The vertical profiles of the mean di¤erences of (a) temperature, (b) relative humidity, and (c) watervapor mixing ratio between the present and the near future climate (dashed lines) and between the presentand the future climate (solid lines), and the T values for (d) the temperature di¤erence, (e) the relative hu-midity di¤erence, and (f ) the mixing ratio di¤erence obtained from the undisturbed conditions in the GCMsimulations. In (a)–(c), the positive di¤erence means that the future (near-future) case is larger than thepresent case. In (d)–(f ), the threshold for the statistical significance (T ¼ 1.99) is indicated by dotted lines.


is expected to be more active; however, the value ofTLR unanimously shifts toward unstable on the Rdays in each climate period.

The t-test statistic was carried out for the di¤er-ences in the means of the stability parameters be-tween the N and R days, as was done in NT11.The T values for the di¤erences in the parametermeans between the rain categories in each simu-lated period are summarized in Table 2. The pa-rameters exceeding the significance level for all thesimulated periods are CAPE, LFC, SSI, LI, KI,TT, TLR, and PW. NT11 showed that these pa-rameters had statistical significance in distinguish-ing their mean values on the no-rain and rain days.The analysis for the present climate indicates that

the results obtained from the GCM outputs are con-sistent with the results obtained from the analysesof the radiosonde and MSM data. This also givesreliability on the analyses for the near-future andthe future climates.

The analyses of the statistical significance shownin Table 2 further indicate that the environmentalstability settings in terms of the stability parametersthat distinguish no-rain and rain events are basi-cally unchanged between the present and the futureclimate period. In other words, it is suggested thatthe environmental characteristics favorable for af-ternoon precipitation under the synoptically un-disturbed conditions will not change under globalwarming.

Fig. 14. The frequency distribution of (a) TLR, (b) PW, (c) LI, and (d) CAPE for the synoptically un-disturbed conditions in the present (blue), the near-future (green), and the future (red) climate simulations.


Fig. 15. The frequency distributions of (a) CAPE, (c) TLR, and (e) PW in the present (blue), the near-future(green), and the future (red) climates for the N days, and those of (b) CAPE, (d) TLR, and (f ) PW in the Rdays.


4. Discussion

The present analyses on the environmental stabil-ity under the synoptically undisturbed conditionsfrom the GCM outputs showed that with the futureperiod the temperature lapse rate in the lower tro-posphere becomes more stable while precipitablewater vapor and CAPE become larger (Fig. 14).The increase in precipitable water vapor was dueto the increase in water vapor mixing ratio at allthe vertical levels (Fig. 13c), while the decrease intemperature lapse rate was due to the larger increasein temperature at higher levels (Fig. 13a). Thesechanges were found to be statistically significant. Asstated in Section 3.3, these environmental changesappear to be more enhanced under the synopticallyundisturbed conditions than under the compositedAugust conditions in each climate period.

Figure 16 demonstrates the vertical distributionsof the di¤erences of temperature and moisture be-tween the present and the future (near future) cli-

mates for all the August days at 0900 JST. Thereare slight increases in temperature and water vapormixing ratio from the present to the future cli-mate: temperature increased by less than 2 K at allthe levels; and mixing ratio increased by less than1.5 g kg�1 at the lowest level and by less than1.0 g kg�1 above the 800-hPa level. In addition,the di¤erences between the present and the near-future climate are very small. Although statisticalsignificance test indicated that the di¤erences intemperature and mixing ratio were significant,those di¤erences for the August composites are sig-nificantly smaller than the di¤erences for the un-disturbed conditions. Furthermore, the temperaturedi¤erences are almost the same at all the levels be-tween 975 hPa and 500 hPa, which means thatTLR is not changed with the future time for all thecases in August. Therefore, the decrease of the tem-perature lapse rate in the lower troposphere in thefuture climate seems to be a unique feature for theundisturbed conditions in August.

Table 2. The T values for the di¤erences in the stability parameters between the N and R days in the present, the near-future, and the future climate simulations. The significance level is T ¼ 1.97. The bold digits satisfy this significancelevel.

CAPE CIN LCL LFC LNB SSI LI KI TT TLR PW

Present 6.32 1.74 0.72 3.80 0.27 4.96 6.86 3.90 5.69 5.68 3.43

Near future 5.26 0.93 1.03 3.31 0.82 5.57 6.26 5.04 7.12 6.73 2.08

Future 4.51 1.27 1.15 5.39 0.57 5.92 5.22 5.50 6.91 5.66 4.04

Fig. 16. The vertical profiles of the mean di¤erences of (a) temperature, (b) relative humidity, and (c) watervapor mixing ratio between the present and the near future climates (dashed lines) and between the presentand the future climates (solid lines) for all the days in August during each climate period.


As demonstrated in Fig. 8, the undisturbed con-ditions in August are resulted from the existence ofa subtropical high-pressure system over the Pacific.Synoptic-scale geopotential heights for the N andR days during the near-future and future periodsare exhibited in Fig. 17. Compared with the fieldshown in Fig. 8, the spatial patterns appear to beoverall similar among the simulated climate peri-ods. The area of interest, the Kanto Plain, is cov-ered with the subtropical high. For the same raincategory, the geopotential heights over the KantoPlain increase with the climate period, which is in-

dicated to be a manifestation of the future changesin meridional circulation rooted in the Tropics. It isclearly seen that the a¤ected region is not limited tothe Kanto Plain but also extended to the southernpart of the Japanese islands. Therefore, the changesof the environmental conditions in the Kanto Plainregion will appear in the same way in the southernpart of the Japanese islands. The decrease of themagnitude of TLR in the future periods is consid-ered to be a¤ected both by the warming in the trop-ical upper troposphere in the warming climate andby subsidence adiabatic warming in the middle to

Fig. 17. The same as Fig. 8, except from the near future-climate simulation for (a) the N days and (b) the Rdays and from the future-climate simulation for (c) the N days and (d) the R days.


upper troposphere over the subtropical-high region.The precipitation characteristics that are implied

from the projected changes in the environmentalstability are discussed here. Takemi (2007, 2010) in-vestigated the precipitation characteristics withinmesoscale convective systems by changing tempera-ture lapse rate and moisture content as well asshear profile in an idealized simulation setup wherethe base-state atmosphere is horizontally uniformwithout any external forcing. This idealized setupis regarded as synoptically undisturbed, since onlyconvective forcing is an initial thermal (and the re-sultant self-organizing e¤ects). The weak shear con-dition (i.e., 5 m s�1 di¤erence only in the lowest2.5 km) examined in Takemi (2007, 2010) corre-sponds to the synoptically undisturbed conditionsexamined in this study. Their results indicated thatwith the same CAPE environment convection in-tensity (represented as updraft strength) and meanprecipitation decrease as temperature lapse rate de-creases, and also indicated that with the similarlapse rate convection intensity and mean precipita-tion increase as CAPE increases. In the projectedenvironments in the future periods, it was foundthat temperature lapse rate decreases but CAPE in-creases. These contrasting e¤ects of the lapse ratedecrease and the CAPE increase will be canceledwith each other in the future-climate environments.Based on the results of Takemi (2007, 2010), con-vection intensity and mean precipitation will notdecrease. Thus, it is implied that precipitationamount would be maintained or even intensified inthe future climates.

An interesting result in Takemi (2010) was thatthe maximum precipitation intensity increases withthe decrease in temperature lapse rate while withCAPE being unchanged. On the other hand, thepresent analyses of the GCM data indicated thatboth static stability and CAPE increases. Applyingthe result of Takemi (2010) to the projected changesin the environmental stability, the precipitation in-tensity in short time-scales is suggested to increase.This point should be quantitatively evaluated fromthe results of non-hydrostatic RCM simulationswithout cumulus parameterizations.

5. Conclusions

This study investigated the characteristics of theenvironmental stability for afternoon convectiveprecipitation that would develop under synopticallyundisturbed conditions in summer with the use ofthe outputs of GCM climate simulations and evalu-

ate the changes in the stability conditions underwarming climates at a regional-scale. The GCMsimulations were done by a super-high-resolutionatmospheric GCM, MRI-AGCM3.2S, at about20-km resolution (Kitoh et al. 2009). The climatesimulation outputs that were used in this study cor-responded to three 25-year periods: 1980–2004 forthe present climate; 2020–2044 for the near-futureclimate; and 2075–2099 for the future climate. Thesimulations for future climates were done with theIPCC A1B emission scenario that induces globalwarming. The Kanto Plain was chosen as the areaof interest, because there have been a su‰cientnumber of the previous studies that examined theenvironmental stability over the region and it wasuseful to evaluate the performance of the GCMoutputs against the observations and the analyses.To diagnose the environmental conditions, some ofthe commonly used stability parameters and indiceswere used. The future projections of the changes inthe environmental stability for the afternoon pre-cipitation events over the region were provided.

The comparison of the GCM present climate ex-tracted as the synoptically undisturbed conditionswith the corresponding states revealed by the radio-sonde observations at Tateno in the Kanto Plainand the MSM analyses indicated that the GCMpresent climate well reproduced the characteris-tics of the temperature and moisture profiles, thestability parameters, and also daily precipitationamounts up to 25 mm. These favorable agree-ments of the GCM outputs with the real, existingclimates enabled us to investigate and evaluate thechanges in the environmental stability under globalwarming.

In the future climates, temperature lapse ratedecreased in the lower troposphere, while watervapor mixing ratio increased throughout the deeptroposphere. The changes in the temperature andmoisture profiles resulted in the increase in bothprecipitable water vapor and CAPE, which wereevaluated as statistically significant. These pro-jected changes were indicated to be enhanced withthe future period.

Furthermore, the di¤erences of the stability pa-rameters between the no-rain and the rain daysunder the synoptically undisturbed condition ineach simulated climate period were examined. Thestatistical analyses for these di¤erences indicatedthat the environmental conditions in terms of thestability parameters that distinguish no-rain andrain events are basically unchanged between the


present and the future (including the near-future)climate. This result suggests that the environmentalcharacteristics favorable for afternoon precipitationin the synoptically undisturbed environments willnot change under global warming. The present re-sults further suggest that the potential for thunder-storm activity is not significantly a¤ected in thefuture climates despite the increase of moisture con-tent; this is considered to be due to the contrastinge¤ects of the moisture increase and the lapse ratedecrease.

The present study focuses on the characteristicsand future changes of the environmental stabilityfor convective precipitation, but not on the precipi-tation amount itself. Precipitation characteristicsin future climates under global warming have beenactively investigated by non-hydrostatic regionalclimate models such as in Kanada et al. (2008,2010b). Combining those RCM results with thepresent analyses will give more in-depth under-standing on the behavior of precipitation underglobal warming.

Acknowledgments

We would like to thank Drs. Sachie Kanada andMasuo Nakano at Japan Agency for Marine-EarthScience and Technology for their comments on thepresent study. Prof. Eiichi Nakakita at DisasterPrevention Research Institute (DPRI), Kyoto Uni-versity is greatly acknowledged for his dedicated ef-forts in promoting interactions between the DPRIand the MRI research groups. We would also liketo thank two anonymous reviewers for their helpfulcomments in improving the original manuscript.This work was conducted under the framework ofthe ‘‘Projection of the change in future weather ex-tremes using super-high-resolution atmosphericmodels’’ supported by the KAKUSHIN programof the Ministry of Education, Culture, Sports,Science, and Technology (MEXT) of Japan. Thiswork was also supported by a Scientific Researchgrant from Japan Society for Promotion of Sciences(JSPS). The GCM computations were carried outon the Earth Simulator.

References

Bengston, L., and K. I. Hodges, 2011: On the evaluationof temperature trends in the tropical troposphere.Clim. Dyn., 36, 419–430.

Bluestein, H. B., 1993: Synoptic-Dynamic Meteorology in

Midlatitudes, Vol. II, Oxford Univ. Press, 594 pp.

Bluestein, H. B., and M. H. Jain, 1985: Formation ofmesoscale lines of precipitation: Severe squall linesin Oklahoma during the Spring. J. Atmos. Sci., 42,1711–1732.

Christy, J. R., W. B. Norris, R. W. Spencer, and J. J.Hnilo, 2007: Tropospheric temperature changesince 1979 from tropical radiosonde and satellitemeasurements. J. Geophys. Res., 112, D06102,doi:10.1029/2005JD006881.

Chuda, T., and H. Niino, 2005: Climatology of environ-mental parameters for mesoscale convections inJapan. J. Meteor. Soc. Japan, 83, 391–408.

Del Genio, A. D., M.-S. Yao, and J. Jonas, 2007: Willmoist convection be stronger in a warmer climate?Geophys. Res. Lett., 34, L16703, doi:10.1029/2007GL030525.

Fu, Q., S. Manabe, and C. M. Johanson, 2011: On thewarming in the tropical upper troposphere: Modelsversus observations. Geophys. Res. Lett., 38,L15704, doi:10.1029/2011GL048101.

Fujibe, F., 1998: Spatial anomalies and long-termchanges of precipitation in Tokyo. Tenki, 45, 7–18(in Japanese).

Fujibe, F., N. Yamazaki, M. Katsuyama, and K. Ko-bayashi, 2005: The increasing trend of intenseprecipitation in Japan based on four-hourly datafor a hundred years. SOLA, 1, 41–44.

Fujibe, F., N. Yamazaki, M. Katsuyama, and K. Ko-bayashi, 2006: Long-term changes in the diurnalprecipitation cycles in Japan for 106 years (1898–2003). J. Meteor. Soc. Japan, 84, 311–317.

Fujibe, F., H. Togawa, and M. Sakata, 2009: Long-termchange and spatial anomaly of warm season after-noon precipitation in Tokyo. SOLA, 5, 17–20.

IPCC, 2007: Climate Change 2007: The Physical Science

Basis. Cambridge University Press, Cambridge,United Kingdom, 996 pp.

Iwasaki, H., 2010: Recent positive trend in heavy rain-fall in eastern Japan and its relation with varia-tions in atmospheric moisture. Int. J. Climatol.,doi: 10.1002/joc.2269.

Iwashima, T., and R. Yamamoto, 1993: A statisticalanalysis of the extreme events: long-term trend ofheavy daily precipitation. J. Meteor. Soc. Japan,71, 637–640.

Kamiguchi, K., A. Kitoh, T. Uchiyama, R. Mizuta, andA. Noda, 2006: Changes in precipitation-based ex-tremes indices due to global warming projected bya global 20-km-mesh atmospheric model. SOLA,2, 64–67.

Kamiguchi, K., O. Arakawa, A. Kitoh, A. Yatagai, A.Hamada, and N. Yasutomi, 2010: Developmentof APHRO_JP, the first Japanese high-resolutiondaily precipitation product for more than 100years. Hydro. Res. Lett., 4, 60–64.

Kanada, S., C. Muroi, Y. Wakazuki, K. Yasunaga, A.Hashimoto, T. Kato, K. Kurihara, M. Yoshizaki,


and A. Noda, 2005: Structure of mesoscale convec-tive systems during the late Baiu season in theglobal warming climate simulated by a non-hydrostatic regional model. SOLA, 1, 117–120.

Kanada, S., M. Nakano, S. Hayashi, T. Kato, M. Naka-mura, K. Kurihara, and A. Kitoh, 2008: Reprodu-cibility of maximum daily precipitation amountover Japan using a high-resolution non-hydrostaticmodel. SOLA, 4, 105–108.

Kanada, S., M. Nakano, and T. Kato, 2010a: Changes inmean atmospheric structures around Japan duringJuly due to global warming in regional climate ex-periments using a cloud-system resolving model.Hydro. Res. Lett., 4, 11–14.

Kanada, S., M. Nakano, and T. Kato, 2010b: Climato-logical characteristics of daily precipitation overJapan in the Kakushin regional climate experi-ments using a non-hydrostatic 5-km-mesh model:Comparison with an outer global 20-km-mesh at-mospheric climate model. SOLA, 6, 117–120.

Kanae, S., T. Oki, and A. Kashida, 2004: Changes inhourly heavy precipitation at Tokyo from 1890 to1999. J. Meteor. Soc. Japan, 82, 241–247.

Kawano, K., Y. Hirokawa, and H. Ohno, 2004: Diagno-sis of airmass thunderstorm days using radiosondedata: The summer Kanto area under the Pacificsubtropical anticyclone. Tenki, 51, 17–30 (in Japa-nese).

Kim, S., Y. Tachikawa, E. Nakakita, and K. Takara,2010: Hydrologic evaluation on the AGCM20 out-put using observed river discharge data. Hydro.

Res. Lett., 4, 35–39.Kimoto, M., N. Yasutomi, C. Yokoyama, and S. Emori,

2005: Projected changes in precipitation character-istics around Japan under the global warming.SOLA, 1, 85–88.

Kitoh, A., and T. Uchiyama, 2006: Changes in onset andwithdrawl of the east Asian summer rainy seasonby multi-model global warming experiments. J.

Meteor. Soc. Japan, 84, 247–258.Kitoh, A., T. Ose, K. Kurihara, S. Kusunoki, M. Sugi,

and KAKUSHIN Team-3 Modeling Group, 2009:Projection of changes in future weather extremesusing super-high-resolution global and regional at-mospheric models in the KAKUSHIN program:Results of preliminary experiments. Hydro. Res.

Lett., 3, 49–53.Kusunoki, S., J. Yoshimura, H. Yoshimura, A. Noda, K.

Oouchi, and R. Mizuta, 2006: Change of Baiu rainband in Global warming projection by an atmo-spheric general circulation model with 20-km gridsize. J. Meteor. Soc. Japan, 84, 581–611.

Mizuta, R., K. Oouchi, H. Yoshimura, A. Noda, K. Ka-tayama, S. Yukimoto, M. Hosaka, S. Kusunoki,H. Kawai, and M. Nakagawa, 2006: 20-km-meshglobal climate simulations using JMA-GSM

Model—Mean climate states—. J. Meteor. Soc.

Japan, 84, 165–185.Murakami, H., and B. Wang, 2010: Future change of

north Atlantic tropical cyclone tracks: Projectionby a 20-km-mesh global atmospheric model. J. Cli-mate, 23, 2699–2721.

Murakami, H., B. Wang, and A. Kitoh, 2011: Futurechange of western North Pacific typhoons: Projec-tions by a 20-km-mesh global atmospheric model.J. Climate, 24, 1154–1169.

Ninomiya, K., 2009: Characteristics of precipitation inthe Meiyu-Baiu season in the CMIP3 20th centuryclimate simulations. J. Meteor. Soc. Japan, 87,829–843.

Nomura, S., and T. Takemi, 2011: Environmental stabil-ity for afternoon rain events in the Kanto plain insummer. SOLA, 7, 9–12.

Santer, B. D., T. M. L. Wigley, C. Mears, F. J. Wentz,S. A. Klein, D. J. Seidel, K. E. Taylor, P. W.Thorne, M. F. Wehner, P. J. Gleckler, J. S. Boyle,W. D. Collins, K. W. Dixon, C. Doutriaux, M.Free, Q. Fu, J. E. Hansen, G. S. Jones, R. Ruedy,T. R. Karl, J. R. Lanzante, G. A. Meehl, V. Ram-aswamy, G. Russell, and G. A. Schmidt, 2005:Amplification of surface temperature trends andvariability in the tropical atmosphere. Science,309, 1551–1556.

Santer, B. D., P. W. Thorne, L. Haimberger, K. E. Tay-lor, T. M. L. Wigley, J. R. Lanzante, S. Solomon,M. Free, P. J. Gleckler, P. D. Jones, T. R. Karl,S. A. Klein, C. Mears, D. Nychka, G. A. Schmidt,S. C. Sherwood, and F. J. Wentz, 2008: Consis-tency of modelled and observed temperature trendsin the tropical troposphere. Int. J. Climatol., 28,1703–1722.

Sato, M., and M. Takahashi, 2000: long-term changesin the properties of summer precipitation in theTokyo area. Tenki, 47, 643–648 (in Japanese).

Sugi, M., A. Noda, and N. Sato, 2002: Influence of theglobal warming on tropical cyclone climatology:an experiment with the JMA global model. J. Me-

teor. Soc. Japan, 80, 249–272.Taguchi, A., K. Okuyama, and Y. Ogura, 2002: The

thunderstorm activity observed by SAFIR and itsrelation to the atmospheric environment over theKanto area in the summer. Part II: Thunderstormprediction by stability indices. Tenki, 49, 649–659(in Japanese).

Takemi, T., 2010: Dependence of the precipitation inten-sity in mesoscale convective systems to temperaturelapse rate. Atmos. Res., 96, 273–285.

Takemi, T., 2007: Environmental stability control of theintensity of squall lines under low-level shear con-ditions. J. Geophy. Res., 112, D24110, doi:10.1029/2007JD008793.

Thorne, P. W., J. R. Lanzante, T. C. Peterson, D. J.


Seidel, and K. P. Shine, 2010: Tropospheric tem-perature trends: history of an ongoing controversy.WIREs Climate Change, 2, 66–88.

Wakazuki, Y., M. Yoshizaki, K. Yasunaga, C. Muroi, S.Kanada, A. Hashimoto, T. Kato, K. Kurihara,and A. Noda, 2005: Changes in the characteristicfeatures of disturbances appearing in the Baiu fron-tal zone over western Japan due to global warm-ing. SOLA, 1, 129–132.

Yasunaga, K., M. Yoshizaki, Y. Wakazuki, C. Muroi,K. Kuruhara, A. Hashimoto, S. Kanada, T. Kato,S. Kusunoki, K. Oouchi, H. Yoshimura, R. Miz-uta, and A. Noda, 2006: Changes in the Baiu fron-tal activity in the future climate simulated bysuper-high-resolution global and cloud-resolving

regional climate models. J. Meteor. Soc. Japan,84, 199–220.

Yonetani, T., 1982: Increase in number of days withheavy precipitation in Tokyo urban area. J. Appl.Meteor., 21, 1466–1471.

Yonetani, T., 1975: Characteristics of atmospheric verti-cal structure on days with thunderstorms in thenorthern Kanto Plain. J. Meteor. Soc. Japan, 53,139–148.

Yoshizaki, M., C. Muroi, S. Kanada, Y. Wakazuki, K.Yasunaga, A. Hashimoto, T. Kato, K. Kurihara,A. Noda, and S. Kusunoki, 2005: Changes ofBaiu (Mei-yu) frontal activity in the global warm-ing climate simulated by a nonhydrostatic regionalmodel. SOLA, 1, 25–28.


title stability for summertime afternoon precipitation...

Documents