Madden-Julian Oscillation (MJO) is the most significant atmospheric variation in the tropics, and its reproduction is one of the major issues for current General Circulation Models (GCMs). Using the output data of the MJO 2006 experiment conducted by Miura et al. (2007) in a global cloud resolving Non-hydrostatic Icosahedral Atmospheric Model (NICAM, Satoh et al. 2008), the author analyzed the Convective Momentum Transport (CMT) in relation with the phase of a successfully reproduced MJO. It was found that, at the altitude of 2km-6.5km, CMT in ensemble worked effectively to slow down the eastward propagation of the westerly wind involved with the MJO convection.

Convective momentum transport of subgrid-scale cloud systems are a well known source of uncertainty in current GCMs. In GCMs, CMT are usually, if at all, parameterized as a mixing component which operates to reduce environmental wind shear (down gradient). However, it has long been known that organized Meso-scale Convective Systems (MCSs) often involve CMT that produce or intensify environmental wind shear (up gradient); squall line type MCSs are typically up gradient in the direction perpendicular to the line alignment. By directly calculating the clouds without using cumulus parameterization, NICAM permits such up gradient CMT, and has possibly overcome this uncertainty due to inaccurate representation of CMT effects. Since NICAM succeeded in reproducing the MJO, it is worth examining what roles CMT has played in the achievement. Despite a successful reproduction of the MJO with NICAM, accurate parameterization of CMT effects remains to be a major issue, since the state-of-the-art super computer resources are not sufficient to perform long term integrations using global cloud resolving models, while satisfactory MJO reproduction is necessary in meeting social expectations such as long-range weather forecast or prediction of global climate change.

Setting the long term goal on the development of a cumulus scheme relevant to up gradient CMT, the objectives of this study were (I) to clarify the distribution structure of CMT acceleration involved in rainbands within the MJO convection, (II) to quantify their ensemble impacts on the environmental zonal wind, and (III) to analyze what components account for the ensemble. The results of analysis aimed on (I), (II), and (III) are respectively shown in the 4th, 5th, and 6th chapter of this paper, succeeding preliminary results shown in chapter 3.

In chapter 3, center of the MJO convection was located; meso-scale rainbands produced by NICAM and those retrieved from the Tropical Rainfall Measurement Mission Microwave Imager (TRMM/TMI) were compared; MJO wind structures produced by NICAM and re-analysis were compared. The average propagation speed of the MJO center between 100。 - 170。 E was 1.08。 / (6hours) =4.8m/s. In the output data of NICAM, 162542 rainbands were identified, and the frequency distributions of the rainbands were in good agreement with those observed by TMI. Area within 100。 - 170。 E, 12。 S-12。 N was selected for further analysis due to the large number of rainband cases identified in it and their relatively homogeneous distribution. In this analysis region, 60997 cases were found, 15221 cases of which came along with complete 3-d dataset necessary to calculate CMT. The axis alignment of NICAM rainbands showed an overall preference to the east-west direction, but for those having large area and located between -20。 to 0。 from the MJO center tended to be closer to southeast-northwest direction. Similar preferences were observed in TMI rainbands, but the data amount of TMI was not enough to show the significance of the latter tendency. Average wind structure of MJO produced in NICAM and that of JCDAS agreed reasonably.

In chapter 4, an MJO-relative composite structure of the acceleration due to CMT is displayed, after a check on temporal representability of 6-hourly CMT data. Six-hourly snapshot data proved to be sufficient to represent the characteristics of CMT and its upscale effect on the environmental horizontal wind. Equatorial rainbands that involve strong upscale acceleration to the environment were most frequently found between -20。 to 0。 from the MJO center. Their upscale zonal acceleration ensemble formed a three-storied structure: positive at lower levels (below 1.6km); negative at mid levels (2km-6.5km); postive at upper levels (above 11km). At 7km-10km, both negative and positive acceleration were found and the inclination was unclear.

In chapter 5, the impact of upscale acceleration of the CMT ensemble on the environmental zonal wind was quantified, and under an assumption that the MJO structure does not change, the author estimated the expected modification in phase speed of the MJO when the CMT effects were ignored. The upscale acceleration due to CMT accounted for -160% on the 2km-6.5km averaged wind difference between +20。 and -30。 from the MJO center. In the absence of CMT effects, would increase to 260% of its original value; provided that the MJO structure does not change, the eastward propagation of the westerly wind at 2km-6.5km would become 2.6 times faster. The accelerated eastward propagation of the westerly wind (2km-6.5km) is likely to lead to earlier triggering of new convection to the east of the MJO, and in a view that the wind structure of the MJO is a response to convective heating, it is suggested that the eastward propagation speed of the entire MJO may also increase by 2-3 times of the original speed.

In chapter 6, the CMT components that account for the three-storied structure were explored; rainbands were sliced into height-longitude sections, each 2-d structure of up/down drafts in the height-longitude sections were identified, and the 2-d structures were further divided and classified into groups according to the types of CMT they involve. While more than half of the negative acceleration of 2.5km-6km were due to the contribution from updrafts involving up gradient CMT, the contribution from updrafts that mix the mid-level (2.5km-6km) with the upper levels (above 7km) and contribution from rear-inflow type down drafts were also relatively large.

It is concluded in correspondence with the objectives of this study (I) - (III) that:

(1) CMT involved with rainbands within the MJO has a clear preference to construct into a three-storied structure which positively/negatively/positively accelerate environmental zonal wind at the lower/middle/upper levels.

(2) The acceleration due to CMT has large impact on the change of environmental zonal wind (up to -160% at mid-levels), possibly having strong control over the phase speed of the MJO.

(3) Both up/down gradient CMT largely account for such strong impacts.

It is strongly suggested that cumulus parameterization relevant to the effects of up gradient CMT is necessary for GCMs in order to reproduce the MJO in satisfactory. According to the past works done by others and to the preference for the southeast-northwest direction found in the axis alignment of large rainbands located between -20。 to 0。 relative to the MJO center, it is speculated that the up gradeint CMT were involved with rainbands having north-south component in their alignment; they are likely to evolve under conditions in which a low-level (1000hPa-800hPa) zonal wind shear exist. However, such relationships of environment and CMT cannot yet be discussed quantitatively enough, and the direct connection between the environmental condition and CMT ensemble effects remains as a major issue to be clarified.

マッデン=ジュリアン振動(MJO) とは赤道域で数1000kmスケールの対流活発域を伴う波数1の擾乱が東進し、30-60日で地球を1周する現象である。MJOは熱帯域で最大の振幅をもつ季節内変動であり、台風やエルニーニョの発生、中緯度の気象にも影響を与え、中期の気象予測にとって重要な現象である。MJOは、衛星観測により、その中に対流雲群から構成される西進雲クラスターを内在する、スーパークラスターと呼ばれる東進する大規模な雲クラスターを伴うなどの複雑な階層構造を持つことが知られている。しかしながら、小規模で複雑な挙動をする対流雲が重要な役割を演じているため、その力学や東進の機構の理解は不十分で、全球大気の数値モデルによる再現精度も低い。

申請者は、最近、対流雲群から構成される雲システムを解像する画期的な全球大気モデル(Non-hydrostatic Icosahedral Atmospheric Model: NICAM)により、世界で初めてMJOが再現されたことに注目し、この出力データを用いて、MJO内の対流雲システムの運動量輸送(Convective Momentum Transport: CMT) を解析した。その結果、CMT は集団として背景東西風に対して大きな加減速の効果を持つこと、とりわけMJO に伴う西風域の東進を大きく遅らせる働きをしている可能性があることを明らかにした。

第1章では、対流雲群で構成される降水バンドによる運動量輸送及びMJOに関する包括的レビューが行われ、第2章では、本研究で用いるデータの詳細と、MJO中心の決め方、降水バンドの客観的な抽出方法などの解析手法が記述されている。

第3章では、最初に、NICAMが再現した降水バンドと熱帯降雨観測衛星搭載のマイクロ波放射計で観測された降水バンドの地域別の頻度分布が概ね良く対応することが確認され、降水バンドの事例数が豊富でかつ頻度分布の偏りが小さい100。 - 170。E, 12。S-12。N が主要な解析領域として選定される。NICAM の降水バンドは全体としては東西方向に長軸を持つものが支配的であったが、MJO の中心から西20。の範囲では南東-北西方向の長軸を持つ傾向が見られた。続いて、NICAM が再現したMJO の風速場が、気象庁気候データ同化システム(JCDAS) による再解析データとも良く対応していることが確認された。

第4章では、NICAMの出力データを用いて、MJO 内でのCMTによる加速の分布構造が求められ、強い加速をもたらすCMT を伴う降水バンドは赤道域ではMJO 中心から西20°にかけて最も多く存在することが示された。これらの降水バンドに伴うCMTは集団としては、背景東西風に対して、下層で正、中層で負、上層で正の加速という3 階建て構造の加速特性を持っていることも示された。

第5章では、MJO の構造がCMTの有無に依らないとの仮定のもとに、CMTの効果を除外した場合に期待されるMJO の位相速度が見積られている。その結果によると、CMTの減速効果が無ければ、MJOの東進速度は観測される値の2~3 倍となる可能性が示唆された。

第6章ではCMT による加速の3 階建て構造を構成する要素について、それぞれの降水バンドを東西鉛直断面に分解して調べている。各断面において上昇流/下降流を識別し、更にこの上昇流/下降流をCMT の性質によって分類して、各分類グループ別に3 階建て構造への寄与を集計したところ、中層における減速の半分以上は勾配拡散型でないCMTの寄与によるが、中層と上層を混合するタイプの上昇流など勾配拡散型のCMT も寄与していることが示された。

このように本論文は、MJOの東進に降水バンドに伴うCMT による減速効果が強く効いている可能性が高いこと、この減速には勾配拡散型とそうでないCMTの両方が寄与していることを明らかにしたものである。本研究に用いたNICAMの空間解像度は個々の積雲対流を解像するのには不十分な7kmであるため、この結果が現実大気のCMTにどう対応しているかどうか、またCMTと降水バンドとの走向はどのような関係にあるか等については今後も検証が必要であるが、MJOの再現に成功したNICAMにおいてCMTが重要な役割を演じていることを明らかにしたことは、MJOの力学や東進機構の理解に繋がる重要な手懸かりを与えた研究として高く評価される。また、全球大気モデルにおける大きな不確定要素の1つであるCMTのパラメタリゼーションの開発にも貴重な示唆を与えるものである。

なお、本論文は高薮 縁氏(指導教員)、佐藤正樹氏、那須野智江氏、三浦裕亮氏との共同研究に基づくが、論文提出者が主体となって解析を行ったものであり、論文提出者の寄与が十分であると判断する。

よって、博士(理学)の学位を授与できると認める。