Coastal trapped wave (CTW) is defined as the wave in a stratified ocean over sloping topography and propagates along a coast on its left (right) in the southern hemisphere (northern hemisphere).The characteristics of the CTW have a hybrid nature between coastal Kelvin waves [Thompson,1879] and continental shelf waves [Robinson, 1964] and the wave is mainly excited by alongshore component of winds. A large number of observed evidence for the existence of the CTWs are reported from the coastal area around the world, e.g., [Shoji, 1961; Cutchin and Smith, 1973; Smith,1978; Brink et al., 1978].

　For Australian coasts, the studies of the CTW are initiated by a seminal observational work of Hamon [1962]. Since then, vertical structures and phase speed of the CTW have been locally investigated in detail by several intensive observational studies, such as the Australian Coastal Experiment [Freeland et al., 1986; Church et al., 1986]. These studies, however, treat the waves mainly as the local phenomena along the eastern coast of Australia, and the origin of the waves and behaviour of the waves during their propagation along the coast are not discussed in detail.

　In this study, therefore, we investigate the propagation characteristics and structures of the CTWs around Australia from the viewpoint of the continental scale using observed sea level data and results from ocean general circulation models.

　The daily mean sea level data at 20 stations around Australia (Fig.1) are obtained from National Tidal Facility of Australia and the University of Hawaii Sea Level Center. The variations of the sea level are dominated by the seasonal variability at the stations located on the northern coast, while short term variations of the sea level with a period of shorter than one month are dominant along thecoast in southwestern, southern and eastern regions. It also turns out that the short term variations show strong seasonality, with large amplitude during austral winter. Since observed data are limited only for 4 years, we focus on the variation with the period shorter than a month.

　The high pass filtered sea level variability during the austral winter clearly shows anticlockwise propagation of the sea level signals from the western/southwestern coast to the eastern coast (Fig.2).However, the amplitude of the waves abruptly weakens around Tasmania in many cases. It is found that typical phase speed of the wave differs between southern and eastern coast; the phase speed in the southern coast is faster than 5m/s but it is 2-4 m/s in the eastern coast. Spectrum analyses indicate that dominant time-scale of the variation of the sea level in the southwestern coast is around 10-14 days, which is also a peak in spectrum for alongshore wind variability in the southwestern coast. The association between the two fields is quite high, suggesting the wind forced wave excitation in this region.

　To check a possibility for the waves to propagate freely from southwestern region to the eastern coast and to evaluate roles of topographic features on the characteristics of the wave propagation, a regional OGCM around Australia is developed and used for sensitivity studies. The model is based on Princeton Ocean Model, which incorporates a free surface and a terrain-following σ coordinates in the vertical direction. An idealised alongshore wind stress limited only within the southwestern coast with a period of two weeks is applied as a forcing of the coastal waves.

　Simulated sea surface height (SSH) variability from results with realistic bottom topography(Control Run) captures the major characteristics obtained from the observed sea level data (Fig.3a).The SSH signal propagates anticlockwise from the southwestern coast to the eastern coasts, with the phase speed of 5.6 m/s and 2.8-3.7 m/s in the southern and eastern coast, respectively. The model also reproduces the weakening of the wave amplitude around Tasmania. In order to explore reasons for the above characteristics of the waves, especially for the change in the phase speed and the amplitude, two sensitivity experiments are conducted with different bottom topography around the southeastern and eastern coasts. In the first experiment, Case 1, a wider continental shelf as in the southern coast is applied to the eastern coast. The phase speed of the waves along the eastern coast increases significantly to the value similar to that in the southern coast (Fig.3b). Modal decomposition of alongshore currents to eigenvalue solutions of the CTWs confirms that the first mode CTW explains more than 60 % of total variability both in the southern and eastern coasts.However, the difference in the width of the continental shelf, the wide shelf in the south and narrow shelf in the east, results in the different phase speed; the narrower the shelf is, the slower the phase speed becomes. In the second experiment, Case 2, the width of the continental shelf in the eastern coast remains narrow as in Control Run, but Tasmania island at the southeastern corner of Australia is removed. In Case 2, the simulated amplitude and the phase speed of the waves are almost the same as in Control run (Fig.3c). It turns out from the two experiments that the narrow continental shelf in the eastern coast is responsible for the relatively slow CTWs and the weakening of the amplitude in the eastern coast.

　To examine the behaviour of the CTWs in more realistic conditions, results from a high-resolution ocean general circulation model (OFES) are explored in detail. The OFES results also reproduce the anticlockwise propagation of the CTWs around Australia. Spatial structures of the CTWs, derived from a composite of the simulated positive SSH events, demonstrate that the waves are trapped within about 200 km (80 km) from the coast in the southern (eastern) region with much larger alongshore wavelength of a few thousand kirmetres. The same eigenmode analyses for Control Run of the regional OGCM are applied to the OFES result. It is found that the first CTW mode is dominant both in the southern and eastern coasts, but the phase speed in the southern coast is faster than that in the eastern coast, which is also consistent with the observed results.

　However, OFES does indicate the large amplitude responses in the eastern coast in many times,which has strong association with the wind forcing at the southeastern coast of Australia. An additional sensitivity experiment with wind forcing at the southern part of the eastern coast is carried out. The result indicates that the forcing in the eastern coast is necessary for the waves to have the amplitude similar to those observed in the sea level data and simulated in OFES.

　The above results from the observed data and the numerical models indicate that the CTWs observed in the eastern coast of Australia can be excited in the southwestern coast by theatmospheric synoptic disturbances and that the propagation characteristics of the CTWs are determined mainly by the width of the continental shelf in the southern and eastern coasts. In addition, the wind forcing in the southern part of the eastern coast plays a key role in strengthening of the amplitude or excitation of the CTWs along the eastern coast of Australia.

Fig.1 ; Locations of sea level stations around Australia (Red dots)

Fig.2 ; Hovmoller diagram of high-pass filtered (a) sea level anomaly from observation and (b) alongshore component of the wind in 1995. Time spans from 1 June to 31 August (during austral winter). Horizontal axis indicates the distance measured anticlockwise from Broome (northwestern part of Australia) to Wyndham (northern part).

Fig.3: Hovmoller diagram of sea surface height from the results of the numerical experiments for (a)Control Run, (b) Case 1; a wider shelf in the eastern coast and (c) Case 2; without Tasmania island.Horizontal axis indicates the distance measured anticlockwise from a point in the southwestern coast(origin of the wave). A few letters in the horizontal axis denote abbreviations of some locations (refer to Fig.1.)

　大陸沿岸に沿って伝播する海洋波動は、沿岸域の海況変動をもたらすだけでなく、大洋規模の海洋気候変動の一部をなす要素として重要な役割を果たしている。このような沿岸捕捉波の励起源や伝播特性を調べることは、沿岸海洋変動と大規模な気候変動との関連性を解き明かし、さらには海洋変動予測の精度向上にも貢献するものである。沿岸捕捉波の研究は、日本沿岸では異常潮位や急潮との関連において、また北米および南米大陸沿いではエルニーニョ現象との関連において多数行われてきている。しかし、インド洋と太平洋の境に位置するオーストラリア亜大陸周りの沿岸捕捉波については、南半球の中緯度と熱帯域との橋渡しの役割を担う可能性があるにもかかわらず、局所的な視点から扱われた断片的な研究に限られていた。本論文は、オーストラリア亜大陸周りの観測データと数値実験を通じて、沿岸捕捉波の強制域と伝播過程を大陸規模の視点から明らかにすることを目的としたものである。

　本論文は5つの章から成り立っている。

　まず、第1章は導入部であり、海洋の沿岸波動、特に沿岸捕捉波研究の歴史、およびオーストラリア亜大陸周辺での沿岸捕捉波の研究の現状が紹介されるとともに、本論文の内容と目的が述べられている。

　第2章では、オーストラリア沿岸の潮位計データを解析し、水位変動の統計的特性の抽出を試みている。その結果、オーストラリア亜大陸北部沿岸域では季節的な海面高度変動が卓越するが、西部-南部-東部では一ヶ月よりも短周期の変動が南半球の冬場に卓越し、反時計回りに伝播していることが示された。また、波動伝播が顕著な南岸と東岸域とでは、沿岸捕捉波の位相速度が大きく異なるとともに、水位変動の振幅もオーストラリア南東岸付近で急激に減少していることも明らかにされた。この成果は、従来、局所的な解釈しかなされてこなかったオーストラリア沿岸の沿岸捕捉波が、大陸規模の伝播を伴っていることを初めて明らかにした点で高く評価できる。

　第3章では、第2章で明らかにされた短周期水位変動の励起源として南半球のストームトラックに関連する大気擾乱を仮定し、オーストラリア亜大陸の南西岸で吹く風応力が励起する沿岸捕捉波の時空間的な伝播特性を調べている。比較的簡略化した設定下での海洋循環モデルを用いた数値実験を通じて、南岸と東岸のどちらにおいても沿岸捕捉波の第1モードが卓越していることを示し、両岸での位相速度の違いは大陸棚の幅の違いに伴う力学特性の相違が原因であることを明らかにした。これは、従来のオーストラリア東岸における沿岸捕捉波の解釈に大幅な修正を促す重要な結果である。さらに、オーストラリア南東岸付近を境にした水位変動の振幅の急減少に関しても考察を進め、タスマニア島の存在よりも、南岸から東岸に向かって大陸棚の幅が急激に減少することにその主原因が求められることを明らかにした。

　第4章では、より現実的な状況下での波動の伝播特性を調べるため、高解像度海洋大循環モデルの結果の解析を行っている。第3章の簡略化したモデルと異なり、平均流や中規模渦擾乱などの影響を含んでいるにもかかわらず、第3章で明らかにされたものと同様なオーストラリア南西岸からの沿岸捕捉波の伝播特性を確認することができた。その一方で、第3章の簡略化モデルでは再現することのできなかった東岸域での振幅の大きな沿岸捕捉波の伝播も現実的なモデルでは再現されていた。そこで、簡略化したモデルによる感度実験を改めて繰り返すことによって、この東岸を伝播していく振幅の大きな沿岸捕捉波は、オーストラリアの南西岸付近で励起されたものではなく、南東岸付近における第2の風強制によって励起されたものであることを確認した。

　このように、本論文はオーストラリア亜大陸周りで発生する一ヶ月以下の周期帯の沿岸捕捉波に注目し、その励起源および伝播特性の詳細を大陸規模の視点から明らかにすることによって、沿岸捕捉波の研究に大きく貢献した。特に、従来の局所的な枠組みの内にとどまっていたオーストラリア周りの沿岸捕捉波の解釈を大きく見直す必要のあることを示した点は、今後の沿岸海洋波動に関する研究を新たに方向付けた成果として高く評価できる。

　なお、本論文の第2、3、4章は山形俊男教授、升本順夫助教授との共同研究であるが、いずれも論文提出者が主体となって研究を行ったものであり、その寄与が十分であると判断できる。

　したがって、審査員一同は、博士(理学)の学位を授与できると認める。