Background and Objectives. There are a wide variety of subduction-related orogens in the world. One extreme is the Marianas type orogens, which are characterized by back-arc spreading; the other extreme is the Andean type orogens, which are characterized by a fold-and-thrust belt on the back-arc side. The Northeast Japan arc has changed from a Mariana-type orogen in Miocene time to an Andean-type orogen since Pliocene time to the present. Therefore, detailed studies on the tectonic evolution of the Northeast Japan arc, with the use of exceptionally dense observational data on it, would place an important constraint in understanding subduction-related orogeny in general.

It has been widely accepted that fore-arc deformation is important in the evolution of subduction-related orogens, and therefore a lot of previous studies have focused on the forearc. However, recent studies on the Andean Orogen have revealed that large horizontal shortening in the back-arc region plays an important role in thickening the crust beneath, and hence uplifting isostatically, the Andean Mountains. Deformation styles in back-arc regions are divided into two types: the thick-skinned type and the thin-skinned type. The most significant difference between these two types is in the amount of shortening; thin-skinned type deformation can result in a large amount of horizontal shortening, whereas thick-skinned type deformation causes much less shortening.

In Northeast Japan, these two types of deformation have never been distinguished clearly. We revealed in this study that the study area is divided into two tectonic domains; (1) the thin-skinned domain, including the Uetsu-Northern Fossa Magna basin and its seaward extensions, and (2) the thick-skinned domain, including the Sado Ridge and its northeast extension. We then demonstrate a potential role of Miocene extension structures, which have been inherited to, and have been reactivated to produce contractive deformation in the present-day Northeast Japan arc.

Faults Distribution in the Study Area. We reprocessed multi-channel seismic reflection data and gravity data that were obtained by the Japan National Oil Corporation (JNOC). We also used the single-channel seismic profiles that were obtained by the Advanced Industrial Science and Technology (AIST). Based on the interpretation of these seismic profiles, first we mapped faults and fault-related deformation; the following results were obtained: The faults in the north-northeastward extension of the Sado Ridge are mainly west-dipping, active, reverse faults, some of which clearly show tectonic inversion. The normal and reverse faults in the north-northeastward extension of the Sado Ridge have small amounts of slip during both the rifting stage and the post-inversion stage; the syn-faulting sedimentary unit associated with each fault is about 2 km at maximum. In contrast, the reverse faults in the seaward extension of the Uetsu-Northern Fossa Magna basin have much more amounts of slip. In this rift basin, a thick pile of Miocene sediments have been deformed into fault-bend folds and fault-propagation folds with large horizontal shortening since Pliocene time.

In terms of the nature, amplitude and distribution of the faults, we divided the study area into two tectonic domains: (1) the thin-skinned domain, which includes the Uetsu-Northern Fossa Magna basin and its seaward extensions, and (2) the thick-skinned domain, which includes the Sado Ridge and its northeast extension. In the former, we found such tectonic features that are reasonably explained by the presence of a detachment fault, upon which upper crustal layer can deform independently from the lower part of the lithosphere. That is why we use the term "thin skinned". On the other hand, there is no indication of intra-crustal decoupling in the latter domain, which therefore is termed "thick skinned".

Crustal Extension and Shortening in the Back-Arc Region. Then we examined the amounts crustal extension and shortening along four transects (A~D from north to south). We found tectonic features characteristic to thin-skinned domain; they are fault-bend folds, rotated fault blocks, isolated fault blocks, rollover anticlines, and breakaways.

In transect A extending from Line-09 to the Yokote basin, there exists a deep (~10km) and wide (~40 km) basin filled with a thick pile of Miocene and younger sediments. Post-inversion structures in this transect include fault-bend folds within the rift basin, the Kitayuri Thrust Zone at the rift flank, and the Sen'ya Fault in the east. We reconstructed a pre-inversion cross-section by restoring thrust slip on faults in this transect. The total amount of horizontal shortening along the whole length of this transect was estimated by comparing the restored and present-day cross-sections, and is found to be 14.3 km. The pre-inversion cross-section thus reconstructed shows characteristic features of an asymmetric rift, including a rollover anticline on the east and a breakaway at the west. Then, we reconstructed a pre-rift cross-section by restoring normal slip on the detachment fault. The amount of extension during the Miocene rifting is estimated by comparing the post-rift (pre-inversion) and pre-rift cross-sections, and is found to be as large as 39-43 km.

In transect B along Line-14 off Murakami, we identified a series of (17-33°) basement blocks bounded by normal faults, forming a rotated domino structure. Sediments on top of each block dip progressively steeper with increasing depth. Such strong rotation is best explained by deformation due to normal slip on listric faults, which are likely to sole onto a detachment fault at depth. In this transect, compressional deformation since Pliocene time is hardly identified. Therefore, we made a direct reconstruction of pre-rift geologic section from the present-day geologic section, without considering post-inversion deformation. Then, the amount of extension in Miocene time is estimated to be 10.7-12.5 km.

In transect C from Sado via Mt. Kakuda to the Niitsu anticline, we identified a wide rift basin, within which two isolated fault blocks exist. These isolated blocks show dramatically-strong rotation and hence were interpreted to have separated from the hanging wall during Miocene rifting. We also revealed post-inversion deformation, including 2.4 km of vertical displacement on the East Boundary Fault of Mt. Kakuda, and 5.8 km of shortening across the Niitsu Anticline. Total amount of shortening since Pliocene time and extension in Miocene time are estimated to be 8.2 km and 56 km, respectively.

In transect D from the Itoigawa-Shizuoka Tectonic Line (ISTL) via the Sai-gawa Hills to the Nagano Basin, again we identified a wide rift basin. Post-inversion deformation is concentrated near the both margins of the rift: At its east margin is the Western Boundary Fault Zone of the Nagano Basin, which is interpreted to be a back thrust that was formed on top of the rollover anticline bulldozing rift-fill. At the west margin are the ISTL and two subparallel faults, which are interpreted to merge downdip onto the reactivated detachment fault. By restoring the post-inversion deformation, the shortening since Pliocene time is estimated to be 7.3 km. The pre-inversion cross-section thus reconstructed shows characteristic features of an asymmetric rift, including a rollover anticline on the east and a breakaway on the west. The Miocene extension in this transect is estimated to be 30.9 km.

Conclusions. We demonstrated that, in terms of the nature, amplitude and distribution of faults, the back-arc region of Northeast Japan is divided into two tectonic domain: the thin-skinned domain and the thick-skinned domain. In the thin-skinned domain, a deep (~10 km) and wide (30-60 km) asymmetric rift basins were formed in Early-Middle Miocene time due to slip on low-angle detachment faults, the existence of which is evidenced by rollover anticlines, breakaways, and dramatically rotated isolated fault blocks. These rift basins in the thin-skinned domain were filled with a thick pile of sediments, which have subsequently been deformed into fault-bend folds and fault-propagation folds since Pliocene time. Although these structures kinematically require large slip on the causative faults, significantly large vertical offset does not exist across the Uetsu-Northern Fossa Magna basin since Pliocene time. This suggests that the causative faults of fold-and-thrust structure merge onto a low-angle detachment fault at depth. Thus we interpret that, after ~10 Myr of tectonic quiescence, the low-angle detachment fault beneath the Early to Middle Miocene rift basin has reactivated in Pliocene time; since that time, reverse slip on the reactivated detachment fault have resulted in the fold-and-thrust structure within the rift-fill sediments. By comparing the present and restored cross-sections, a large amount of shortening (10-15km) in Pliocene to Quaternary time and a large amount of extension (30-60 km) in Early to Middle Miocene time were estimated across the Uetsu-Northern Fossa Magna basin. The total amount of the Pliocene-Quaternary shortening is much smaller than that of Miocene extension. Therefore, the present-day crustal necking observed beneath the back-arc region was caused principally by large Miocene extension.

The strains rates since Pliocene time derived from the four transects of this study are 1.2~2.3×10(-8) yr(-1), which are one order of magnitude smaller than geodetic strain rates (on an order of 10(-7) yr(-1)). This suggests that the rapid strain accumulation in the last ~100 years is mostly elastic, and would be released in association with a large decoupling event on the subduction zone; only a fraction (~10%) of the geodetically observed strain would accumulate within the arc on a geologic time scale.

Comparisons both in structure and evolution between the Northeast Japan arc and the Andean orogen suggest that the Northeast Japan arc can be viewed as an incipient analogue of the Andean orogen. We speculate that the present-day westward underthrusting of the continental lithosphere beneath the Andean Cordillera is inherited from the Mesozoic rifting, which could have been asymmetric and associated with crustal-scale detachment. Because the early stage of the Andean evolution has been obscured by subsequent strong convergence, the present study would make a significant contribution to understanding the Andean-type orogeny in general.

本論文は,東北日本背弧域における中新世から現在に至る地殻構造発達史を解明することを試みたものである.東北日本弧の変形は背弧域に集中しており,中新世には伸張変形が,鮮新世以降現在にかけては短縮変形が生じている.東北日本弧は沈み込みに伴う地殻変形様式の両極端を経験していることになり,本研究の結果は沈み込み型造山運動一般を理解する上で重要な拘束を与える.

本論文は5章からなる.第1章はイントロダクションであり,Andes山脈の背弧域における既存研究のレビューから,沈み込みに伴う島弧の造山運動において重要な役割を果たしているのは,背弧域の大きな水平短縮量を伴うthin-skinタイプの地殻変形であることに着目した.それに対して,同じく背弧側に存在するthick-skinタイプの地殻変形は,水平短縮量が小さく島弧の造山運動にほとんど関与しないことを指摘した.本研究の対象地域である東北日本背弧域では,これら2つの地殻変形様式の違いが明確に認識されていない現状が示された.

第2章では,海域および陸域反射法地震探査の解析とその丹念な解釈,さらにはブーゲ重力異常の解析から,thin-skinタイプとthick-skinタイプの変形を初めて明確に区分した.その結果,北部フォッサマグナから羽越褶曲帯とその海域延長部ではthin-skinタイプの地殻変形が,佐渡海嶺とその延長部ではthick-skinタイプの地殻変形が生じていることを明らかにした.これら両地域の地質構造の違いは明瞭であり,thin-skinタイプの変形領域では,非常に厚い(最大8 km以上)中新世以降の堆積層が存在し,その内部に大きな水平短縮を伴う構造が発達するとした.一方,thick-skinタイプの変形領域では,先新第三紀の基盤を切る主に高角・西傾斜の断層が分布し,それらの断層が変形を与えている堆積層の厚さは最大でも2 km程度であり,変形の規模もthin-skinタイプのそれと比較すると非常に小さいことが示された.

第3章は,海域および陸域の反射法地震探査の解析と解釈,地表地質データ,および坑井データを紹介し,それらに基づいて作成されたthin-skinタイプの変形領域を横切る4本の地質構造断面が示された.さらにこれらの断面から,balanced cross section法を用いて中新世以降の地殻水平伸張量と鮮新世以降の地殻水平短縮量を見積もった.それによると,中新世における羽越褶曲帯およびその海域延長部の地殻伸張は東に傾く低角なdetachment断層によって生じ,その結果極めて深く非対称な断面形の堆積盆が形成されたことが明らかになった.これらの堆積盆を充填する中新世以降の厚い堆積物は,大きな水平短縮量を伴う褶曲変形を受けている.この短縮は,中新世に生じた低角なdetachment断層が鮮新世以降に再活動することにより生じたものと推定された.thin-skinタイプの変形領域が担う中新世における伸張量および鮮新世以降の短縮量は,それぞれ30-60 kmおよび10-15 km程度と見積もられた.羽越褶曲帯の全長にわたって,このように大きな水平伸張と水平短縮が生じたことを定量的に明らかにしたことは,本論文の特筆すべき成果である.

第4章は,屈折法地震探査によって明らかにされた東北日本弧の地殻構造と,本研究で得られた地殻伸張・短縮量との比較が行なわれている.上述のように東北日本弧背弧域における中新世の地殻水平伸張量は鮮新世以降の地殻水平短縮量を大きく上回っており,それが東北日本背弧域における地殻のneckingを生じさせていると推定された.また,中新世における大きな水平伸張と鮮新世以降の水平短縮とがほぼ同一の場で生じていることは,下部地殻から最上部マントルにおける物性の不均質構造が東北日本弧の地殻変形を支配する重要な要因であったことを示唆するとした.以上の結果は,沈み込み型造山運動一般を理解する上で,重要な拘束をあたえる可能性がある.

第5章の結論では,本研究の成果を中心に東北日本背弧域における中新世から現在に至る地殻構造発達史を定量的かつ総合的にまとめた結果を述べている.このように本論文は東北日本弧を例として沈み込み型造山運動の理解に重要な貢献をしたと評価できる.

なお,本論文における浅層反射法地震探査データの取得に関しては,池田安隆,越後智雄,戸田 茂,木村治夫,加藤 一,石山達也,楮原京子,森下信人,奥田里奈,渡邉太樹,小田 晋との共同研究であるが,論文提出者である岡田真介が,データ取得からデータ解析およびその解釈の全般にわたって中心的に携わっており,寄与が十分であると判断される.

したがって,博士(理学)の学位を授与できると認める.