Uplift of Tainan Tableland (SW Taiwan) revealed by SAR interferometry Bénédicte Fruneau, Erwan Pathier, Daniel Raymond, Benoit Deffontaines, C.T. Lee, H.T. Wang, Jacques Angelier, Jean-Paul Rudant, Chung-Pai Chang

To cite this version:

Bénédicte Fruneau, Erwan Pathier, Daniel Raymond, Benoit Deffontaines, C.T. Lee, et al.. Uplift of Tainan Tableland (SW Taiwan) revealed by SAR interferometry. Geophysical Research Letters, American Geophysical Union, 2001, 28 (16), pp.3071-3074. ￿10.1029/2000GL012437￿. ￿hal-02882387￿

HAL Id: hal-02882387 https://hal.archives-ouvertes.fr/hal-02882387 Submitted on 8 Feb 2021

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. GEOPHYSICAL RESEARCH LETTERS, VOL. 28, NO. 16, PAGES 3071-3074, AUGUST 15,2001

Uplift of Tainan Tableland (SW Taiwan) revealed by SAR interferometry B. Fruneau, E. Pathier,e D. Raymonde B Deffontainese C T Lee3 H. T. Wang,4 J. Angelier,e J.P. Rudant,1 and C. P. Change

Abstract. Interferometric processingof five SAR-ERS im- resultingfrom plate collisionsince about 5 Myr [Ho, 1986]. ages reveals uplift of the Tainan Tableland (SW of Tai- The Tainancity (1.2 millioninhabitants, 4•h city of Tai- wan) during the period 1996-1998.The maximummeasured wan) is located in the coastalplain of SW Taiwan, between ground motion for these two years is 2.8 cm along the radar the continentalshelf of South China to the West (the Tai- line of sight towards the satellite, indicating for the displace- wan Strait) and the collisionbelt to the East. The N070øE ment vector a vertical component of 3.2 cm, and a horizontal trending Tainan Basin extends onshore in the coastal plain component of 1.6 cm towards the WSW considering addi- where it is filled with pre-Pliocene sediments of the pas- tional information from (]PS data. The reconstructed dis- sive margin and overlain with siliciclastic sediments of the placement field is consistentwith the geologicalinterpreta- Pliocene and Quaternary, linked to the growth and erosion tion of the Tainan Tableland as an actively growing anticline of the Taiwan belt. The recent geological evolution of SW connected to the Taiwan fold-and-thrust belt. This implies Taiwan was controlled by the position of the advancing oro- that the deformation front is located farther west than usu- genic front. East of Tainan, the Meilin thrust was regarded ally assumed in the Tainan area. The large Tainan city is as the deformation front of the SW Taiwan fold-and-thrust thus located in an active deformation zone. Seismic hazard belt [Leeet al., 1993, 1995; Gonget al., 1995](fig.lb). assessment is however dif•cult because the mechanisms and In the Tainan area, the coastal plain, generally fiat low- kinematics are not known in detail. land, exposesHolocene coastal deposits. At the Tainan city sandy sediments cover a structural high: the major part 1. Introduction of the city is built on a 30 m high elongated tableland, 12.5 km long and 4 km width with a N020øE trending axis The Tainan area belongs to the coastal plain of south- (fig. 2a). The Tainan Tableland showsa westwardconvex western Taiwan. Until recently, this domain was considered shape and an east-west asymmetry, its western part dipping as a relatively stable foreland in the front of the Taiwan gently westward, while the eastern one is being steeper. The fold-and-thrust belt well documented East of the Tainan central area shows a 2 km wide fiat top. Based on analyses Tableland [Gong et al., 1995]. Some authors have how- of aerial photographs,Sun [1964]described the easternedge ever proposed that the elongated tableland on which the of the Tainan Tableland as a N020øE trending fault with the Tainan city was built results from recent folding and uplift downthrown block to the East corresponding to the Tawan [Deffontaineset al., 1997; Lacombeet al., 1999]. If this is Lowland. Later, Hsieh [1972]interpreted the Tainan Table- correct, the Tainan Tableland belongsto the active deforma- land as an anticline abovea diapir. Lee et al. [1993, 1995] tion front, rather than to the stable foreland. As Synthetic considered this area as an uplifted block bounded by two Aperture Radar interferometry(InSAR) allowedsuccessful normal faults, corresponding to an extensional feature of reconstruction of earthquake deformation, volcano inflation the offshore Tainan basin. In contrast, Deffontaines et al. or deflation and dyke intrusion, landsliding or subsidence [1997] and Lacombeet al. [1999]proposed to considerthis [Massonnetand Feigl, 1998;Amelung et al., 1999],we used tableland as the surface expressionof a ramp anticline above it to determine the present-day ground deformation in the a west-vergingthrust, like a pop-up system (fig. 3b). Ac- Tainan area. cording to this interpretation, this anticline represents part of the deformation front of the Taiwan belt, farther west 2. Geological setting than the Meilin thrust. This is consistent with the offshore In the Taiwan collision zone, fast convergenceoccurs be- location of the deformation front SW Tainan suggested by tween the Philippine Sea plate to the East and Eurasian Liu et al. [1997](fig. lb). plate to the West [8.1 cm/yr, [Yu et al., 1997]] (fig. la). Onshore, most of the Taiwan belt is an accretionary prism 3. SAR interferometry

l Institut Franciliendes G60sciences,Universit6 de Marne-la- 3.1. Data acquisition and processing Vall•e, France. Because of its large urban surface, Tainan area does not 2Laboratoirede Tectonique,Universit• Pierre et Marie Curie, suffer from INSAR coherenceloss, usually observedin tropi- Paris, France. 3Institute of applied Geology, Central University, Chungli, cal countrieslike Taiwan [Fruneauand Sarti, 2000]. We used Taiwan. the 2-pass InSAR approach which combines a pair of images 4Remote SensingDepartment, Central University, Chungli, of the samearea acquiredat differenttimes (by the satellites Taiwan ERS-1 and ERS-2 operatingwith a 56mm wavelength)with topographicinformation [Massonnet and Feigl, 1998]. The Copyright2001 by theAmerican Geophysical Union. interferograms were processed by the DIAPASON¸ soft- Papernumber 2000GL012437. ware of the FrenchCNES (National Centre for Spatial Stud- 0094-8276/01/2000GL012437505.00 ies). The topographiccontribution was removed using a

3071 3072 FRUNEAU ET AL.: UPLIFT OF TAINAN TABLELAND (SW TAIWAN)

80 m x 80 m Taiwanesedigital elevationmodel (DEM) with a 5 m average height accuracy retrieved from 1:10000 topo- graphic maps. We selectedfive images (table 1), suitable .?• -• ;--.4:'•'•f'•?•'•x'•••'•••.r•.. . ••l for ten interferometric combinations. The time interval of 2 years (1996-1998) is sufficientto detect a deformation sig- ...... :x:-:::----..-.-..:.-.-: .. ".:-•?:• '::::::?' '.:<:•:-- -,.'-:..: .... .:...g::•- -'?<•" "'-:•:: .... _•U ...... '•: ...... ' 'C•:: ---:::•:'::•'• <'::" ::,%.' :•-':'::-.-'...... nal and short enough to maintain the phase coherenceof the groundsurface The valuesof the altitude of ambiguity (ha), i.e. the DEM error required to generate one interferometric

fringe, range from 57 m to 14094 m.0

3.2. Results The six long-term interferograms(A-D, A-E, B-D, B- E, C-D and C-E) reveal the same significant fringe (i.e. UrbanArea' /• ½ one colour cycle) pattern, illustrated in figure 2b and 2c, ofTainan kO • whereas no significant signal is detected in the four short- term interferograms(A-B, A-C, B-C and D-E). The pattern //- Xanan 2,0, of the fringe is especially highlighted by the red band which roughly follows the outline of the Tainan Tableland, except in the South West where the phasecoherence is lost (fig. 2). Thrust •.•x,'I[ Note that the noisy aspect of interferogram B-D is caused by atmospheric artefacts on the image D which affect ev- ery interferogram using this image. As shown by narrower . • • /x½• 22.8 colour bands, the gradient of phase difference is steeper at the periphery than on the Tainan Tableland. • g SouthChina A fringe corresponds to a phase variation of 2•r radians Synchne• Sea ••b•Xxl10 km I that depicts displacements at sub-wavelength accuracy in the radar line of sight, unmodelled topography and possi- axis' 12•.1• •• 120.3I •x x • I ble artefacts of atmospheric origin. Topographic effect can be regarded negligible because interferograms having very different ha values show the same fringe pattern (fig. 2). Possible atmospheric effects deserve attention ; they may Figure 1. a) Taiwan geodynamicalsetting. b) Southwestern taiwan geological structural setting. Inferred deformation front appear when the atmospheric conditions during the two im- position accordingto Lee et al. [1995] (white triangles) and ac- age acquisitionsdiffer enough to modify the propagation de- cordingto Deffontaineset al. [1997](black triangles). Box shows lay of electromagneticwaves [Zebker et al., 1997]. Similar- the extent of figure 2. ity between fringe pattern and relief, related to atmospheric

Figure 2. a)Topographyand shadedrelief of the Tainan area. b) InterferogramA-E (701 days, ha=-1141m). c) InterfergramB-D (665 days, ha=57m). Both independentinterferograms AE and BD having different ambiguity altitude valuesshow one fringe which well agrees with the outline of the tableland. The fringe indicates an uplift of the Tainan Tableland in the 2 years time interval. The noisy aspect of interferogram BD is due to atmospheric artbracts on image D. Hydrography and coastline are in black thick line, the white circle with arrow represents 1004 GPS station and its displacement. Dashed line represents the outline of the urban area of Tainan. FRUNEAUET AL.: UPLIFT OF TAINANTABLELAND (SW TAIWAN) 3073 effects, may appear where topography contrasts are strong, suchas for the Etna volcano[Beauducel et al., 2000]. The very low topography of the Tainan Tableland cannot pro- duce such an atmospheric effect. Using the pair-wise logic developedby Massonnetand Feigl [1998], we comparedinde- pendent long-term interferograms having no common images and obtainedthe samefringe pattern (fig. 2). The probabil- ity that a same change in atmospheric state occurs exactly between the two image acquisitions of the independent in- terferograms is very low; moreover, the fringe appeared only in the long-term interferograms, not in the short-term ones. We conclude that the signal representsground deformation. An interferogram provides only relative value of displace- ment along the radar line of sight. A fringe correspondsto a range change of 2.8 cm along this line between ground and satellite. Assuming a null range change for the phase- coherentzone NW of the Tainan Tableland (dark blue, up- per left-hand corner of figure 2b), we found out that the Figure 3. Three schematicstructural models of the Tainan Tableland. a) Blind thrust model. b) Fish-tail model. c) Pop-up Tainan Tableland was getting closer to the satellite between model. 1996 and 1998 passes,with a maximum range change of 2.8 cm reachedat the middle of the tableland (dark blue area in the middle of figure 2b and 2c). It is reasonableto ex- (the generalpattern of GPS displacementsindicates veloci- trapolate this deformation to the southern part of the relief ties that rapidly decreasefrom SE to NW). where coherence is lower. In comparison, preliminary assessmentof the vertical mo- tion of the I004 station gives a 1.26 cm/yr uplift (S. B. 3.3. Comparison with other sources of Yu, personalcommunication, 2000). Chen and Liu [2000] information showthat their data, basedon radiocarbon agesand relative sea-levelcurves, are consistentwith an 0.5 cm/yr uplift of In order to estimate the vertical displacement of the the Tainan Tableland since 10,000 years. Furthermore, they Tainan Tableland, we consider additionnal information from showa 0.1 cm/yr subsidencein the Tawan Lowland, a small GPS data. One station of the Taiwan GPS network (I004, depressioneast of the Tainan Tableland (fig. 2). This phe- trianglein figure2) is locatedon the Tainan Tableland,close nomenonis presenton all our long term interferograms(less to our largest range change zone. At this station, the East than one fringe with a 3 km elongated concentric pattern lo- and North velocity componentsare-0.76 and-0.27 cm/yr, cated on the Tawan Lowland,see fig. 2). We alsocollected respectively[Yu et al., 1997]. Thesehorizontal components field data in and around Tainan city, paying particular at- of displacement are given relative to the Penghu Islands in tention to cracks in buildings. Because most buildings were the stableforeland of the Taiwan Strait (fig. 1). Consider- built later than 1950 and the deformation is spreaded out ing both the range change of 2.8 cm revealed by interferome- throughout Tainan city, we found few evidencesfor present- try and the unit vector from groundto satellite (east=0.28, day deformation(west of Tainan city, North of the Tai-Pin north=-0.03, up=0.96), we computedthe vertical compo- bridge on the Yen Suah River): a concretefloor in front nent of displacement near the GPS station, that is 3.2 cm of of a warehouse affected by tension fractures and a narrow two years uplift from 1996 to 1998. This implies the reason- concrete bridge which is broken are consistent with present- able assumption that insignificant motion occurs between day tectonic compressiongenerating uplift. All these inde- the coastal plain NW of Tainan and the Penghu Islands pendent results concur to indicate present-day uplift of the Tainan Tableland.

Table 1. Left: SAR ERS Images (ERS-1 for A, ERS-2 for 4. Discussion others) with orbit number and date of acquisition (local time is 2:31 AM, track 232, frame 3141). Right: Interferograms with A crucial problem deals with mechanism and kinematics altitude of ambiguity (ha) and interval of times. of deformation revealed by our Insar analysis. Chen and Liu [2000],from Hsieh'shypothesis [1972], advocate for a diapir Images Interferogramms origin of the Tainan Tableland. Diapiric phenomena cer- Orbite Date ha(m) time(day) tainly occuroffshore [Liu et al., 1997]and onshorein south- (A) 23767 31-jan-96 A-B -89 1 western Taiwan where thick late Cenozoic mudstone forma- (B) 4094 01-feb-96 A-C -1241 71 tions are present. However, these phenomena do not occur (C) 5096 11-apr-96 A-D 160 666 alone and they are closely related to compressive tectonics. A-E -1141 701 The existence of present-day compression is evidenced by B-C 96 70 GPS-measured shortening between the Tainan station and B-D 57 695 the PenghuIsland of the TaiwanStrait [ Yu and Chen,1998]. B-E 97 7OO The N020øE trending axis of the tableland, parallel to the C-D 142 595 major structures of the collision belt, also suggestsa struc- (D) 13613 27-nov-97 C-E -14097 630 tural control. We conclude that although diapirism cannot (E) 14114 01-jan-98 D-E -140 35 be excluded in the Tainan Tableland, compressive tectonics 3074 FRUNEAU ET AL.- UPLIFT OF TAINAN TABLELAND (SW TAIWAN) and horizontal shortening certainly control its development. Council of Taiwan (NSC), by the PNRN (French National Pro- Therefore, our results suggest that the Tainan Tableland gram on Natural Hazards), by the CNES and by the European belongs to the active fold-and-thrust belt, as proposed by Spatial Agency. We thank Falk Amelung and an anonymous ref- eree for their constructive comments. Deffontaineset al. [1997]and Liu et al. [1997]. The uplift of the Tainan Tableland revealed by our InSar References analysis cannot be of co-seismic origin because no earth- quake with a local magnitude larger than 4 has occurred Amelung, F., D. L. Galloway, J. W. Bell, H. A. Zebker, and R. J. from 1996 to 1998 in the Tainan area (accordingto data Laczniak, Sensingthe ups and downs of Las Vegas: InSAR re- of the Central Weather Bureau of Taiwan). Note that be- veals structural control of land subsidenceand aquifer-system deformation, Geology, 27, ,6 483-486, 1999. cause of the concentric pattern of interferometric fringes, Beauducel, F., P. Briole, and J. L. Froget, Volcano wide fringesin the deformation must be local in origin, and cannot be ERS synthetic aperture radar interferograms of Etna: Defor- attributed to distant earthquakes. Coming to the inter- mation or tropospheric effect?, J. Geophys. Res., 105, 16,391- pretation in terms of interseismic deformation, the uplift 16,402, 2000. could be explained by accumulation of elastic strain. Ac- Chen, Y. G., T. K. Liu, Holocene uplift and subsidencealong cording to the well-documented structural style of the SW- an active tectonic margin southwestern Taiwan, Quaternary Science Reviews, 19, 923-930, 2000. Taiwan Foothills [$uppe,1981], a reasonableassumption is Deffontaines, B., O. Lacombe, J. Angelier, H. T. Chu, F. that the Tainan Tableland involves a d•collement-related Mouthereau, C. T. Lee, J. Deramond, J. F. Lee, M. S. Yu, and ramps style. Accordingly, we discuss below the three ma- P.M. Liew, Quaternary transfer faulting in Taiwan Foothills: jor styles of deformation front which could account for the Evidence from a multisource approach, Tectonophysics,27•, Holoceneuplift of the Tainan Tableland(fig. 3). The sim- 61-82, 1997. Fruneau, B., F. Sarti, Detection of ground subsidencein the city plest model is a single thrust dipping to the East, the second of Paris using radar interferometry: isolation of deformation one a fish-tail, and the third one a pop-up structure. The from atmospheric artifacts using correlation, Geophys. Res. distribution of the InSAR deformation field shows that sur- Lett., 27, 3981-3984, 2000. face deformation remains continuous and must be local in Gong, S. Y., T. Y. Lee, J. C. Wu, S. W. Wang, and K. M. Yang, origin. This suggestseither the upper parts of the faults are Possible links between Plio-Pleistocene reef development and locked or strain accumulation is located at the tip of creep- thrust migration in the Southwestern Taiwan, Proc. $rd Sino- French Symposium, Taipei, March 1995, ACT Syrup., 113- ing blind faults. The inferredsingle blind thrust (fig. 3a) is 119, 1995. not consistent with the asymmetry of the Tainan Tableland Ho, C. S., A synthesis of the geologic evolution of Taiwan, (with steeperslope to the East). The latter suggestsrather a Tectonophysics,125, 1-16. 1986. west dipping backthrust upper part of a fish tail structure in Hsieh, S. H., Subsurfacegeology and gravity anomalies of Tainan agreementwith the regionalfold-and-thrust structure (fig. and Chungchou structures of the coastal plain of southwestern 3b). But, the stronger fringe gradient at the periphery of Taiwan, Petrol. Geology of Taiwan, 10, 323-338, 1972. Lacombe, O., F. Mouthereau, B. Deffontaines, J. Angelier, H. the tableland relative to the central area is better explained T. Chu, C. T. Lee, Geometry and quaternary kinematics of by a pop-up structure (fig. 3c). In this case,the asymmetry fold-and-thrust units of southwestern Taiwan, Tectonics, 18, is attributed to the different dip angles of the thrust and 6, 1198-1223, 1999. backthrust. Lee, T.Y., C. H. Tang, J. S. Ting, and Y. Y. Hsu, Sequence Considering our assumption that the uplift is only due to stratigraphy of the Tainan Basin, offshore SW Taiwan, Petrol. elastic strain accumulation, it is however noteworthy how Geology of Taiwan, 28, 119-158, 1993. Lee, T. Y., Y. Y. Hsu, and C. H. Tang, Structural geometry well the fringe pattern agree with the outline of the table- of the deformation front between 22øN and 23øN and migra- land. This suggeststhat the defc;rmationcould partly be tion of the Penghu canyon, offshore southwestern Taiwan arc- a non-recoverable inelastic deformation absorbed by fold- continent collision Zone, Proc. $rd Sino-French Symposium, ing. Thus, the permanent cumulated uplift of the tableland Taipei, March 1995, ACT Syrup., 219-227. 1995. would fit the pattern of the interseismic deformation field. Liu, C. S., I. L. Huang, and L. S. Teng, Structural features off That is why we suggestthat part of the present-daydefor- southwesternTaiwan, Marine Geology, 137, 305-309, 1997. Massonnet, D., and K. Feigl, Radar Interferometry and its appli- mation of the Tainan Tableland is inelastic and irreversibly cation to changes in the earth's surface, Rev. Geophys., 36, 4, absorbed by an actively growing anticline. 441-500, 1998. From the seismichazard point of view (a major concern Sun, S.C., PhotogeologicStudy of the Tainan-Kaohsiung Coastal in Taiwan), stressand strain build up duringthe interseismic Plain Area, Taiwan, Petrol. Geology of Taiwan, $, 39-51, 1964. period of seismiccycle, and the elastic part of this deforma- Suppe, J., Mechanics of mountain-building and metamorphism in Taiwan,Mere. Geol. Soc. China, •, 67-89, 1981. tion is released by earthquake. The size of earthquake is Yu, S. B., H. Y. Chen, and L. C. Kuo, Velocity field of GPS depending on the amount of elastic strain stored. To this stations in the Taiwan area, Tectonophysics,œ7•, 41-59, 1997. respect, any definite conclusion from this work in terms of ¾u, S. B.,and H. Y. Chen, Strain accumulation in south-western seismic hazard would be premature. This is an important Taiwan, TAO, 9, 31-50, 1998. issue because 1.2 million inhabitants live in Tainan city. To Zebker, H. A., P. A. Rosen, and S. Hensley, Atmospheric effects in interferometric synthetic aperture radar surface deforma- better investigate this essential aspect, further works, in- tion and topographic maps, J. Geophys. Res., 102, 7547-7563, cluding more InSar analyses, should certainly be carried out 1997. in the near future. B. Fruneau, Institut Francilien des Gdosciences,Universit• de Marne-la Vall•e, 5 bd. Descartes, 77454 Marne-la-Vall•e Cedex 2 Acknowledgments. Specialthanks to H. Chakroune, S. France. (e-mail: [email protected]) B. ¾u, F. C. Li, Hsu H. H., and J. F. Brouillet for their help. This work was supported by the Taiwan-France cooperation program (ReceivedOctober 3, 2000; revised March 2, 2001; of the French Institute in Taipei (IFT), by the National Science acceptedMarch 29, 2001.)