觀測大地（Ｉ）：土地與海岸Observing the Earth (I): Land and Coastlines

摘要 Abstract：

Surface deformation during interseismic periods is the key for understanding fault behaviors and seismic hazard potential. The interseismic aseismic slip of active faults would not only continually damages man-made structures but also influences the estimation of asperity. Southwestern Taiwan is located on the orogenic front under the arc-continent collision between the Eurasian Plate and the Philippines Sea Plate, and many active faults in this area pose a threat to the infrastructure and the safety of local residents. Various geodetic observations have been applied to capture the surface deformation induced by these active structures. Recently, Synthetic Aperture Radar (SAR) images have become an increasingly influential way for accessing high spatial resolution deformation patterns in a region. However, there is a current gap in research utilizing multiple sets of L-band SAR images, which are capable of penetrating vegetation, to investigate active structures in Southwestern Taiwan.

In this presentation, I will show the processed results of L-band ALOS and ALOS-2 satellite data. I have integrated the deformation time series from SAR with ground deformation data from continuous GNSS stations in the study region, constructing a ground deformation dataset for 2007-2011 and 2015-2022. According to the result, the spatial distribution of active structures in the study region can be observed in detail, allowing the further insights into the hazard potential of Southwestern Taiwan. 

關鍵字 Keywords：Interseismic Deformation, InSAR, GNSS

摘要 Abstract：

Understanding the distribution of co-seismic slip patterns becomes a key observation to understand the fault geometry and tectonic implications. On 17 and 18 September 2022, a series of earthquakes with Mw6.4 and Mw6.8 magnitude occurred in Guanshan and Chihshang area in central Longitudinal Valley, and have distributed surface rupture along the Yuli fault and Longitudinal Valley fault. To detect the high spatial resolution distribution of surface displacement, optical satellite images are needed because of the large and wide coverage surface deformation area. However, image subpixel correlation measurement, the most common surface displacement detection method by using optical image, became a challenge in the 2022 Taitung earthquake because of the limitation of fast vegetation change and capture recurrence time. Thus, a different method is needed in this event. In this study, we generate a detailed surface displacement field around the earthquake affected area from orthoimages captured by the high-resolution satellite, Pléiades, both before and after the earthquake. Utilizing the K-means image classification method, a machine learning (ML) algorithm, we classify the ground features (ex: paddy and building boundaries) in the orthoimages and employ the boundary locations of these features to measure surface displacement.

關鍵字 Keywords：surface displacement, image classification, 2022 Taitung earthquake sequence

摘要 Abstract：

臺灣政府近年積極推動能源轉型計劃，促使太陽光電裝置容量從2015年的884千瓩快速增加至2021年的5,903千瓩，並預計在2025年讓累計裝置容量能達到2,000萬瓩的規模；其中，屋頂型的目標為800萬瓩，而地面型則為1,200萬瓩。然而，大量設置的地面型太陽光電場改變了原有的土地利用模式，也間接引發社會對於能源轉型與國土保育間的兩難爭議，並呼籲政府應訂定景觀及生態環境審定原則，避免不當的場址開發與衝擊。

人類活動對於土地利用型態的改變，是造成地表溫度與微氣候變化的主因，更會直接影響既有的生態、景觀及環境品質。本研究希望探究太陽光電場及其週邊土地利用的搭配，對於地表溫度的影響；首先，利用地理人工智慧 (GeoAI) 的方法，以深度學習的模式從高解析影像中探勘出太陽能板的分佈範圍，並以該分佈範圍為基礎搭配國土利用調查資料，以建立一標準化的土地利用型態 (land use pattern)。後續，則是藉由比較：標準化土地利用型態與Landsat8所反演的地表溫度 (Land Surface Temperature, LST)，探勘不同光電場址與土地利用型態的搭配，對於週邊地表溫度的影響。

整體而言，太陽光電的設置規模與地表溫度上升的幅度成正比，且可能與週邊土地產生溫度升高的效益，進而形成一個小型的熱島環境。此外，在同一土地利用型態上，規模越大的太陽光電設施，其所引起的溫升效應越是明顯。然而，受限於衛星資料的時空解析度與相關限制，若要進一步驗證太陽光電對於週邊微氣候所產生的影響，仍需利用感測器進行現地量測，始能進一步釐清。

關鍵字 Keywords：太陽光電、地表溫度、地理人工智慧、土地利用型態

摘要 Abstract：

Coastal land subsidence has been identified as one of the major global problems, which may induce various coastal hazards along with socio-economic losses. In Taiwan, because of rapid development and over-pumping of groundwater, many areas along the west coast are also affected by land subsidence. However, the spatiotemporal patterns of the coastal subsidence are less understood. Therefore, this study aims to reconstruct a quantitative history of coastal land subsidence in western Taiwan over the past decades. To obtain ground displacement time series with fine spatial resolution, this study applies Small Baseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR) technique based on multi-mission datasets. Further, the performance of tropospheric corrections and different phase deramping manners are also assessed. 

Preliminary results show that Changhua County and Yunlin County were the two major subsidence zones. Besides, Chiayi County, northern Tainan City, and Pingtung County underwent significant coastal subsidence as well. During the period 1995-1999, a maximum LOS (line-of-sight) deformation rate of 6.0 cm/yr was observed in Dacheng Township, Changhua County, and 5.3 cm/yr in the northern part of Sixth Naphtha Cracker Complex (which was constructed on reclaimed land), Yunlin County. Between 2006 and 2008, the subsidence center in Changhua County moved inland while the LOS deformation rate near the coast reduced to about 2.7 cm/yr. The subsidence center in the Sixth Naphtha Cracker Complex moved southward, having a maximum rate of 7.5 cm/yr. As for Pingtung and the northern part of Tainan, deformation velocities decrease during this period.

關鍵字 Keywords：coastal land subsidence, SBAS-InSAR, ERS-1/2, Envisat ASAR, Sentinel-1, MintPy

摘要 Abstract：

Surface fault creep is an important fault behavior associated with seismic hazard assessments. Surface creep releases energy through aseismic slip, but sometimes it is capable of generating earthquakes if the fault is not entirely creeping. The Longitudinal Valley fault (LVF) in eastern Taiwan is one of famous creeping faults in the world. The fault has been observed surface creep with oblique motion along its southern portion from Yuli in the north to Lichi in the south based on varied geodetic data. As the fault is located within the active collision boundary between the Philippine Sea and Eurasian plates, the reported surface creep rate is ~10-30 mm/yr at individual sites. Surface velocities derived from InSAR time series using Envisat and ALOS-1 images also show significant surface creep across the southern Longitudinal Valley fault. Although the geodetic data show widely distributed surface creep along the southern portion of the Longitudinal Valley fault, in-situ observations are inconsistent with the imaging geodetic observations as discernible surface creep is localized at few specific localities. In addition, fault creep derived from InSAR time series show different creeping pattern with time. Therefore, in this study, a spatial–temporal analysis of the surface creep in southern of LVF was performed with the help of the Small Baseline Subset (SBAS) InSAR. We applied the ISCE software for the pre-process, co-registration and DInSAR of 30 ALOS-2 SAR images, which are in the ascending orbit and stripmap (SM3) mode, and apply the split-spectrum method for ionospheric correction. The results show an obvious line-of-sight (LOS) rate boundary in the southern valley, and the LOS velocity difference is about 10 to 30 mm/yr. A clear change in LOS velocity, about 8 mm/yr, is observed across the northern segment of the Luyeh fault. In contrast, the southern segment of the Luyeh fault does not exhibit a noticeable velocity difference. Instead, the adjacent Pinglang fault to the west shows a gradual velocity change of approximately 5 to 8 mm/yr.

關鍵字 Keywords：interseismic deformation、active tectonics、Chihshang Fault、surface deformation