各個地區自然災害的發生與破壞，常常會造成重大損失。其中以人員傷亡、環境變遷與地形劇變等等為災害常見的情況。而台灣有著極端的氣候變化，其中最常見的天然災害為颱風，就每年七月至十月而言常有四個颱風通過台灣。因地表變形和上述災害有正向相關，故可藉由地形變化偵測來分析颱風造成災害的分布情況。因此，本研究是以2009年八月七日莫拉克颱風造成巨大的損害為例，運用JAXA的ALOS PALSAR影像來評估颱風造成的地表變形。運用方法即為多軌跡的差分合成孔徑干涉雷達(DInSAR)來建立不同時間影像的干涉條紋，再利用干涉條紋估算地表變形量。步驟分別為建立干涉條紋、消除地球曲率、影像遮罩處理、相位解纏等，最後得到數值地形模型(DEM)和地表變形。在多軌跡的方法中，使用莫拉克颱風前後的地形所產生的相位差估算地表變形，藉此得到颱風前後的變化。而在結果中可以清楚看到，颱風造成特定區域變形較大，可能有土石流或滑動的跡象，也許不太適合居住。最後，期望這些資訊能夠系統化的提供給政府做都市規劃和災害防治等評估，以期得到良好成效。;Natural disasters cause damages and occur in many countries. When hit by natural disasters, the countries would suffer from losing a large number of residents, environmental damage and surface deformation. Taiwan has a humid tropical climate and it frequently pass through by 4 typhoons between July and October each year. One of the biggest typhoons was Typhoon Morakot, which struck Taiwan from 7–9 August in 2009. ALOS PALSAR time series images by JAXA were employed in this study to examine land deformation that might be occurred after Morakot Typhoon in the study area. Multi-pass approach DInSAR was applied and the interferogram of each pair were generated, that each of them gave the information about surface changes of two images took from two different times. The processing steps including the interferogram generation for each pair, interferogram flattening and filtering, phase-unwrapping, DEM and displacement map generation. In multi-pass approach, topography-related phase can be estimated using DEM generated from an independent interferometric pair, we prefer to estimate it from the image pair acquired before Morakot happened, and finally, the displacement map and its exact value were obtained. The purpose of this study was to investigate the deformation that affected by Morakot Typhoon over the study area (in the middle of Taiwan). The locations with higher surface deformation are obviously are unsafe for residential purpose. In addition, the result of this study hopefully can provide meaningful information for the government to do the urban planning and disaster management in the future.