結合高精度無人載具測量及差分干涉合成孔徑雷達與NDSI分析於土砂量體的評估方法

A DEBRIS ESTIMATION METHOD COMBINING HIGH ACCURACY UAV, DINSAR AND NDSI ANALYSIS

王國隆(Kuo-Lung Wang) ； 林俊廷(Jun-Tin Lin) ； 張峻瑋(Jun-Wei Zhang) ； 林妍琇(Yen-Hsiu Lin) ； 洪志祥(Jr-Shiang Hung) ； 洪政義(Jheng-Yi Hung) ； 邱啓芳(Ci-Fang Ciou) ； 傅桂霖(Kuei-Lin Fu)

《農業工程學報》 65卷4期 (2019/12) Pp. 26-39

https://doi.org/10.29974/JTAE.201912_65(4).0003

摘要

台灣的高屏溪2009年的莫拉克颱風發生嚴重的坡地災害，流域內產生大量土砂堆積，堆積的土砂運移，造成下游河道變化，並引發二次災害，估計這些土砂量體及影響變成重要的課題，衛載合成孔徑雷達影像（Synthetic Aperture Radar，SAR）為近年來高度發展的遙測技術，具有不受天候甘擾、大面積探測、固定周期等特性，現已廣泛運用於坡地的災害調查，SAR影像的紀錄包含強度、極化、相位等資訊，其中強度與極化主要運用於地形判釋及分類，相位資訊則可經差分干涉後得到厘米級的變形量資訊，本研究提出結合強度判釋技術及差分干涉技術進行莫拉克災害崩塌地的殘餘土砂量評估，運用於災後無法立即獲得災害範圍及難以追蹤土砂流失量等問題，使用NDSI分析於研究區內可成功自動化擷取77%崩塌範圍，透過DInSAR分析可以將崩塌地轉化的土砂量估計出來，此外於莫拉克颱風災後已經接近10年，災時所產生的崩塌土方的移動仍影響目前的工程整治，本研究選定高雄市桃源區復興地區為目標，進行短時間多幅雷達影像分析土方變遷趨勢，輔以無人載具進行四期正射影像及數值地表模型製作，驗證於2017年間進行之清疏治理工程成效。

關鍵字

無人載具 ； 合成孔徑雷達 ； 土砂量評估

並列摘要

Gaoping Creek in Taiwan suffered from severe landslide disasters in the typhoon Morakot in 2009. A large amount of debris stays in the basin, and the debris transportation causing changes in the downstream river channel and secondary disasters. It is estimated that these debris volumes and influences become essential. The subject of the Synthetic Aperture Radar (SAR) is a highly developed remote sensing technology in recent years. It has the characteristics of being unaffected by weather, large-area detection, and fixed revisit period. It has been widely used in landslide disaster investigation. The records of SAR images include information such as intensity, polarization, and phase. The intensity and polarization are mainly used for terrain interpretation and classification. The phase information can be obtained by DInSAR to centimeter-level deformation. The radar intensity identification and DInSAR are used to evaluate the residual debris content of the Morakot disaster landslide site, and it is impossible to obtain the disaster locations immediately after the disaster, and it is difficult to track the amount of debris loss. The NDSI analysis can successfully automate capture 77% landslides in the study area, and the DInSAR analysis can estimate the amount of soil sand converted from landslides. In addition, it has been close to 10 years after the Morakot typhoon disaster. The movement of the landslide debris still affects the current engineering remediation. Fushing area in Taoyaun District, Kaohsiung is selected for this study. Time series DInSAR analysis in one year and a half supplemented by four periods of unmanned vehicle's orthophoto and digital surface model to verify the effectiveness of the dredging project of governance in 2017.

並列關鍵字

UAV ； SAR ； Debris estimation