以高解析度衛星影像快速調查崩塌地

Rapid Investigation for Landslides by Using High Resolution Satellite Image

周晏勤(Y.C. Chou) ； 蔡尹萱(Y.H. Tsai) ； 陳昭旭(C.S. Chen) ； 陳時祖(S.H. Chen)

《鑛冶：中國鑛冶工程學會會刊》 50卷4期 (2006/12) Pp. 117-124

https://doi.org/10.30069/MM.200612_50(4).0007

摘要

在地震、豪雨過後，由於山區交通常中斷，使用衛星影像判釋可能是最快速的調查山崩災害方法。本文以阿里山公路部份路段為研究區域，使用三張不同拍攝日期的福衛二號（（Formosat-2）影像進行常態化差異植生指標（NDVI）計算，再求每個像元之NDVI差值。然後，建立差值的最佳門檻值，以能最正確地指出新生崩塌地位置。另將不同時間之多光譜影像進行人工比對，加上現地勘查，來找出新生崩塌地以與上述方法比對。結果顯示，NDVI差值分析與人工判釋所圈選之區域吻合度相當高。可知NDVI差值分析法配合上Formosat-2影像，應用於山區的新生崩塌地快速調查，確有實用上的參考價值，並可具有Formosat-2每日再訪性、高解析度、廣域拍攝、易取得、價格便宜等優點。

關鍵字

遙感探測 ； 福衛2號 ； 常態化植生指標 ； 崩塌地調查

並列摘要

The purpose of this study is to establish a method to quickly locate the newly formed landslide sites in the mountain area by using Formosat-2 images, which has 2 m and 8 m ground resolutions for the panchromatic images and multi-spectral images respectively. The chosen study area is the nearby slopes along part of Ali Shan highway. Three satellite images with different imaging dates were used. The normalized differential vegetation index (NOVI), which represents the density of vegetation, of each pixel was calculated. Then, the difference of NOVI values of each pixel between two different date images was calculated. Finally, a best threshold NOVI difference value was calculated, and it can locate the newly formed landslide sites with best accuracy. Human image interpretation and field trip were employed to check the above automatic image processing results, the agreement is good.

並列關鍵字

Formosat-2 satellite images ； NDVI ； landslide sites