行政院農業委員會水土保持局委辦計畫摘要報告
坡地土砂災害觀測資訊進階加值分析與應用
計畫編號：SWCB-109-277 執行單位：國立臺灣大學 計畫主持人：劉格非教授
摘要
目前土石流觀測站鮮少被應用於預警或其它加值研究使用，因此本計畫擬透過觀 測站最常使用的兩種觀測資料：影像與地聲，發展可用於預警的加值運算，並以整合 式互動網站呈現最後應用方式。影像分析部分，本計畫完成以總灰階值偵測土石流抵 達時間，並以多區域法估算土石流波速，並提出以灰階值的多次傅立葉轉換方法，估 算畫面中巨石粒徑。地聲振動訊號方面，本計畫完成以振動訊號時變率來偵測土石流 抵達，以震聲波理論推估現地地聲檢知器安裝的環境可能產生的干擾訊號，同時建立 地聲訊號與流量與流速的初步關係。所有的偵測與計算方式，都加入監測加值，以神 木村的現場觀測資料建立互動式網站，同時建立神木監測站整個集水區的相關資料，
包含：歷年航照、衛星照比對、崩坍沖蝕變化、360 度虛擬實相片、過去事件整理與示 範等。
關鍵字: 影像分析、震聲波、預警、土石流
Added-value analysis and application of monitoring data on slopeland disaster
ABSTRACT
Debris flow monitoring system is rarely used for warning or research. This project will utilize the most used equipment: CCD and geophone as the main focus for added value operation.
The results will be shown through an interactive website. For CCD images, the total variation of the grey level is used to calculate the debris flow arrival time, velocity, celerity of debris flows, and multiple Fast Fourier Transform is applied to estimated the diameter of out of scale huge boulder. For geophone data, multi-scale perturbation is used to construct the signal variation through media, and the possible resonant frequency is found. Variation of the amplitude of the vibration signal is used to detect the arrival of debris flows. The relationship between the amplitude of the vibration signal is correlated to flow rate and surface velocity.
All added value results will be focused on Shen–Mu monitoring station. All data and research results will be shown and used for Shen-Mu international research center. As a base for whole information, results of landslides and erosion amount distribution is computed using aerial photos. The 360 virtual photos for different years were uploaded and the past event data is constructed.
Keywords: Images analysis, Seismo-acoustic wave, Warning, Debris flows
1
