Title: | 測深光達波形於水深演算之研究 The Estimation of Water Depth from Bathymetric Lidar Waveforms |
Authors: | 林暐尊 Lin, Wei-Tsun 史天元 Shih, Tian-Yuan 土木工程學系 |
Keywords: | 測深光達;波形;Bathymetric Lidar;Waveform |
Issue Date: | 2010 |
Abstract: | 測深光達採用雷射掃描方式對水體覆蓋區域進行水深量測,藉由雷射波形之回波偵測定義水面及水底,最終求得水深。由商業系統掃描並解算出之水深具有相當精度,而其計算原理及處理方法皆屬於未公開技術。本研究以HawkEye II系統掃描於東沙環礁及澎湖地區之測深光達波形為研究數據,由實際處理波形資料及回波萃取進行水深演算,並與商業系統軟體計算成果比對。
東沙環礁區域共選取7個區域之波形資料,區域間水深具有相當差異,欲藉此觀察隨水深變化之演算成果,水深最淺小於1m,最深則有至40m水深;澎湖地區則以一區塊作為測試。波形資料使用整體平均經驗模態分解法(ensemble Empirical Mode Decomposition,EEMD)進行濾波處理。濾波後波形以Weibull分佈函數進行擬合,找出回波位置。
針對HawkEye II系統提供波形於不同波段間具有時間偏移問題處理後,將時間偏移改正後由偵測之回波位置演算水深,水深成果與商業軟體成果相比,平均誤差與標準偏差皆與水深大致呈現正相關。東沙地區成果於水深超過約20m時,平均誤差開始明顯增大,至40m水深高達4.06m;標準偏差於淺水區約0.5m,至40公尺水深時達4.92m;澎湖地區演算成果略差於東沙地區,使用紅外光波段為水面波段時,水深之平均誤差為-0.115m,標準偏差0.700m;使用拉曼紅光波段時,水深之平均誤差為0.547m,標準偏差1.552m。 Airborne bathymetric Lidar measure the water depth in laser scanning manner. Through the return echo detection to define the water surface and bottom, the depth can be derived. The accuracy of depths from the commercial system and software performs well result, but the processing techniques still in private. In this research, bathymetric Lidar waveforms derived from the HawkEye II in Dongsha Atoll and Penghu area are used to the water depth estimation and compared with the result from commercial software. Waveforms of Dongsha Atoll are chosen from 7 areas with large depth range, from smaller than 1m up to 40m. And only one area taken from Penghu for testing. After filtering the waveforms by the EEMD method, filtered waveforms are fitted with the Weibull distribution function for the echo definition and depth calculation. The mean error and standard deviation of depth are roughly in positive correlation with the depth. The mean error significantly increase when the depth larger than 20m, up to 4.06m with depth about 40m. The standard deviations are about 0.5m in the shallow area, up to 4.92m with depth about 40m. In Penghu area, the mean error and standard deviation of depth by using the in-fared as surface channel are -0.115m and 0.700m, and which are 0.547m and 1.552m by Raman channel. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT079816580 http://hdl.handle.net/11536/47331 |
Appears in Collections: | Thesis |