標題: 多重高度空載重力測量應用於計算大地起伏和黑潮
Multiple-altitude airborne gravity surveys and applications to geoid determination and Kuroshio Current
作者: 施亘昶
Shih, Hsuan-Chang
黃金維
Hwang, Cheinway
土木工程學系
關鍵字: 空載重力;大地起伏;黑潮;剩餘地形效應;向下延續;最小二乘配置法;airborne gravity;geoid;kuroshio Current;residual terrain model (RTM) effects;downward continuation;least squares collocation (LSC)
公開日期: 2009
摘要: 本論文的主要目的為結合多重高度的空載重力資料和地表面的重力資料於計算台灣區域的大地起伏模型,並利用此模型計算 T/P 和 Jason-1 測高衛星之交叉點的黑潮流場。由內政部提供的空載重力資料主要分佈於高度約 5000 公尺和 1500 公尺,其涵蓋範圍為台灣本島、台灣東部的黑潮流域、台灣海峽和東沙群島。地表面的重力資料包含陸測重力資料、船載重力資料和測高重力資料(DNSC08)。本論文建立一個適用於空載重力測量的重力值約化流程,以 Bernese 5.0 定位軟體求解航機的動態位置,以此動態位置計算航機的速度量與加速度量,整合重力儀的觀測資料,利用低通濾波(low-pass filter)的方式消除原始重力資料的高頻雜訊,最後將此流程實際應用於多重高度的空載重力測量之空中重力值的約化並分析精度。為獲得到最佳的空中重力值,本研究分析了 Gaussian 濾波、 Butterworth 濾波和另一為分析非線性訊號設計的 Hilbert-Huang 轉換法做為高頻雜訊消除的適用性,經實際的應用後發現Gaussian 濾波罩窗 150-180 秒為一適當的大小;設計由原始重力資料偵測粗差和修補的技術,將其應用於實際的空載重力測量航線後,發現可提升精度約 0.69 mgal,為執行偵測前的原始精度之 19 %。將空中重力異常值與向上延續之重力異常值比較後發現,較大的差異量多分佈於高山地區,主要原因為此區域缺乏地表面重力資料。利用交叉點分析、重複性分析和相關頻譜分析後,顯示空載重力資料的精度約在 2-4 mgal,在約 1500 公尺高度的空間解析度為 6 公里,而在約 5000 公尺高度的空間解析度為 8 公里。 本文提出一有限頻譜的最小二乘配置法(Band-limited LSC),可依照不同空間解析度的重力資料給予不同階數的協方差函數(covariance function),有效的結合不同空間解析度的重力資料用於計算大地起伏模型和其他相關應用。在計算大地起伏的過程中,本研究分析了最新的 EGM08 全球重力場,顯示其精度有了大幅度的改善,主要的改善原因來自於 2160 階;組成最新的台灣地區數值地形模型(TWDTM2008),網格的解析度為 3 秒,以此模型利用高斯求積法計算空中的剩餘地形效應;結合空載與地表面重力資料所計算得的大地起伏,在平地的精度小於 5 公分,而部分山區可達到 10 公分以內的精度。 將上述計算得到的大地起伏模型應用於計算 T/P 和 Jason-1 測高衛星之交叉點的黑潮二維流場,以目前大地起伏模型和海水面高度(SSH)的精度推估後發現,必須使用一大於 100 公里的空間濾波,方可求得一 10 公分精度的地轉流(geostrophic current),本研究共計算了台灣附近海域上之 3 個交叉點的地轉流場,其中 A 點和 B 點的流場與海洋模型推估的流場較為相近,並且加入 WOCE 的浮標資料進行比對。另有一 C 點位於台灣海峽,其深度僅 79 公尺,由交叉點法求得之地轉流與海洋模型推估的流場相去甚遠,證明了此法目前適用於深海的範圍。
Multiple-altitude airborne gravity data are now available due to the effort of Ministry of the Interior, Taiwan and are distributed at the altitudes of 5000 m and 1500 m, covering the Taiwan Island, Kuroshio Current east of Taiwan, Taiwan Strait and Dongsha Atoll. The computer programs for the data reduction and accuracy analysis of scalar airborne gravimetry have been developed and produced reliable gravity data at the three airborne gravity surveys. Several de-noising tools, including the Gaussian filter, Butterworth filter and EEMD of HHT, were evaluated using a survey line from the Kuroshio Current airborne gravity survey. A filter width between 150 to 180 s is appropriate between noise reduction and gravity signal preservation. The analysis of a survey line from the Kuroshio survey shows the outlier detection and down-weighting technique yields an improvement of 0.69 mgal (compared with upward-continued surface gravity). The accuracy of the gravity anomalies from the three airborne gravity surveys is about 2 to 4 mgal. The spatial resolutions are 6 and 8 km (half-wavelength) at the altitudes of 1500 m and 5000 m, respectively. The airborne gravity anomalies agree well with the upward-continued surface gravity anomalies, except over high mountainous areas and locations with sparse surface gravity. For one cm-level geoid model a band-limited LSC, combining airborne and surface gravity data, in the RCR procedure was used. Evaluation of EGM08 using GPS/leveling-derived geoid undulations and surface gravity anomalies around Taiwan shows the improvements both on the gravity anomaly and height anomaly. The grid of TWDTM2008 was assessed by the leveling-derived heights at the benchmarks and shows a standard deviation of 5.25 m. The geoid model, Model 7, show yields standard deviations of 2.2, 6.6, 7.6 and 4.8 cm along the north, east, central and south first-order GPS/leveling routes, respectively. A crossover method for computing zonal and meridional ocean current components was developed. The error analysis concludes that a DOT at mm-level accuracy is needed to obtain ocean current velocities at the 10 cm•s-1 accuracy and 6 km resolution. To achieve a 10 cm•s-1 accuracy, the DOT must be filtered to a 100 km spatial resolution or coarser. Ocean currents at the T/P and JASON-1 crossover A and B were in-situ compared with the model-derived velocities and show a good agreement.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT079316805
http://hdl.handle.net/11536/40542
Appears in Collections:Thesis


Files in This Item:

  1. 680501.pdf