標題: 以DCW高程資料製作台灣地區網格式數值地形模型之探析
A Study on the Grid Digital Elevation Model Generation of Taiwan Area from DCW Hypsographic Data
作者: 邱永祺
Young-Chyi Chiu
史天元
Tian-Yuan Shih
土木工程學系
關鍵字: 網格式數值地形模型;Grid Digital Elevation Model
公開日期: 1994
摘要: 利用航空攝影測量方法,在立體模型內測取高程點,繼而內插求取其他點 位,如網格點之高程,是製作數值地形模型的標準方法。然而,由於現有 數化圖很多,從舊有圖籍之等高線資料,獲得數值地形模型,對於許多應 用範疇而言,這種製作方式也是很重要的。本文謹就DCW圖庫粹取數化等 高線資料,應用在台灣地區網格式數值地形模型的製作進行探討。 DCW 為美國國防製圖局所編製的通用性全球圖庫。台灣地區高程資料是取自百 萬分之一戰略航圖之數化等高線而來,等高線間距為一千英呎(304.8公 尺)。本文選用:一、克利金圓模型法,二、克利金指數模型法,三、克 利金高斯模型法,四、克利金線形模型法,五、克利金球形模型法,六、 不規則三角網線性法,七、不規則三角網五面體法,八、附加限制條件的 不規則三角網線性法,九、附加限制條件的不規則三角網五面體法等兩類 九種方法加以研究。製作成果則以圖形法與數值法加以檢定,其中,數值 法所選用的精度指標包括:最大誤差、百分之九十排序誤差、均方根誤差 、絕對平均誤差與平均誤差五種,而檢核點的來源是獲取自 DCW內含之獨 立高程點資料。經由評估結果顯示:在台灣地區,DCW可產製30〞×30〞 網格數值地形模型,附加限制條件之不規則三角網線性內插法成果,其高 程無偏均方根誤差為 176.836公尺,無偏百分之九十排序誤差為213.149 公尺。 Photogrammetry is the major procedure for generating grid Digital Elevation Model (DEM). The elevation data is sampled with stereoplotter first, then the elevation for grid points is interpolated. In some cases, however, it would be quite p- ractical to generate DEM from existing topographic maps. This study investigates the generation of grid-DEM of Taiwan area with the DCW hypsographic data. Nine procedures based on two methods were applied to inte- rpolate elevation. These are, Kriging with circular, exponent- ial, gaussian, linear, spherical models, and TIN-linear, TIN- quintic, constrained TIN- linear, constrained TIN-quintic mode- ls. The result is evaluated graphically and also five numeri- cal indices. These indices are the maximum error, ninetieth percentile error, root mean squares error, mean absolute error , and mean error. The spot heights from DCW are used as the c- heck points. It is found that for the resulting 30" x 30" gr- id-DEM of Taiwan area generated with constanined TIN-linear m- odel, the root mean square error is 176.836 meters and the 90% ranking error is 213.149 meters with the bias removed.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT830015047
http://hdl.handle.net/11536/58740
顯示於類別:畢業論文