水位數據品管檢核之研究

Study on Accuracy Verifying Procedure of Water Level Data

顔沛華(Pei-Hwa Yen) ； 蔡宗旻(Tsung-Min Tsai) ； 李友平(Youe-Ping Lee)

《農業工程學報》 55卷4期 (2009/12) Pp. 80-89

https://doi.org/10.29974/JTAE.200912.0007

摘要

對於現場蒐錄之水位數據，本研究提出觀測資料必須符合資料合理性、連續性及關聯生三項檢驗原則之品管程序，以90%信賴區間法濾除水位數據之異常值，對於缺漏或經品管檢核遭剔除的資料，則採用時序列卡門濾波法進行補遺以儀器規格及探頭（感測器）高程作爲合理性之比對基準，以馬可夫鏈模式來作連續性品管，以不同物理量闊的關係作爲關聯性檢測的依接。對於如颱風等異常氣候造成資料無法通過檢核峙，則改採人工檢核，令現場量測之水位數據更爲準確可靠。本研究以灣內橋及達德安攔砂壩三測站之實測值作資料合理性、連續性及關聯性三項檢驗之測試，據分析結果顯示，總缺遺筆數不應超過總檢核筆數之7.32%（8%以下），否則經補選之水位數據與實測飽間可能有超過10%的誤差。

關鍵字

卡門濾波 ； 馬可夫鏈 ； 信賴區間 ； 品管程序

並列摘要

Accuracy data verifying procedures such as observation rationality, data continuity and phenomena relativity tests are proposed in this paper. Water level data being collected including abnormal value in data bank should be screened out by statistical consideration of 90% confidence interval and the Kalman filtering algorithm was adopted here as the supplement technique for data missing. Limitations of device specifications and sensor elevation were regarded as the basis of observation rationality test. The Markov chain model (AR model) was used for data continuity verification. Physical relation between two parameters may be used as the phenomena relativity test. As to extreme events (say, typhoon or earthquake) resulted in failure testing through the accuracy data verification procedure, the manual inspection should he running as an option to ensure data accuracy and avoid fault applications. In the aspect of data verification procedure, testing data collected from Wan-Ray Bridge and Dar-Ter-An check dam stations were used for observation rationality, data continuity and phenomena relativity testing and analyzing result revealed that the supplementary data should not exceed the total amount data by 8%, else more than 10% error may be expected in between measuring and supplementary data values.

並列關鍵字

Kalman Filtering ； Markov Chain ； Confidence interval ； Accuracy verifying procedure