Determination of the Groundwater Data Missing Data by Regression Model And Artificial Neural Network

• Main Authors: Hun-ling Cho, 卓惠玲
• Other Authors: Tien-Yin Chou
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/95747458022670548988

碩士 === 逢甲大學 === 土地管理所 === 96 === Due to global climate change in recent years, continuous droughts occur frequently. The short of water resource will certainly affect people''s living and economics. The groundwater provides a great help when drought happens. Recently, many government organizations and research institution have focused on understanding the water resource management and estimation. During estimation, lack of data for some spans is crucial to reseatch results and further decision making. For this reason, this study is focused on interpolation of groundwater elevation for discontinuous measurement and to develop complete and continuous groundwater elevation data. In this study, both regression analysis and artificial neural network methods were used for groundwater data imputation. The groundwater elevation data over the past ten years (from 1996 to 2005) at Beigang(2) station of the Chostui River have collected for this study. With these data, annual, seasonal periods are analyzed by multiple regression analysis and artificial neural network. The result showed that artificial neural network is suitable for trend of large variation with long period of data imputation, yet regression analysis is suitable for trend of small variation with short period of data imputation. The result also revealed that artificial neural network has the capability for noise filtering. The results of this study will provide a solid reference in dealing with the issue of land subsidence prevention.