Journal of the Japan Society of Erosion Control Engineering,
Vol.56,No.5,2004
A study on setting a critical line of landslide disaster for warning and evacuation and effect evaluation of groundwater drainage works based the prediction model of groundwater level by neuralnetwork
Hiroaki TAKEMOTO, Atsukuni KAJIMA, Kazumasa KURAMOTO, Tsutomu HONJO and Kohei FURUKAWA
Abstract
Some methods had previously been proposed to model the relationship between groundwater level and rainfall. These methods, however, involve parameter setting by trial and error and may require considerable experience in determining the structure of an appropriate prediction model. Groundwater level modeling can be simplified by using neural networks, which excel in nonlinear analysis, without considering complex parameters. This study constructs a neural]network]based water level fluctuation model from rainfall and groundwater level data in a landslide area and demonstrates that the model has the ability to predict groundwater levels. Not only rainfall but also other factors that influence groundwater level are used to improve the precision of analysis. In addition, disaster prediction and the effect of countermeasures are discussed by using optimized models.
Key words:landslide, groundwater level, Neural Network
Japan Society of Erosion Control Engineering
Sabo Kaikan, 2-7-5 Hirakawa-cho, Chiyoda-ku,
Phone +81-3-3222-0747 Fax +81-3-3230-6759
http://www.jsece.or.jp/
Mail jimu@jsece.or.jp