近年來,受到都市人口成長壓力與經濟產業發展帶來生活型態結構變遷的影響,原本屬於較低經濟價值與脆弱之邊際土地資源,皆成為土地開發的主要對象。而日益密集的土地使用型態與不適性的開發,使大地反撲且嚴重威脅民眾生命財產的安全。 本研究目的在以液化潛勢微分區探討土地開發適宜性,透過Matlab程式語言之類神經網路結合微分區之劃設,探討彰化縣員林地區之液化潛勢。本研究運用921地震中所蒐集員林地區之現地鑽孔點位資料作運算,並結合倒傳遞類神經網路加以訓練與測試,結果顯示此法有不錯之推估結果,可應用於土地開發前之參考用。 本研究亦利用地理資訊系統軟體ArcGIS之空間分析功能及IDW法,展繪出液化潛勢分區,並再加入土地使用現況及其他評估參數綜合分析,以探討其液化敏感區整體之土地開發適宜性,並在法令考量下,針對所評估出來之土地開發適宜性等級,作住商密度等級之建議,以作為未來相關單位開發時參考之用。
Taiwan is located at a complex juncture between the Eurasian and Philippine Sea plates. Seismicity is extremely active on this island. Earthquakes that occur frequently cause not only countless structure damages, but also soil liquefaction. Sinking of soil layer, gush of sand, side direction displacement, etc. were observed. Therefore, the development of evaluation of liquefaction potential in different area must be done in earthquake mitigation. The liquefaction potential and corresponding risk in land development of Yuanlin area are studied in this research. The purpose of this research is to probe into land development suitability by delimiting of microzonation of liquefaction area. Neural network combined with microzonation are employed. The field data in Yuanlin gathered from the Chi-Chi earthquake is used to perform the verification of evaluation method. The function of spatial analysis of ArcGIS is used to show zonation of area with liquefaction potential. The soil parameter and geology of study area, and the distance to river and fault are considered as inputs of artificial neural networks. The liquefaction potential index evaluated by the real liquefaction characterization is used as output. With discretization and IDW method in analysis module in geographic information system, the result of evaluation using artificial neural networks can be transformed to a liquefaction potential map. The result shows that this system works well and can be put to practice in land development. In the future, the current usage and other parameters will be added to assess land suitability.