1999年9月21日發生芮氏規模7.3的強震,依據90年7月行政院主計處統計之建物全半倒印領清冊,全台共計有105630筆,其中以南投縣受創最為嚴重,計有57191筆,其次為台中縣與台中市。本研究範圍為建物受創最為嚴重之南投縣及台中縣市,以距車籠埔斷層2公里範圍內之全部建物基礎,透過全半倒建物座落點位及建物各項屬性條件,與環境地理資訊進行空間結合,並參考建物震損相關研究,歸納建物毀損之成因,以決策樹之C4.5與CART理論建立建物震損推估模式,在近斷層下盤地區之準確度可達90%。故藉由空間方式呈現建物震損模式之推估結果,並透過Moran's I指標掌握建物震損推估模擬,可瞭解其隱含之空間意義。本研究以空間資料挖掘技術建立建物震損推估,可作為都市防震減災規劃之基礎。
A catastrophic earthquake with a magnitude of 7.3 on the Richter scale occurred in Chi-Chi and the Sun-Moon Lake areas of Nantou County in the early morning (01:47 local time) on Sept. 21, 1999. Located in central Taiwan, Nantou Country received the severest impact from the earthquake. The Directorate-General of Budget, Accounting, and Statistics (DGBAS) records indicate that more than 10,000 buildings were partially damaged or completely destroyed by this earthquake. This study examined all buildings within 2 KM of the Chelungpu fault, including Nantou County, Taichung County and Taichung City. Building risk evaluation was conducted using decision tree algorithms of C4.5 and classification and regression trees (CART), spatial autocorrelation analysis and the geographic information systems (GIS) technique. Model accuracy was 90% for damaged buildings in the footwall of the Chelungpu fault. Furthermore, the distribution of predicted results obtained by risk assessment of damage was simulated using Moran's I analysis. The seismic building damage model estimated by the spatial data mining approach