空間資料庫檢索系統開發與關聯樣式探勘之研究

Abstract

隨著資訊科技的迅速發展，地理資訊系統在環境議題的研究上扮演越來越重要的角色。航遙測空間資料包含網格資料與向量資料，因圖資分散在各處找尋不易，為解決此一問題，故需建立一個即時且有效的圖資檢索系統。 本研究開發以Web GIS 為基礎的查詢介面，讓使用者可以在圖面上藉由放大、縮小、平移及框選等工具來查詢圖資，並建立圖資的詮釋資料。這些圖資經過歸納、分類整理與壓縮後，提供使用者透過網際網路檢索及運用。 又為滿足空間決策支援上的需要，傳統的資料庫管理系統或地理資訊系統已無法滿足需求，所以空間資料探勘技術近年來逐漸被重視。本研究以93年艾利颱風引致的山崩為案例，將相關資料建立成提供資料探勘使用的資料集。再利用關聯法則分析等資料探勘技術，探索隱含於地震山崩與相關環境因子間的關聯樣式，獲致良好的結果，相信這樣的方法可以應用於其它地理資料在環境的議題上。

Geographic Information Systems (GIS) is growing popularity and gaining importance in environmental applications, as the technology advances in spatial data handling and analysis. Aerial photos and Remote Sensing data are include raster-based data and vector-based data. Due to spatial data scattered everywhere is not easy to find, therefore we should construct a gis-based indexing system for quickly and correctly searching those data. In this study, we construct a web gis-based indexing system. This system provides users to zoom in, zoom out, and AOI (area of interesting ) search on the map. We generaliz, classify and compress those geospatial data to let users access and apply those data via internet. GIS technology still fails to meet the demand for spatial decision support fully. Recently, geospatial data mining emerges as a promising approach to meet the challenge. In this study, we analyze the association pattern between landslides induced by Typhoon Aere in 2004 and environmental variables to demonstrate the use of spatial data mining technique, association rules mining, to gain insights into environment settings vulnerable to landslides after Typhoon Aere. While the case study is specific, our approach is applicable to associate environmental variables that may address other geospatial data on environmental issues of interest.