應用多重感應器監測雲林地區三維變形

Abstract

本研究應用多重感測系統對雲林地區地面變形進行監測，監測方法包括全球定位系統(GPS)、水準測量、磁感應環分層式地層下陷監測井（以下簡稱地陷監測井）與合成孔徑雷達差分干涉術DInSAR(Differential Interferometric Synthetic Aperture Radar）。 水準測量成果顯示，雲林地區呈現碗狀的下陷，沉陷中心集中在虎尾鎮、土庫鎮與元長鄉，而雲林縣之持續下陷面積與降雨量呈現逆相關，其原因為雲林地區地面水源不足，主要水源大幅依賴地下水，在未來需求水量增加，但新增水源尚未完成之前，一旦出現乾旱，則雲林地區發生大幅地層下陷的機率將會增高。 分析土庫國中地陷監測井之地層壓縮的單位厚度壓縮速率成果顯示，土庫地區主壓縮層發生在第四含水層（地下250公尺以下），封填周圍之深層抽水井之後，深層土壤壓縮出現逐漸減緩的趨勢，顯示未來政府單位若能有效管控深層的抽水行為，將可有效降低地表之下陷量。 應用GPS固定站資料輔助地陷監測井資料分析土壤應力應變分析與計算體積壓縮係數，由其成果顯示，西港地區因長期地層下陷，地層變形慢慢由塑性進入彈塑性階段，未來年平均壓縮速率將逐漸變小，應用ARIMA預測模式分析，2009年西港地區之年平均下陷速率為2.9公分/年。 利用精密軌道資料、Goldstein radar interferogram filter與stack方法，提高DInSAR的測量精度，並利用水準測量與DInSAR的成果交叉驗證，其成果顯示均方根誤差約在1∼2公分以內。另外利用融合技術（Fusion），整合水準測量與DInSAR成果，經實驗結果顯示其均方根誤差降至0.7∼0.9公分。

This study uses a multi-sensor system for monitoring land deformation in the Yunlin area. The system includes Global Positioning System (GPS), leveling, multi-level compaction monitoring well and Differential Interferometric Synthetic Aperture Radar (DInSAR). The results from the leveling surveys show that the subsidence in the Yunlin area is bowl-shaped. The center of the subsidence bowl is in Huwei Township, Tuku Township, and Yuangchang Township since this area mainly depends on ground water due to the lack of other water resources; therefore, whenever drought comes, severe land subsidence would occur in this area. The analysis of the compaction rate indicates that the major compaction layer occurs in the forth aquifer (under 250 m). After the government plugging the nearby deep wells, the deep sediment compaction has been eased in Tuku area, which suggests that the subsidence rate can be effectively reduced with the government’s effective monitoring and management policies in deep well extraction. GPS continuous station data are used to simulate the multi-level compaction well data to analyze the stress-strain relationship of sedimentary strata and the parameter of mass compaction. The result has indicated that due to the chronic land subsidence in Sigang area, the compaction of sedimentary strata has become from inelastic to mostly inelastic, and the future yearly average compaction rate will gradually decrease. Through ARIMA prediction model analysis, the result also shows that the average subsidence rate in Sigang area is 2.8 cm/year in 2009. The precise orbits, Goldstein radar interferogram filter and stake methods are applied to increase the accuracy rate of DInSAR. The results from the cross validation between leveling and DInSAR methods show that RMSE is about 1-2 cm. The fusion method is also used to integrate the results from leveling and DInSAR. The fusion reduces RMSE to 0.7-0.9 cm.