- 學位論文
以永久散射體雷達干涉法監測哈德遜灣附近冰後回彈
Detecting post-glacial rebound around Hudson Bay using Persistent Scatterer SAR Interferometry
摘要
加拿大哈德遜灣附近由於環境惡劣，連續觀測GPS站難以設立，因此該地區缺少地表變形觀測量，使得冰川均衡调整(glacial isostatic adjustment, GIA)模型預估的冰後回彈於此區域有改善空間。為了解決地表變形量缺乏的問題，本研究使用2017年至2019年的六月至十月的Sentinel-1B 衛星影像，應用永久散射體雷達干涉法(Persistent Scatterer Interferometric Synthetic Aperture Radar, PSInSAR)，且為了克服雷達干涉法在測量空間長波長變形量的限制，本研究使用了天氣預測模型ERA5模擬對流層改正量，且進一步的應用估計空間變化尺度因子的方法改正ERA5的對流層改正量，並使用stacking的方法將面狀觀測結果轉為緯度相關的條狀結果以消除殘餘誤差，最終測得該區域的地表變形量。研究結果顯示，在light of sight (LOS) 和垂直方向結果中，計算的地表變形速度除了通過與GPS比較的廣義假說檢定，且其root-mean-squared (RMS) difference也僅為2.57 mm/year 和5.05 mm/year，證明本研究的成果除了在統計上與GPS的觀測結果一致，其精度也相當高。與GIA模型中的ICE-6G_D比較中本研究的RMS difference在LOS與垂直方向上與ICE-6G_D僅為3.36 mm/year 和6.14 mm/year，十分接近GPS與ICE-6G的差異，1.33mm/year 和1.91mm/year，顯示本研究觀測冰後回彈的變形量與GPS的觀測結果差異不大。由於GPS於研究區域分布的較不平均，為了檢視其他地區的PSInSAR結果是否確實有觀測到冰後回彈，本研究將PSInSAR結果的空間長波長變形量與ICE-6G_D的結果比較，其相關係數達0.99。總和以上所述，本研究證實使用PSInSAR可以觀測得加拿大哈德遜灣附近的冰後回彈變形量，可望補足該地區地表變形量不足的問題。
並列摘要
Current glacial isostatic adjustment (GIA) models such as ICE-6G_D may still need more ground constraints around the Hudson Bay for model improvements. To provide constraints for improved GIA models, we used the Sentinel-1B SAR images in the summer months (June to October) over 2017–2019, and the method of Persistent Scatterer Interferometric Synthetic Aperture Radar (PSInSAR) to detect surface deformations around Hudson Bay. It is challenging for InSAR to measure long wavelength surface deformation such as post-glacial rebound (PGR). We used the spatially varying scaling method to tropospheric correction estimated from ERA5 and stacking to minimize the tropospheric residuals. The stacked PSInSAR-derived line of sight (LOS) and vertical velocities match those from GPS observations to 2.57 mm/year and 5.05 mm/year, respectively. The results from a hypothesis test suggests that the stacked PSInSAR-derived velocities are consistent with the GPS result. The root-mean-squared (RMS) difference between the stacked PSInSAR-derived and ICE-6G_D-derived velocities is 3.36 mm/year in LOS and is 6.14 mm/year in the vertical direction. These RMS differences are slightly greater than the RMS differences between corresponding values between GPS and ICE-6G_D, which are 1.33mm/year and 1.91 mm/year in the LOS and vertical directions. Thus, the velocities from PSInSAR and GPS are consistent to few mm/year levels. The long wavelength component of the stacked PSInSAR-derived deformations are consistent with that from ICE-6G_D with a correlation coefficient of 0.99. PSInSAR successfully detects PGR signatures around the Hudson Bay, and may be additional surface deformation measurements for GIA modeling.