921震災崩塌地監測及植生復育模式之研究－以台灣中部山區爲例

Assessment and Monitoring of Vegetation Recovery at Landslides Caused by the Chi-Chi Earthquake in Central Taiwan

林文賜(Wen-Tzu Lin) ； 鄧亞恬(Ya-Tien Teng)

《水保技術》 5卷3期 (2010/07) Pp. 134-141

https://doi.org/10.30040/JSWCT.201007.0002

摘要

921集集大地震造成台灣中部地區坡地遭受嚴重崩塌裸露，以草屯到埔里間烏溪沿岸的九九峰地區及南港溪上游支流韭菜湖溪流域的九份二山地區破壞規模最爲嚴重。由於震災至今已滿十年，瞭解該區之崩塌地變化及植生復原狀況有其必要性，本研究藉由遙測技術及非監督性類神經網路模式，評估九九峰及九份二山地區多期SPOT衛星影像之崩塌地變遷，並以植生指標監測崩塌區位植生復育情形。分析結果顯示，經過十年之自然植生演替，九九峰之崩塌面積已由857.55公頃縮減爲221.87公頃，復原比例爲74.13%（635.68公頃）；而九份二山之崩塌面積已由212.03公頃縮減爲59.95公頃，復原比例爲71.73%（152.08公頃）。在崩塌地植生復育評估，九九峰及九份二山地區之植生復育率分別達66.24%及64.59%，顯示在不受人爲干擾情況下，崩塌之裸露地表已逐年恢復植被。

關鍵字

崩塌地 ； 非監督性類神經網路模式 ； 植生復育率

並列摘要

Massive landslides, caused by the catastrophic Chi-Chi earthquake in 1999, occurred in central Taiwan, especially at the Jou-Jou Mountain area in Wu-Chi river and the Chiufenershan area in Nantou county. There have been over ten years since the earthquake. Therefore, it is necessary to assess the landslide change and its vegetation recovery condition. In this study, multi-temporal SPOT images coupled with unsupervised neural network model were used to identify the post-quake landslides. Then, vegetation recovery index was proposed to quantify the vegetation recovery condition at landslides. The analyzed results show the landslide areas had been reduced from 857.55 ha to 221.87 ha, about 74.13% (635.68 ha) of landslide restored, in the Jou-Jou Mountain area and from 212.03 ha to 59.95 ha, about 71.73% (152.08 ha) of landslide restored, in the Chiufenershan area, over ten years of monitoring. The vegetation recovery rate in the Jou-Jou Mountain area and Chiufenershan area reached up to 66.24% and 64.59%, respectively. Without human interference, the denudation surfaces have been gradually covered with native pioneer plants.

並列關鍵字

Landslide ； Unsupervised Neural Network Model ； Vegetation Recovery Rate