SPOT衛星影像於墾丁國家公園銀合歡入侵分布之繪製

Using SPOT Imagery to Map the Invasive Distribution of Leucaena lucocephala in Kenting National Park

鍾玉龍(Yuh-Lurng Chung) ； 呂明倫(Ming-Lun Lu)

《台灣林業科學》 21卷2期 (2006/06) Pp. 167-177

https://doi.org/10.7075/TJFS.200606.0167

摘要

外來物種的入侵是某一地區生物多樣性喪失的因素之一，藉由衛星遙測可提供地表空間資訊，有助於經營外來植物入侵對於原生棲地之衝擊。本研究利用SPOT衛星影像繪製墾丁國家公園轄區之銀合歡入侵分布圖，藉由瞭解銀合歡於乾、濕季節所表現之生理及形相特性的差異，經t檢定結果顯示，濕季之植生量明顯高於乾季，因此運用SWIR/NIR比值運算來突顯銀合歡於不同季節之影像特徵，並據以選取訓練樣區，如此可有效提升銀合歡分類結果之準確度（生產者精確度80.71%、使用者精確度82.48%），所繪製之銀合歡空間分布圖配合研究區物理環境資料分析結果顯示，道路與海拔高為銀合歡重要之環境控制因子，可提供管理單位進行銀合歡長期監測之有效資訊。

關鍵字

外來物種 ； 生物多樣性 ； 遙測 ； 監測

並列摘要

The ecological invasion of an alien species is 1 of the major factors in the loss of biodiversity worldwide. Remote sensing data can provide spatial information of alien plant invasion into native habitats; this offers managers an effective tool to monitor this problem. In this research, we used SPOT imagery to derive an invasive distribution map of Leucaena lucocephala in Kenting National Park, southern Taiwan. First, the normalized difference vegetation index (NDVI) was calculated to understand the seasonal biomass and plant morphological differences of Leucaena lucocephala between the dry and wet seasons; results of the t-test showed that the biomass in the wet season was greater than that in the dry season. The characteristic shown from the subtraction operation between the short-wave infrared/near infrared ratio image had good information from which to select the training sample for Leucaena lucocephala, and therefore could increase the image classification accuracy (the producer's accuracy was 80.71%, while the user's accuracy was 82.48%). Analysis of the invasive distribution map combined with physical environmental data showed that roads and elevation are important limiting factors for the distribution of L. lucocephla, and this map may provide managers with useful information for long-term monitoring.

並列關鍵字

alien species ； biodiversity ； remote sensing ； monitoring