頁籤選單縮合
題 名 | 天然闊葉林冠層孔隙分布空間統計分析=Spatial Statistical Analysis on Pattern of Canopy Gaps of Natural Broad-Leaved Forests |
---|---|
作 者 | 謝漢欽; 張哲彰; | 書刊名 | 國家公園學報 |
卷 期 | 20:4 2010.12[民99.12] |
頁 次 | 頁45-62 |
分類號 | 436.27 |
關鍵詞 | 林冠孔隙; 地景生態學; 空間統計; 空間相關; Canopy gaps; Landscape ecology; Spatial statistics; Spatial correlation; |
語 文 | 中文(Chinese) |
中文摘要 | 為了瞭解天然闊葉林不同時期林冠孔隙於地景層級與組類層級尺度變遷 的分布格局與空間相關性,藉由不同空間統計方法分析結果,探討不同分析方法的 適用性。本研究使用1998 年及2002 年兩期林業試驗所蓮華池試驗林150 公頃闊葉 天然林數位航空照片,以數位立體判釋方式,求得兩期的林冠孔隙空間分布主題 圖。應用一個整合了相關、因素、兩階段聚集及判別分析的多變數統計分析程序, 將林冠孔隙分類成3 個主要生態干擾類型聚集組,並進行孔隙變遷分析與探討。在 林冠孔隙空間相關分析方面,本研究使用與取樣距離有關的半變異距分析、6 種空 間相關統計分析方法,針對兩期孔隙的全域、組類及局部尺度,進行綜合分析及方 法比較。結果發現不同空間統計方法,依其演算方法從簡單到複雜,會導致空間相 關分析結果相當的差異;此外以同樣方法在地景層次及組類層次的分析結果也受到 尺度效應的影響,所得分析結果也有所不同。使用局部的空間相關分析模式,可以 地理資訊圖層顯示孔隙空間相似性與高低聚集的位置,可解決空間相關分布格局本 身具有空間異質性的問題。 |
英文摘要 | In this study, applicability of various spatial statistical methods was examined in order to understand spatial distribution patterns and correlations of natural broad-leaved forests canopy gaps at landscape and class levels in different periods. An experimental area of 150ha in the Experimental Forest of Taiwan Forestry Research Institute was studied. Nnatural broad-leaved canopy gap thematic maps of the experimental area in 1998 and 2002 were made by adopting a digital aerial photograph stereo interpretation approach. An integrated multivariate statistical procedure including correlation analysis, factor analysis, two-stage cluster analysis and discriminate analysis was applied to obtain optimal three clustered classes as the main canopy types in the two periods. Canopy gap change analyses and discussions between the two periods were also conducted. Sampling distance was determined using the ranges from the analyses of semivariogram, and six different spatial statistical analytical methods were adopted to execute comprehensive analysis and method comparison. The results show that using simple to complex algorithm, different spatial statistical methods would lead to considerably different spatial correlation. Furthermore, scale can also influence the results obtained at the landscape level and class level. Local spatial correlation analysis model can be used to display canopy gap distribution with geographical information layers, find high and low clustered positions and avoid spatial heterogeneity problems in distribution pattern. |
本系統中英文摘要資訊取自各篇刊載內容。