應用航測資料與數學模式監測民墾地變遷之研究

Monitoring of Landscape Change on Unlawful Cultivated Forestland Using Photogrammetric Data and Mathematical Models

吳治達(Chih-Da Wu) ； 鄭祈全(Chi-Chuan Cheng)

《航測及遙測學刊》 10卷4期 (2005/12) Pp. 373-384

https://doi.org/10.6574/JPRS.2005.10(4).6

摘要

森林地景變遷監測之目的在掌握過去及現在地景結構的變化過程，並進一步推估未來演變趨勢，以提供森林經營管理之依據。本研究應用航測技術製作1971與1998年之土地利用類型圖，量化與人為干擾有關之Shannon多樣性指數分析蓮華池試驗林地景結構的變遷情形，並以Shannon多樣性指數t檢定法，評估1971年及1998年期間民墾干擾程度之差異，最後整合Markov鏈模式與二項式logit廻歸之分析結果，預測未來地景變遷模擬民墾地發生之空間分佈情形，進而探討民墾地干擾對未來地景之影響效應。由研究結果可知，二期Shannon多樣性指數雖有差異，但其地景變遷在5%顯著水準時，均未達顯著差異；至於在使用Markov模式預測2052年、2106年及2160年之地景變遷，則發現民墾地面積由1971年的0.39%和1998年的3.39%，增加為民國2052年的7.85%、2106年的12.08%和2160年的16.09%，明顯呈現上到趨勢；應用二項式Logit模式分析民墾地發生之空間分佈結果顯示，民墾地的發生與海拔高、坡度、及距民墾地、道路和河流的距離均有密切關係，其中以距民墾地和距道路的遠近最有相關。最後整合兩模式的結果以模擬未來民墾地之空間分佈結果指出，民墾地之分佈確實有逐漸擴張之趨勢，並且其干擾程度已達統計上之顯著水準。綜合以上研究結果可知，整合Markov鏈模式和二項式logit廻歸分析，預測未來之地景變遷和模擬民墾地發生之空間分佈情形，確實為一有效、可行的方法，此外，Shannon多樣性指數之t測驗檢定亦可用以監測和評估民墾地干擾對未來地景之影響效應，因此，本研究所得結果可提供蓮華池試驗林經營管理之參考。

關鍵字

Markov模式 ； Logit模式 ； 地景變遷 ； 民墾地 ； 空間分佈

並列摘要

The purpose of monitoring forest landscape change is not only to understand the change of past and present landscape structure, but also to predict the future trend for providing the reference of forest management. This study applied Shannon diversity index, which was related to human disturbance, to qualify the landscape change of Lien-Hwa-Chih Experimental Forest between 1971 and 1998. The t-test based on Shannon diversity index was also used to examine the disturbance differences caused by unlawful cultivated forestland. In addition, the mathematical models such as Markov chain and binomial logit regression analysis were applied to predict the future overall landscape structure and to simulate the spatial distribution of unlawful cultivated forestland. Again, the t-test was applied to monitor the future effect of disturbance caused by unlawful cultivated forestland. The result indicated that, using Shannon t-test, the landscape change between 1971 and 1998 had no differences at the 5% significance level. That is, there was no significant differences about the disturbance caused by unlawful cultivated forestland. The prediction of landscape change in 2052, 2106, and 2160 using the Markov chain model indicated that the area of unlawful cultivated forestland was increased from original 0.90% in 1971 and 3.39% in 1998 to 7.85% in 2052, 12.08% in 2106, and 6.09% in 2160, respectively. As for the spatial distribution using binomial logit regression analysis, the result showed that the occurrence of unlawful cultivated forestland had the relationship with elevation, slope, and the distance of road, river, and previous unlawful cultivated forestland. However, it was strongly correlated with the distance of road and previous unlawful cultivated forestland, and less correlated with elevation, slope, and the distance of river. Finally, the spatial simulation of unlawful cultivated forestlands illustrated that the distribution of unlawful cultivated forestlands was gradually expanded, and the disturbance level was differences at the significance level. From the result, obviously the integration of Markov chain and binomial logit regression analysis is a feasible approach to predict the overall landscape change in future and to simulate the spatial distribution of unlawful cultivated forestland. Also, the Shannon t-test is a good way to monitor and assess the disturbance effect of unlawful cultivated forestland on future landscape. The above result can be provided to Lien-Hwa-Chih Experimental Forest for the reference of forest management in order to prevent the occurrence of unlawful cultivated forestland.

並列關鍵字

Markov model ； binomial logit model ； landscape change ； unlawful cultivated forestland ； spatial distribution