使用細胞自動化模擬烏石坑地區之地景動態

Modeling Landscape Dynamics using a Cellular Automata in Wu-Shyr-Keng Area of Taiwan

預測自然地景的動態發展有利於保育工作的推動。本研究以細胞自動化（cellular automata, CA）為基礎，結合馬可夫模式與邏輯斯迴歸，發展一模擬烏石坑地區地景動態之預測模式。＂Satellite Pour L'observation de la Terre＂遙測影像所建立2006與2012年之地景分類圖資進行變遷監測發現，本區部份的森林覆蓋已呈現回復的情形，進而將馬可夫模式所預測的地景面積，以及邏輯斯迴歸推估之空間分布機率，整合至CA演化規則中，建構模擬森林回復之地景動態模式。模擬的結果對照遙測分類結果發現，兩者各地景類型的空間分布的概況相去不遠，有助於模擬地景動態。總體而言，本研究所發展之CA模式可作為生態環境與資源經營管理之參考，另一方面，亦可作為其他自然地景之研究範本。

Forecasting a reliable dynamic process in respect of natural landscape is central to drive conservation efforts. In this study, a Markov model and logistic regression imbedded cellular automaton (CA)-based model are developed to predict the landscape dynamics in Wu-Shyr-Keng area of Taiwan. We observed landscape change between 2006 and 2012 using remotely sensed data from ＂Satellite Pour L'observation de la Terre＂. The result showed that a portion of forest covers has been restored, and the spatial distribution of forest resilience was ingested into a Markov and logistic model to build transition rules for CA simulation. We found a satisfactory comparison between the simulation result and remote-sensing estimate in the corresponding year. The developed protocol may not only provide insights for conservation and resource management, but also be extended for use in other ecosystems.