紅樹林生態系統扮演著地球上重要的角色,其擁有自然生態價值、環境保護價值、經濟價值…等,但因人為砍伐與自然因素進而造成紅樹林快速減少,驅使人們開始重視紅樹林生態系統保育及復育工作。利用遙測科技可以獲得大範圍且長期的紅樹林分佈變化資訊,以利紅樹林生態系統長期監測,配合應用Landsat衛星影像可有效地了解紅樹林分佈變化情形。 本研究之研究區為伊洛瓦底江三角洲,為緬甸國內紅樹林之主要分佈地。本研究採用Landsat衛星影像分析紅樹林分佈變化,利用變遷偵測後影像分類方法提高影像分類成果精度,藉此獲得1989年、2001年、2014年紅樹林分佈,並加以比較三期分類成果以了解本研究區域內紅樹林分佈變化,從1989年至2014年紅樹林約減少了12%之多。為評估未來紅樹林分佈變化趨勢,本研究使用Probablistic Landscape Modeling and Simulation Tool (ProLAMS)來預測未來之紅樹林分佈變遷情形。此預測可分為三個步驟:首先,利用分類成果影像與數值地形模型找出可能影響紅樹林分佈變遷之影響因子,包含環境因子、區域性因子、相鄰近關係、光譜資訊等。由於紅樹林的分佈變遷具有空間異質性,本研究區內紅樹林面積增加與減少的區域分佈在不同的位置,本研究透過羅吉斯迴歸分析分別建立紅樹林消失與增加之空間機率模式,以獲得紅樹林消失與增加之機率圖。其次,根據已獲得之紅樹林消失與增加之機率圖,利用細胞自動機進行紅樹林分佈變遷模擬,藉由歷史資料律定模式參數並驗證模式之模擬能力,驗證精度(AUC)分別為0.81與0.79,其表示此模式適合用來模擬紅樹林分佈變化情形。最後,運用該模式模擬本研究區域內紅樹林分佈之未來變化。本研究認為使用ProLAMS可正確評估研究區域內紅樹林過去變化與未來趨勢,能提供相關管理者對紅樹林復育與保育之參考依據。;Mangrove forest is a globally important ecosystem. It provides ecological, environmental and economic services for human society. In the Irrawaddy Delta, Myanmar, the overly agricultural development to meet the human pressing basic needs has driven the deforestation of mangrove forests, consequently causing environmental impacts. Thus, spatiotemporal understanding of changes in mangrove forests is important to evaluate current management practices of mangrove forests in regard to formulating a better long-term management strategy of mangrove ecosystem. This study focuses on not only evaluating current mangrove deforestation but also predicting changes of mangrove forests in the future. This study used classification after change detection method and the Probabilistic Landscape Modeling and Simulation Tool (ProLAMS) to evaluate current mangrove deforestation and to predict future mangrove distribution, respectively. ProLAMS aims to integrate remote sensing (RS) and geographic information system (GIS) in landscape modeling and prediction. Applying ProLAMS in mangrove prediction basically requires two components: a statistic model for evaluating the probability of mangrove change evolution, and a dynamic algorithm to simulate the mangrove change evolution. Specifically, ProLAMS uses logistic regression analysis for statistic modelling and cellular automata (CA) for simulating the mangrove change. The results proved the ability of using ProLAMS for predicting mangrove forests distribution in the future.