應用灰色馬可夫模式預測森林面積

Prediction of Forest Area With Grey Markov Model

薛怡珍(I-Chen Hsueh) ； 陳麗惠(Li-Hui Chen) ； 李國忠(J. Kou-Chung Lee) ； 邱祈榮(Chyi-Rong Chiou) ； 賴明洲(Ming-Jou Lai)

《中華林學季刊》 37卷1期 (2004/03) Pp. 49-59

https://doi.org/10.30064/QJCF.200403.0004

摘要

森林永續經營管理或生態系經營管理的宗旨為維持森林生態系統的健康與恢復，追求生態系統整體所提供的全部效益和價值，並要在地景的水平層次上長期保持森林的活力、生產力、生物多樣性及再生能力。森林是否達到生態系經營管理的成效則必須加以定期監測，而監測的指標與經營管理的指標上「森林面積」的變化消長則是主要的重點，進行森林面積（森林覆蓋率總量）的動態監測也是當前森林生態系經營管理的主要任務之一。本研究應用灰色馬可夫模式進行台大實驗林第25～42林班森林面積之預測，因為灰色馬可夫模式，兼具灰色預測GM(1,1)模式和馬可夫模式的優點，能充分利用歷史數據資料所提供的訊息，對隨機波動性較大的數據進行預測。此方法用於長期預測時，可將預測年預測的狀態和預測值計入數據序列重新構置馬可夫轉置矩陣逐年預測，應用此方法進行長期預測其精確度也明顯高於馬可夫轉置矩陣模式及灰色預測GM(1,1)模式。研究結果顯示以灰色馬可夫模式進行預測之平均精確度高達99.99%，以灰色預測模式進行預測之平均精確度為96.70%。

關鍵字

灰色馬可夫模式 ； 灰色預測模式 ； 馬可夫模式 ； 地景變遷 ； GM(1,1)模式

並列摘要

The objective of forest sustainable management or ecosystem management is to maintain the health and restoration of forest ecosystem, to achieve all benefits and values which are provided by ecosystem, and to keep vitality, productivity, biodiversity and regenerative capability of forest at landscape level. It is necessary to set a forest environment monitoring system for evaluating the efficiency of ecosystem management. The key point of monitoring as well as management indication is the fluctuated change of forest areas (the total amount of forest coverage). Currently, one of the main tasks of the EM (ecological management) is to carry out the monitoring of change of forest areas. We applied the grey Markov model to predict the values of the forest areas in 25-42 compartments of the Experimental Forest, National Taiwan University. The grey Markov model characterized by GM(1,1) model of grey prediction and Markov model was applicable to the previous data to predict the value by random and fluctuant data. Its precision was comparatively accurate than GM(1,1) model or Markov model. The data of grey Markov model was based on the previous data. When performing grey Markov model by more collective pervious data, the precision of the predicted value would be more reliable. The results showed that the precision was 99.99% by grey Markov model and the precision was 96.70% by GM(1,1) model. It demostrated that the precision of grey Markov model was higher than GM(1,1) model.

並列關鍵字

Grey Markov model ； grey prediction ； Markov model ； landscape change ； GM(1,1) model