Predicting Land Use Changes Using Dyna-CLUE Model in Shihmen Sub-Watershed, Taiwan

• Main Authors: LaxmanMaharjan, 雷諮曼
• Other Authors: Pao-Shan Yu
• Format: Others
• Language: en_US
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/n9h45v

碩士 === 國立成功大學 === 自然災害減災及管理國際碩士學位學程 === 105 === Anthropogenic activities combined with the physiographical attributes in the watershed can largely affect the land use change pattern. Land use change in one hand indicates the human development whereas in other hand if overexploited cause harm to ecosystem causing global warming, climate change or even trigger natural disasters. The purposes of this study are: (1) to analyze the relationship between driving factors and their response to land use change and (2) to investigate the possible land use changes in Shihmen sub-watershed by using Dynamic Conversion of Land Use and its Effects (Dyna-CLUE) model. Firstly, the land use pattern in 2004 is recognized as the base map from Landsat satellite image by using a classification algorithm. Five different land use types are classified which are water, forest, built-up, grassland and bare land. Secondly, a logistic regression model is built to quantify the relationship between estimated driving factors and land use types. Simple linear extrapolation method is adopted to calculate the future demand of each land use types. Then, spatial allocation model, Dyna-CLUE is used to simulate the evolvement of land use pattern from 2004 to 2011 and the parameters of Dyna-CLUE (i.e., conversion elasticities for various land use types) are tuned to match with the observed land use map in 2011. Conversion elasticities of 1, 0.95, 0.8, 0.3, and 0.05 for water, forest, built-up, grassland and bare land, respectively, give the most reliable result. Relative operating characteristics (ROC) curve, kappa and accuracy indices are selected for statistical validation of the generated maps. Finally, the study used tuned Dyna-CLUE to project the land use pattern in 2020 for four different scenarios: (1) Linear trend of land use demand without restriction areas, (2) Linear trend of land use demand with restriction areas, (3) Input of minimum and maximum time steps of conversion sequence in the conversion matrix, and (4) Higher rate of land transformation. All the results show an increase in built-up and bare land area on the expenses of forest and grassland area in 2020.