Land-Cover Simulation and Forcast of Melmeg Wetland Using Cellular Automata

This paper presents a model to simulate and forecast wetland land-cover changes with geographic cellular automata.As one of the three ecosystems,Wetland ecosystem has so many functions in regulating the climate and water circulation on earth and wetland is the important resource for human being.But in recent years,the function of wetland has been degenerating severely with the increased human activity.How to protect the wetland resource has become a hot issue in the wetland research domain.Cellular automata,a "Bottom-to-up" dynamic system with powerful function in simulation have been used to simulate complex phenomena in many domains.The article expounds the importance of cellular automata in protecting the wetland resource in many aspects,especially in dynamically forecasting and simulating the trend of wetland land-cover,which will provide the important reference for the management and development of wetland.Considering the traditional square cell of CA has intrinsic deficiency of anisotropy because of its geometry shape,the model replaces it with hexagon cell which has better character in isotropy.The model,which uses the back-propagation neural network to mine transition rules of cellular automata,has a lot of merits that not only reduces the difficulty to establish the transition rule because of many kinds of land-cover types,but also reduces subjective idea and then improve the simulation accuracy.Finally,this model is applied to the Melmeg National Reserve Wetland.The result,which owned accuracy of about 80%,shows that the model is suitable for simulating wetland land-cover change.The forecast and simulation is based on the hypothesis that the development velocity in study area is the same in a short period.By simulation,it's clear that,in Melmeg wetland,the area of farmland will increase while wetland will decrease,and the grassland and forest may disappear in the coming years.The environment of wetland ecosystem of Melmeg will be worsened according to the simulation.