| With the rapid growth of the urban population and extent,the sustainability-enhancing urban management requires to conduct annual analyses of land cover dynamics in urban areas,in order to provide a thorough understanding of the urbanization effects on environment and valuable information for the improvement of urban growth modeling.Although the urban land cover dynamics are widely studied using remote sensing data,most current studies focus on major land cover changes.Limited by the spatio-temporal inconsistency of land cover time series data,few research focuses on characterizing the long-term and high temporal frequency land cover dynamics in cities,ignoring the abundant continuous spatio-temporal information of land cover changes.This study proposed a post classification method,the Spatio-Temporal Land Cover Filter(STLCF),designed to remove the illogical land cover change events with small probabilities in the time series of land cover maps,and then analyzed the annual land cover dynamics in urban areas.The knowledge of illogical land cover change event was ‘learned' from the land cover maps through the actual spatio-temporal transition probability matrix,instead of experts' knowledge.The illogical event was modified into the land cover type with the maximum joint transition probability calculated by the na?ve Bayesian equation.This method was highly automated and simply parameterized,which provides a new idea for the study of long-term and high temporal frequency land cover dynamics.The STLCF was applied to Wuhan,a typical densely urbanized Chinese city,to test and evaluate its performance.The classic C4.5 decision tree was employed to classify the study area into built-up areas,bare land,vegetation,and water bodies.The initial land cover classification maps with high accuracy from 2000 to 2013 were produced,which provided a high-quality data base for the STLCF.Results showed that the mean overall accuracy of annual change detection for the land cover maps was 5.89% higher than those before the implementation of the STLCF.In addition,the amount of land cover trajectories with unrealistically frequent changes was significantly decreased.The static trajectories with the stable land cover types accounted for 91.57% of all pixels in the study area,7.86% of the pixels experienced one land cover change,0.50% of the pixels experienced twice land cover changes,and about 0.07% of the pixels experienced more than twice land cover changes.The annual analyses of land cover change rate and its temporal trend demonstrated the nonlinear increasing tendency in urbanization.Since 2004,urbanization accelerated with the pronounced increasing trend in the conversions from vegetated areas and water bodies to built-up areas.The trend of the vegetation loss and the conversion to water bodies was roughly in agreement with that of the urbanization.The analyses of the distribution of land cover trajectories showed that the most significant land cover change in the study area was the conversion from vegetation to built-up areas,which can be found on the fringe of the metropolitan area.In the northwest of the study area,vast vegetated areas were converted to water bodies,from the development of planting industry to fishery industry.The study also found the conversion from built-up areas to other land cover types,mainly caused by the restoration of built-up areas to the park or green belt/wedges along rivers and new roads in the metropolitan areas,and to the cropland and woods in the rural areas.Thus this study explored the reversibility of built-up areas.Results of this study showed that ‘learning' from land cover maps of the study area was of great significance for the spatio-temporal consistency check.What's more,the modified long time series land cover trajectories better described the spatial and temporal development pattern of Wuhan,and provided insights for the land cover dynamic research and urban sustainable development. |
PDF Full Text Request