基于自组织映射法的时间序列土地利用变化的时空可视化

发布时间：2018-08-19 14:23

【摘要】：精细尺度下多时间序列土地利用时空演变分析是当前研究的一个趋势,本研究基于2005、2007、2009、2011、2013年5期土地利用数据采用自组织映射方法分析了北京市乡镇级多时间序列土地利用的时空演变规律,实现了乡镇尺度下多时间序列土地利用数据的时空一体化表达和对比分析。通过构建自组织映射神经网络,利用其聚类和降维可视化功能对5个监测时期的土地利用数据同时进行训练,在其输出面板可以发现不同土地利用类型的分布聚集模式以及相互之间的结构比例关系,并对输出神经元进行二次聚类以及土地利用变化轨迹分析,展示出北京市乡镇级5个监测时相的土地利用时空演变规律。结果揭示出北京市平原区、山区及二者过渡的山前结合带的各自不同的土地利用时空变化轨迹与模式:北京市平原区向高建设用地比例的土地利用结构方向演变,山区向高林地比例的土地利用结构方向演变,而山前结合带的土地利用时空演变较为复杂。
[Abstract]:Multi-time series land use spatio-temporal evolution analysis in fine scale is a trend of current research. Based on the data of land use in 2005 / 2007, 2009 / 2011 and 2013, the temporal and spatial evolution of land use in multi-time series at township level in Beijing was analyzed by using self-organizing mapping method. The spatiotemporal integration and comparative analysis of land use data of multi-time series in township scale are realized. By constructing a self-organizing mapping neural network, the land use data of five monitoring periods are trained simultaneously by using its clustering and dimensionality reduction visualization functions. In the output panel, the distribution and aggregation patterns of different land use types and their structural and proportional relationships can be found, and the output neurons can be analyzed by quadratic clustering and land use change trajectory analysis. The temporal and spatial evolution of land use in 5 monitoring phases at township level in Beijing was demonstrated. The results reveal the different loci and patterns of land use in Beijing plain area, mountainous area and the transition of the two areas in front of the mountains: the land use structure of high proportion of construction land evolves in the plain area of Beijing. The land use structure of the mountain area changes to the high proportion of forest land, while the land use time and space evolution of the foreside conjunctive zone is more complex.
【作者单位】：北京师范大学遥感科学国家重点实验室;北京师范大学环境遥感与数字城市北京市重点实验室;北京师范大学地理学与遥感科学学院;
【基金】：国家自然科学基金项目(40671127) 国土资源部公益性行为科研专项(201411015-03)
【分类号】：F301.2;P208

【相似文献】