松嫩高平原土地利用格局空间尺度识别及其关联关系研究

研究目的:实现由传统的人工识别到自动识别土地利用格局空间尺度的方法创新,为优化区域土地利用格局提供科学依据。研究方法:基于分形理论,采用二阶导数对计盒维数模型进行改进和优化。研究结果:(1)研究区土地利用格局在900 m×900 m尺度上具有明显的分形特征;(2)1991年、2003年和2015年研究区土地利用格局表现为各土地利用类型之间空间作用力、分布均匀程度和内部联结性的大小不同,并呈现出空间相互关联关系复杂性;(3)1991—2015年,研究区土地利用格局空间相互关联关系变化特征明显,表现为土地利用格局中竞争强度强弱的差异。研究结论:(1)改进的计盒维数模型能够实现土地利用格局空间尺度的自动识别;(2)建立的计盒维数自动识别尺度模型,确定研究区土地利用格局空间尺度900 m×900 m是独立最小地理单元,也是分析土地利用格局变化的最优尺度。

Study on Spatial Scale Recognition and Its Relationships of Land Use Pattern in Songnen High Plain

The purpose of the paper is to provide the scientific reference for optimizing regional land use pattern by realizing a research method innovation from traditional artificial recognition to automatic identification in regard to land use pattern on the spatial scale. The method of the paper is that based on the fractal theory,the box counting dimension model is improved in three aspects by means of the second derivative. The results show that firstly, automatic identification of the optimal spatial scales of land use pattern is 900 m×900 m, indicating that the land use pattern has obvious fractal characteristics in the study area. Secondly, land use pattern presented as space forces among different land use types, the extent of equal distribution and internal association, showing the complexity of spatial relationships in study area in 1991, 2003 and 2015. Thirdly, the spatial relationships of land use pattern had the obvious variation, suggesting the differences of competition intensity about land use pattern in study area from 1991 to 2015. It concludes that the improved box-counting model can realize the automatic recognition of spatial scale of land use pattern; the automatic dimensioning model of box-counting dimension is established to determine the land use pattern of the study area; and the independent minimum geographic unit of the spatial scale is 900 m×900 m which is the optimum scale for spatial analysis.