基于CLUE-S模型的榆神府地区土地利用变化模拟研究

发布时间：2019-07-05 21:23

【摘要】：土地利用／土地覆盖变化(LUCC)旨在揭示人类生产系统与地球环境系统之间的相互作用关系,其作用机理可以通过分析和模拟土地利用类型的变化来进行解读。土地利用变化是一个复杂的地理过程,它既包含自然演变过程,也包含人类发展过程,基于引发两种过程的驱动力,模拟土地利用变化格局,从而明确区域土地利用变化的成因、过程以及变化趋势,是目前国际上研究LUCC的重要手段。本文采用目前国际上使用较为广泛的CLUE-S模型,以资源型城市榆林市的榆神府地区为研究区,首先应用趋势预测法、灰色GM(1,1)预测法以及线性回归预测法对榆神府地区2000-2010年的土地利用需求量进行拟合,并对应各地类选取误差最小的方法模拟2010-2020年的土地利用需求量,然后应用Logistic二元回归模型计算出研究区各地类与自然、经济、社会、资源等各方面驱动因子之间的回归方程,运用ROC曲线检验多尺度下的回归精度,得出模拟效果最好的研究尺度,最后按照CLUE-S模型的输入规则,将文件输入模型,模拟出研究区2020年的土地利用空间格局。研究结果说明了该模型对于榆神府地区的土地利用空间格局模拟效果较好,该研究结果也可以为榆神府地区及类似资源开发地区的土地利用规划决策提供更为科学的依据。论文从土地利用变化与驱动因素作用过程入手,模拟土地利用变化的空间格局,从而探究土地利用变化的内在机理,主要研究工作及结论如下：(1)对于模型中的土地需求模块,使用趋势预测法、灰色GM(1,1)预测法以及线性回归预测法对研究区的土地需求量进行模拟和预测,确定了2010-2020年的土地需求预测量,结果表明,对于研究区的土地需求量数据,耕地、草地、城乡居民建设用地、未利用土地使用线性回归模型进行预测效果更好,而水域使用灰色GM(1,1)预测模型,林地使用二次多项式进行预测效果更好。(2)利用解译后的土地利用图,提取对土地利用变化有影响的驱动因素,包括农田、河流、水库、道路、工矿区、植被、城镇和村庄等因素,再结合DEM中的坡度、坡向以及社会经济数据中的人口、GDP等因素,通过二元L ogistic回归分析,得出各地类的ROC检验结果较为理想,说明驱动因素与地类之间的相关性较高,拟合效果较好。(3)本文运用多尺度分析法计算研究区的土地利用模拟精度,结果表明,各尺度下ROC值均大于0.8,处于较高精度,说明驱动因素的选取可以有效的模拟土地利用变化情况,并且对于榆神府研究区,尺度在500*500以下均可获得精度较高的回归结果。(4)进一步比照模拟结果与解译图,发现存在一定的位移偏差,从而明确了总增减变化量与总转移变化量两个概念,从空间格局上明晰了土地利用变化的内在机理,结果表明,CLUE-S模型能够更好的消除由于解译误差引起的位移偏差,实现更为精确的模拟结果。(5)从模拟结果上看,耕地、草地、林地、水域、城乡居民建设用地、未利用地等地类在模拟结果上的空间布局符合资源型城市发展特点,同时,也存在着一定的问题,对此,提出以下三点建议：一是大力推进土地整治工作；二是科学开展土地开发研究；三是积极完善生态安全建设。
文内图片：

图片说明：图３－１输神府地区不意图逡逑该地区北部则隶属毛乌素沙漠南缘的风沙草滩区，，而经过治理后的沙滩地植被遍逡逑布、郁郁葱葱；海子（湖泊）星罗棋布，其中中小地貌类型多样，有风沙滩地貌、沙逡逑
[Abstract]:Land-use/ land cover change (LUCC) is designed to reveal the interaction between the human production system and the Earth's environmental system, and its mechanism of action can be interpreted by analyzing and simulating the change of land-use types. The land-use change is a complex geographical process, which includes both the natural evolution process and the human development process, and based on the driving force of the two processes, the land-use change pattern is simulated, so that the cause, the process and the changing trend of the regional land-use change are clearly defined, It is an important means to study the LUCC at present. In this paper, a wide variety of CLUE-S models are used in this paper to study the land-use demand in the elm-Shenfu area from 2000 to 2010 by using the trend prediction method, the grey GM (1,1) prediction method and the linear regression prediction method, using the CLUE-S model which is widely used at present. and the regression equation between the various parts of the study area and the driving factors such as natural, economic, social, resource and the like is calculated by using the logistic binary regression model, According to the input rule of CLUE-S model, the land-use spatial pattern of the research area by 2020 is simulated by using the ROC curve to check the regression accuracy of the multi-scale. The results of the study show that the model can provide a more scientific basis for the land-use planning decision-making in the region of the Ulmus and the similar resources. This paper starts with the process of land use change and driving factors, and simulates the spatial pattern of land use change, so as to explore the internal mechanism of land use change, and the main research work and conclusion are as follows: (1) For the land demand module in the model, the trend prediction method is used, The grey GM (1,1) prediction method and the linear regression prediction method are used to simulate and forecast the land demand of the study area, and the land demand forecast quantity for 2010-2020 is determined. The results show that the land demand data, the cultivated land, the grassland, the urban and rural residents construction land for the study area, The prediction effect is better without using the linear regression model of the land, and the grey GM (1,1) prediction model is used in the water area, and the prediction effect of the forest land using the quadratic polynomial is better. (2) The driving factors, including farmland, river, reservoir, road, mining area, vegetation, town and village, are extracted by using the interpreted land-use map, and then combined with the slope, slope and the population in the social and economic data in the DEM, Based on the two-way logistic regression analysis, the ROC test results of all regions are ideal, and the correlation between the driving factors and the ground class is high, and the fitting effect is good. (3) The method of multi-scale analysis is used to calculate the accuracy of the land-use simulation in the study area. The results show that the ROC values of each scale are more than 0.8, and the results show that the selection of the driving factors can effectively simulate the land-use change. Regression results with higher accuracy can be obtained at the scale of 500 * 500. and (4) according to the simulation result and the interpretation map, a certain displacement deviation is found, so that the two concepts of the total increase/ decrease change amount and the total transfer change amount are defined, the intrinsic mechanism of the land use change is clearly defined from the spatial pattern, the result is that, The CLUE-S model can better eliminate the displacement deviation caused by the interpretation error and realize the more accurate simulation result. (5) From the simulation results, the spatial layout of cultivated land, grassland, forest land, water area, the construction land of urban and rural residents and unused land in the simulation results is in accordance with the development characteristics of resource-based cities. At the same time, there are some problems. In this regard, the following three suggestions are put forward: The first is to push forward the land regulation work; secondly, to carry out the land development research; and thirdly, to actively improve the ecological security construction.
【学位授予单位】：西北大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F301

【相似文献】