【作者】 黄强；

【导师】 黄海；

【作者基本信息】 重庆交通大学 ， 地图学与地理信息系统， 2019， 硕士

【摘要】 土地资源是一种稀缺且不可再生的资源,是维持人类生存和生态系统长期稳定的重要因素。但随着社会经济的快速发展,其供需矛盾也显现出来,目前普遍存在土地利用结构不合理,土地利用水平低,土地资源质量下降等严重问题。土地资源利用优化配置是促进土地资源集约利用与实现可持续发展的重要途径。当前,研究者已经研发了较多的土地利用优化配置模型。但是,总体上这些优化配置模型的自适应能力和动态调控能力较弱,优化配置的结果也往往以数量和空间的两维特征为主,不考虑优化配置结果实现的时序过程,不考虑土地利用系统中人与地、人与人之间的空间决策行为,难以保证了模型的合理性与实用性,导致土地利用优化配置结果在实现的过程中较为困难。所以本文利用多主体系统理论来认识和解释土地利用优化的一般规律,并且结合系统动力学、地理信息科学、计算机科学以及智能科学的相关理论和技术,建立可以清晰表达主体决策时空显性的土地利用优化模型,基于多主体系统与系统动力学的土地利用优化模型（Land use optimization model based on multi-agent system and system dynamics,LUOM）,以天津市蓟州区为研究区域,对区域土地利用进行优化分析,探索区域土地利用优化的内在机制和过程。进而为土地资源利用和管理政策的制定提供基础技术支持。其主要研究结果如下:1)本研究将土地利用系统视为复杂空间系统,从复杂性的视角审视区域土地利用优化问题。并将土地利用系统简化为政府主体、部门主体（农业部门、经济部门、社会部门、优化部门）、居民主体（城市土地、农村居民点、耕地、林地、草地和果园）三类。政府是单一主体,感知所有主体决策行为与宏观全局,制定的各种宏观总体规划,以最大空间效益为准分配土地资源。部门主体使用离散选择模型进行优化位置的选择,在追求部门效用最大化假设的前提下,指挥居民主体执行优化操作。居民主体则借助于动态随机效用模型计算每一个候选位置的适宜度值,并传递至上层主体。影响多主体系统的变量包括自然变量、环境变量、位置因素变量、交通可达性变量、社会经济变量5种共14个空间指标。2)研究为了达到更好的土地优化效果,使用农村居民点分类分区整理的策略。利用系统动力学建立了一个农村居民点分类模型。根据研究情景,城市扩张率、人口增长率等社会经济参数,将单一的农村居民点类型分为城镇化型农村居民点、重点发展型农村居民点、限制发展型农村居民点与迁弃型农村居民点四类。在多主体系统中依据农村居民点类型,生成对应的居民主体类型,并对于不同农村居民点主体给与不同的优化策略,实现农村居民点的精细整理。3)研究结果中,生态优先情景下区域有机碳储量由2015年的585.23万吨增长到842.15万吨,增加了43.9%,区域经济总产出由6.981×10¹⁰元增加到8.66×10¹⁰元,增加了24.1%,土地利用集约程度值由0.394增长到0.455,增加了15.4%。经济优先情景中,相较于生态优先情景,区域有机碳储量下降为730.95万吨,减少了13%,相较于基准年上涨了30%,区域经济总产出提升至1.09×10¹¹元,增加了25.9%,土地利用集约程度值为0.54,增加了18.7%。生态经济协调情景下区域有机碳储量、经济总产出与土地利用集约程度值分别为783.62万吨,1.05×10¹¹元与0.533,相较于原土地利用结构分别增加了33.9%、44.1%与35.4%,与情景1与情景2相比,生态用地与建筑用地的增长都较为缓和。4)土地利用优化过程中与敏感性分析中,三个情景的农村居民点用地适宜性折线均为震荡上扬,并在迭代中期开始分化,震荡反应出现实社会中土地利用优化的不确定与不稳定性,上扬也能折射出宏观调控带来的趋势变动。三个情景的聚集程度折线前期均为平滑下跌,后期震荡上扬,主要由于是在优化的前期需要拆除部分建筑用地,后期会对拆除的用地进行修复。景观格局指数的折线图中,斑块密度、形状指数、分维数前期上升后期下跌,斑块平均面积前期下跌后期上升。生态经济协调情景的形状指数折线走势与分维数不同,在迭代最后期形状指数是低于其他情景。考虑到形状指数和分维数分别代表的是斑块与等面积圆形和正方形的相似度。即在经济优先的情况下,模型优化的方向更贴近与正方形,使得运算中以栅格为基准的斑块更加聚集。在生态优先的情景下更贴近圆形,减少斑块棱角,缓和对于景观基质的破坏。更多还原

【Abstract】 Land resources are scarce and non-renewable resources and an important factor in maintaining human survival and long-term stability of ecosystems.However,with the rapid development of the social economy,the contradiction between supply and demand has also emerged.At present,there are widespread problems such as unreasonable land use structure,low land use level,and declining quality of land resources.The optimal allocation of land resource utilization is an important way to promote the intensive use of land resources and achieve sustainable development.Currently,researchers have developed more land use optimization configuration models.However,in general,the adaptive and dynamic control capabilities of these optimized configuration models are weak.the result of optimizing the configuration is also often a two-dimensional feature of quantity and space.The timing process implemented by optimizing the configuration results is not considered.The spatial decision-making behavior between people and land and between people in the land use system is not considered.It is difficult to ensure the rationality and practicability of the model,which makes the land use optimization configuration results more difficult in the process of implementation.Therefore,this paper uses multi-agent system theory to understand and explain the general law of land use change,and combines system dynamics,geographic information science,computer science and intelligent science related theories and techniques to establish a land use optimization model that can clearly express the subject decision（Multi-agent system for Land Use Optimization Allocation,LUOM）.Taking Yinzhou District of Tianjin as the research area,quantitative analysis of regional land use optimization was carried out to explore the internal mechanism and process of regional land use optimization.It provides basic technical support for the formulation of land resource utilization and management policies.The main findings are as follows:1)This study considers the land use system as a complex space system and examines the problem of regional land use optimization from the perspective of complexity.The land use system is simplified into three categories:government agent,department agent（agricultural department,economic department,social department,optimization department）and resident agent（urban land,rural residential area,cultivated land,forest land,grassland and orchard）.The government is a single subject,perceives all the decision-making behaviors of the main body and the macro-level overall situation,and formulates various macro-level多主体系统ter plans,and allocates land resources based on the maximum space efficiency.The departmental agent entity uses the discrete selection model to optimize the location selection,and under the premise of pursuing the assumption of departmental utility maximization,directs the resident agent entity to perform optimization operations.The resident agent entity calculates the suitability value of each candidate position by means of the dynamic random utility model.The variables affecting the multi-agent system include natural variables,environmental variables,location factor variables,traffic accessibility variables,and socio-economic variables.2)In order to achieve better land optimization results,the study used the strategy of sorting and arranging rural residential areas.A rural residential point classification model was established using system dynamics.According to the research situation,the social expansion parameters such as urban expansion rate and population growth rate,the single rural residential areas are divided into urbanized rural residential areas,key development rural residential areas,restricted development rural residential areas and abandoned rural areas.In the multi-agent system,according to the type of rural residential areas,the corresponding types of resident entities are generated,and different optimization strategies are given to different rural residential areas to achieve fine finishing optimal of rural residential areas.3)Among the research results,the regional organic carbon stocks increased from 5.8523million tons in 2015 to 8.421 million tons,an increase of 43.9%,and the total regional economic output increased from 6.981×10¹⁰ yuan to 8.66×10¹⁰ yuan,an increase of 24.1%,the value of land use intensification increased from 0.394 to 0.455,an increase of 15.4%.In the economic priority scenario,compared with the ecological priority scenario,the regional organic carbon stocks decreased to 7,305,500 tons,a decrease of 13%,which was 30%higher than the base year,and the regional economic output increased to 1.09×10¹¹ yuan.At 25.9%,the land use intensification level was 0.54,an increase of 18.7%.The regional organic carbon stocks,total economic output and land use intensification values under the eco-economic coordination scenario were 7,836,200 tons,1.05×10¹¹ yuan and 0.533,respectively,which increased by 33.9%,44.1%and 35.4%.4)In the land use optimization process,the suitability lines of the three scenarios are all oscillating and start to differentiate in the middle of the iteration.The shock response shows the uncertainty and instability of land use optimization in real society,and the rise can also reflect the trend changes brought about by macro-control.The aggregation degree of the three scenarios is a smooth decline in the early stage,and the volatility rises in the later period,mainly because some construction land needs to be demolished in the early stage of optimization,and the demolished land will be repaired later.In the line chart of the landscape pattern index,the plaque density,shape index,and fractal dimension decreased in the early stage and the late stage,and the average area of the plaque increased in the early stage.The shape index polyline trend of the eco-economic coordination scenario is different from the fractal dimension,and the shape index in the final period of the iteration is lower than other scenarios.Considering that the shape index and the fractal dimension respectively represent the similarity of the plaque to the equal area circle and square.That is,in the case of economic priority,the direction of model optimization is closer to the square,so that the plaques based on the grid are more concentrated.In the ecological priority scenario,it is closer to the circle,reducing the edges and corners,easing the damage to the landscape matrix.更多还原