IMHE OpenIR  > 山区发展研究中心
基于农户-地块尺度的山区耕地利用ABM模型研究
Alternative TitleStudying plot-scale agricultural land use change in mountain areas by using agent-based modeling and simulation
李明
Subtype博士
Thesis Advisor王玉宽
2018
Degree Grantor中国科学院大学
Place of Conferral北京
Degree Discipline自然地理学
Keyword土地利用强度 耕地扰动强度 农户行为 农村环境保护 ABM
Abstract土地利用变化既是全球变化的重要组成部分,也是全球环境变化的主要原因。然而,土地变化科学发展至今,存在明显的重数量变化、轻质量变化研究的缺陷,而这两方面的变化均可能产生明显的生态环境效应。对于我国山区农村而言,受农村经济改革、城镇化发展、退耕还林政策和土地流转政策等因素影响,农户的生计策略和生产行为发生了显著改变,导致耕地的利用方式和利用强度发生了明显变化,这是引起农村环境变化的直接原因。本文主要围绕山区耕地“种不种、种什么、怎么种,以及其环境效应如何”这一系列问题,基于智能体模型(Agent based model, ABM)开展耕地利用方式变化模拟研究,并结合ABM和耕地扰动强度评估,模拟耕地利用强度的变化趋势,以期为山区农村环境治理和山区可持续发展提供科学参考。主要内容和结论如下:(1)运用多维偏好分析和聚类分析方法,开展了基于耕地利用行为影响因素的农户类型定量划分研究。结果表明:家庭劳动力整体受教育程度、耕地面积、劳均耕地面积、退耕还林地占比、家庭稳定性现金收入、非农收入占比、畜牧业收入占比和补贴占比等8个指标是影响农户行为的关键变量。基于聚类分析结果和分类树方法,分别依据非农收入占比(≥90%和50~90%)、养殖占比(≥50%)、耕地面积(≥30亩)和补贴占比(≥50%),将研究区的农户划分为非农型、兼业型、养殖型、种植大户、补贴型和传统型共6种类型。最后对比分析了上述6种类型农户的耕地利用行为差异。(2)为了解决山区耕地“种不种”和“种什么”的问题,本研究从农户行为的视角,通过构建具有“自下而上”建模机制的ABM模型,模拟和预测未来不同发展情境下耕地利用方式的变化。本模型包括模型初始化、农户类型判断、耕地弃耕预测和作物选择决策等子模块,共涉及18个关键步骤。智能体的学习过程基于经历-权重吸引模型(EWA)模拟;农户的弃耕行为通过综合考虑各类型农户的弃耕意愿和各地块的被弃耕概率模拟;农户的作物选择行为基于信念-愿望-意图(BDI)开放性决策框架模拟。(3)以宝兴县咔日村为例,对2000~2015年的耕地弃耕过程和地块作物演变过程进行模拟试验,并选取多个指标对模型结果进行检验。结果显示,本模型对弃耕地数量和耕地地块作物的播种面积模拟精度较高(误差分别为1%和18%),但在空间格局上存在一定的差异。这主要是由于模型对耕地权属的分配与现实之间存在差异。基于率定后的模型,分别设置了历史发展、生态保护和适度发展3个未来不同发展情景,模拟耕地利用方式的变化趋势。结果表明,生态保护情景下,农户的生计策略由种养混合型转变为维持经济作物种植型;相对于历史发展和适度发展情景,退耕还林地面积增加1.9%,主要作物类型由饲料作物转变为经济作物,牲畜数量明显减少(95.0%),且农户的生计质量能够得到明显改善。(4)为了量化山区耕地“怎么种及其环境效应如何”,本研究创新性地提出耕地扰动强度(CLDI)的概念,并给出地块尺度的定量评估方法。本文将耕地扰动类型区分为物理扰动和化学扰动,其对应的主要环境问题分别为水土流失和面源污染。对我国西南山区的宝兴县、普格县的耕地扰动强度研究发现:耕地受到的化学扰动强度(d_C)明显高于物理扰动强度(d_P),中高等d_C的地块占60.5%,中高等d_P的地块占43.4%;以非农活动为主的区域,耕地的d_C更高,而传统农耕区耕地的d_P更高;翻地阶段和施肥阶段分别对耕地的d_P和d_C的贡献程度最大(分别为69.5%和49.1%);不同作物类型和不同轮作模式的耕地地块受到的理化扰动差异显著;耕地扰动强度显著受农户和地块层面因素的共同影响,农户参与非农活动导致耕地的d_C增大d_P减小,距农户150 m和800 m的地块受到的d_P和d_C最高。(5)最后,结合ABM模型研究和耕地扰动强度研究结果发现,历史发展情景下,研究区有69.1%的耕地地块受到中-高等强度的物理扰动,23.5%的耕地受到中-高等强度的化学扰动。而在生态保护情景下,仅有14.3%的耕地受到中等程度的物理扰动,无耕地地块受到中-高等强度的化学扰动。本研究表明,对于当前的西南山区农村环境治理工作而言,水土流失和面源污染防治的重点区分别为距农户150 m和800 m范围的耕地地块,最有效的非工程防治措施为推广少耕或免耕法、施用农家肥和新型高效有机复合肥。对于未来的农村环境治理工作,除上述建议外,政府加强对部分环境友好型经济作物的引导,合理调控种植业和畜牧业的比重,将是减缓区域环境风险和实现山区可持续发展的有效途径。
Other AbstractAs an important part of global change, land use change (LUC) have also been recognized as major causes of global environmental change. The quality and the quantity aspect of land use change were regarded as the most important components of LUC, with the former mainly expressed as the land-use/ land-cover change (LUCC) and the latter expressed as the land use intensity (LUI), respectively. Both aspects reflect the differences and changes of human land use decision making behavior. In rural China, the cropland assignment strategy follows the principle of the collocation of cropland plots with both combined cropland location and soil quality, which resulted in extremely dispersed cropland plots for each household. On the other hand, rural China has experienced a series of policy reforms including rural economic reform, urbanization, the Sloping Land Conversion Program (SLCP) and land transfer policy. Both aspects resulted in the differences and changes of human land use decision behavior, which determined the spatial pattern and trends of rural environmental issues (such as soil erosion, non-point source pollution, and arable land degradation). For rural environmental protection, the critical works are to figuring out the following problems: whether or not the cropland plots will be planted? What type of crops will be planted? How to plant? And what are the potential environmental effects?The main purpose of this study was to study plot-scale agricultural land use change in mountain areas by using an agent-based modeling and simulation approach (ABM). The main research contents and research conclusions are as follows. (1) Since farmer's decision-making objectives and land use behavior differed greatly among household types, ABM studies are usually based on agent typology analysis. In this study, farm households were classified into six types based on the screened impacting factors of farming practices by using the CAPTCA, cluster analysis and classification-tree methods. Critical variables used for the typology analysis including: the average education level of household agricultural labor force, cropland total area, cultivated land area per agricultural labor, SLCP land ratio, cash income, non-farm income ratio, livestock income ratio and subsidy income ratio. Farm households in southwest China were classified into six types namely non-farm household, concurrent-business household, livestock-dependency household, scale plantation household, subsidy-dependency household, and traditional household. These household types were obtained based on the classification-tree method according to the criterion of non-farm income ratio (≥90% and 50~90%), livestock income ratio(≥50%), cropland area (≥30 acres) and subsidy income ratio(≥50%), respectively. (2) For the purpose of figuring out “whether or not the cropland plots will be planted? What type of crops willed be planted?”, this study developed an ABM model to simulate plot-scale agricultural land use changing under different development scenarios in the future. The ABM contains 18 key steps, including model initialization, household type identification, cropland abandonment simulation, farmers’ crop choice behavior simulation, etc. The process of farmers’ livelihood strategy decision-making were simulated by using the Experience-Weighted Attraction model (EWA). In addition, farmers’ cropland abandonment behavior were simulated by the comprehensive considering of household types and plot-level factors by using binary logistic regression model. Furthermore, farmers’ crops selection behavior at the plot-level was simulated by the Beliefs–Desires–Intentions model (BDI). The ABM model was developed based on the Repast platform, with the spatial resolution of cropland plot level. The cropland plot size was allocated based on the per capita cultivated area. (3) The Kari village located in southwest China was selected as the case study area. It covers an area of 3000 ha. The ABM was simulated for the period of 2000 to 2014, and validated in 2015. It was prediction for the period of 2016 to 2030. Results show that the ABM simulation has high precision rate for the amount of abandoned farmland and the sown area of certain crops (errors are 1% and 18%, respectively), however there is a certain error for the spatial pattern simulation. It is mainly because of cropland ownership was randomly assigned according to the location of households, which is different from reality. Based on the calibrated model, this study simulated the cropland change trends under three development scenarios in the near future, namely the historical development scenario, the ecological conservation scenario, and the moderate development scenario. Results show that under the ecological conservation scenario, the leading household livelihood strategy will be changed from aquaculture to economic crops planting. Comparing with the historical development scenario and the moderate development scenario, the SLCP land area will be increased by 1.9%. The main crop types will be changed from forage crops to cash crops, and the number of livestock will be significantly reduced (95.0%). The livelihood quality of local farmers will also be significantly improved. (4) To figure out the question “How to plant? What are the potential environmental effects”, this study presents a methodology for the development of a novel index, specifically targeted at the assessment of plot-scale cropland disturbance intensity (CLDI). Different farming practices during each crop management stage that potentially induce both chemical disturbance (d_C) and physical disturbance (d_P) were systematically evaluated. The rough set method was utilized to avoid subjectivity during weight allocation. Furthermore, an ordered logit model was applied to analyze critical factors that affect CLDI as well as to identify potential areas of rural environmental protection in the mountainous regions of southwestern China. Results indicate that cropland plots in study area mainly suffered from medium- to high-levels of disturbances, with higher d_C than d_P. For catchments with popular off-farm work and broad participation in the SLCP (e.g. Xiangxing), cropland plots received the most intensive chemical disturbance. For traditional farming areas (see Kari), cropland plots received more intensive physical disturbance. The model results show that both household and plot level variables significantly influenced the CLDI (R2 = 0.65, P < 0.01). At the household level, critical variables that positively affected the CLDI includes the scale of the agricultural laborer, cash income, and cultivated land area per agricultural laborer. The intensity of chemical disturbance increased with increasing off-farm work. At the plot level, distance from the household negatively impacted CLDI, while the distance to the nearest forest posed a positive influence. (5) Combining the results of the ABM model with the CLDI assessment, under the historical development scenario, 69.1% of the cropland plots in the study area will be subjected to medium- to high-level of physical disturbance intensity, and 23.5% of the cropland plots will be suffered from medium- to high-level of chemical disturbance. However, under the ecological conservation scenario, only 14.3% of the cropland will be disturbed moderately, and no cropland plots will be suffered from medium- to high- level of chemical disturbance. Based on the results of the ABM and the CLDI, this study proposed some suggestions for the rural environmental protection and sustainable development in the mountainous regions of southwestern China. Currently, the application of reduced tillage, farm manure, and long-lasting organic fertilizers represents effective methods in the study area. Besides, for the soil erosion reduction and non-point source pollution control in the study area, we suggest to prioritize cropland plots within a radial distance of 150 and 800 meters from households, respectively. Likewise, the promotion of some environment- friendly cash crops, as well as the rationally controlling of the proportion of planting and animal husbandry could be effective ways for achieving the win-win goal of environment protection and economic growth in the near future. 
Pages146
Language中文
Document Type学位论文
Identifierhttp://ir.imde.ac.cn/handle/131551/24795
Collection山区发展研究中心
Affiliation中国科学院成都山地灾害与环境研究所
First Author Affilication中国科学院水利部成都山地灾害与环境研究所
Recommended Citation
GB/T 7714
李明. 基于农户-地块尺度的山区耕地利用ABM模型研究[D]. 北京. 中国科学院大学,2018.
Files in This Item:
File Name/Size DocType Version Access License
基于农户-地块尺度的山区耕地利用ABM模(5693KB)学位论文 开放获取CC BY-NC-SAView Application Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[李明]'s Articles
Baidu academic
Similar articles in Baidu academic
[李明]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[李明]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 基于农户-地块尺度的山区耕地利用ABM模型研究.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.