【作者】 焦继宗；

【导师】 王乃昂；

【作者基本信息】 兰州大学 ， 人文地理学， 2012， 博士

【摘要】 我国西北干旱地区生态环境极其脆弱,民勤绿洲作为阻挡巴丹吉林和腾格里两大沙漠合拢的天然屏障,具有重要生态意义。多年来,随着人类活动的持续影响,民勤绿洲生态环境持续恶化,直观的表现就是土地利用／覆盖变化(Land use/cover change, LUCC)。主要表现为大量开垦土地造成天然植被的破坏并由此导致沙漠化,以及重灌轻排导致耕地土壤盐碱化、沙漠化等现象。本研究应用多种方法对1991-2010年间的LUCC进行分析。通过对TM和SPOT影像的融合提取高精度土地利用分类数据；利用遥感、GIS技术及多种土地利用动态变化参数模型揭示了民勤绿洲土地利用演变时空特点和规律；利用主成分分析法及多元统计回归探究了引起土地利用时空演变的社会经济因素及原因；通过建模及模拟技术手段,预测了民勤绿洲土地利用结构未来的空间格局及演变趋势。本研究初步揭示了民勤绿洲生态环境问题及其根源所在,为防止民勤成为“第二个罗布泊”相关政策的制定提供理论依据。本文主要研究结论如下：1)1991年-2010年间,耕地、未利用地、盐碱地相互转换频繁,其中耕地变化率最高,耕地转换为未利用地主要坝区明显,耕地转换为盐碱地在湖区显著。2000年以前,耕地变化率、土地利用动态度在湖区较高,坝区较低,2000年后,相关指数在坝区继续增高,并且在湖区有这大幅增加。2000年前,土地利用程度综合指数(人类活动影响程度)坝区数值较高,湖区略小；2000年后,该指数在坝区继续增高,并且在湖区也有大幅增加。湖区土地利用类型多样性高于坝区；土地利用类型空间重心呈分散趋势,耕地、居民地向坝区方向偏移明显,未利用地、盐碱地向湖区方向偏移显著。多种景观指数分析发现：绿洲区域土地利用类型多样化,破碎度和不规则性提高。总体来看,自1991年以来,民勤绿洲受人类活动影响非常显著,由点向面逐渐扩展；同时土地覆盖状况进一步恶化,生态环境脆弱性不断加剧。2)利用相关分析及主成分分析法,以受人类活动影响的44个民勤绿洲社会经济指标作为候选驱动因子,通过计算得到影响绿洲土地利用变化的4个主成分：经济发展、农业技术进步、贫富状况及人口变化。建立土地利用类型与主成分之间的多元回归方程,定量分析了土地利用变化的诱因：耕地的变化与经济发展、农业技术进度及人口变化呈正相关；与贫富状况呈的负相关。未利用地与经济发展呈负相关,与其他主成分呈正相关；盐碱地变化与人口变化呈负相关,与其他主成分呈正相关。其他土地利用类型也在不同程度上受四个主成分的影响并且相关系数不同。3)一方面利用数学建模方法,开发了基于Matlab的灰色—马尔科夫程序模块,实证数据检验结果分析说明该模块较Gray预测数量模拟误差率较小,符合模拟精度要求；另一方面通过多尺度的土地利用变化与驱动力之间的二元Logistic回归分析发现：在200米时,其ROC拟合检验值最高；最后利用Autologistic模型进行自相关驱动力计算,结果表明ROC拟合检验较Logistic回归有明显的提高。应用以上多种方法对CLUE-S模型进行改进,然后以1991年土地利用为初始数据,模拟2010年的土地利用及空间格局。模拟结果实证分析表明：新构建的混合模型不仅提高了模拟精度,同时其模拟结果的位置差异率下降、数量一致率上升,明显改善了模拟精度和效果。4)应用改进后的混合模型,设置了3种情景模拟,分析了未来民勤土地利用变化情况：(1)继续1991年-2010年的发展模式和速度：耕地面积将持续大量增加,未利用地进一步减少,大部分耕地的增加由未利用地转换而来；(2)实施关井压田的政策：林地和草地有一定的增加,盐碱化趋势有一定的减缓,但城镇居民地仍然保持快速增长趋势,未利用地变化趋于平缓；(3)加强退耕还林还草政策：林地和草地大幅增加,耕地明显下降,未利用地降幅和盐碱地增幅趋势减缓。情景1下,由于耕地具有易盐碱化和沙尘化的不稳定性,大量的耕地会造成严重的生态后果和灾难,务必要引起相关部门的重视。情景2下,总体生态环境恶化趋势较第一种情景有一定的遏制和改善,但改善力度有限。情景3下,生态环境明显趋于好转。因此,针对民勤绿洲生态环境恶化现状,采取一定的生态环境保护和社会经济措施显得十分必要。建议严格控制耕地的增加,坚决贯彻实施关井压田的政策。大力推广节水灌溉,尽可能减少机井灌溉,减少土地的盐碱化、沙漠化。继续实施生态移民工程,减少人类活动对土地利用变化的影响,使绿洲土地利用／覆盖变化走上良性发展的道路。更多还原

【Abstract】 Arid areas of northwest China ecological environment is extremely fragile, Minqin oasis as a barrier of the Badain Jaran and the Tengger desert to stop it merger is very important. In recent years, with the continuing impact of human activities in Minqin oasis ecological environment continued to deteriorate, mainly for a large number of land reclamation caused the destruction of natural vegetation which led to the desertification and pay attention Irrigation but ignore Drain led to soil Salinization. land use/cover change (LUCC) is intuitive show of ecological environment change in Minqin oasis. In this study, by20years of LUCC analysis between1991-2010revealed the source of the Minqin oasis ecological environment problem. TM and SPOT Image fusion based on wavelet theory to extract high-precision land-use classification data; Reveals the Minqin oasis spatial and temporal Rules and Features of land use changing by using of remote sensing and GIS technology and a variety of land use dynamic change parameters; Explore the socio-economic factors and reason that caused the land-use spatial and temporal changing by using principal component the analysis and multivariate statistical regression. Predict the future trend of evolution in Minqin oasis land use and spatial pattern by using modeling and simulation techniques. The results of this study provide a theoretical basis for "policy-making that prevent the Minqin become second Lop Nur.The main conclusions of this study as follows:Both arable land, unused land and saline land conversion frequent between1991to2000; The rate of change of arable land the highest in variety of land types, its mainly focus on Dam area before2000but focus on Lakes area after2000. Dynamic degree of land changing focus on Lakes region before2000, but concentrate on Dam area after2000meantime its Continue increasing in lakes region. The year2000, composite index of the land use degree (human activities affect degree) is very obvious in the dam area and relatively weak in Lakes area before2000, in addition to the continues to strengthen in Dam, Lake area increased significantly after2000. Lakes diversity of land-use type is higher than dam area; space center of gravity of land use type was scattered trend, Arable land moving direct to Lake and unused land and residents moving to county direction. The diversity, fragmentation and irregular increased in Minqin oasis in the view of variety of Landscape index, so these changes seriously affect the security of Minqin oasis land use and ecological environment. Overall, Minqin oasis affected by human activities is very significant from the point to the surface gradually extended since1991. Land cover changes in further deterioration of the fragile ecological environment is growing. Use44digitized socio-economic indicators as a candidate driving factors, calculated four principal component driving factors that affect land use change by use of correlation analysis and principal component analysis, these are economic development, technological advances in agriculture, rich or poor condition, the population changed. Quantitative analyze the reason that cause the change in land use by multiple regression model, it Positively associated with Economic development the progress of agricultural technology and population change, and negatively correlated with wealth status. By the calculate of driving factor basis is found that rivers, deserts, roads and urban have certain affect on arable land change, spatial factors is significant reason that caused the land use change.Developed Matlab gray-Markov program based on Mathematical modeling, test results by the actual data analysis showed that the number of simulation error rate was small, fully comply with the requirements of the simulation accuracy. Take advantage of Autologistic autocorrelation driving force analysis, combined with the CLUE-S model to simulate the2010land use and spatial pattern make the1991land use as the initial data. Further analysis revealed that the newly constructed hybrid model simulation results not only improve the simulation accuracy, the location of the simulation results differences was declined, the number of consistent rate was increased, it significantly improved the simulation results.Analyzed three kinds of scenario simulation of land use/cover change result in2030based on the Application of Improved Gray-Markov combination of CLUE-S model.to continue the development of the mode and speed in1991to2010, arable land further increase of6.5×104ha, unused land further reduced;2) the implementation of the policy of Close motor-pumped well and reduce arable land. a certain increase of woodland and grassland, a certain reduction in the trend of salinization, but the settlements continues to grow, without use of change to flatten;3) Reforestation:woodland and grassland will be a significant increase in arable land dropped significantly, the use of land and saline significant trend. Under Scenario1, due to the instability of the arable land, a lot of arable land will result in extremely serious ecological consequences and disaster, be sure to arouse the attention of the government departments. Under Scenario2than the first scenario, the trend toward ecological deterioration containment and improve eco-environmental improvement efforts. Under Scenario3, the trend towards an improvement of ecological environment has clearly improved.更多还原