【作者】 陈四清；

【导师】 刘纪远； 庄大方；

【作者基本信息】 中国科学院研究生院（遥感应用研究所） ， 地图学与地理信息系统， 2002， 博士

【摘要】 土地利用/土地覆盖变化和碳循环问题是与当今人类生存和发展休戚相关的全球气候变化研究中的热点问题。内蒙古锡林河流域草原是我国北方干旱/半干旱温带草原的重要组成部分，地处我国草原从东部半湿润草甸草原区向西北干旱荒漠和山地草原区的过渡地带，在研究土地利用/土地覆盖动态变化、草原退化、草原保护和草地资源可持续利用等方面有特殊的意义。本文运用遥感、GIS技术和生态模型方法，以内蒙古锡林河流域草原生态系统为例，研究我国北方干旱/半干旱草原地区土地利用/土地覆盖变化和碳循环问题。文章首先对内蒙古锡林河流域四个时期的Landsat TM/ETM+影像进行土地利用/土地覆盖分类、成图；通过对比，分析了锡林河流域近20年的来的土地利用/土地覆盖变化；并进一步运用GIS方法研究了锡林河流域草地退化的演化路径；最后运用CENTURY模型模拟了内蒙古锡林河流域大针茅草原、羊草草原的碳循环过程，绘制了其碳循环模式图；并进一步分析了锡林河流域典型草原生态系统对大气碳库的源/汇功能。全文的主要结论如下： 1)锡林河流域土地利用/土地覆盖分类。分别用1978、1991、1997、2000年Landsat TM/ETM+数据对内蒙古锡林河流域作了土地利用/土地覆盖分类和分类精度分析。四期影像的总分类精度分别是1987年的81.0％、1991年的81.7％、1997年的80.1％和2000年的78.2％。分别对四个时期的土地利用/土地覆盖分类图进行了去云、去阴影等优化处理，便于分析土地利用/土地覆盖动态变化。 2)锡林河流域土地利用/土地覆盖变化。锡林河流域20年来土地利用/土地覆盖变化的主要特征为草甸草原、典型草原面积的大幅减少和荒漠草原、农田和沙漠化土地面积的大幅增加及城镇的扩张。其中面积增加最大的是荒漠草原，增加了2328km~2；相当于1987年荒漠草原面积的56％。农田和城镇面积呈逐年增大的趋势，分别从1987年的114.3km~2和25.2km~2增加到2000年的332.1km~2和43.6km~2。面积减少最多的是羊草+丛生禾草、羊草+杂类草等优良高产温带典型草原类型，20年来面积共减少2040km~2。草甸草原面积亦呈逐年减少的趋势，从1987年的1103km~2减少到2000年375km~2，面积减少了65.9％。优质高产草原类型面积的减少，以及荒漠类低产劣质草原面积的增加，预示我国北方优良天然草场资源正在急剧退化。非草原的土地覆盖类型增加了62.5％，这其中主要是农田、沙漠化和城市化面积的增加，显示了以草地开垦、过度放牧和城市化等人类活动对草地生态系统的巨大影响。 3)锡林河流域草地系统的退化演化分析。从遥感影像的分类结果入手，用GIS方法分析了锡林河流域草地系统的退化演化路径。分析结果显示，锡林河流域在过度放牧压力下，经历了和正在经历着由草甸草原(线叶菊草原、贝加尔针茅草原)向典型草原(羊草+杂类草草原、羊草+丛生草草原、羊草+大针茅草原、大针茅+丛生草草原、羊草+冷蒿草原)、荒漠草原(克氏针茅草原、冷蒿草原)和沙化地盐碱地(碱斑地)甚至裸地的退化过程。这再次证明了人类活动对脆弱的草地系统巨大影响，为草地工作者提供了一种新的研究草地退化演化的方法。 4) CEN刊RY模型对锡林河流域典型草原碳循环的模拟。运用CENTURY生物地球化学模型模拟锡林河流域大针茅草原和羊草草原的碳循环过程。结果显示，地上生物量的模拟结果与实际观测值相符较好。线性回归分析表明，大针茅草原和羊草草原观测值(y)和模拟值(x)存在较好的相关关系，其回归方程和系数分别为y=0 .ss76x+12.59(rZ一0.503，p一0.0001)和y=o.o96x+13.805(产一0.751，p-0.0001)。在Century模型模拟的各碳库中，植被碳库以地下生物量碳库为最大。土壤碳库以缓性碳库为最大。以2000年的模拟结果为例，就大针茅草原而言，植被碳库和土壤碳库以缓性碳库分别683.28岁衬，1564.289/m2，就羊草草原而言，二者分别为69s.ssg/mZ和1441.269/扩。 5)锡林河流域典型草原碳循环研究。运用CENTURY模型模拟的锡林河流域草原各地上、地下碳库大小及碳素在各碳库之间流动过程中伴随的CO:的排放形成的碳损失量，建立了大针茅草原和羊草草原的碳循环模式图。其中NPP是植被一土壤系统从大气的碳输入量，各种CO:的呼吸消耗是植被一土壤系统向大气排放CO:的方式。NPP和呼吸作用的碳排放差即是草原生态系统对大气碳库的净碳源或碳汇。2000年羊草草原的NPP为274.839/m2，而各呼吸作用的co:排放量之和为159.979/m2，羊草草原对大气碳库而言是碳汇，其大小为116.86 g/mZ。大针茅草原2000年NPP为273 .01留耐，而各呼吸作用的coZ排放量之和为1 60.85留扩，大针茅草原对大气碳库而言是碳汇是1 12.169/m2。 6)锡林河流域典型草原对大气碳库的源/汇功能。根据1987、1991、1997、2000年的遥感影像的分类结果，计算了相应年份锡林河流域大针茅草原和羊草草原的总碳源(汇)大小。1987、1991和2000年锡林河流域羊草草原和大针茅草原相对于大气碳库均为碳汇，羊草草原净碳汇分别为71.85XI护gC、66.58xl护gc、so.3lxlo6ge;大针茅草原净碳汇分别为30，63Xlo6ge、69.3oXlo6ge、sl.59xIO6gC，而199更多还原

【Abstract】 Land-use/land-cover change and carbon cycle form two of the most substantial aspects of global climate change on both global and regional scales. It is expected that changing land use / land cover pattern will be one of the driving forces of environment change at regional scale superimposed on the natural changes at the time scale from decade to century. At the same time, the carbon cycle at mid-latitudes of North Hemisphere still remains unknown, which leads people to nowhere in gaining a deep understanding of the mechanism of global change. In Xilin River Basin, Inner Mongolia, which is right located at the mid-latitudes of North Hemisphere, the high speed of social-economic development shows high rate and strong intensity to land use / land cover change in the past two decades.In this paper, remote sensing, GIS and ecological modeling techniques were combined to study the land use / land cover change and carbon cycle of Xilin River Basin. The land use/land cover change of Xilin River Basin in the past two decades was investigated through land use / land cover classification of multi-temporal Landsat TM/ETM+ images. Century model was used to simulate the carbon cycle of the typical grassland in Xilin River Basin. The main conclusions of this study were as follows:1) Land-use/land-cover classification of Xilin River Basin. 4 sets (each set contains 2 scenes) of Landsat TM/ETM+ images acquired on Jul.31, 1987, Aug. 11, 1991, Sep. 27, 1997 and May 23, 2000, respectively, were used to classify land use / land cover of Xilin River Basin, Inner Mongolia. The overall classification accuracy was 81.0% for 1987, 81.7% for 1991, 80.1% for 1997 and 78.2% for 2000. The classification maps were optimized for later land use / cover change analysis.2) Land-use/land-cover change of Xilin River Basin. The main characteristics of land use/land cover change in Xilin River Basin over the past two decades were significant decrease in area of meadow grassland, temperate grassland vs. significant increase in area of cropland, desert grassland, urban area and desertilized land. For the latter, the desert grassland had the biggest increase in area, i.e. 2328 km2, equal to 56% of the total area of desert grassland in 1987. The cropland and urban area had increased from 114.3 km2 and 25.2 km2in 1987 to 332.1 km2’and 43.6 km2in 2000, respectively. The A. lymus + bunchgrass steppe, A. lymus + forbs steppe had the greatest decrease in area, i.e. 2040 km2. The decrease in area of meadow grassland and temperate grassland with high production vs. increase in area of desert grassland with low production indicated a dramatic degradation of the grassland in North China and the overwhelming impacts of human activities superimposed on the natural grassland ecosystem.3) Evolution route of degradation of the grassland ecosystem in Xilin River Basin. The degradation evolution route of the grassland ecosystem in Xilin River Basin was analyzed utilizing GIS techniques from the view of the point of land use/land coverclassification of multi-temporal Landsat TM/ETM+ images. Driven by overgrazing, the grassland ecosystem in Xilin River Basin, Inner Mongolia had undergone and was undergoing degradation evolution; the evolution route was from meadow grassland (F. sibiricum Steppe, S. bacalensis Steppe), via temperate grassland (A. lymus + bunchgrass Steppe, A. lymus + forbs Steppe, A. lymus + S. grandis Steppe, S. grandis + bunchgrass Steppe, S. grandis + forbs Steppe and A. lymus+ Ar. Frigida Steppe) to desert grassland (S. krylavii Steppe and Ar. frigida Steppe).4) Simulating the carbon cycle of the grassland ecosystem in Xilin River Basin with CENTURY model. The carbon cycle of the grassland ecosystem in Xilin River Basin was simulated with CENTURY model, using monthly precipitation accumulator, monthly mean minimum temperature, monthly mean maximum temperature, soil PH, soil texture, etc. as basic input variables. The model was evaluated and validated by comparing one output variable of the model, agcacc, i.e. aboveground live bio更多还原