【作者】 龚文峰；

【导师】 李凤日； 范文义； 叶荣华；

【作者基本信息】 东北林业大学 ， 森林经理学， 2007， 博士

【摘要】 川西北亚高山林区是我国云、冷杉暗针林集中分布区，也是天然林保护工程的重点实施区域和长江上游的生态屏障。本文以1985 TM、2005 SPOT5遥感图像、森林分布图、森林资源二类调查数据、地形图、土地利用规划图和外业调查数据和其它文字材料为基础，在遥感(RS)和地理信息系统(GIS)技术的支持下，结合数字高程模型(DEM)，应用景观生态学的相关原理及方法，从时间序列上和空间尺度上研究区域的景观格局及其变化，详细系统的分析了研究区域的景观生态结构、景观格局现状、动态变化和生态评价过程，在土地利用变化的基础上进行生态区划，提出景观生态分类，建立以生态恢复重建为目的区域景观优化配置，探索研究区域天然林景观生态恢复途径和方法，对我国天然林保护地区的生态系统恢复和管理具有重要意义。在此过程中，本文主要取得以下的成果和进展。(1)对研究区域的地形图进行校正、矢量化、错误修改、拼接，在ARC／INFO通过线性内插(LINEAR)和非线性内插(QUINTIC)中生成DEM，用检查点法和剖面分析法进行精度检验。结果表明：利用检查点的方法检验时，非线性插值精度较高；在利用剖面分析法时，任意方向上的误差较小，Y方向上的误差较大；地势陡峭的山区，非线性插值精度较高，在地势较为平坦的区域，两者精度相差不大。(2)在ArcGIS下，对DEM进行坡度、坡向的提取进行高程分带处理，研究区域的陡坡和平坡所占的比例较大，其次是斜坡。平坡主要集中在丘陵和河流地区，陡坡的比例稍高于平坡，表明了研究区域的地势比较陡峭。(3)对SPOT5遥感影像的处理中，利用研究区域的DEM数据对遥感影像实施正射纠正，很好地消除地形对遥感影像的影响。因此，对我国地形复杂山区林地而言，利用DEM数据对遥感影像进行正射纠正具有很重要的作用；对传感器波段特性、遥感影像统计特征分析、计算各波段的熵值、标准差和相关系数的基础上定量得出：研究区SPOT5多光谱以412的波段组合效果最佳。(4)借助于ERDAS IMAGE 9的AutoSync模块，把已经过几何精校正的SPOT5(10×10m)多光谱影像作为参考图像，对TM遥感数据进行地理坐标匹配和边缘匹配，大大提高了TM几何校正的精度，通过统计和计算各个波段的标准差和相关系数，计算OIF来确定最佳的波段组合，并对图像进行有监督分类和精度检验，分类结果满足生产需要。(5)不同景观类型在同一高程带的景观格局不同。随高程的增加，景观类型随着海拔的变化呈现出明显的垂直分异，在各高程带中，面积最大的3种景观依次为：〈1985〉裸崖地、高山草甸、高山灌丛(Ⅰ)—高山草甸、高山灌丛、森林(Ⅱ)—森林、高山草甸、高山灌丛(Ⅲ)—森林、草地、灌丛(Ⅳ)—森林、草地、灌丛(Ⅴ)—森林、草地、灌丛(Ⅵ)；〈2005〉裸崖地、高山灌丛、高山草甸(Ⅰ)—高山草甸、高山灌丛、森林(Ⅱ)—森林、草地、灌丛(Ⅲ)—森林、灌丛、草地(Ⅳ)—森林、草地、灌丛(Ⅴ)—森林、灌丛、草地(Ⅵ)。(6)随高程的增加，景观的多样性指数、优势度呈现一定的分布规律：多样性指数在第六高程带最大，在第四高程带最小，随后随着高程的增加，景观多样性指数有所增加，但是总体趋势随着高程的增加而减小，景观优势度分布趋势和景观多样性相反。(7)选取平均斑块面积、多样性指数、优势度、斑块密度四个景观指数，研究景观格局与坡度、坡向之间的相互关系。结果表明：对陡坡区域，所占面积比例上较大，斑块密度和多样性指数最小，斑块平均面积和景观优势度最大；缓坡区域的斑块密度、多样性指数最大，其斑块平均面积和景观优势度指数最小；阴坡区域的斑块密度最小，斑块平均面积最大，景观优势度和多样性指数较小；半阳坡斑块密度和景观优势度最大，斑块平均面积和多样性指数最小。(8)对重新分区后的高程、坡度和坡向3个地形因子与景观空间格局的相互关系进行分析。结果表明：景观格局的分布与海拔高度、坡度和坡向均有一定的相关性，农(弃)耕地多随高程和坡度的增加，分布的面积逐渐减少；有林地、灌木林地是阴陡坡中地的主要景观类型；草地是阳缓坡中地的主要景观类型。但是，各景观类型在坡向分布上差别不是太明显，由此说明地形对景观的宏观分布有一定的影响。(9)把格网样点提取的森林景观类型和DEM相叠加，在空间尺度上描述各森林景观的分布格局、地理位置或相对位置，在此基础上，对各景观要素在进行空间格局趋势分析。从而探讨森林景观分布格局的形成机制、分布规律及其空间格局的动态变化，为区域天然林生态系统的保育、恢复与可持续经营提供理论依据。(10)在水平趋势面分析：水平位置的变化对景观格局的演变产生了一定的系统影响，景观要素水平空间分布趋势在东西方向或南北方向上的地带分异不显著，景观要素水平分布趋势实质上反映了研究地区地形地势对景观格局的控制作用。(11)海拔和坡向趋势面上分析：在低海拔带上，阴坡上SI较低，且变化剧烈，该区域分布着桦木林、山杨等相对纯林的不稳定景观类型，处于景观演替的前期，演替指数相对较小；在阳坡上变化相对不大，该区域主要分布着云冷杉和桦木和的混交林景观，是演替过程中到达顶极景观格局的重要转折阶段；在高海拔地带上，阴坡的SI较大，该区域主要分布着云、冷杉林及其混交林景观，处于稳定结构状态，景观演替指数比较大；阳坡区域SI指数变化比较大，只有少数耐干早耐物分种布在高海拔地带，森林景观分布较少；在海拔大于3200米的高程带上，随着坡向的增加，SI呈现马鞍状分布；在海拔小于3200米的高程带上，随着坡向的增加，SI呈现“⌒”分布。(12)坡度和坡向趋势面上分析：坡度低于20°时，SI相对较小，随着坡向的增加，呈现一个波形的分布曲线；坡度20°-60°范围内，随着坡向的增加，SI的变化不大，呈现平稳分布；坡度60°-90°范围，随着坡向的增大，SI不断的增大，变化比较剧烈，呈现一个“∩”型的分布曲线。(13)通过矩阵转移法来研究特定时段不同土地利用类型之间数量时空变化，结果表明：有林地和灌丛景观是研究区域比较稳定的景观类型，保持着高转入贡献率和低产出贡献率，有林地的转入奉献率稍高灌丛；草地一直同时保持着高转入和转出贡献率，但是转出奉献率要高于转入奉献率。其它景观组分的转入奉献率和转出奉献率相对变化不大。(14)选取景观结构指标和生态指标，通过层次分析法和模糊评价，借助于ArcGIS的强大空间分析功能，对研究区域进行总体景观生态评价，并把评价结果和地形因素(高程分带、坡度和坡向)相叠加。结果表明：有林地的景观生态评价指数较大，在第二高程带到达最大；灌丛景观次之；其次是草地景观；裸崖地和荒山荒地的景观生态评价指数较小，裸崖地的指数在第三高程带达到最小。(15)把高程模型(DEM)、景观格局和景观生态评价结果进行叠加，在土地利用变化的基础上，进行景观生态分类的研究，从空间和时间上研究不同地形的景观格局的生态过程和景观生态恢复措施，建立以生态恢复重建为目的区域景观优化配置和生态重建，探索研究区域天然林景观生态恢复途径和方法。更多还原

【Abstract】 Subalpine region plays principal role in Natural Forest Protection Project in Western Sichuan. Based on TM image in 1985, SPOT5 image in 2005, forest distribution map, forest resource secondary surveying, topographic map, land use plan map, survey data and others, with the support of Remote sensing (RS) and Geographical Information System (GIS), Combined with the Digital Terrain Model (DEM), the theory and method of landscape ecology were used in this paper. The landscape pattern and dynamic processes were completely analysed in serial time and spatial space. The landscape ecology construction, landscape present situation, dynamic processes and ecology estimation were studied in detail. Based on the change of land use to make landscape ecology plan, the landscape ecology restoration region was raised and rebuilt, which could get better landscape allocation, it is very important to manage and restore ecology system of natural forest protection region in China by studying the method and approach of landscape ecology restoration. During the processing, following results and evolvement were gained in this paper.1. The topographic maps of study area were rectified, vector, modified errors and merged. The DEM in ARC/INFO was generated by the methods of linear and nonlinear interpolation. The accuracy of DEM was checked by the methods of spoint check and profile analysis. The results showed that: using point check, the accuracy of nonlinear interpolation was slightly higher than linear interpolation; using profile analysis, the errors of arbitrary direction was smaller than Y direction, the accuracy of nonlinear interpolation was higher at steep hill, but, the accuracy was very similar at flat region by using linear and nonlinear interpolation.2. With the support of ArcGIS, the slope and aspect were extracted and slice level was studied from DEM, the proportion of flat and steep region was larger, the proportion of iean region was smaller. The flat region was focused on the tong and parts of the river, the proportion of steep region was higher than that of fiat region, which showed that the study area was steep.3. During the processes of SPOT5 image, the data was orthorectified by using the DEM of the study area, the influence of the terrain was well eliminated. So it was very import to orthorectified the image by using DEM for forest land of complicated mountain region in China. Based on the property of remote sensor and the statistical characters of image, the quantitative conclusion could be drawn that the best multiple spectral bands combined result was 412 of SPOT5 by calculating the entropy, standard deviation and correlation coefficient in each band.4.With the support of AutoSync model in ERDAS IMAGE 9, the rectified SPOT5 of multiple spectral bands (10×10m) was used as referenced image to make match the geometric coordinate and edge match for TM data, which could improve the accuracy of TM greatly. The correlation coefficient and standard deviation were calculated and statistics, the best bands combined result could be ascertained by calculating the OIF. The supervised classification and accuracy check were made and the result of classification could be accepted.5. Different landscape type had different landscape pattern at the same elevation level. With the elevation increased, the change of landscape type and elevation was clearly shown. At different elevation level, three landscape types that had biggest area as flow: (1985a)bare land, alpine meadow, alpine shrub(I)-alpine meadow, alpine shrub, forest(Ⅱ)-forest, alpline meadow, alpine shrub (Ⅲ) forest, grassland, shrub (Ⅳ)-forest, grassland, shrub (Ⅴ) -forest, grassland, shrub (Ⅵ)(2005a):bare land, alpine meadow, alpine shrub(I)-alpine meadow, alpine shrub,forest(Ⅱ)- forest, alpine grassland, alpine shrub(Ⅲ)-forest, shrub, grassland(Ⅳ)-forest, grassland, shrub(Ⅴ) -forest, shrub, grassland(Ⅵ)6. With the elevation increased, the landscape diversity index and dominance index emerged on a certain distribution law. The landscape diversity index was biggest at 6th elevation level, but the smallest landscape diversity index was 4th elevation level, after that, The landscape diversity index increased a little, Generally, the landscape diversity index decreased with the elevation increased, on the contrary, landscape dominance index had adverse law with landscape diversity index.7. The landscape index of mean patch area, diversity index, dominance and patch density was selected to study the relationship among the landscape pattern and slope and aspect. The results showed that: the steep region had greater advantage of area proportion, the patch density and diversity were smallest, but the path mean area and landscape dominance were biggest; in fiat region, the patch density and diversity index were biggest and its landscape dominance and diversity index were smallest; in shady region, the patch density was smallest, patch mean area was biggest, the landscape dominance and diversity were much smaller, in sunny region, patch density and landscape dominance were biggest and patch mean area and diversity index were smallest.8. The relationship was analysed among the landscape spatial pattern and elevation, aspect and slope of reclassification. The results showed that: landscape distribution had certain correlation with elevation, aspect and slope, with the elevation and slope increased, the distribution of cropland (abandoned cropland) decreased, forest and shrub land were the main landscape type of shady steep slope, grassland was the main landscape type of sunny flat slope, but the difference of the landscape type distribution in different aspect was not obvious, It seemed that terrain had a certain influence on macro-distribution of landscape pattern. 9. The forest landscape type extracted by the grid sample and DEM was overlay, which could described the distribution pattern, geography and relative location of forest landscape in spatial scale. According to that results, the landscape element was analysed by using spatial pattern trend surface analysis. The formation, distribution law and the dynamic spatial pattern of forest landscape distribution pattern were studied, which could provide the theory for the restoration and sustainable of the natural forest ecology system of study area.10. The change of horizontal location made a certain influence on the landscape pattern evolution by using horizontal trend surface analysis, the difference of the trend surface of horizontal landscape element distribution in west-east and north-south direction was not obvious. In fact, the trend surface of horizontal distribution of landscape element could show that the terrain had control function on local landscape.11. In low elevation, the SI of shady slope was smaller and changed greatly, because the relative pure of broadleaf landscape distributed this region and belonged to the unstable landscape type, which was the earlier stage of landscape succession, its SI was smaller; the change of SI of sunny slope was stable, the mixed coniferous and broadleaf forest landscape distributed there and belonged to the important phase from the landscape succession process to climax pattern. In high elevation, the SI in shady slope was bigger, because the mixed coniferous forest landscape distributed here, its landscape construction was stable, the change of SI of sunny slope was great, because the amount of forest landscape of this region was litter. When the elevation greater than 3200 meter, With the aspect increased, the distribution shape of SI index was saddle-shapo, on the contrary, the distribution’shape of SI was“⌒”12. When the slope was below 20°, with the aspect increased, the relative SI was smaller, the distribution shape of SI was wave form, when the slope was between 20°and 60°, the distribution shape of SI was stable, when the slope was between60°and 90°,the SI increased gradually and changed greatly, the distribution shape of SI was“⌒”13. The spatial change of different land use type was studied by using matrix transfer in certain period in this paper. The result showed that: the forest and shrub land were stable landscape type of the study area, which kept the higher contribution rates of conversion-in and lower conversion-out, but the contribution rates of conversion-in of the forest was much higher than shrub, the contribution rates of conversion-in and conversion-out of the grassland were very high, but the contribution rates of conversion-in was greater than conversion-out. The contribution rates of conversion-in and conversion-out of other landscape was not changed a lot.14. The index of ecology and landscape structure was selected, with the support of spatial analysis of ArcGIS, the landscape ecology estimation was carried out by using AHP method and fuzzy analysis. The terrain factor(slice level, aspect and slope) and the estimation result were overlay. The result showed that: the forest land lafidscape ecology estimation index was bigger and got the biggest index at second elevation level; the forest land landscape ecology estimation index of shrub and grassland was second and third, the landscape ecology estimation index of barren land wasteland and barren hills was much smaller and the result of barren land was smallest at third elevation level.15. The distribution map of landscape pattern and ecology estimation was overlay with DEM, based on the land use, the landscape ecology classification was studied, the ecology process and restoration measure of landscape pattern of different terrain was studied in time and space, the landscape ecology restoration region was raised and rebuilt, which could explore the measure and method of natural forest landscape ecology restoration.更多还原