【作者】 吴大千；

【导师】 张新时； 王仁卿；

【作者基本信息】 山东大学 ， 生态学， 2010， 博士

【摘要】 黄河三角洲有着中国暖温带最广阔、最完整的新生河口湿地生态系统,同时也是东北亚内陆和环西太平洋鸟类迁徙的重要越冬栖息地和繁殖地,是生物多样性保护的热点地区和生态保育的难点地区。黄河三角洲特有的多种物质交汇、多种动力系统相互作用的多重生态界面决定了黄河三角洲植被的特殊性、自然灾害频繁、生态系统脆弱的特点。因此利用遥感与地理信息系统(Geographic information system, GIS)技术迅捷、精确、有效的特点开展黄河三角洲植被研究有着重要的现实和理论意义。黄河三角洲是研究多尺度植被环境关系的理想区域。通过遥感技术和地理信息系统技术解译黄河三角洲地区的植被信息,搜集、获取区域重要的环境变量,在GIS平台上对植被环境关系进行多尺度的排序分析,结果表明：土壤水分与盐分的交互作用是所有研究尺度上植被分布的决定性变量；在小尺度上,土壤表面蒸发是土壤水分和盐分调控主要机制；在大尺度上,地形因素参与水分再分配的过程,并极有可能通过地表径流和地下水再分配过程对土壤水分和盐分进行调控：排序轴与部分人类干扰变量显著相关,表明区域内的人类干扰已经对植被格局产生了明显的作用。通过基于三期Landsat TM和ETM+影像的NDVI数据,进一步探讨黄河三角洲植被和地形要素间的尺度依赖关系以确定在大尺度上是否存在水分再分配的过程：黄河三角洲四种主要植物群落间的NDVI值差异显著,可以作为指示该地区不同植物群落的良好指标；空间滞后响应模型显著地降低了模型残差的Moran’s I系数,可以较好地抑制空间自相关对于模型的不利影响,从而得到更加真实可靠的生态学结论；多尺度的回归分析表明地形湿润度指数和坡度均在750m尺度左右表现为显著或者接近于显著,表明地形要素极有可能在该尺度上通过水分再分配过程对土壤盐分和水分进行调控；基于样线的小波分析表明降水再分配的过程虽然在较小的尺度上也存在,但是一般分布在较大尺度上。选择多种监测方法和阈值判别方法,利用随机生成的目视判别点筛选适合黄河三角洲的基于遥感特征的环境变化监测方法,结果表明：基于最大期望理论的阈值确定方法与简单的植被指数差分技术结合可以取得良好的监测效果；NDVI或者OSAVI指数与基于最大期望理论的阈值确定方法相配合是监测黄河三角洲区域环境变化方法中最优的,而基于遥感光谱特征的人工神经网络方法也有不错的表现。利用1992、1996、2001、2005和200等5期遥感影像,通过遥感与地理信息系统技术对黄河三角洲1992-2008年间土地利用状况进行解译,获取了土地利用的动态变化特征：刁口河附近海岸线蚀退速度较为缓慢,在2001年以后海岸线基本上处于一种相对平衡的状态；1996年后形成的新河口的南汊部分逐步蚀退,而北汊部分的西侧则表现出逐步扩展的趋势；1992-2008年间,研究区域内土地利用变化迅速而复杂,灌草地、耕地、裸地、水域和滩涂地的状态转换频繁,水资源条件和人类活动是区域土地利用的最主要驱动要素；1992-2008年间,强烈变化区域和较强变化区域主要分布在人类干扰较为频繁的区域,而新河口地区也有一定面积的强烈变化和较强变化区域。空间自相关是生态学中的常见问题,但是在元胞自动机研究中,空间自相关问题并没有得到应有的关注。利用元胞自动机模型模拟柽柳(Tamarix chinensis Lour.)种群的时空动态,并在模型中考虑空间自相关要素,比较加入空间自相关后元胞自动机参数的变化和模型的表现。结果表明,普通Logistic方程的残差中存在着显著的空间自相关效应,且普通Logistic方程不能获取所有的空间效应：Autologistic回归方程的AICc较低,模型的解释力较高,且模型残差的空间自相关效应不显著；Autologistic回归有效的降低了模型残差中的Moran’ s I系数,消除了空间自相关对模型的统计学影响；基于Autologistic回归的ALCA模型与普通的基于普通Logistic回归的OLCA模型相比,其模拟精度显著提高,且在统计学上稳健,是一种非常有效的景观模拟模型。本文利用遥感和GIS技术较为系统地研究了黄河三角洲植被的空间格局、动态监测和模拟,结果表明：土壤水分和盐分的交互作用是黄河三角洲植被环境关系的决定性要素,植被空间格局在大尺度上存在基于地形要素的水分再分配调控作用；NDVI和OSAVI与基于最大期望理论的阈值确定方法相配合的方法是目前检测的最优监测方法；1992-2008年间黄河三角洲土地利用动态变化快速而复杂,主要是灌草地、耕地、裸地、水域和滩涂地几种类型间的频繁转换,主要的驱动力为区域水资源条件和人类互动；基于Autologistic回归的元胞自动机不仅模拟精度显著提高,且在统计学意义上稳定,是一种非常有效的景观模拟模型。在GIS和遥感的新平台下利用传统生态学方法分析黄河三角洲的多尺度植被关系,检验研究区域是否存在景观尺度上的水分再分配过程,筛选适宜黄河三角洲的遥感检测指数,明确了研究区域土地利用变化的时空变异并构建考虑空间自相关的元胞自动机,研究方法具有较强的创新性,不仅可以提供对生态系统的多尺度认识,还可以有效地监测和模拟景观尺度上的环境变化和时空动态。在利用遥感和GIS时应当着重注意尺度和空间自相关的问题,进行多尺度分析可以更好的了解格局背后的多尺度调控关系,而在研究中考虑空间自相关特性可以获得更加真实可靠的生态学结论和更加精确模拟结果。更多还原

【Abstract】 The Yellow River Delta is one of the most active regions of land-ocean interaction among the large river deltas in the world. The area of perennial waterlogged wetlands in the Yellow River Delta including rivers, lakes, estuary waters, pits and ponds, reservoirs, channels, salt lakes, shrimp and crab pools, and tidal flats is 964.8 km2, accounts for 63.06% of total area. The Yellow River Delta is regarded as the largest and most well-preserved wetland in Chinese temperate area. The Yellow River Delta also has become an important over-wintering and breeding site for migrating birds in Northeast Asian Inland and Western Pacific Rim. And the delta is a complex depositional environment, with multiscale sedimentological, hydrological and ecological progress and georelational environment variables, which makes the Yellow River Delta to be a typical site for biodiversity conservation, environmental monitoring, multiscale analysis on vegetation-environmental relationships and spatially-explicit simulation using combined RS and GIS methodology.We tend to derive a holistic understanding of vegetation-environment relationships at mesoscales using an integrated remote sensing and GIS method. Canonical correspondence analysis (CCA) was employed to specify the relationships between vegetation pixels that obtained from high-resolution imagine and a group of biophysical, geographical and anthropogenic variables at different aggregation levels in the Yellow River Delta. Our study found that the statistic correlations between vegetation composition and environment variables increased as the grain sizes increased. The first CCA axes were closely related to the environmental variable of SS (soil salinity) and temperature vegetation dryness index (TVDI) at all the five scales, and mainly reflected the gradient of soil moisture-salinity interaction. The changes in other environmental variables that significantly related to the first axes at different scales may associate with the different processes and mechanisms that dominate on soils moisture and salinity. Several environmental variables used to depict anthropogenic activity were closely related to the second CCA axes, which indicated that the human disturbances had imposed obvious impacts on vegetation pattern in the delta. Our study confirmed that the relationships between vegetation and environmental factors at mesoscales in the delta were scale-dependent. The integration of GIS and remote sensing could be a promising method to detect relationships between vegetation and environment at different spatial scales.At larger scale, the elevation gradient and the redistribution of water and soil solutes always interact as a whole to determine regional vegetation pattern, especially in the region with small variation in elevation. The objective of this section is to test if water redistribution affects vegetation pattern at large scales in a coastal ecosystem in the Yellow River Delta, using an integrated remote sensing and GIS method with the Landsat images and a series of topographic variables. Results showed that:(1) NDVI was highly different among four prevailing communities, which was determined by the differences in habitat structure of coastal plant; (2) the spatial lagged model was statistically sound to obtain less significant Moran’s I index in model residuals and reasonable and robust ecological conclusions; (3) topographical variables affected the vegetation pattern via the scale-dependent adjustment and handle on soil moisture and salinity. At small scales, topographic factors regulate soil water and salinity through the evaporation from soil surface. At large scales, however, topographic factors redistribute the soil water and salinity through the runoff and groundwater system.A couple of vegetation index and other derived index, coupled by threshold determination methods, were selected to monitor regional environmental changes, with the randomly selected points in 1992,1996,2001 and 2005 as the validation points. The result showed that the combination of two vegetation index (NDVI and OSAVI) and Expectation Maximization (EM) provided considerable performance in environmental monitoring. The result also showed that the adoption of ANN also showed good performance in the study.With the aim to investigate the dynamic changes in land use, the combination of remote sensing and geographic information system were adopted to interpret the land use dynamics of the Yellow River Delta from five Landsat images from 1992 to 2008. As shown by the land use maps, the coastline in Diaokou river mouth was relatively gently, achieving to equilibrium since 2001, while, the coastline near the new river mouth showed rapid expansion. From 1992 to 2008, the land use changes in the delta were rapid and complex, with frequent transition among tide land, shrub and grass land, bare land and water body. Condition of water resources and anthropic activities were the main driving forces for land use changes. The most dynamic area and less dynamic area mainly located in the areas where the anthropic activities were intense. And there were also dynamic and less dynamic areas distributing in the new river mouth.Spatial autocorrelation (SAC) is frequently encountered in most spatial data in ecology. Cellular automata (CA) models have been widely used to simulate complex spatial phenomena. However, little has been done to examine the impact of incorporating SAC into CA models. Using image-derived maps of Chinese tamarisk (Tamarix chinensis Lour.), CA models based on ordinary logistic regression (OLCA model) and autologistic regression (ALCA model) were developed to simulate landscape dynamics of T. chinensis. In this study, significant positive SAC was detected in residuals of ordinary logistic models, whereas non-significant SAC was found in autologistic models. All autologistic models obtained lower AICc (Akaike’s information criterion corrected for small sample size) values than the best ordinary logistic models. Although the performance of ALCA models only satisfied the minimum requirement, ALCA models showed considerable improvement upon OLCA models. Our results suggested that the incorporation of the autocovariate term not only accounted for SAC in model residuals but also provided more accurate estimates of regression coefficients. The study also found that the neglect of SAC might affect the statistical inference on underlying mechanisms driving landscape changes and obtain false ecological conclusions and management recommendations. The ALCA model is statistically sound when coping with spatially structured data, and the adoption of the ALCA model in future landscape transition simulations may provide more precise probability maps on landscape transition, better model performance and more reasonable mechanisms that are responsible for landscape changes. Our holistic and systemtic research on the spatial pattern, monitoring and dynamic simulation relied on the combination of remote sensing and GIS techniques. The result indicated that the integrated application of remote sensing and GIS not only provided scale-dependent knowledge on multiscale vegetation-environment relationships, but also provide effective ways to monitor, detect and simulate environmental changes and spatiotemporal dynamics at landscape scale. The results also clearly showed the necessity to pay special attention to scale and spatial autocorrelation when using integrated remote sensing and GIS. The analysis that carried at multiple scales would provide more holistic and comprehensive knowledge on the underlying operating processes. The incorporation of spatial autocorrelation into ecological studies would eliminate the disturbance of spatial autocorrelation, and provide more sound and reliable ecological conclusions and more accurate model performance.更多还原