流域地形自动分割研究

本文提供了基于栅格数字高程模型（ＤＥＭ）提取流域空间结构,进行流域自动分割的技术方法。该方法包括地貌形态结构定义、流域结构定义和算法设计３个部分。基于本文设计的算法,可以从ＤＥＭ提取完整的汇流网络和分水线网络,两者组成的多边形集把流域分割成具有拓扑结构的多个子汇流区,这个子汇流区集作为存贮流域信息的基本地貌单元,是流域信息进入地学数据库的框架,也是进行空间分布式机理──过程流域分析模型的基础。并以山西离石黄土丘陵沟壑区的王家沟流域为试验区,获得了较为满意的结果。     

基于地理信息系统的数字环境模型研究

本文提出了以遥感与地理信息系统为技术手段,用有序数值阵列的方法建立数字环境模型的新思路,具体阐述了建立数字环境模型的意义、方法和应用范围。并在建立数字环境模型的基础上,对西藏“一江两河”地区的环境综合状况进行了定量评价。     

An analysis of changes in the urban landscape composition and configuration in the Sancaktepe District of Istanbul Metropolitan City,Turkey using landscape metrics and satellite data

Changes in landscape composition and configuration patterns of Sancaktepe Municipal District in the Asian side of Istanbul Metropolitan City of Turkey were analysed using landscape metrics. Class-level and landscape-level metrics were calculated from the land cover/land use data using Patch Analyst, an extension in the Arc View GIS. The land cover/land use data were derived from classified satellite images of Landsat Thematic Mapper of 2002 and 2009 for Sancaktepe District. There was evidence of increase in agglomeration process of built-up patches as indicated by the increases in mean patch size, decrease in total edge and number of patches between 2002 and 2009. The urban expansion pattern experienced overall was not fragmented but concentrated due to infilling around existing patches. Changes in Area-Weighted Mean Shape Index and Area-Weighted Patch Fractal Dimension Index indicated that the physical shapes within built-up, forest and bareland areas were relatively complex and irregular. A conclusion is made in this study that spatial metrics are useful tools to describe the urban landscape composition and configuration in its various aspects and certain decisions whether to approve a specific development in urban planning could, for example, be based on some measures of urban growth form or pattern in terms of uniformity and irregularity, attributable to the dynamic processes of agglomeration and fragmentation of land cover/land use patches caused by urban expansion.     

An assessment of chlorophyll-a concentration spatio-temporal variation using Landsat satellite data,in a small tropical reservoir

Traditional approaches to monitoring aquatic systems are often limited by the need for data collection which often is time-consuming, expensive and non-continuous. The aim of the study was to map the spatio-temporal chlorophyll-a concentration changes in Malilangwe Reservoir, Zimbabwe as an indicator of phytoplankton biomass and trophic state when the reservoir was full (year 2000) and at its lowest capacity (year 2011), using readily available Landsat multispectral images. Medium-spatial resolution (30 m) Landsat multispectral Thematic Mapper TM 5 and ETM+ images for May to December 1999–2000 and 2010–2011 were used to derive chlorophyll-a concentrations. In situ measured chlorophyll-a and total suspended solids (TSS) concentrations for 2011 were employed to validate the Landsat chlorophyll-a and TSS estimates. The study results indicate that Landsat-derived chlorophyll-a and TSS estimates were comparable with field measurements. There was a considerable wet vs. dry season differences in total chlorophyll-a concentration, Secchi disc depth, TSS and turbidity within the reservoir. Using Permutational multivariate analyses of variance (PERMANOVA) analysis, there were significant differences (p < 0.0001) for chlorophyll-a concentration among sites, months and years whereas TSS was significant during the study months (p < 0.05). A strong positive significant correlation among both predicted TSS vs. chlorophyll-a and measured vs. predicted chlorophyll-a and TSS concentrations as well as an inverse relationship between reservoir chlorophyll-a concentrations and water level were found (p < 0.001 in all cases). In conclusion, total chlorophyll-a concentration in Malilangwe Reservoir was successfully derived from Landsat remote sensing data suggesting that the Landsat sensor is suitable for real-time monitoring over relatively short timescales and for small reservoirs. Satellite data can allow for surveying of chlorophyll-a concentration in aquatic ecosystems, thus, providing invaluable data in data scarce (limited on site ground measurements) environments.     

短波红外CO2反演过程约束研究及初步反演结果

短波红外通道卫星CO2遥感是近年国际研究热点。首先,开展了卫星观测对气溶胶及大气温度的敏感性研究;其次,针对基于最优化理论的非线性迭代反演方法反演过程中的不收敛问题,提出了修正的阻尼牛顿方法(MDNM),并利用模拟数据评估了MDNM方法的有效性;最后,利用GOSAT卫星数据反演CO2的垂直混合比浓度,并与地基TCCON站点数进行比对。研究结果表明:短波近红外CO2卫星遥感受气溶胶散射及温度影响明显;通过地基比对初步验证了MDNM具有良好的精度,两者的相关性R2达到了0.729。     

元胞自动机最短路径算法优化

概括了当前GIS中最短路径算法,分析了元胞自动机在最短路径分析算法中的原理及应用现状,并从两个方面对基于元胞自动机的最短路径算法进行优化即直线优化的元胞自动机最短路径算法。(1)将A*算法中的启发函数引入元胞自动机模型,提出了直线优化元胞自动机最短路径模型;(2)考虑道路网特征对最短路径算法的影响,得出具有道路网自适应性的最短路径分析模型。最后选取不同形态特征的shp道路网数据,验证了优化算法在实际应用中的适用性和高效性。     

尺度效应的叶面积指数产品真实性检验方法

真实性检验是评价遥感反演产品质量和验证遥感应用产品是否准确、真实地反映实际情况的重要途径。叶面积指数(LAI)是表征陆地植被结构和长势的关键参数,全面准确评价和验证LAI产品是产品用于陆面过程模型的前提。本文以MODIS LAI与GLASS LAI产品为研究对象,在尺度效应和尺度转换的基础上,建立了针对非均匀像元的低分辨率LAI产品真实性检验方法。在考虑空间异质性和植被长势差异的情况下,借助中分辨率的遥感影像,分别利用1 km像元平均叶面积指数和反演表观叶面积指数实现了对LAI算法和产品的真实性检验。为了比较作物长势差异和地表非均匀度对产品的影响,本文选择有代表性的河南鹤壁和甘肃张掖两个地区进行两种LAI产品真实性检验研究。研究结果表明,GLASS LAI和MODIS LAI产品均存在明显的低估现象。这并不是产品算法的问题,而是由于地表异质性和非均匀度的影响。在异质性更显著的张掖盈科灌区,低估现象更明显。GLASS LAI产品是多种LAI产品的融合,它的平均LAI比MODIS更接近真实情况,但是LAI的动态范围比MODIS窄。     

融合像素—多尺度区域特征的高分辨率遥感影像分类算法

针对基于像素多特征的高分辨率遥感影像分类算法的"胡椒盐"现象和面向对象影像分析方法的"平滑地物细节"现象,提出了一种融合像素特征和多尺度区域特征的高分辨率遥感影像分类算法。(1)首先采用均值漂移算法对原始影像进行初始过分割,然后对初始过分割结果进行多尺度的区域合并,形成多尺度分割结果。根据多尺度区域合并RMI指数变化和分割尺度对分类精度的影响,确定最优分割尺度。(2)融合光谱特征、像元形状指数PSI(Pixel Shape Index)、初始尺度和最优尺度区域特征,并对多类型特征进行归一化,最后结合支持向量机(SVM)进行分类。实验结果表明该算法既能有效减少基于像素多特征的高分辨率遥感影像分类算法的"胡椒盐"现象,又能保持地物对象的完整性和地物细节信息,提高易混淆类别(如阴影和街道,裸地和草地)的分类精度。     

多尺度字典学习的高光谱图像压缩算法

结合小波变换及字典学习提出了一种针对高光谱图像的压缩算法。该算法首先通过小波变换构建多尺度样本集,在小波域使用K-均值奇异值分解(K-SVD)方法学习得到原子尺寸不同的多尺度字典,然后在稀疏表示的过程中,定义一个原子使用频次筛选因子,通过统计局部最优波段稀疏表示时原子使用情况,结合筛选因子对字典原子进行优化筛选,使用精简后的字典对其余波段进行稀疏求解,最后针对不同尺度的表示系数采用自适应的量化编码。实验结果表明,与目前常用的3D-SPIHT和其他的多尺度字典学习算法相比,本文算法在中低比特率下,具有更好的重建性能。     

基于时间序列统计特性的森林变化监测

森林动态变化分析对揭示生态系统环境变化及植被恢复和布局重建等具有重要意义,时间序列的遥感数据为森林监测提供了基础数据。本文根据森林植被的统计学特性,在暗目标法的基础上,利用归一化植被指数NDVI实现森林样本自动选择;并融合NDVI构建了新的综合森林特征指数(Integrated Forest Z-Score,IFZ);以时间序列的IFZ分析森林动态信息,实现森林变化动态监测。以三峡大坝及周边区域森林为研究区,利用2001年至2012年每年生长季节(5月—10月)的Landsat TM影像检验本文算法。基于2002年、2006年和2010年三期7月—9月的Quick Bird影像的精度分析结果发现:研究区森林变化检测的总体精度可达96.53%,Kappa系数为0.9512。在添加NDVI指数后构建的IFZ提高了总体监测精度。其中,毁林类别的检测精度提高显著,漏检率和误检率分别为2.74%和3.64%;干扰后重建的森林类别的检测精度有一定提高,其漏检率和误检率分别为10.79%和10.51%。研究结果表明,改进暗目标法能提高森林样本的选样效率,添加NDVI的IFZ能提高森林动态变化的识别度。此外,本算法不仅能定性识别森林变化,而且能定量提供森林干扰发生时间和干扰强度。