Modelling soil erosion in tropical environments using remote sensing and geographical information systems

Abstract

Soil erosion probability maps were produced under various case scenarios by accounting for uncertainties in the data and in the decision rule, using the Universal Soil Loss Equation (USLE), remote sensing and geographical information systems (GIS). This objective was realized by applying the Bayesian Probability Theory within IDRISI, a raster based GIS. The outcomes were two continuous probability soil erosion maps ranging from zero to 1. Comparing these maps with an earlier study indicates that accounting for the uncertainties has, in general, decreased the probability of soil erosion. Based on average readings for specific sites on the maps, increases in erosion risk under the second case scenario have had the highest impact on the highlands that is in the central, eastern, and northern regions of Langkawi Island, Malaysia. Assuming a 10% risk, this impact has increased by 11.98, 11.83 and 5.741% for high, medium and low soil erosion risk areas on the island respectively.     

Spatio-temporal changes in land cover and aquatic macrophytes of the Danube floodplain lake

The aquatic vegetation of ?í?ov Lake in the Danube floodplain, which is listed in the Ramsar Convention, was investigated to address three main questions: (1) how have landscape composition and the structures of the lake and its buffer zone changed from the mid-20th century; (2) how have species richness and the abundance of the aquatic macrophyte assemblage in this lake ecosystem changed over the last 34 years; and (3) which landscape metrics can best explain these temporal changes for floating-leaved macrophytes? Two methodological approaches, remote sensing and botanical field surveys, were applied. Historical (1949, 1970, 1990) and contemporary (2006) aerial photographs were analysed to determine land cover. Landscape configuration and structure were analysed using eight landscape metrics selected in advance to measure spatio-temporal changes and the fragmentation of the lake ecosystem and its corresponding buffer zone. The species diversity, abundance and distribution of true aquatic macrophytes were surveyed eleven times in five survey stretches between 1973 and 2007.At the landscape level, a decrease in the area covered by floating-leaved macrophytes, as well as an increase in open water surface and fragmentation of the land cover classes in the lake ecosystem, were recorded from 1949 to 2006. Overall, 30 true aquatic macrophytes were found from 1973 to 2007. Species richness did not change considerably, but the abundance of aquatic species fluctuated over the years. Three groups of true aquatic vegetation, based on common structural characteristics, were found in 1973–1983, 1989–2002, and 2004–2007 over the last 34 years. The landscape metrics NP, PD, LPI, and SHDI, which all express patterns of landscape fragmentation mostly indicate temporal changes in floating-leaved macrophytes.     

Total evaporation estimation for accurate water accounting purposes: an appraisal of various available estimation methods

The estimation of total evaporation is fundamental for water accounting, considering its influence on water availability. Moreover, the current increase in water consumption (e.g. in sub-Saharan Africa and the world over), land cover/use changes, deteriorating water quality and the climate change projections in most regions of the world underscore the need to understand water loss. So far, different approaches have been developed and implemented in estimating the variations of total evaporation, with varying accuracies. The aim of this work was therefore, to provide a review of these different approaches for estimating total evaporation, as well as a detailed discussion of their strengths and weaknesses. Findings from this review have shown that total evaporation estimates derived, using ground-based meteorological and micro-meteorological methods are inadequate for representing its large-scale spatial variations. On the other hand, remote sensing technology, which acquires data at different resolutions (i.e. radiometric, spectral, spatial and temporal), provides timely, up-to-date and relatively accurate spatial estimates of total evaporation over large geographic coverage, for sustainable and effective water accounting, which is key for well-informed and improved management of water resources at both catchment and regional scales. In this regard, more details on the remote sensing-based methods of estimating total evaporation are provided, especially considering the robust technological advancements and its potential in characterizing earth features over time and space. This work has also managed to identify research gaps and challenges in the accurate estimation of total evaporation, using remote sensing, especially with the emergence of more advanced sensors and the characteristics of the landscape.     

关于测绘数字成果数据远程容灾的研究

解决数据本地集中带来的风险,是测绘资料管理部门面临的重要问题。本文通过对容灾关键技术的研究和对测绘数字成果特点的分析,给出了基于远程容灾技术,建立测绘数字成果异地远程容灾体系的思路。     

遥感技术估算森林生物量的研究进展

结合机载LiDAR数据,提出了一种改进的GLAS光斑点冠层高度地形校正模型,以校正后的GLAS光斑点作为输入样本,结合MODIS遥感影像,利用支持向量回归（SVR）的方法对研究区森林冠层高度进行分生态区估测,并利用野外调查数据和机载LiDAR冠层高度结果对估测结果进行验证。结果显示：研究区的坡度等级直接影响GLAS光斑点森林冠层高度估测精度,改进的地形校正模型可以较好的减小坡度对GLAS光斑点森林冠层高度估测的影响,模型精度RMSE稳定在3.25~3.48 m;不同生态分区的SVR模型估测精度较为稳定,其RMSE=6.41~7.56 m;与算数平均高相比,样地的Lorey's高与制图结果拟合最好,不同生态分区平均估测精度为80.3%。机载LiDAR冠层高度结果的验证平均精度为79.5%,和Lorey's高验证结果呈现较好的一致性。     

江西龙南—石城地区干热岩资源优选

通过总结前人的研究成果,划分出江西3种类型干热岩潜力区：燕山期含高放射性产热型花岗岩出露区、燕山期火山岩区（火山岩盖层+燕山期高放射性产热型隐伏花岗岩）、白垩系红盆区（沉积岩盖层+燕山期高放射性产热型隐伏花岗岩）。结合地质、地热、物探和遥感等方法,选定江西龙南南部、安远、信丰、周田和石城5个干热岩潜力区。在最具干热岩潜力的龙南南部地区开展大地电磁测深工作,推断深部发育3处隐伏花岗岩岩体。依据地质、物探和遥感成果圈定干热岩远景区面积298.44 km2,保守估算4.6～8.0 km深度的静态干热岩潜在资源总量为220.35×1018J,折算成标准煤为75.28亿t。     

基于稀疏表示理论的优化算法综述

近年来,稀疏表示理论在信号处理、图像处理和计算机视觉等领域引起了广泛关注。很多研究人员提出了基于稀疏模型的算法,从不同视角对稀疏表示进行分类,如稀疏约束中使用的不同范数的方法可分为l₀范数最小化、l_p范数(0<p<1)最小化、l₁范数最小化、l₂范数以及l_2,1范数最小化5类。此外,还可对求解上述稀疏模型中不同的优化算法进行分类,包括贪婪策略、约束优化策略、逼近算法和同伦算法。通过介绍不同的稀疏模型及分析各类优化算法的基本原理,充分揭示出稀疏表示理论的潜在性质,为相关研究人员提供指引。     

基于ATP的微地形三维建模方法研究

基于人工倾斜摄影测量(artificialtiltphotogrammetry,ATP)的方法,通过设置两种拍摄路径、3种拍摄高度、3种分辨率相机的组合,测量并建立微地形地表三维模型,并对其进行精度评价。结果表明,所有三维模型总体平面误差为±8.98mm,高程误差为±7.79mm;较“○”形路径、1.0m和1.5m测量高度、1230万和610万像素相机的测量方法而言,“”形路径、0.5m测量高度、2430万像素相机所生成的三维模型平面误差为±4.01mm,高程误差为±1.81mm,能够更加精确地表达地表的细节特征,为定量化研究微地形侵蚀地表界面过程提供较高精度的数据支持和三维建模方法参考。     

基于Landsat-8卫星OLI遥感影像和AdaBoost算法的水体信息提取

以红碱淖地区为研究对象,基于AdaBoost算法,将Landsat-8卫星OLI遥感影像谱间关系、NDVI、NDWI和K-T变换wetness等特征相结合,构造提取水体的强分类器,实现水体的快速精确提取。结果表明,组合后的分类器能有效提高单个分类器的分类精度,平均分类精度提高了5.15%。该算法说明了Landsat-8遥感影像的水体信息提取的可行性,为快速精确提取水体信息提供了新思路。     

基于MongoDB的遥感影像存取方案的探讨与实现

针对传统关系型数据库存储遥感影像的一些不足,探讨如何设计非关系型数据库影像存取管理系统,既能实现遥感影像的存储与管理,又可以节省开支与资源,力求以很小的代价实现遥感影像的有效存储与管理。通过不同数据库的对比验证,基于MongoDB的遥感影像的存储效率更高,并且支持对任意格式影像文件的高效存储,明显优于关系型数据库。