基于超像元词包特征和主动学习的高分遥感影像变化检测

为解决高分辨率遥感影像变化检测中存在底层特征缺乏语义信息、像元级的检测结果存在“椒盐”现象以及监督分类中样本标注自动化程度较低,本文提出一种基于超像元词包特征和主动学习的变化检测方法。首先采用熵率分割算法获取叠加影像的超像元对象;其次提取两期影像像元点对间的邻近相关影像特征（相关度、斜率和截距）和顾及邻域的纹理变化强度特征（均值、方差、同质性和相异性）,经线性组合作为像元点对的底层特征;然后基于像元点对底层特征利用BOW模型构建超像元词包特征,并采用一种改进标注策略的主动学习方法从无标记样本池中优选信息量较大的样本,且自动标注样本类别;最后训练分类器模型完成变化检测。通过选用2组不同地区的GF-2影像和Worldview-Ⅱ影像作为数据源进行实验,实验结果中2组数据集的F1分数分别为0.8714、0.8554,正确率分别为0.9148、0.9022,漏检率分别为0.1681、0.1868,误检率分别为0.0852、0.0978。结果表明,该法能有效识别变化区域、提高变化检测精度。此外,传统主动学习方法与改进标注策略的主动学习方法的学习曲线对比显示,改进的标注策略可在较低精度损失下,有效提高样本标注自动化程度。     

结合POI数据的道路自动选取方法

人们对道路重要性的认知与道路周边设施有重要关系,针对现有地图综合道路选取方法中对语义特征考虑不足的问题,将POI数据引入到道路的语义特征分析中,提出一种综合考虑道路空间特征和语义特征的道路综合自动选取方法。首先,结合POI位置数据构造了道路语义特征度量的3个新参量：设施点密度、重要设施比率和专题设施比率;然后,与道路长度、连接值、总深度值、平均线密度等反映道路几何、拓扑和分布特征的度量参量一起,通过归一化和熵值法赋权进行整合计算,得到道路重要性值;最后,综合考虑道路重要性值、道路stroke构成和stroke连通度的约束条件进行道路的分步选取。实验结果表明,该方法在保留主要道路、保持道路分布疏密特征和道路连通性的同时,较好地顾及了道路的语义特征信息。     

断层氢气观测技术实验

采用ATG-6118H痕量氢在线自动分析仪，在朱夏-商丹断裂带南阳地区布设独山、马山口2个氢气观测台，研究设备取气深度、抽气时间设置长短对观测断层氢气浓度的影响，从而为利用痕量氢设备进行断层氢气观测提供参考依据。实验结果表明：①取气深度越浅，观测到的氢气浓度越高，日变化幅度越大；②集气管内氢气浓度偏低时，抽气时间设置略高于导气管路排空时间，氢气浓度明显变化，否则变化不明显。     

设计反应谱标定方法的研究现状与讨论

设计反应谱及其标定方法的研究是工程抗震研究领域的基本问题之一。在现有研究成果的基础上，归纳和总结了设计反应谱及其标定方法的研究历史和现状；评述了研究进程中的若干节点问题；介绍了设计反应谱的标定原理和当前几种有代表性的设计反应谱的标定方法；分析了设计反应谱标定参数的影响因素。在此基础上，对设计反应谱及其标定方法研究中的强震资料积累问题、谱形状问题、标定参数的确定问题以及标定方法等问题进行了讨论并提出了改进和进一步研究的建议。     

基于一维综合孔径微波辐射计的海面温度反演研究

Due to the low spatial resolution of sea surface temperature(T_S) retrieval by real aperture microwave radiometers,in this study, an iterative retrieval method that minimizes the differences between brightness temperature(T_B)measured and modeled was used to retrieve sea surface temperature with a one-dimensional synthetic aperture microwave radiometer, temporarily named 1 D-SAMR. Regarding the configuration of the radiometer, an angular resolution of 0.43° was reached by theoretical calculation. Experiments on sea surface temperature retrieval were carried out with ideal parameters; the results show that the main factors affecting the retrieval accuracy of sea surface temperature are the accuracy of radiometer calibration and the precision of auxiliary geophysical parameters. In the case of no auxiliary parameter errors, the greatest error in retrieved sea surface temperature is obtained at low T_S scene(i.e., 0.710 6 K for the incidence angle of 35° under the radiometer calibration accuracy of0.5 K). While errors on auxiliary parameters are assumed to follow a Gaussian distribution, the greatest error on retrieved sea surface temperature was 1.330 5 K at an incidence angle of 65° in poorly known sea surface wind speed(W)(the error on W of 1.0 m/s) over high W scene, for the radiometer calibration accuracy of 0.5 K.     

Establishment and application of an intelligent treating method for oil spill identification

In the identifying process of an oil spill accident, manual integral and artificial visual comparison are commonly used at present to determine the oil spill sources, these methods are time-consuming and easily affected by human factors. Therefore, it is difficult to achieve the purpose of rapid identification of an oil spill accident. In this paper, an intelligent method of automatic recognition, integration and calculation of diagnostic ratio of Gas Chromatography-Mass Spectrometer (GC/MS) spectrum are established. Firstly, four hundreds of samples collected around the world were analyzed using a standard method and Retention time locking technology (RTL) was applied to reduce the change of retention time of GC/MS spectrum. Secondly, the automatic identification, integration of n-alkanes, biomarker compounds, polycyclic aromatic hydrocarbons and calculation of the diagnostic ratios were realized by MATLAB software. Finally, a database of oil fingerprints were established and applied successfully in a spill oil accident. Based on the new method and database, we could acquire the diagnostic ratios of an oil sample and find out the suspected oil within a few minutes. This method and database can improve the efficiency in spilled oil identification.     

Global Statistical Assessment and Cross-Calibration with Jason-2 for Reprocessed HY-2 A Altimeter Data

HY-2 A (Haiyang-2 A) satellite was launched on August 16, 2011 and radar altimeter is one of its main payloads. We reprocessed two years of HY-2 A altimeter sensor geophysical dataset records (SGDR) data. This paper presents the main results in terms of reprocessed HY-2 A altimeter data quality: verification of data availability and validity, monitoring several relevant altimeter parameters, and assessment of the HY-2 A altimeter system performances. A cross-calibration analysis of reprocessed HY-2 A altimeter data with Jason-2 was conducted. The reprocessed HY-2 A altimeter data show good quality and have a low level of noise with respect to Jason-2. The same geophysical correction methods were used to calculate the sea surface height (SSH) for the two missions. The mean standard deviations of the crossover differences for HY-2 A and Jason-2 are 5.24 cm and 5.34 cm, respectively. The mean standard deviation of the crossover differences between HY-2 A and Jason-2 is 5.37 cm. These show that HY-2 A can provide SSH measurements at almost the same level of accuracy as Jason-2. The relative SSH bias between HY-2 A and Jason-2 due to the Ultra Stable Oscillator (USO) drift is obviously observed, and it can affect the calculation of mean sea level and should be further studied and corrected.     

Calibration method affects the measured δ2H and δ18O in soil water by direct H2Oliquid–H2Ovapour equilibration with laser spectroscopy

The direct H₂O_liquid–H₂O_vapour equilibration method utilizing laser spectroscopy (DVE-LS) is a way to measure soil pore water stable isotopes. Various equilibration times and calibration methods have been used in DVE-LS. Yet little is known about their effects on the accuracy of the obtained isotope values. The objective of this study was to evaluate how equilibration time and calibration methods affect the accuracy of DVE-LS. We did both spiking and field soil experiments. For the spiking experiment, we applied DVE-LS to four soils of different textures, each of which was subjected to five water contents and six equilibration times. For the field soil experiment, we applied three calibration methods for DVE-LS to two field soil profiles, and the results were compared with cryogenic vacuum distillation (CVD)-LS. Results showed that DVE-LS demonstrated higher δ²H and δ¹⁸O as equilibration time increased, but 12 to 24 hr could be used as optimal equilibration time. For field soil samples, DVE-LS with liquid waters as standards led to significantly higher δ²H and δ¹⁸O than CVD-LS, with root mean square error (RMSE) of 8.06‰ for δ²H and 0.98‰ for δ¹⁸O. Calibration with soil texture reduced RMSE to 3.53‰ and 0.72‰ for δ²H and δ¹⁸O, respectively. Further, calibration with both soil texture and water content decreased RMSE to 3.10‰ for δ²H and 0.73‰ for δ¹⁸O. Our findings conclude that the calibration method applied may affect the measured soil water isotope values from DVE-LS.     

Estimating groundwater recharge for a freshwater lens in an arid region: Formative and stability assessment

The formation of subsurface freshwater lenses on top of brackish groundwater is a fascinating hydrologic phenomenon that creates groundwater supplies of great potential value in arid regions. Information on the recharge quantity and mechanism of these lenses is both scarce and uncertain. This study examines the formation and macroscale stability of the Rawdatain freshwater lens in Kuwait, for which significant pre-development data are available. The Rawdatain is a large (150 million m³) subsurface freshwater lens overlying brackish groundwater compared to the other freshwater lenses in the Arabian Peninsula. In this study, a three-dimensional (3D) density-dependent groundwater flow model is tested against the following data targets to estimate long-term diffuse and focused groundwater recharge: (i) groundwater head, (ii) total dissolved solids (TDS) groundwater concentration, (iii) volume and vertical thickness of stored groundwater of three different water quality TDS ranges (0–700, 700–1000 and 1000–2000 mg/L) and (iv) geometrical shape features of the lens along cross-sections. To better represent the spatial variation in TDS, six different recharge zones were assigned to allocate diffuse and focused recharge conditions. Twelve recharge rate scenarios, encompassing a wide range of feasible long-term average annual recharge values (200,000–5,000,000 m³/year), were tested against the multiple targets and compared with the groundwater age of the Rawdatain lens. Based on comparison with data targets, the long-term average annual recharge is estimated to be 500,000 m³/year. Scenarios of reduced recharge, which may occur due to changes in land-use or climate, demonstrate the extremely slow response of the lens, which is in agreement with the slow development and formation of the lens (>2,000 years). Within a 100-year time frame, a 50% reduction in annual recharge reduces the lens volumes by 21, 17 and 9% for the three water quality categories, respectively. This study demonstrates the stability of freshwater lenses in arid regions and also provides methodology for similar focused rainfall recharge freshwater lenses.     

多源高分辨率DOM批量快速制作方法研究

随着遥感数据在自然资源调查中的广泛应用，对多源高分辨率遥感影像“一张图”批量快速制作有了更高的需求。本文以广西壮族自治区北海市银海区作为试验区，利用Pixel Engine软件的“RS+GIS+Photoshop”算法，首先，针对多源高分辨率遥感卫星数据特点，制定相应的数据处理流程，自动化制作单景DOM；然后，以县级辖区为单位，批量快速制作县级辖区DOM；最后，从制作精度和速度两个方面，将该方法制作得到的DOM成果与人工处理成果进行对比，分析该方法的可行性，提出了一种快速、有效的多源高分辨率DOM制作方法。