Ecological classification of European transitional waters in the North-East Atlantic eco-region

A new methodology to classify European North-East Atlantic transitional waters into ecological types has been developed based on the most important hydrological and morphological features that are likely to determine the ecology of aquatic systems in transitional waters. Hydrological indicators help identifying if a transitional water area is dominated by fresh or sea water and/or by intertidal or subtidal areas, while morphological indicators allow an estimation of the complexity of the transitional water and the diversity of the habitats involved. Twelve transitional waters of the southern Bay of Biscay were classified using this methodology and the five hydro-morphological types obtained were validated with benthic macro-invertebrate data. Transitional waters with a complex morphology showed the highest values of species diversity, while those with a smaller tidal influence showed lower species diversity. The ‘Scrobicularia’ and ‘Abra’ assemblages, previously identified in the study area, were found to be related to different types of transitional waters. The ‘Abra’ assemblage only appeared in estuaries with a complex morphology and dominated by tidal influences, while the ‘Scrobicularia’ assemblage was detected in all the transitional waters except for a single coastal lagoon. This classification of transitional waters may therefore be useful to establish the biological reference conditions needed for European Directives.     

Capabilities of the bathymetric Hawk Eye LiDAR for coastal habitat mapping: A case study within a Basque estuary

The bathymetric LiDAR system is an airborne laser that detects sea bottom at high vertical and horizontal resolutions in shallow coastal waters. This study assesses the capabilities of the airborne bathymetric LiDAR sensor (Hawk Eye system) for coastal habitat mapping in the Oka estuary (within the Biosphere Reserve of Urdaibai, SE Bay of Biscay, northern Spain), where water conditions are moderately turbid. Three specific objectives were addressed: 1) to assess the data quality of the Hawk Eye LiDAR, both for terrestrial and subtidal zones, in terms of height measurement density, coverage, and vertical accuracy; 2) to compare bathymetric LiDAR with a ship-borne multibeam echosounder (MBES) for different bottom types and depth ranges; and 3) to test the discrimination potential of LiDAR height and reflectance information, together with multi-spectral imagery (three visible and near infrared bands), for the classification of 22 salt marsh and rocky shore habitats, covering supralittoral, intertidal and subtidal zones. The bathymetric LiDAR Hawk Eye data enabled the generation of a digital elevation model (DEM) of the Oka estuary, at 2 m of horizontal spatial resolution in the terrestrial zone (with a vertical accuracy of 0.15 m) and at 4 m within the subtidal, extending a water depth of 21 m. Data gaps occurred in 14.4% of the area surveyed with the LiDAR (13.69 km²). Comparison of the LiDAR system and the MBES showed no significant mean difference in depth. However, the Root Mean Square error of the former was high (0.84 m), especially concentrated upon rocky (0.55–1.77 m) rather than in sediment bottoms (0.38–0.62 m). The potential of LiDAR topographic variables and reflectance alone for discriminating 15 intertidal and submerged habitats was low (with overall classification accuracy between 52.4 and 65.4%). In particular, reflectance retrieved for this case study has been found to be not particularly useful for classification purposes. The combination of the LiDAR-based DEM and derived topographical features with the near infrared and visible bands has permitted the mapping of 22 supralittoral, intertidal and subtidal habitats of the Oka estuary, with high overall classification accuracies of between 84.5% and 92.1%, using the maximum likelihood algorithm. The airborne bathymetric Hawk Eye LiDAR, although somewhat limited by water turbidity and wave breaking, provides unique height information obscured from topographic LiDAR and acoustic systems, together with an improvement of the habitat mapping reliability in the complex and dynamic coastal fringe.     

基于随机森林的光子计数激光雷达点云滤波

新一代星载激光雷达卫星ICESat-2首次采用了微脉冲光子计数激光雷达技术,由于单光子探测的灵敏性导致数据在大气和地表下层产生了大量噪声,因此对光子计数激光雷达点云数据实现信号和噪声的分离是开展进一步应用研究的前提和基础。本文选择美国俄勒冈州和弗吉尼亚州2个研究区,采用MATLAS数据,根据光子点云数据的特点构造了12个光子点云特征,对所构造的特征利用随机森林进行变量筛选,用机器学习方法对光子点云进行分类,并将建立好的模型推广到整个研究区。研究结果表明,本文构建的分类器分类总精度达到了96.79%,Kappa系数为0.94,平均生产者精度和用户精度分别为97.1%和96.8%。在相对弱噪声、平坦地形区域和强噪声、复杂地形区域都取得较好的分类结果。本文结果显示了基于少量样本通过机器学习的方法构建模型,可以推广到较大范围区域的光子点云分类应用中。     

高分六号遥感卫星新增波段下的树种分类精度分析

高分六号卫星具有覆盖广、多种分辨率、波段多的优势,能为遥感解译提供更丰富的信息。为探究高分六号卫星新增波段在森林树种识别上的应用,本文以覆盖根河市阿龙山林业局的一期高分六号宽幅影像为数据源,基于特征优化空间算法（Feature Space Optimization,FSO）和最大似然分类法,分别利用高分六号的前4个波段和所有波段（8波段）的光谱、纹理等特征进行了森林树种分类,并逐一添加新增波段特征确定了各波段的贡献率排名。结果表明：在加入了优选出的均匀性纹理、均值纹理和角二阶矩纹理3种纹理特征后,前4波段和8波段的分类精度比只基于光谱特征时的精度分别高出13.23%和24.63%;利用8波段信息比只利用前4波段在基于光谱特征上的精度高11.88%,在基于光谱+纹理特征上则高23.24%;基于8波段光谱+纹理特征的树种分类精度最高,达到68.74%,新增4波段的贡献率排名为B6>B5>B8>B7,说明新增红边波段对于本次树种分类试验的贡献率最高,能为北方树种识别提供有效帮助。     

桂林-阳朔地区DEM地形特征与岩性相关性分析及分类研究

地形地貌是岩性解译的重要信息,地形因子作为描述DEM数字曲面几何特征的定量指标参数,可用来定量化表达不同岩性所在地区地形地貌特征。本文以桂林-阳朔地区为研究区,研究地形因子数学、地质意义,建立岩性与地形因子组合间的定量关联,进而实现岩石类型划分。本文基于ASTERGDEM提取坡度、起伏度等12个地形因子,在分析各个地形因子地质意义基础上,通过聚类分析及方差分析的多元统计分析方法,研究各岩性地形因子特性及其关联性,建立研究区岩性之间的定量差异;此外,利用因子分析方法研究岩性分类过程中的主导因素,确定适宜岩性分类方法以实现定量化岩性分类。实验结果表明：不同岩性、不同地形地貌的地形因子（组合）之间具有显著差异,基于因子分析得到的宏观地形复杂度指数（MTI）以及微观曲率指数（MCI）对岩石类型的分类精度达77.36%。研究表明,地形复杂度等地形因子可用于岩性分类,采用因子分析方法可获取反映地形地貌宏观、微观特征的定量指标,且岩性分类效果良好。     

Extending self-organizing maps for supervised classification of remotely sensed data

An extended self-organizing map for supervised classification is proposed in this paper. Unlike other traditional SOMs, the model has an input layer, a Kohonen layer, and an output layer. The number of neurons in the input layer depends on the dimensionality of input patterns. The number of neurons in the output layer equals the number of the desired classes. The number of neurons in the Kohonen layer may be a few to several thousands, which depends on the complexity of classification problems and the classification precision. Each training sample is expressed by a pair of vectors ： an input vector and a class codebook vector. When a training sample is input into the model, Kohonen＇s competitive learning rule is applied to selecting the winning neuron from the Kohouen layer and the weight coefficients connecting all the neurons in the input layer with both the winning neuron and its neighbors in the Kohonen layer are modified to be closer to the input vector, and those connecting all the neurons around the winning neuron within a certain diameter in the Kohonen layer with all the neurons in the output layer are adjusted to be closer to the class codebook vector. If the number of training sam- ples is sufficiently large and the learning epochs iterate enough times, the model will be able to serve as a supervised classifier. The model has been tentatively applied to the supervised classification of multispectral remotely sensed data. The author compared the performances of the extended SOM and BPN in remotely sensed data classification. The investigation manifests that the extended SOM is feasible for supervised classification.     

���ڻز�ǿ�ȵĺ��׵��ʷ���ϵͳ�ؼ�����

???????????????????????????????????????????????????ALL????????????????洢????????????????????????ALL???????????????????????????????????????????е???????????γ?????????????????????????????????????????BP?????????????????????к????????????????????????????????Ч????????????????ò??????????????????     

基于Weka的地震数据挖掘系统的设计与实现

基于开源的数据挖掘系统Weka,使用Java语言及面向对象的思想,设计并实现了地震数据挖掘系统。根据地震数据资料的特点,将数据挖掘的核心技术（聚类分析、关联规则分析等）引入到该系统中,其中聚类分析选用DBSCAN作为核心算法,关联规则分析选用Apriori作为核心算法。用户使用该系统只需在交互界面选择相关参数,即可实现调用数据挖掘算法来分析地震数据,发现探索其隐含规律。     

文蛤Meretrix meretrix血细胞形态与分类

光镜观察湛江南三地区文蛤血细胞的形态与结构特征,并进行分类。按细胞大小、形态和结构特征可将文蛤血细胞分为4类:透明细胞、小颗粒细胞、大颗粒细胞和淋巴细胞。细胞直径有统计学上显著性差异(P<0.05),分别为10.11±0.44μm、10.99±0.32μm、9.63±0.41μm和9.22±0.25μm。透明细胞是血细胞中的主要成分,细胞质中无颗粒;小颗粒细胞的细胞质中存在大量细小颗粒;大颗粒细胞细胞质中的颗粒直径有1.0～2.5μm;淋巴细胞是血细胞中含量最少的,有很高的核质比,大约为0.5。     

使用支持向量机识别地震类型的影响因素分析

对山东地区2006~2017年3种地震事件--天然地震、爆破及塌陷的波形记录进行小波变换,对提取出的香农熵特征采用支持向量机LIBSVM方法进行分类识别,并设计一系列实验研究影响最终分类效果的因素。结果表明,5种影响因素--信号窗长度、小波分解方式、小波基类型、向量机算法类型、向量机核函数类型均对地震类型的分类识别结果产生一定的影响;识别率最高的3组处理方式均采用了2 000 s信号窗长度+db7小波基+υ-SVC算法的组合方式。所得的识别率较高的几种影响因素组合,在未来可应用于地震类型的实时识别,进一步提高地震类型的识别率和触发准确率。