多时相特征组合优选的胶州湾湿地分类

基于光谱特征准确提取湿地信息并研究其动态变化,对于湿地环境监测和保护具有重要价值。本研究以山东省胶州湾湿地为研究区,构建面向对象的分类方法并集合多时相Landsat8 OLI遥感影像的多种特征变量:光谱特征,穗帽变换(K-T)特征,纹理特征,植被指数,水体指数等来提取湿地信息以提高中等分辨率影像的分类精度。结果表明,在光谱波段组合中,近红外波段和红光波段是湿地分类的最有效波段,穗帽变换产生的特征分量对湿地分类也较为有效,光谱指数中的Normalized Difference Water Index(NDWI)和纹理特征中的方差(Var)和对比度(Con)特征变量则并不适用于湿地分类,且特征变量的个数在到达24时会使可分离性下降。计算方差变化率(ROC)并结合目视解译确定最佳分割尺度,将面向对象的方法(Object-Based Image Analysis, OBIA)与Random Forest(RF)算法结合,其总体分类精度达到91.21%(Kappa系数=0.9011),较基于像素的分类方法总体精度增加5.18%,Kappa系数增加了0.0583。表明优选特征组合的面向对象的随机森林算法可有效提高湿地分类精度,为湿地信息提取提供一种新方法。     

基于空谱融合与AlexNet算法的滨海湿地植被分类研究

为提高遥感影像融合质量,提升资源一号(ZY-1 02D)高光谱遥感影像滨海湿地植被分类精度,提出将ZY-1 02D高光谱影像与空间分辨率为10 m的哨兵2号(Sentinel-2)影像进行Brovey融合,并通过搭建AlexNet卷积神经网络对ZY-1 02D高光谱影像和Brovey融合影像的滨海湿地植被进行分类,与支持向量机、随机森林和BP神经网络分类算法进行精度对比。研究结果表明:经Brovey融合后,AlexNet、支持向量机、随机森林和BP神经网络算法的植被分类总体精度分别提高15.60%、7.00%、14.80%和10.00%,Kappa系数提高了21.35%、9.93%、18.97%、12.85%;基于Brovey影像融合与AlexNet算法的植被分类精度最高,总体精度为92.40%,Kappa系数为89.42%。空谱融合配合AlexNet卷积神经网络有效解决了高光谱遥感影像在滨海湿地植被分类应用中精度较低的问题,为滨海湿地植被资源动态监测提供技术和方法支撑。     

联合星载光学和SAR影像的漳江口红树林与互花米草遥感监测

文章以福建省漳江口国家级红树林自然保护区为研究区, 提出了一种联合星载光学和合成孔径雷达(SAR)影像的红树林与互花米草分布的自动提取算法, 提取研究区内红树林和互花米草的空间分布。本研究选择2016年、2017年和2018年各一景低潮时的Sentinel-2 A光学影像数据, 获取植被和其他地物的光谱和纹理信息。算法首先基于归一化植被指数、增强型植被指数、地表水分指数以及数字高程模型的相关规则计算红树林和互花米草的潜在分布区; 通过随机森林分类算法区分红树林和互花米草, 2016年、2017年和2018年影像的分类总体精度和Kappa系数分别为98.53%和0.980、96.52%和0.952、98.71%和0.978, 分类效果良好; 使用当年所有Sentinel-1 A/B的SAR影像获得研究区的常年海水分布范围, 使用与海水交界的判据, 实现红树林、互花米草提取范围的优化。研究表明, 研究区2016年、2017年和2018年红树林总面积分别为56.85hm²、59.88hm²和58.61hm², 互花米草总面积分别为109.23hm²、124.00hm²和142.39hm², 与前人的研究成果在红树林和互花米草的空间分布和面积量级上有较好的一致性。     

黄河口湿地典型地物类型高光谱分类方法

黄河口湿地地物类型具有复杂多样的特点,本文将线性光谱混合分析模型与归一化植被指数(NDVI)和归一化水体指数(NDWI)相结合,建立了一种新的滨海湿地遥感影像分类方法;开展了基于CHRIS高光谱影像的黄河口湿地芦苇、柽柳、碱蓬、大米草、潮滩和水体6种典型地物分类实验,整体分类精度为77.33%,Kappa 系数为 0.71,与经典的最大似然分类(MLC)方法相比较,整体分类精度提高1.6%,Kappa 系数提高0.02,尤其是芦苇、碱蓬、大米草和潮滩的分类精度明显提高。     

基于多时相Sentinel-2影像多特征优选的湿地信息提取

本文提出了一种随机森林(random forest,RF)模型和Pearson相关系数相结合的RF-Pearson模型特征优选方法。以多时相Sentinel-2影像为数据源,提取多时相多特征;利用RF-Pearson模型进行特征选择,筛选出特征重要性得分较高且相关性较小的特征作为优选特征,参与黄河三角洲湿地信息提取;最后将分类结果与多时相全特征和随机森林模型优选特征进行比较。实验表明:特征优选能够提高湿地信息的提取效果,基于RF-Pearson模型特征优选方法的分类精度最高,表明了特征优选方法的有效性以及特征优选在湿地分类方面的优势。     

一种融合纹理特征与NDVI的随机森林海冰精细分类方法

海冰的精准分类对于掌握海冰生长发育状况,保障航海安全等具有重要意义。由于受数据源和分类方法等影响,使得海冰分类精度提高受限。本文面向高空间分辨率的光学遥感影像,提出了一种融合纹理特征和归一化差分植被指数（NDVI）的海冰精准分类方法,运用随机森林分类器构建海冰分类方法。以青岛胶州湾为实验区,高分二号（GF-2）为实验数据,进行了海冰类型提取,并与其他分类方法进行对比。结果显示:针对GF-2高分辨率光学遥感数据,融合纹理特征和NDVI的随机森林方法,相比于传统的随机森林、支持向量机、自动决策树和融合纹理特征的最大似然分类方法,总体分类精度分别提高13.70%、11.60%、19.22%、29.37%。Kappa系数分别提高0.16、0.13、0.22、0.44。相比于融合纹理特征和归一化水指数（NDWI）的随机森林方法,总体分类精度提高了9.67%,Kappa系数提高了0.09。这表明本文构建的海冰分类方法可有效提高海冰分类精度,为海冰的精确分类提供了一种有效的技术手段。     

联合光谱和纹理特征的滨海湿地高光谱深度学习分类—以黄河三角洲湿地为例

本文基于CHRIS高光谱遥感影像,发展了一种结合地物光谱特征和多纹理空间特征信息,采用双全链接的8层深度卷积神经网络分类算法对滨海湿地高光谱影像进行遥感地物分类,并在黄河口滨海湿地进行了应用。结果表明：1）基于测试样本数据,联合光谱特征和K-L变换的纹理特征信息,采用DCNN模型方法展现了高的分类精度,精度高达99%;2）利用光谱特征和全纹理特征的精度比仅使用光谱特征和光谱特征联合K-L变换后纹理特征的分类精度低。利用K-L变换后的光谱特征和纹理特征的DCNN分类精度达到99.38%,相比于使用全纹理特征信息的精度提高了4.15%;3）基于验证图像,发展的DCNN分类方法精度优于其他算法,DCNN方法总体分类精度为84.64%,Kappa系数为0.80;4）相比于浅层分类方法,本文发展的DCNN模型分类算法保证了所有地物类型的分类精度更加均衡,保持了主要地物类型的分类精度几乎不变,同时提高了滩涂和农田的精度。基于DCNN模型,潮滩和农田的分类精度分别达到79.26%和56.72%。比其它浅层分类方法提高了2.51%和10.6%。     

基于Sentinel-2 MSI影像与面向对象相结合的红树林树种精细化分类方法研究

红树林是最典型的滨海生态系统之一，红树林种间类型的精确识别对于红树林生态系统保护、修复及碳储量评估具有重要意义。遥感是开展红树林种间类型识别的有效手段，但传统的遥感红树林分类方法多是基于像元开展的，分类结果“椒盐”现象严重且精度还有很大提升空间。因此，本研究以东寨港红树林保护区为例，基于Sentinel-2 MSI影像，在传统遥感分类方法的基础上引入图像分割技术，分别构建了面向对象的支持向量机（Support Vector Machine,SVM）和随机森林（Random Forest,RF）分类法，并在此基础上对各模型的分类精度和适用性进行了分析。模型对比结果表明：（1）图像分割技术的引入能有效改善分类结果的“椒盐”现象，提升红树林种间类型的识别精度，基于像元使用SVM和RF分类算法总体分类精度分别可达78.82%(Kappa=0.75)和82.94%(Kappa=0.82)，面向对象的SVM和RF模型分类总体精度分别可达81.5%(Kappa=0.78)和92.67%(Kappa=0.88)，相较于以像元为分类对象的模型而言，后者精度分别提高了2.68%和7.43%;(2)从4个模...     

基于高分光学与全极化SAR的海南八门湾红树林种间分类方法

合理的红树种间组成结构是有效发挥红树林湿地生态价值的前提,明确的红树林种间分布信息是开展红树林生态系统治理和规划工作的有效依据。针对海南八门湾红树林湿地,基于高分三号(GF-3)全极化合成孔径雷达(synthetic aperture radar,SAR)和高分六号(GF-6)多光谱遥感数据,本文提取了35个红树林遥感特征,利用极端梯度提升树(eXtremegradientboosting,XGBoost)算法开展了特征重要性排序、特征筛选和红树林种间分类实验,将其与传统的支持向量机(supportvectormachine,SVM)、随机森林(randomforest,RF)机器学习算法进行精度比较,并基于XGBoost算法进行了3种特征组合方式(优选特征、多光谱特征、全极化SAR特征)的分类精度比较,旨在探索XGBoost对红树林种间分类的适用性和光学与全极化SAR数据对红树林种间分类的能力。结果表明:1)识别红树林种类的优势特征依次为多光谱的光谱波段、极化分解参数、光谱植被指数,且仅利用前8个优选特征(绿光波段反射率G、蓝光波段反射率B、Yamaguchi面散射分量Ys、近红外波...     

融合浅层特征的深度卷积神经网络互花米草遥感监测方法

基于2018年10月份黄河口入海两侧的LANDSAT8 OLI影像,提取植被指数和缨帽变换分量共9维光谱特征,构建融合浅层特征的8层深度卷积神经网络(deepconvolutionalneuralnetwork,DCNN)分类模型,开展互花米草(SpartinaalternifloraLoisel)遥感监测的方法研究,并从不同的浅层特征来具体分析互花米草的监测结果。结果表明:(1)在分类方法上,DCNN模型的总体分类精度最高,达到90.33%,与支持向量机(support vector machine, SVM)、随机森林(random forest, RF)分类器相比,精度分别提高4.78%、2.7%,互花米草的生产者精度分别提高了2.56%、0.47%,说明在滨海湿地遥感影像分类中, DCNN有着更好的应用潜力;(2)融合浅层特征后, DCNN的总体分类精度和互花米草的识别精度分别提高了0.34%和3.25%,有效提高了对互花米草的监测能力。其中,融合归一化植被水分指数(NDII)浅层特征的DCNN分类方法中,互花米草的识别精度提高最多,为2.56%,比值植被指数(RVI)次之,为2.32%。研究结果可为互花米草的监测与管理提供技术与数据支撑。     

Sentinel-1卫星合成孔径雷达获取第一幅台风图像

In this note, we present the first Sentinel-1 synthetic aperture radar(SAR) typhoon image acquired in the northwest Pacific on October 4, 2014. The eye shape and sea surface wind patterns associated with Typhoon Phanfone are clearly shown in the high-quality SAR image. SAR winds retrieval procedure was given but the actual wind estimates will only be available after the European Space Agency(ESA) releases the official calibration coefficients in order to accurately derive the SAR-measured normalized radar cross section. This study demonstrates the advantage of Sentinel-1 SAR with regards to imaging fine scale typhoon patterns on the sea surface beneath storm clouds. This paper also advocates the use of Sentinel-1 SAR data that is made freely and openly available worldwide for the first time in civilian SAR history.     

基于Sentinel-2影像的围海养殖信息提取

为了能够利用遥感图像快速准确地提取围海养殖矢量信息,本文选取养殖水体、堤坝及育苗室等交错分布的海参围海养殖区域作为研究区域,根据研究区域Sentinel-2遥感影像的光谱特征,选用归一化差异水体指数（Normalized Difference Water Index,NDWI）、改进归一化差异水体指数（Modified Normalized Difference Water Index,MNDWI）和增强水体指数（Enhanced Water Index,EWI）三类水体指数,分别进行提取实验,利用同时期高空间分辨率的高分二号卫星（GF-2）影像作为参考,验证不同方法的提取精度,精度评价结果表明：相较MNDWI和EWI两类水体指数,NDWI的分类精度更高,且利用NDWI提取研究区域的围海养殖信息的效果更好,所以该方法可在养殖区域的动态监测和规划管理中发挥数据支撑作用。     

基于Sentinel-3载荷OLCI和SRAL数据的内波同步探测研究

The ocean and land color instrument(OLCI) and synthetic aperture radar altimeter(SRAL) installed aboard the Sentinel-3 satellite have been in orbit for operational uses. In this study, data collected from Sentinel-3 are used to investigate internal waves in the South China Sea. An internal wave is detected using an OLCI image with a resolution of 300 m, and an analysis was performed with a quasi-synchronous moderate-resolution imaging spectroradiometer(MODIS) image. The opposite characteristics of OLCI and MODIS images of the same internal wave are explained by the critical angle in brightness reversals. The unique observational geometry of the OLCI image and its influence on observations of internal waves are discussed. The distribution of σ0 and sea surface height anomalies(SSHAs) induced by internal waves are studied using SRAL records. The σ0 records of SRAL occasionally show less sensitivity to the modulation of internal waves, which may be attributed to the observational geometry, while SSHAs show obvious variations. The synchronous pairing of OLCI images and SRAL records are analyzed to extract the three-dimensional sea surface signatures induced by internal waves. The analysis demonstrates that the profile of SSHAs in the surface shows an opposite phase to the profiles of internal waves in the ocean. The opposite phase relationship, observed in the remote sensing view, is also confirmed with a laboratory experiment.     

联合升降轨Sentinel-1A的地表形变监测技术研究

河谷型、海岸带、海岛礁以及滩涂等区域地表形变信息的准确获取,迫切且关键。利用25景升轨和28景降轨Sentinel-1A数据,基于SBAS-InSAR技术,获取兰州城区2014年10月~2016年9月的地表沉降速率图和沿视线方向(line-of-sight,LOS)的时相位移量,探讨升降轨数据在河谷型城市地表形变监测中的适用性。结果表明:升降轨Sentinel-1A数据之间的时相位移量具有高度关联性,其中升轨数据在地形起伏较大区体现出良好的干涉性,而降轨数据在平坦区表现出较好的监测能力。兰州城区大部分区域处于稳定状态,但沙井驿、九州、小达坪、东岗、伏龙坪、西客站以及周边区域地表存在沉降趋势,最大沉降速率可达30 mm/年。联合升降轨Sentinel-1A数据监测河谷型、海岸带、海岛礁以及滩涂等区域地面沉降具有可行性和可靠性。     

使用参数化初猜测谱算法提取VV极化Sentinel-1 SAR图像的海浪要素并与ECMWF再分析数据以及浮标数据进行验证对比

The purpose is to study the accuracy of ocean wave parameters retrieved from C-band VV-polarization Sentinel-1Synthetic Aperture Radar(SAR) images, including both significant wave height(SWH) and mean wave period(MWP), which are both calculated from a SAR-derived wave spectrum. The wind direction from in situ buoys is used and then the wind speed is retrieved by using a new C-band geophysical model function(GMF) model,denoted as C-SARMOD. Continuously, an algorithm parameterized first-guess spectra method(PFSM) is employed to retrieve the SWH and the MWP by using the SAR-derived wind speed. Forty–five VV-polarization Sentinel-1 SAR images are collected, which cover the in situ buoys around US coastal waters. A total of 52 subscenes are selected from those images. The retrieval results are compared with the measurements from in situ buoys. The comparison performs good for a wind retrieval, showing a 1.6 m/s standard deviation(STD) of the wind speed, while a 0.54 m STD of the SWH and a 2.14 s STD of the MWP are exhibited with an acceptable error.Additional 50 images taken in China's seas were also implemented by using the algorithm PFSM, showing a 0.67 m STD of the SWH and a 2.21 s STD of the MWP compared with European Centre for Medium-range Weather Forecasts(ECMWF) reanalysis grids wave data. The results indicate that the algorithm PFSM works for the wave retrieval from VV-polarization Sentinel-1 SAR image through SAR-derived wind speed by using the new GMF C-SARMOD.     

基于Sentinel-2A卫星数据面向对象的水边线提取——以如东县为例

使用基于面向对象的方法提取水边线, B分量作为阈值分割条件, Sentinel-2A数据作为提取水边线的影像,通过多尺度分割与光谱差异分割组合的方式。对如东沿海的淤泥质海岸、交通围堤海岸和养殖围堤海岸3种不同类型的海岸水边线进行提取。通过提取的水边线与影像叠加,并对受潮汐影响较小的水边线做精度验证。总体来说,提取的水边线较为准确。水边线的快速准确提取,对监测海岸带动态变化具有重要意义。     

Sea surface wind speed retrieval from Sentinel-1 HH polarization data using conventional and neural network methods

Conventional retrieval and neural network methods are used simultaneously to retrieve sea surface wind speed(SSWS) from HH-polarized Sentinel-1(S1) SAR images. The Polarization Ratio(PR) models combined with the CMOD5.N Geophysical Model Function(GMF) is used for SSWS retrieval from the HH-polarized SAR data. We compared different PR models developed based on previous C-band SAR data in HH-polarization for their applications to the S1 SAR data. The recently proposed CMODH, i.e., retrieving SSWS directly from the HHpolarized S1 data is also validated. The results indicate that the CMODH model performs better than results achieved using the PR models. We proposed a neural network method based on the backward propagation(BP)neural network to retrieve SSWS from the S1 HH-polarized data. The SSWS retrieved using the BP neural network model agrees better with the buoy measurements and ASCAT dataset than the results achieved using the conventional methods. Compared to the buoy measurements, the bias, root mean square error(RMSE) and scatter index(SI) of wind speed retrieved by the BP neural network model are 0.10 m/s, 1.38 m/s and 19.85%,respectively, while compared to the ASCAT dataset the three parameters of training set are –0.01 m/s, 1.33 m/s and 15.10%, respectively. It is suggested that the BP neural network model has a potential application in retrieving SSWS from Sentinel-1 images acquired at HH-polarization.     

Three-Dimensional Geographically Weighted Inverse Regression (3GWR) Model for Satellite Derived Bathymetry Using Sentinel-2 Observations

Current trends of development of satellite derived bathymetry (SDB) models rely on applying calibration techniques including analytical approaches, neuro-fuzzy systems, regression optimization and others. In most of the cases, the SDB models are calibrated and verified for test sites, that provide favorable conditions for the remote derivation of bathymetry such as high water clarity, homogenous bottom type, low amount of sediment in the water and other factors. In this paper, a novel 3-dimensional geographical weighted regression (3GWR) SDB technique is presented, it binds together methods already presented in other studies, the geographically weighted local regression (GWR) model, with depth dependent inverse optimization. The proposed SDB model was calibrated and verified on a relatively difficult test site of the South Baltic near-shore areas with the use of multispectral observations acquired by a recently launched Sentinel-2 satellite observation system. By conducted experiments, it was shown that the proposed SDB model is capable of obtaining satisfactory results of RMSE ranging from 0.88 to 1.23[m] depending on the observation and can derive bathymetry for depths up to 12m. It was also shown, that the proposed approach may be used operationally, for instance, in the continuous assessment of temporal bathymetry changes, for areas important in the context of ensuring local maritime safety.