共查询到16条相似文献，搜索用时 734 毫秒
1.
渤海是我国重要的经济区,海冰灾害严重威胁着人类生产活动。合成孔径雷达具有全天候成像能力,研究渤海区域的SAR图像海冰检测具有重要意义。传统海冰检测方法受限于特征提取方法和建模方式,检测精度有待提升。深度学习具有极强的特征自学习能力,适用于图像检测问题。本文基于深度学习框架U-Net,以Sentinel-1双极化（VV和VH）合成孔径雷达图像为输入信息,设计混合损失函数优化传统U-Net模型,形成了基于混合损失U-Net的渤海海冰检测模型。将本文模型与传统海冰检测方法[脉冲耦合神经网络（PCNN）、马尔科夫随机场（MRF）和分水岭算法]和基于深度卷积神经网络（CNN）的深度学习方法进行了对比。实验结果表明:本文基于混合损失U-Net的海冰检测模型在重叠度、F1分数、精确度和召回率4项度量指标上分别达到了97.567%、98.769%、98.767%和98.771%,检测效果明显优于对比方法;双极化信息输入的检测结果比VV单极化输入的检测结果在F1分数、精确度、召回率和重叠度上分别提高了0.375%、0.111%、0.639%和0.740%;混合损失函数的检测结果比非混合损失函数的检测结果在F1分数、精确度、召回率和重叠度上分别提高了1.129%、0.947%、1.794%和2.231%;模型能对冰水沿线、冰间水道、冰间隙等细节进行有效检测;可应用于渤海区域整幅SAR图像的海冰检测,为海冰监测、海冰变化分析、海冰预报提供技术支撑。 相似文献
2.
结合纹理特征的SVM海冰分类方法研究 总被引：5，自引：4，他引：1
海冰分类是遥感监测领域中的重要应用之一，海冰分类的准确性对于评估海冰冰情、保证航海安全和开辟北极航道具有重要的意义。针对海冰分类问题，本文选用Sentinel-1遥感数据，结合纹理特征分析，提出了一种改进的SAR海冰分类方法。该方法选用灰度共生矩阵提取特征值，通过实验得到适宜用于海冰分类的多特征组合，在此基础上利用支持向量机开展SAR海冰类型的分类研究。实验结果表明，该方法可以实现对海冰SAR图像中一年冰、多年冰和海水3种类型识别，与传统的海冰分类方法神经网络和最大似然法相比较，使用SVM分类方法，结合纹理特征开展海冰类型监测是可行的，同时也表明多特征组合有利于提高SAR图像的分类精度，从而验证了本方法的有效性，为海冰分类提供了一种新思路。 相似文献
3.
正海冰是北极变化的"指示器"。准确的海冰覆盖信息对于北极环境与气候变化认知、极地航行安全保障等方面具有重要作用。星载合成孔径雷达(Synthetic aperture radar,SAR)以全天候、全天时和高空间分辨率的特点,在极地海冰观测展示了突出优势。前期针对高分3号SAR数据,我们建立了基于支持 相似文献
4.
地球系统模式FIO-ESM对北极海冰的模拟和预估 总被引：5，自引：3，他引：2
评估了地球系统模式FIO-ESM(First Institute of Oceanography-Earth System Model)基于CMIP5(Coupled Model Intercomparison Project Phase 5)的历史实验对北极海冰的模拟能力,分析了该模式基于CMIP5未来情景实验在不同典型浓度路径(RCPs,Representative Concentration Pathways)下对北极海冰的预估情况。通过与卫星观测的海冰覆盖范围资料相比,该模式能够很好地模拟出多年平均海冰覆盖范围的季节变化特征,模拟的气候态月平均海冰覆盖范围均在卫星观测值±15%范围以内。FIO-ESM能够较好地模拟1979-2005年期间北极海冰的衰减趋势,模拟衰减速度为每年减少2.24×104 km2,但仍小于观测衰减速度(每年减少4.72×104 km2)。特别值得注意的是:不同于其他模式所预估的海冰一直衰减,FIO-ESM对21世纪北极海冰预估在不同情景下呈现不同的变化趋势,在RCP2.6和RCP4.5情景下,北极海冰总体呈增加趋势,在RCP6情景下,北极海冰基本维持不变,而在RCP8.5情景下,北极海冰呈现继续衰减趋势。 相似文献
5.
北极海冰正处于快速减退时期，北极海冰体积变化是全球气候变化的重要指示因子。本文利用两种卫星高度计数据（ICESat和CryoSat-2）反演得到的海冰厚度数据，结合星载辐射计提取的海冰密集度数据以及海冰年龄数据，估算了近期的北极海冰体积以及一年冰和多年冰体积变化。CryoSat-2观测时段（2011-2013年）与ICESat观测时段（2003-2008年）相比，北极海冰体积在秋季（10-11月）和冬季（2-3月）分别减少了1 426 km3和412 km3。其中，秋季和冬季的一年冰的体积增加了702 km3和2 975 km3。相反，多年冰分别减少了2 108 km3和3 206 km3。多年冰的大量流失是造成北极海冰净储量下降的主要原因。 相似文献
6.
1979-2012年北极海冰运动学特征初步分析 总被引：3，自引：3，他引：0
利用美国冰雪数据中心(NSIDC)发布的海冰速度和范围数据,本文分析了1979—2012年间北极海冰的运动学特征,以及北极海冰运动与分布范围演变之间的关系。结合欧洲中期天气预报中心(ECMWF)发布的2007和2012年高分辨率的气压场、风场数据,探讨了北极风场和气压场与海冰运动、辐散辐合和海冰面积的关系。结果表明,在1979-2012年间北极海冰平均运动速度呈显著增强的趋势,冬季海冰平均运动速度增加趋势明显强于夏季;北极、波弗特-楚科奇海域和弗拉姆海峡的冬、夏季海冰平均运动速度的增加率分别为2.1%/a和1.7%/a、2.0%/a和1.6%/a以及4.9%/a和2.2%/a。1979-2012年北极海冰平均运动速度和范围的相关性为-0.77,二者存在显著的负相关关系。北极冬季和夏季风场的长期变化趋势与海冰平均运动速度的变化趋势一致,冬季和夏季的相关系数分别为0.50和0.48。风场和气压场对海冰的运动、辐散及重新分布发挥着重要作用。2007年夏季,第234~273天波弗特海域一直被高压系统控制,波弗特涡旋加强,使得波弗特海域海冰聚集在北极中央区;顺时针的风场促使海冰向格陵兰岛和加拿大北极群岛以北聚合。2012年,白令海峡和楚科奇海域处于低压和高压系统的交界处,盛行偏北风,海冰从北极东部往西部输运,加拿大海盆的多年海冰因离岸运动而辐散,向楚科奇海域的海冰输运增加,受太平洋入流暖水影响,移入此区域的海冰加速融化,从而加剧海冰的减少。 相似文献
7.
灰度共生矩阵纹理特征对SAR海冰漂移监测的增强性能研究 总被引：1，自引：0，他引：1
海冰漂移监测对气候变化分析、船只航行、海上石油平台等海上活动安全作业具有重要意义。当前主流的SAR海冰漂移监测方法多是基于SAR灰度图开展的,其受噪声、环境等因素的影响较大,导致其在海冰漂移探测时,特征失配率高,匹配正确率低。针对这一问题,本文尝试利用SAR海冰纹理特征来增强海冰漂移探测性能。首先对比分析了8种纹理特征对海冰漂移探测中特征匹配的增强性能,筛选出能够有效增强特征匹配性能的最优纹理特征;其次进一步分析了海冰类型、入射角和分辨率对基于纹理特征的海冰漂移探测性能增强的影响。实验结果表明,均值是最优的纹理特征,与SAR强度图相比,特征匹配正确率提高了约7%。 相似文献
8.
高分3号星载合成孔径雷达极地海冰自动检测方法研究 总被引：3，自引：3，他引：0
随着全球变暖等一系列气候变化的发生，极地海冰成为人们日益关注的焦点。由于不受光线和云雨影响，合成孔径雷达（SAR）可以进行全天时全天候的观测。高分3号是我国高分系列卫星中的一颗星载合成孔径雷达成像卫星，具有多种成像模式，可以在全球获取SAR数据。全天时全天候的工作特性和高空间分辨率的优势，使得高分3号星载SAR在极地海冰遥感监测中发挥重要的作用。本文基于高分3号水平-垂直（Horizontal-Vertical，HV）极化数据，提出了一种基于支持向量机的无需人工干预的海冰检测方法，实现海水和海冰的自动分离。利用该方法得到的海冰和海水分离结果同辅以人工解译的半监督分类结果相比较为吻合，为高分3号服务于极区海冰监测奠定了良好的基础。 相似文献
9.
Nudging资料同化对北极海冰密集度预报的改进 总被引：2，自引：2，他引：0
北极夏季海冰的快速减少使得北极航道提前开通成为可能。为了给北极冰区船运活动提供及时可靠有效的海冰预报保障,急需提高海冰预报水平。本文基于麻省理工大学通用环流模式(MITgcm),使用牛顿松弛逼近(Nudging)资料同化方法将德国不莱梅大学的第二代先进微波辐射成像仪(AMSR2)海冰密集度资料同化到模式中,建立了北极海冰数值预报系统。设计试验对比3种不同Nudging系数计算方案的改进效果,结果表明选择合适参数后,不同方案均能显著改进海冰密集度初始场。通过设计有无Nudging同化的两组预报试验,结合卫星遥感海冰密集度及中国第五次北极科学考察期间"雪龙"船的走航海冰密集度观测数据,定量分析了Nudging同化方案对北极海冰密集度的24~120 h预报结果的改进效果。结果表明,Nudging同化对120 h内全北极海冰密集度的空间分布和移动单点目标的海冰密集度预报结果均有显著改善;但在海冰变化很小的情况下,Nudging同化试验的24~120 h预报结果均劣于惯性预报结果,说明基于Nudging同化的数值预报系统还需进一步提高预报技巧。 相似文献
10.
利用SMAP卫星雷达资料与美国国家冰雪数据中心(NSIDC)发布的近实时逐日极区网格化海冰密集度数据建立匹配数据集,分析了海冰和海水的L波段雷达后向散射特性差异,建立了基于线性判别分析算法的海冰检测算法。选择Sentinel-1A SAR极地地区的海冰影像对SMAP卫星雷达资料海冰检测产品进行实验验证,结果显示二者的海冰边缘线一致,说明SMAP海冰检测算法具有较高的精度。利用SMAP卫星雷达资料制作了北极和南极地区海冰覆盖图,计算了海冰覆盖面积,通过与美国国家冰雪数据中心(NSIDC)海冰覆盖面积比较发现,SMAP检测的北极地区海冰面积略大于NSIDC,相对偏差为3.3%,SMAP检测的南极地区海冰面积略小于NSIDC,相对偏差为1.8%,表明二者的覆盖面积基本一致,证实了SMAP海冰检测算法的精度。 相似文献
11.
A classification of sea ice in the Arctic by age (multiyear; first-year; and first-year deformed ice, nilas, etc.) is developed based on the Bayesian approach using satellite radar data and taking into account regional peculiarities of these types of ice for different sectors of the Arctic. Estimations of a priori probabilities for each ice type, which are required for the use of the Bayesian classification, are obtained by the analysis of ice charts in the Arctic seas developed at the AARI in 2008–2013 using satellite data. A posterior probabilities are estimated visually by an expert. Types of sea ice distinguished by the expert on satellite images make it possible to create sample values of the radar-scattering cross section (RSCS). Examples of the proposed Bayesian classification of ice in the Laptev Sea according to Envisat satellite data are given. 相似文献
12.
The ability to use radar to discriminate Arctic Sea ice types has been investigated using surface-based and helicopter-borne scatterometer systems. The surface-based FM/CW radar operated at 1.5 GHz and at multiple frequencies in the 8-18-GHz region. Measurements were made at angles of10deg to70deg from nadir. The helicopter-based radar operated at the 8-18-GHz frequencies with incidence angles of0deg to60deg . Extensive surface-truth measurements were made at or near the time of backscattar measurement to describe the physical and electrical properties of the polar scene. Measurements in the 8-18-GHz region verify the ability to discriminate multiyear, thick first-year, thin first-year, and pressure-ridged sea ice and lake ice. The lowest frequency, 9 GHz, was found to provide the greatest contrast between these ice categories, with significant levels of separation existing between angles from15deg to70deg . The radar cross sections for like antenna polarizations, VV and HH, were very similar in absolute level and angular response. Cross-polarization, VH and HV, provided the greatest contrast between ice types, The 1.5-GHz measurements showed that thick first-year, thin first-year, and multiyear sea ice cannot be distinguished at10deg to60deg incidence angles with like polarization, VV, by backscatter alone; but that undeformed sea ice can be discriminated from pressure-ridged ice and lake ice. The effect of snow cover on the backscatter from thick first-year ice was also investigated. It contributes on the order of 0 to 4 dB, depending on frequency and incidence angle; the contribution of the snow layer increased with increasing frequency. Snow cover on smooth lake ice was found to be a major backscatter mechanism. Summer measurements demonstrate the inability to extend the knowledge of the backscatter from sea ice under spring conditions to all seasons. 相似文献
13.
利用C波段RADARSAT-2单极化SAR图像的方位向截断波长估计有效波高 总被引：1，自引：1，他引：0
This paper proposes two simple models, look-up table(LUT) model and empirical model, to directly retrieve significant wave height(Hs) using synthetic aperture radar(SAR) azimuth cutoff(λc). Both models aim at C-band VV, HH, VH, and HV single-polarization SAR images. The LUT model relates Hs to λc, while the empirical model relates Hs to both λc and SAR range-to-velocity(β). The LUT model coefficients are derived by simulation under different sea states and observation conditions, which depend on incidence angle(θ), wave direction(dw), and βbut are independent of polarization. The empirical model coefficients are obtained by fitting the collocated data,which only depend on polarization. To fit empirical model coefficients and validate the two models, C-band RADARSAT-2 fine quad-polarization(VV+HH+VH+HV) single-look complex(SLC) SAR images and collocated buoy data are collected. Retrieved Hs, using Yang model and the two models proposed in this paper from four kinds of polarization SAR data, are compared with buoy Hs. Results show that both LUT and empirical models have the capacity of retrieving Hs from C-band RADARSAT-2 co-polarization SAR data, while Yang model is not suitable for these kinds of SAR data. Moreover, the empirical model is also valid for cross-polarization SAR data showing clear ocean wave stripes. 相似文献
14.
CICE5.0与BCC_CSM2.0模式的耦合及对北极海冰的模拟评估 总被引：2，自引：1，他引：1
本文将美国Los Alamos国家实验室发展的最新海冰模式CICE5.0引入国家气候中心气候系统模式BCC_CSM2.0,替代原有的海冰模式SIS,形成一个新的耦合模式。在此基础上,评估新耦合模式对1985-2009年北极海冰的模拟性能,检验引入CICE5.0后对耦合模式中北极海冰、海洋和大气模拟结果的改进。结果表明,引入CICE5.0后,模式能较好地模拟出北极海冰的空间分布、季节以及年际变化特征。相比于旧版本耦合模式,新耦合模式模拟的北极多年冰增多、一年冰减少,同时,海冰增厚、海冰流速减慢,模拟效果得到显著改进,对波弗特涡流模拟的改善尤为明显。进一步分析发现,相比于SIS,CICE5.0对北极海冰特别是海冰厚度模拟性能的提升,在耦合进入BCC_CSM2.0后,会触发冰-温的正反馈机制,改进了模式对海平面气压场、表层气温和海表温度的模拟,由此进一步提高了模式对北极海冰的模拟能力。 相似文献
15.
The physical structures of snow and sea ice in the Arctic section of 150°-180°W were observed on the basis of snow-pit, ice-core, and drill-hole measurements from late July to late August 2010. Almost all the investigated floes were first-year ice, except for one located north of Alaska, which was probably multi-year ice transported from north of the Canadian Arctic Archipelago during early summer. The snow covers over all the investigated floes were in the melting phase, with temperatures approaching 0℃ and densities of 295-398 kg/m3 . The snow covers can be divided into two to five layers of different textures, with most cases having a top layer of fresh snow, a round-grain layer in the middle, and slush and/or thin icing layers at the bottom. The first-year sea ice contained about 7%-17% granular ice at the top. There was no granular ice in the lower layers. The interior melting and desalination of sea ice introduced strong stratifications of temperature, salinity, density, and gas and brine volume fractions. The sea ice temperature exhibited linear cooling with depth, while the salinity and the density increased linearly with normalized depth from 0.2 to 0.9 and from 0 to 0.65, respectively. The top layer, especially the freeboard layer, had the lowest salinity and density, and consequently the largest gas content and the smallest brine content. Both the salinity and density in the ice basal layer were highly scattered due to large differences in ice porosity among the samples. The bulk average sea ice temperature, salinity, density, and gas and brine volume fractions were-0.8℃, 1.8, 837 kg/m3 , 9.3% and 10.4%, respectively. The snow cover, sea ice bottom, and sea ice interior show evidences of melting during mid-August in the investigated floe located at about 87°N, 175°W. 相似文献
16.
AbstractSea ice type is one of the most sensitive variables in Arctic sea ice monitoring, and it is important for the retrieval of ice thickness. In this study, we analyzed various waveform features that characterize the echo waveform shape and Sigma0 (i.e., backscatter coefficient) of CryoSat-2 synthetic aperture radar altimeter data over different sea ice types. Arctic and Antarctic Research Institute operational ice charts were input as reference. An object-based random forest (ORF) classification method is proposed with overall classification accuracy of 90.1%. Accuracy of 92.7% was achieved for first-year ice (FYI), which is the domain ice type in the Arctic. Accuracy of 76.7% was achieved at the border of FYI and multiyear ice (MYI), which is better than current state-of-the-art methods. Accuracy of 83.8% was achieved for MYI. Results showed the overall accuracy of the ORF method was increased by ～8% in comparison with other methods, and the classification accuracy at the border of FYI and MYI was increased by ～10.5%. Nevertheless, ORF classification performance might be influenced by the selected waveform features, snow loading, and the ability to distinguish sea ice from leads. 相似文献