基于机器学习的重庆长江航道雾图像特征识别研究     

王远谋  李家启  陈施吉  唐家萍  夏佰成  韩世刚 《气象与环境学报》2021,37(1):106-112

基于重庆市境内长江航道雷达站拍摄的雾天气过程影像资料,利用K最近邻、支持向量机、BP神经网络、随机森林等机器学习算法,对无雾和5类有雾天气个例进行图像识别训练,构建雾图像识别模型,并检验了识别准确率。结果表明：机器学习能够有效识别雾图像,随机森林算法的识别效果优于其余3种算法。对于能见度超过1500 m的无雾天气,模型的识别准确率为100%,对于能见度在1000—1500 m范围内的轻雾、能见度低于50 m的强浓雾,模型的识别准确率在90%以上,对于能见度在50—1000 m范围内的雾、大雾和浓雾,识别准确率超过70%。  相似文献   

基于智能物联网技术的天津城市积水监测预警系统     

侯天宇  梁好  霍凯  赵敏  陈子煊  张春莉  苑超 《气象研究与应用》2021,(1):85-89

通过智能物联网技术实时获取积水监测实况数据,利用天津市气象精细化格点预报产品和城市自动雨量观测站实况数据,以机器学习、神经网络模型和天津市城市内涝风险等级划分原理为基础,研究基于用户实时位置的城市内涝预报预警技术,研发天津市城市自动化积水监测预警系统。结果表明,该系统具备一定的城市内涝风险监测预警预报能力,并在2018—2020年多次重大天气过程中应用,积水深度预报结果与监测结果基本一致,应用数据表明验证结果良好,系统可以为政府防灾减灾决策、指挥调度提供精准、及时的气象数据支撑。  相似文献   

A Machine Learning-based Cloud Detection Algorithm for the Himawari-8 Spectral Image     

Chao LIU  Shu YANG  Di DI  Yuanjian YANG  Chen ZHOU  Xiuqing HU  Byung-Ju SOHN 《大气科学进展》2022,39(12):1994-2007

Cloud Masking is one of the most essential products for satellite remote sensing and downstream applications. This study develops machine learning-based (ML-based) cloud detection algorithms using spectral observations for the Advanced Himawari Imager (AHI) onboard the Himawari-8 geostationary satellite. Collocated active observations from Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) are used to provide reference labels for model development and validation. We introduce both daytime and nighttime algorithms that differ according to whether solar band observations are included, and the artificial neural network (ANN) and random forest (RF) techniques are adopted for comparison. To eliminate the influences of surface conditions on cloud detection, we introduce three models with different treatments of the surface. Instead of developing independent ML-based algorithms, we add surface variables in a binary way that enhances the ML-based algorithm accuracy by ~5%. Validated against CALIOP observations, we find that our daytime RF-based algorithm outperforms the AHI operational algorithm by improving the accuracy of cloudy pixel detection by ~5%, while at the same time, reducing misjudgment by ~3%. The nighttime model with only infrared observations is also slightly better than the AHI operational product but may tend to overestimate cloudy pixels. Overall, our ML-based algorithms can serve as a reliable method to provide cloud mask results for both daytime and nighttime AHI observations. We furthermore suggest treating the surface with a set of independent variables for future ML-based algorithm development.  相似文献   

Climatology of Shear Line and Related Rainstorm over the Southern Yangtze River Valley Based on an Improved Intelligent Identification Method     

刘金卿  陈鹤  徐靖宇 《热带气象学报(英文版)》2022,28(4):413-424

Based on four reanalysis datasets including CMA-RA, ERA5, ERA-Interim, and FNL, this paper proposes an improved intelligent method for shear line identification by introducing a second-order zonal-wind shear. Climatic characteristics of shear lines and related rainstorms over the Southern Yangtze River Valley (SYRV) during the summers (June-August) from 2008 to 2018 are then analyzed by using two types of unsupervised machine learning algorithm, namely the t-distributed stochastic neighbor embedding method (t-SNE) and the k-means clustering method. The results are as follows: (1) The reproducibility of the 850 hPa wind fields over the SYRV using China’s reanalysis product CMA-RA is superior to that of European and American products including ERA5, ERA-Interim, and FNL. (2) Theory and observations indicate that the introduction of a second-order zonal-wind shear criterion can effectively eliminate the continuous cyclonic curvature of the wind field and identify shear lines with significant discontinuities. (3) The occurrence frequency of shear lines appearing in the daytime and nighttime is almost equal, but the intensity and the accompanying rainstorm have a clear diurnal variation: they are significantly stronger during daytime than those at nighttime. (4) Half (47%) of the shear lines can cause short-duration rainstorms (≥20 mm (3h)-1 ), and shear line rainstorms account for one-sixth (16%) of the total summer short-duration rainstorms. Rainstorms caused by shear lines are significantly stronger than that caused by other synoptic forcing. (5) Under the influence of stronger water vapor transport and barotropic instability, shear lines and related rainstorms in the north and middle of the SYRV are stronger than those in the south.  相似文献   

面向高分遥感影像道路提取的轻量级双注意力和特征补偿残差网络模型     

陈振  陈芸芝  吴婷  李佳优 《地球信息科学学报》2022,24(5):949-961

针对高分辨率遥感影像背景复杂,道路提取容易受阴影、建筑物和铁路等背景信息干扰的问题,提出一种带有轻量级双注意力和特征补偿机制的DAFCResUnet模型。该模型在ResUnet的基础上,通过增加轻量级的双注意力和特征补偿模块实现模型在性能和时空复杂度上的平衡。其中,双注意力模块可以增强模型的特征提取能力,特征补偿模块可以融合网络中来自深浅层的道路特征。在DeepGlobe和GF-2道路数据集上的实验结果表明,DAFCResUnet模型的IoU和F1-score可以达到0.6713、0.8033和0.7402、0.8507,模型的整体精度优于U-Net、ResUnet和VNet模型。与U-Net和ResUnet模型相比,DAFCResUnet模型仅增加了少量的计算量和参数量,但IoU和F1-score均有较大幅度的提高;与VNet模型相比,DAFCResUnet模型在计算量和参数量远低于VNet的情况下取得了更高的精度,模型在精度和时空复杂度两方面均有优势。相比其他对比模型,DAFCResUnet模型具有更强的特征提取和抗干扰能力,能更好解决道路上的干扰物、与道路特征相似地物、树荫或阴影遮挡等造成的道路空洞、误提和漏提现象。  相似文献   

AlexNet支持下的地图建筑物形状分类方法     

焦洋洋  刘平芝  刘爱龙  刘松林 《地球信息科学学报》2022,24(12):2333-2341

地图目标的形状在地图制图综合、空间查询等研究中发挥着重要作用。地图建筑物形状的识别与分类作为建筑物轮廓化简与典型化的基础,一直是制图综合研究的热点问题。目前,主要的建筑物形状识别方法主要依赖对建筑物轮廓的描述,对建筑物等地图面状要素的形态特征有较强的依赖性,通常只在应对特定类型的规则轮廓或直角化轮廓时能发挥较好的效果,对于形状不规则或复杂的情况识别不佳。本文提出一种AlexNet支持下的地图建筑物形状分类方法,将矢量地图中建筑物数据的形状分类问题,转化为建筑物栅格图像的分类问题,通过完成卷积神经网络的图形分类实现建筑物的形状识别。该方法首先结合空间认知规律提出一系列典型建筑物形状类型,然后利用矢量-栅格转换的方法从OSM数据采样单体建筑物栅格图像,通过人工标识获得建筑物形状分类训练样本,训练AlexNet卷积神经网络分类模型,最后利用训练好的模型对大比例尺建筑物数据进行智能形状分类与识别。本文利用北京、香港2个城市的OSM建筑物数据作为样本训练建筑物形状分类模型,并在广州部分城区的OSM建筑物数据上进行验证。相较传统形状相似性度量方法,本文提出的方法对实验区建筑物的识别分类总体查全率提高了2.48%,达到92.32%,对于较为复杂的形状（如T形、十字形）识别也具有更高的精度,查准率分别提高了13.83%和24.53%。实验结果表明本文提出的方法对建筑物形状分类的效果有明显提升,能够实现常见建筑物形状的有效分类,为下一步的建筑物化简、典型化等综合操作打下了基础。  相似文献   

基于改进DeepLabv3+孪生网络的遥感影像变化检测方法     

赵祥  王涛  张艳  郑迎辉  张昆  王龙辉 《地球信息科学学报》2022,24(8):1604-1616

传统遥感影像变化检测方法依赖人工构建特征,算法设计复杂且精度不高;而将2幅不同时相影像叠加后输入神经网络的遥感影像变化检测方法会造成不同时相的特征相互影响,难以保持原始影像的高维特征,且模型鲁棒性较差。因此,本文提出一种基于改进DeepLabv3+孪生网络的遥感影像变化检测方法,以经典DeepLabv3+网络的编解码结构为基础对网络进行改进：① 在编码阶段利用共享权值的孪生网络提取特征,通过2个输入端分别接收2幅遥感影像,以保留不同时相影像的高维特征;② 在特征融合中用密集空洞空间金字塔池化模型代替空洞空间金字塔池化模型,通过密集连接的方式结合每个空洞卷积的输出,以提高对不同尺度目标分割的精度;③ 在解码阶段中针对不同层级特征图信息差异较大,难以融合的问题,引入基于注意力机制的特征对齐模型,引导不同层级的特征对齐并强化学习重要特征,以提升模型的鲁棒性。应用开源数据集CDD验证本文方法的有效性,并与UNet-EF、FC-Siam-conc、Siam-DeepLabv3+和N-Siam-DeepLabv3+网络对比试验。试验结果表明,本文方法在精确率、召回率、F1值和总体精度上达到87.3%、90.2%、88.4%、96.4%,均高于UNet-EF、FC-Siam-conc、Siam-DeepLabv3+网络和N-Siam-DeepLabv3+网络,检测结果较为完整,对边界的检测也更为平滑,且对尺度变化具有更高的鲁棒性。  相似文献   

A brief catalogue of failures: Framing evaluation and learning in fisheries resource management     

Kre Nolde Nielsen  Petter Holm 《Marine Policy》2007,31(6):669-680

We propose a framework for enabling a systematic evaluation of a fisheries resource management system, which we define as a feed-back mechanism coupled to a fishery. The resource management system includes four basic functions: diagnostics, intervention, goal setting, and decision making. This model allows for the development of an evaluation framework for fisheries resource management by facilitating a typology of failures. We suggest that the potential for systemic and interdisciplinary learning will be significantly enhanced through the process of developing such a framework.  相似文献   

A technique for conducting seismic refraction experiments on the ocean bed using bottom shots     

R. E. Kirk  K. Robertson  R. B. Whitmarsh  P. R. Miles 《Marine Geophysical Researches》1991,13(2):153-160

A technique has been devised for firing arrays of bottom shots on the ocean bed in depths upto 4000 m or more. Ten kilogram explosive charges are dropped from the surface while the shooting ship is navigated acoustically. They are detonated at preset times by an electronic timer which initiates an electrical detonator, detonating cord and cast PETN/TNT explosive. Ranges to ocean bottom seismographs, and the shot instants, can be calculated from the arrival-time differences of the direct and surface-reflected water waves. The accuracy, which is dependent on water-depth and range, was better than 22 m in range and 14 msec in shot instant for our experiments.  相似文献   

水下智能潜器的神经网络运动控制   总被引：10，自引：4，他引：10  

彭良  卢迎春  万磊  孙俊岭 《海洋工程》1995,(2)

本文介绍一种基于神经网络的水下智能潜器的运动控制方法，该方法通过在线学习，融控制与滤波为一体。计算机仿真与水池实验验证表明，该方法的控制与滤波性能良好，对环境的学习与适应能力强。该方法事实上可用于一般动力系统的控制。  相似文献