基于YOLOv5算法的飞机类型光学遥感识别

飞机类型检测是遥感影像分析领域的研究热点,在机场监控和情报分析等应用中起着重要的作用。其中,深度学习方法作为遥感影像分析领域广泛应用的手段之一,在飞机类型检测任务中仍面临许多问题,如使用未公开的自制数据集、实验复现困难、无法验证泛化能力等。同时,光学遥感影像易受光照条件、云雨变化等因素影响,使检测任务更加困难。为了解决这些问题,本文首先利用MTARSI数据集对样本进行筛选,再结合Google Earth等开源方法收集飞机影像,采用随机旋转、改变亮度等方法构建新的飞机类型检测数据集。其次,采用YOLOv5作为基础网络框架,针对其多层卷积和池化操作可能会削弱或完全丢失飞机特征的问题,进行多尺度优化训练,有效检测飞机类型特征。最后,利用跨数据集验证模型的泛化能力。实验结果表明,本文方法能准确、有效地检测出光学遥感影像中的飞机的具体类型,具有较强的鲁棒性和泛化能力,跨数据集进行飞机类型检测正确率达到82.12%,可为智能化的飞机目标语义分析、星上应用等研究提供技术支撑。     

Sidewall effects and sand trap efficiency in a large wind tunnel

Aeolian sand transport is a widespread physical phenomenon on the surface of Earth, as well as on Mars and Titan. Accurate measurements of the components of the transport system are necessary if we are to understand the nature of the physical processes. Sand traps are typically used to measure sediment transport rates, and issues associated with the sampling efficiency of traps and the development of reliable traps have received considerable attention in recent decades. In this study, we measured aeolian transport rate at five distances from a wind tunnel sidewall using a vertically‐segmented sand trap. Total transport rates were determined by weighing the bed sediment before and after each experiment, and with and without a trap installed. The following results were obtained: (1) sand transport increased linearly with the distance away from the sidewall, and the appropriate location to measure maximum transport is within the central 20% of the wind tunnel; (2) current methods overestimate the sampling efficiency of sand traps when comparing trap data to transport rate data obtained by weighing sand moved through the entire tunnel because the effects of the sidewalls in decreasing total transport are neglected; (3) the efficiency of the vertically‐segmented trap that we tested ranged from 11.57% to 31.68% using our revised methods, whereas standard methods caused efficiency to be overestimated by 32–72% of the efficiency; (4) using either method, the efficiency of the trap increased exponentially with shear velocity for the range we used. Copyright © 2017 John Wiley & Sons, Ltd.     

Case studies on the parameterization schemes of sea ice fragmentation for ocean waves

Ocean Dynamics - Sea ice on the Southern Ocean has large seasonal variations. Floe size distribution has an important influence on the dynamic and thermodynamic processes of sea ice in the region...     

墨卡托投影最佳基准纬线确定方法

针对海图制图中墨卡托投影以投影区域中纬线作为基准纬线投影变形大的问题,分别在投影区域长度变形最小、面积变形最小和边纬线长度变形相等的条件下确定最佳基准纬度.以北半球为例,3种条件下确定的基准纬度均大于中纬度,南北纬度差为10°、20°、30°、40°时,与中纬度差值最大分别为2.0936°、5.0621°、9.1315...     

辽河东部凹陷新生界火成岩岩性、岩相特征和储层控制因素

基于110口探井1 100 m岩心、500余口探井总长107 000 m火成岩井段资料,通过岩心观察、岩矿鉴定、物性测试和含油气性分析,研究了辽河坳陷东部凹陷新生界不同层系火成岩岩性、岩相发育特征和储层控制因素。结果显示：本区发育火山岩、侵入岩、沉火山碎屑岩3大类16种岩性,其中玄武岩类占91.0%;发育6相16种亚相,其中溢流相占63.0%;发育4类9种储集空间,次生孔隙和次生裂缝占主体。砾-粒间孔隙和裂缝发育的粗面岩类、玄武质火山碎屑（熔）岩和火山口-近火山口相带的侵出相、火山通道相和爆发相是最有利的储集岩性、岩相带。储层主要受岩性、岩相、断裂等多方面控制,岩性决定储集空间类型和储层微观特征,岩相控制储层宏观分布规律,断裂控制岩体分布范围、火山口位置和储层有效性。近油源、靠断裂的优势岩性、岩相带是火成岩油气勘探最有利目标。     

湖相碳酸盐岩白云石化流体来源、作用机制及物质过程

湖相白云岩具有重要的油气地质意义,广泛发育于我国东部新生代地层中。本文通过大量资料统计和分析前人对国内湖相白云岩的研究成果,系统阐述了国内主要湖盆的湖相白云岩的成因、形成流体来源及相关物质的响应。根据白云石化流体的碳氧同位素、锶钡比、有序度、pH值、Eu异常、Ce异常等各类指标,将白云石化流体分为原始湖泊、外来淡水、热水热液和海源流体等类型。不同类型流体来源改造下生成白云岩所伴生的矿物组合不同;不同白云石化流体通过微生物作用、准同生作用、热液作用和埋藏作用等方式对原始碳酸盐岩进行白云石化作用,不同白云石化作用机制直接制约了其物质响应过程。提出了未来利用定量分析技术深入开展白云石化流体来源分析、准确判断成岩流体类型或成岩作用机制是进一步研究的方向。     

胶东中生代耦合金成矿作用分析

该文结合华北东部中生代构造体制转折、郯庐断裂带左行走滑和华北地幔热柱多级演化3次重大地质事件,对胶东金成矿原因进行系统分析,提出胶东中生代金矿同期耦合作用成因的观点,认为胶东金矿是上述三者同期耦合成矿作用的结果,且幔源成矿流体在耦合成矿过程中发生分异富集演化,使得胶东矿集区形成如此大规模、高品位和不均衡分布的金矿。     

Differences in CO2-Water-Rock Chemical Reactions among 'Sweet Spot' Reservoirs: Implications for Carbon Sequestration

The Lucaogou Formation, located in the Jimsar Sag, Junggar Basin, NW China, has great potential for shale oil resources. In the process of CO₂-EOR(CO₂ enhance oil recovery), mineral dissolution, precipitation and transformation, leading to the local corrosion or blockage of reservoirs, have a significant influence on recovery. In this study, a combination of high-temperature and high-pressure laboratory experiments and coupled temperature/fluid-chemistry multifield numerica...     

潮白河中游沉积物氮磷和有机质分布特征及评价

潮白河是京津冀地区的重要河流。本研究以潮白河中游段为研究区,针对该区段沉积物有机质、氮磷污染程度不明的问题,通过对研究区130个采样点表层沉积物中有机质(OM)、总氮(TN)和总磷(TP)的测定,分析了表层沉积物中有机质和氮磷的分布特征及污染来源,将有机指数和有机氮污染指数评价法相结合开展污染及生态风险评价。结果表明:(1)潮白河中游表层沉积物中OM含量变化范围为0.17%~13.15%,平均含量为5.32%; TN的含量为0.005~1.028 g/kg,平均含量为0.192 g/kg;TP的含量为0.367~4.825 g/kg,平均含量为1.662 g/kg。(2)相关性分析表明,OM与TP、OM与TN均呈弱相关,而TN与TP呈强相关。这是因为TN和TP来源相近,主要来自外源,与OM来源有一定差异。根据TOC/TN特征分析,OM主要受藻类、浮游植物和非纤维束植物等影响。(3)有机指数评价结果显示,潮白河中游段表层沉积物监测点中60%为清洁,33.85%为较清洁,4.62%为尚清洁,1.54%为有机污染。有机污染点主要位于运潮减河入潮白河口和田贾庄排干渠入潮白河口。区域有机氮评价结果以清洁级别为主,整体较好。