Calibrating a glaciological model of the Greenland ice sheet from the Last Glacial Maximum to present-day using field observations of relative sea level and ice extent

We constrain a three-dimensional thermomechanical model of Greenland ice sheet (GrIS) evolution from the Last Glacial Maximum (LGM, 21 ka BP) to the present-day using, primarily, observations of relative sea level (RSL) as well as field data on past ice extent. Our new model (Huy2) fits a majority of the observations and is characterised by a number of key features: (i) the ice sheet had an excess volume (relative to present) of 4.1 m ice-equivalent sea level at the LGM, which increased to reach a maximum value of 4.6 m at 16.5 ka BP; (ii) retreat from the continental shelf was not continuous around the entire margin, as there was a Younger Dryas readvance in some areas. The final episode of marine retreat was rapid and relatively late (c. 12 ka BP), leaving the ice sheet land based by 10 ka BP; (iii) in response to the Holocene Thermal Maximum (HTM) the ice margin retreated behind its present-day position by up to 80 km in the southwest, 20 km in the south and 80 km in a small area of the northeast. As a result of this retreat the modelled ice sheet reaches a minimum extent between 5 and 4 ka BP, which corresponds to a deficit volume (relative to present) of 0.17 m ice-equivalent sea level. Our results suggest that remaining discrepancies between the model and the observations are likely associated with non-Greenland ice load, differences between modelled and observed present-day ice elevation around the margin, lateral variations in Earth structure and/or the pattern of ice margin retreat.     

Numerical and Experimental Analysis of A Ducted Propeller Designed by A Fully Automated Optimization Process Under Open Water Condition

A fully automated optimization process is provided for the design of ducted propellers under open water conditions, including 3D geometry modeling, meshing, optimization algorithm and CFD analysis techniques. The developed process allows the direct integration of a RANSE solver in the design stage. A practical ducted propeller design case study is carried out for validation. Numerical simulations and open water tests are fulfilled and proved that the optimum ducted propeller improves hydrodynamic performance as predicted.     

面向地图图片识别的两种卷积神经网络分析

针对目前多种网络基于小数据集或国际通用的一些标准数据集进行训练与测试,无法满足从互联网图片中高效识别出地图图片的需求,该文构建了一套基于卷积神经网络的互联网图片分类体系,对自动采集的92543张互联网图片经类型标注后形成样本库,将样本输入残差网络ResNet50和轻量级网络SqueezeNet中进行训练及测试.结果 显示:在地图类中,ResNet50的精确率、召回率分别比SqueezeNet高2.01％、0.32％;前者所耗费的训练、测试时间分别为后者的2.51倍、6.43倍,将上述指标进行归一化处理来综合评价两种网络在地图图片识别中的优越性.得出结论:SqueezeNet网络在地图识别应用中更具优越性,可有效提升在互联网地图图片中所包含"问题地图"的审查效率和及时响应服务.     

基于Maptalks的外业草图绘制捕捉插件的设计与实现

作为中国为数不多的自主研发的开源地图引擎,Maptalks以其优良的性能和灵活的插件构建方式,在越来越多的项目中得以应用.鉴于其正处于稳步开发的阶段,部分功能还不够完善,还未实现实际项目中的草图绘制编辑所需的捕捉功能.针对此问题,开发了基于Maptalks的捕捉插件,实现了地图编辑中的捕捉功能.此成果一方面满足了外业草图绘制简单捕捉编辑的需求,另一方面完善了Map-talks的功能,扩大了它的应用范围.     

Power Spectrum Density of Stochastic Oscillating Accretion Disk

In this paper, we employ a stochastic oscillating accretion disk model for the power spectral index and variability of BL Lac object S5 0716+714. In the model, we assume that there is a relativistic oscillation of thin accretion disks and it interacts with an external thermal bath through a friction force and a random force. We simulate the light curve and the power spectrum density (PSD) at (i) over-damped, (ii) critically damped and (iii) under-damped cases, respectively. Our results show that the simulated PSD curves depend on the intrinsic property of the accretion disk, and it could be produced in a wide interval ranging from 0.94 to 2.05 by changing the friction coefficient in a stochastic oscillating accretion disk model. We argue that accretion disk stochastic oscillating could be a possible interpretation for observed PSD variability.     

A diagnostic study of the asymmetric distribution of rainfall during the landfall of typhoon Haitang (2005)

The precipitation during landfall of typhoon Haitang (2005) showed asymmetric structures (left side/right side of the track). Analysis of Weather Research and Forecasting model simulation data showed that rainfall on the right side was more than 15 times stronger than on the left side. The causes were analyzed by focusing on comparing the water vapor flux, stability and upward motion between the two sides. The major results were as follows: (2) Relative humidity on both sides was over 80%, whereas the convergence of water vapor flux in the lower troposphere was about 10 times larger on the right side than on the left side. (5) Both sides featured conditional symmetric instability [MPV (moist potential vorticity) <0], but the right side was more unstable than the left side. (6) Strong (weak) upward motion occurred throughout the troposphere on the right (left) side. The Q vector diagnosis suggested that large-scale and mesoscale forcing accounted for the difference in vertical velocity. Orographic lift and surface friction forced the development of the asymmetric precipitation pattern. On the right side, strong upward motion from the forcing of different scale weather systems and topography caused a substantial release of unstable energy and the transportation of water vapor from the lower to the upper troposphere, which produced torrential rainfall. However, the above conditions on the left side were all much weaker, which led to weaker rainfall. This may have been the cause of the asymmetric distribution of rainfall during the landfall of typhoon Haitang.     

低温雨雪过程的粒子群-神经网络预报模型

利用逐日气温和降水量数据、NCEP/NCAR再分析资料以及预报场资料,通过分析提取我国南方区域持续性低温雨雪过程及其预报因子,使用粒子群-神经网络方法建立非线性的统计集合预报模型 (PSONN-EPM),对我国南方区域持续性低温雨雪过程进行预报试验。结果表明:以过程的冷湿程度及影响范围为标准,将低温雨雪过程分为一般过程和严重过程,并建立不同的预报模型效果较好。通过10 d独立样本预报试验看,基于粒子群-神经网络方法建立的集合预报模型比基于逐步回归方法建立的预报模型的预报平均相对误差小,对严重过程预报能力高于对一般过程预报,且这种非线性统计集合建模方法在建模过程中不需要调整神经网络参数,在实际预报业务中值得尝试。