Sub-seasonal Prediction of the South China Sea Summer Monsoon Onset in the NCEP Climate Forecast System Version 2

This study depicts the sub-seasonal prediction of the South China Sea summer monsoon onset(SCSSMO) and investigates the associated oceanic and atmospheric processes, utilizing the hindcasts of the National Centers for Environmental Prediction(NCEP) Climate Forecast System version 2(CFSv2). Typically, the SCSSMO is accompanied by an eastward retreat of the western North Pacific subtropical high(WNPSH), development of the cross-equatorial flow, and an increase in the east-west sea surface temperat...     

Estimation and prediction of plastic waste annual input into the sea from China

Marine plastic debris has been a pervasive issue since the last century, and research on its sources and fates plays a vital role in the establishment of mitigation measures. However, data on the quantity of plastic waste that enters the sea on a certain timescale remain largely unavailable in China. Here, we established a model using material flow analysis method based on life cycle assessment to follow plastic product from primary plastic to plastic waste with statistical data and monitoring data from accurate sources. This model can be used to estimate and forecast the annual input of plastic waste into the sea from China until 2020. In 2011, 0.547 3–0.751 5 million tons of plastic waste entered the seas in China, with a growth rate of 4.55% per year until 2017. And the amount will decrease to0.257 1 to 0.353 1 million tons in 2020 under the influence of governmental management. The amount of plastic waste discharged from coastal areas calculated in this study was much larger than that from river, thus it is suggested to strengthen the governance and control of plastic waste in coastal fishery activities in China in order to reduce the amount of marine plastic waste input.     

灰色稳健总体最小二乘估计及高铁路基变形预测

最小二乘估计和部分变量误差模型的总体最小二乘估计不具备抵御粗差的能力。鉴于粗差可能同时出现在灰色白化微分方程的观测值和系数矩阵中,本文提出基于IGGⅢ抗差方案的部分变量总体最小二乘稳健估计。结合仿真数据和高铁路基观测数据,系统地比较稳健最小二乘、部分变量总体最小二乘、本文算法参数估计结果和算法稳定性。结果表明,本文算法预测精度高,可以应用到高铁路基沉降预测中。     

一种基于熵权法的小波去噪复合评价指标

传统的评价指标在真值未知的情况下不能满足小波去噪质量评价的要求。为此,借助变化率特征重新构建均方根误差变化量和平滑度变化量两个指标,利用熵权法定权将归一化后的两个指标线性组合,所得到的新指标即为复合评价指标。该方法借助指标的变化率随分解层数的增加表现出明显的收敛特性来确定去噪最优分解层数。实验表明,该方法能够在真值未知的情况下准确地指导小波分解,确定去噪最优分解层数,从而达到最优去噪效果。     

室内用户语义位置预测研究

位置预测技术可以提前预知用户下一时刻的位置,在基于位置的服务（Location-based Service,LBS）领域中发挥着极其重要的作用。现有的位置预测技术大多仅使用用户的地理轨迹,仅使用地理轨迹挖掘出来的用户移动模式易受地理特性的限制缺乏深层次的语义信息。本文基于某商场群体用户的室内轨迹数据和语义信息预测用户下一个时刻语义位置。语义位置预测包括停留区域识别、停留区域语义匹配、语义位置建模。在停留区域识别阶段,为减少室内停留时间不固定对停留区域识别的影响,本研究提出了一种新型的时空凝聚层次聚类算法（Spatial-Temporal Agglomerative Nesting, ST-AGNES),该算法具有思想简单、超参数少、自动生成聚类个数等优点。在语义匹配阶段,引入了吸引度规则,充分利用停留区域所有轨迹点与室内高密度的商铺名称信息做匹配。最后,采用长短型记忆神经网络模型（Long Short-Term Memory,LSTM）挖掘群体用户的语义位置模式并预测用户未来的语义位置,实验预测正确率达到61.3%。     

Deterministic non-linear wave prediction using probe data

The accurate prediction of non-linear sea states represents a great challenge, with a number of applications in oceanography, marine engineering, security of people and marine transportation, etc. In this paper, we report on the development of two efficient deterministic prediction models for 2D irregular wave-fields. These models are based on the exploitation of wave elevation time series given by one or more probes and on the use of two different numerical models for the wave-fields simulation. Two effective data assimilation processes are developed to improve the wave-field estimates obtained from the study of one or several probe signals, so that we obtain proper initial conditions for the forecast. The assimilation schemes have been validated on the reconstruction of model-generated observations and accurate predictions of the corresponding synthetic wave-fields evolution have been obtained.     

基于数字高程模型定量分析长江口深水航道工程治理效果

长江口深水航道工程于1998年开工建设,2002年和2005年一期和二期工程先后竣工,分别达到8.5和10.0 m通航水深。自2006年三期工程实施以来,北槽航道中段连续4a发生严重淤积,年疏浚维护量平均达6×107m3,影响三期工程目标的如期实现。通过自主开发的数字高程模型定量分析平台,建立时间序列的空间分布属性数据库,对长江口深水航道工程治理过程前后的河床冲淤变化规律和工程效果进行量化分析,分析结果揭示了长江口南港北槽深水航道近期淤积的泥沙来源、淤积过程、主要淤积原因和淤积部位,从而为工程治理对策提供科学依据。     

Control of Aptian palaeobathymetry on turbidite distribution in the Buchan Graben,Outer Moray Firth,Central North Sea

It is widely recognised that palaeobathymetry is a key control on the distribution of turbidite deposits. Thus, the utilisation of palaeobathymetric surfaces as an input for numerical turbidity current modelling offers a potentially powerful method to predict the distribution of deep marine sands in ancient (subsurface or outcrop) successions. Such an approach has been tested on an Aptian turbidite deposit from the Buchan Graben, UK Central North Sea, where modelled sand distributions could be quality controlled against available well data.     

基于注意力机制与自适应时序分解的气温预报模型

针对传统方法在捕捉气象序列长期依赖关系及泛化性能上的不足,提出了一种基于稀疏注意力与自适应时序分解的气温预报模型（ATFSAS）。该模型整体采用编码器 解码器架构,结合稀疏注意力机制以有效捕捉气象观测数据间的长期依赖性。为减少编码过程中造成的冗余,提出了一种信息蒸馏方法。通过结合多层解码器与自适应时序分解单元,逐步细化预报信号中的周期性和趋势性分量,实现了较为精准的气温预报。基于德国耶拿气象数据集,进行24 h精细化气温预报,其平均绝对误差为1.7108℃。基于中国地面气候资料日值数据集,进行中短期日平均气温预报和多地区单日平均气温预报,相比传统模型LSTM,ATFSAS模型预报结果的平均绝对误差分别提升了35.56%和23.66%。     

内蒙古区域数值预报对探空和地面观测资料的敏感性试验研究

基于内蒙古睿图预报系统的低分辨率版本和WRFDA-FSO诊断工具,评估2021年7月现有探空和地面观测对内蒙古睿图预报系统预报的影响。该方法计算代价相对低廉,并允许根据观测变量、观测类型、气压层次、地理区域等观测的子集对观测影响进行划分。代价函数为以干总能量为度量的背景场和分析场的预报误差之间的差异。结果表明:观测影响的总体总和为负,观测对预报起正贡献作用。对12 h预报误差减小贡献最大的观测来自探空观测的动力变量(U、V风分量)。而单时次单位数量平均观测影响,探空观测的贡献约为地面观测的1/2。探空观测对12 h预报误差减小从近地面层至模式层顶均保持正贡献作用,并在对流层中低层和对流层高空急流层存在两个极大值区域;地面观测在850 hPa以下低层正贡献占比明显。探空观测在被同化系统同化时均总体具有有利的影响,也反映出探空观测数据稳定、质量较高的特征;地面观测对12 h预报误差减小起正贡献作用次数最多的区域在河套地区尤为显著。同时,探讨了需进一步提高地面观测资料同化率的问题。