基于人工神经网络的沿海风速多步预测研究

基于气象历史观测资料,将长短期记忆网络LSTM方法和Transformer模型结合提出了混合短期风速多步预测模型BLSTM TRA。以山东半岛南部沿海6个台站为研究区域,通过气象台站观测数据构建数据集。经与2018年ECMWF模式6 h预报结果对比分析,得出如下结论：构建的BLSTM TRA多步预测模型可大幅度降低风速误差,BLSTM TRA的1 h单步预测结果和ECMWF预报模式结果对比,其RMSE平均降低了58.9%,MAE平均降低了63.2%;风速误差和大风统计过程分析发现,BLSTM TRA模型具有一定的抗干扰能力,可以抓住短时大风等敏感信息,对于大风预报结果明显优于ECWMF模式和传统LSTM模型。     

RESPONSE OF EQUATORIAL PACIFIC UPPER OCEAN CURRENT TO WESTERLY WIND BURSTS

The characteristics of the response of equatorial Pacific upper ocean current to westerly wind bursts(WWB)were analyzed in the frequency domain by using wind and ADCP data collected by the Shiyan3 during TOGA-COARE IOP,1992-1993.The preliminary results showed that the response consistedof an eastward surface jet at shallower than 60m depth,a westward counter current centering near100m and a shear layer between them,with the variations of all three being nonlinear and nearlysynchronous.The oceanic responses in the frequency domain were characterized by occurrences of a remotely forced mixed Rossby-gravity wave with period of 8-10 days in the surface jet andcountercurrent at shallower than 110 m depth,and two locally forced waves with periods of 24 daysand 4-5 days limited in shallower than 70m depth.These fluctuations of the responses depended much more on zonal wind than meridional wind.The results also revealed that the oceanic response toWWB resulted from momentum transport and energy propagation assoc     

体感温度客观分析方法研究

影响人体体感温度的气象要素主要包括温度、湿度、风以及太阳辐射,依据这些气象要素影响人体体感温度的原理,确定了平均气温与体感温度相同时的临界相对湿度、风速及太阳辐射值.以全国有水平面直接辐射曝辐量观测项目的77个气象台站的历年逐日4次定时观测资料为基础,把相邻两个白昼(夜间)的气温差作为因变量,相对湿度差、风速差、垂直辐射差(白昼)作为自变量,以旬为时问分辨率,建立了逐旬多元回归方程.各地所建立的逐句多元线性拟合方程拟合优度相当高,F统计量检验也充分证明线性回归方程总体成立,从气候意义、统计角度来看该方法都具有充分的可行性.     

GPS在地震仪器地理北方位定向中的应用

介绍了用GPS对台站数字化地震仪地理北方向的定位方法。试验表明,定向精度优于2′,完全满足地震仪器定向精度小于6′的要求。     

Zonal displacement of western Pacific warm pool and zonal wind anomaly over the Pacific Ocean

The thermal condition anomaly of the western Pacific warm pool and its zonal displacement have very important influences on climate change in East Asia and even the whole world. However, the impact of the zonal wind anomaly over the Pacific Ocean on zonal displacement of the warm pool has not yet been analyzed based on long-term record. Therefore, it is important to study the zonal displacement of the warm pool and its response to the zonal wind anomaly over the equatorial Pacific Ocean. Based on the NCDC monthly averaged SST (sea surface temperature) data in 2°×2° grid in the Pacific Ocean from 1950 to 2000, and the NCEP/NCAR global monthly averaged 850 hPa zonal wind data from 1949 to 2000, the relationships between zonal displacements of the western Pacific warm pool and zonal wind anomalies over the tropical Pacific Ocean are analyzed in this paper. The results show that the zonal displacements are closely related to the zonal wind anomalies over the western, central and eastern equatorial Pacific Ocean. Composite analysis indicates that during ENSO events, the warm pool displacement was trigged by the zonal wind anomalies over the western equatorial Pacific Ocean in early stage and the process proceeded under the zonal wind anomalies over the central and eastern equatorial Pacific Ocean unless the wind direction changes. Therefore, in addition to the zonal wind anomaly over the western Pacific, the zonal wind anomalies over the central and eastern Pacific Ocean should be considered also in investigation the dynamical mechanisms of the zonal displacement of the warm pool.     

冰雹云雷达回波自动识别系统

利用CAPPI资料对立体风暴进行识别，计算并提供出实用的风暴结构参数，采用矩心踊跃法和矩不变量法相结合对单体风暴和混合性风暴回波进行跟踪；最小二第线性外推预报；根据WSD－88D的冰雹算法，在风暴结构基础上本文建立了Windows98操作平台上冰雹识别系统，经单站1年11次强对流天气过程的资料检验，结果表明：雹云识别精度达82％。系统建立了大量人机对话框以方便用户，增加实用性及推广性。     

THE CHINESE ECOLOGICAL RESEARCH NETWORK

ＴＨＥＣＨＩＮＥＳＥＥＣＯＬＯＧＩＣＡＬＲＥＳＥＡＲＣＨＮＥＴＷＯＲＫＺｈａｏＪｉａｎｐｉｎｇ（赵剑平）（ＢｕｒｅａｕｏｆＲｅｓｏｕｒｃｅｓａｎｄＥｎｖｉｒｏｎｍｅｎｔａｌＳｃｉｅｎｃｅｓ，ｔｈｅＣｈｉｎｅｓｅＡｃａｄｅｍｙｏｆＳｃｉｅｎｃｅｓ，Ｂｅ...     

基于超像元词包特征和主动学习的高分遥感影像变化检测

为解决高分辨率遥感影像变化检测中存在底层特征缺乏语义信息、像元级的检测结果存在“椒盐”现象以及监督分类中样本标注自动化程度较低,本文提出一种基于超像元词包特征和主动学习的变化检测方法。首先采用熵率分割算法获取叠加影像的超像元对象;其次提取两期影像像元点对间的邻近相关影像特征（相关度、斜率和截距）和顾及邻域的纹理变化强度特征（均值、方差、同质性和相异性）,经线性组合作为像元点对的底层特征;然后基于像元点对底层特征利用BOW模型构建超像元词包特征,并采用一种改进标注策略的主动学习方法从无标记样本池中优选信息量较大的样本,且自动标注样本类别;最后训练分类器模型完成变化检测。通过选用2组不同地区的GF-2影像和Worldview-Ⅱ影像作为数据源进行实验,实验结果中2组数据集的F1分数分别为0.8714、0.8554,正确率分别为0.9148、0.9022,漏检率分别为0.1681、0.1868,误检率分别为0.0852、0.0978。结果表明,该法能有效识别变化区域、提高变化检测精度。此外,传统主动学习方法与改进标注策略的主动学习方法的学习曲线对比显示,改进的标注策略可在较低精度损失下,有效提高样本标注自动化程度。     

����ɽ-���׶��ִ��ؿ��α�������Ӧ�䳡�ֲ�

????1998??2011?????????????90??GPS?????????????????????-????????????????????????????????????????????????????GPS????????????1???о??????????????????????????????????????????????????????2)?????????????????????????????????????????????????????????3)????????????????????????й????????????????з??????????????????????????????     

《军官地图集》电子版的设计

本文系统阐述了《军官地图集》电子版的设计和研制中的若干问题,包括《军官地图集》电子版的总体设计、数据库设计、自动分幅设计以及专题图设计等。总体设计主要包括系统的结构设计、地理数据库设计、各功能模块设计等几部分。在地图集系统中,多比例尺海量地理数据的应用必然要求有强大的数据库支持,系统中采用了易于存储、便于管理的数据结构,并对数据库的各类数据描述表进行了详细说明。自动分幅设计是电子地图集的一个新功能,本文提出了电子地图集的分幅方式:经纬线分幅、矩形分幅、名称分幅和区域分幅,并阐述了它们各自的优缺点。《军官地图集》电子版中的专题地图在设计上突出了图集的系统性和军事特色,本文提出了该电子版中专题地图设计的原则和方法。根据该电子版的设计和制作过程,同时提出了电子地图集制作的步骤,并阐明了电子地图集设计中的不足和发展方向。