基于机器学习的格点气温预报订正方法

使用2017年9月至2021年3月国家级业务化运行的智能网格实况分析产品和欧洲中期天气预报中心全球模式(EC)产品,根据湖北省的地理分布特征构建6个分区,采用基于LightGBM机器学习算法建立的气温预报方法,生成湖北省0.05°×0.05°格点气温预报产品。利用2021年4—9月的预报产品和格点实况资料进行检验,结果表明:基于机器学习的气温预报方法(MLT)取得了较好的预报效果,其在0～72 h时效内优于中央气象台下发的气温精细化指导预报(SCMOC)和EC产品;MLT在山区的误差较平原大,但山区的订正幅度大于平原,日最高气温的订正幅度大于日最低气温的订正幅度;4—9月MLT、SCMOC、EC产品的平均绝对误差(MAE)日变化都呈现了白天偏高、夜间偏低、午后凸起的单峰特征,MLT的MAE值较SCMOC和EC产品的更低,并且在转折性天气中仍具有优势;站点检验与格点检验结论一致,基于格点建模的气温预报产品对站点预报同样得到了订正。机器学习在格点气温的模式订正方面可以作为一个行之有效的手段。     

基于地震反演方法的奥陶系顶部含隔水层探测

在山西阳泉泊里矿区,太原组K₂灰岩是15号煤层上部主要的含水层,查明其富水分布特征对上下组煤层安全开采至关重要。为了准确得到K₂灰岩的富水分布区域,首先,利用常规的波阻抗反演获取精确的K₂灰岩空间展布特征。然后,结合皮尔逊相关系数法与交叉验证−逐步回归法优选出9种地震属性,构成网络的训练数据。此外,引入适合于时序数据处理且能够捕捉测井曲线前后相关性的长短期记忆神经网络(LSTM),构建智能化、多变量LSTM视电阻率预测模型,以精确地预测研究区视电阻率进而得到地层富水性分布特征。同时,分别利用常规多属性回归算法与多变量LSTM模型在井点位置建立电阻率测井曲线与地震属性井旁道之间的映射关系。最后,将井点处训练好的网络模型推广至无井区得到全区视电阻率体,根据视电阻率值的高低、矿区地质构造与陷落柱发育情况圈定灰岩富水区。实际数据的测试结果表明：与常规多属性回归算法相比,多变量LSTM模型预测误差小,与测井相关系数高,说明多变量LSTM模型可以更加精确地预测出工区视电阻率,在含煤地层的富水性预测中有较好的应用价值。\t\t\t\t

     

A Method for Automatically Selecting Aids to Navigation Based on their Spatial Influence Domains

To address the limitations of manually selecting aids to navigation (AtNs) on charts, a method for automatically selecting AtNs based on their spatial influence domains (SIDs) is proposed. First, the associations between the spatial attributes of an AtN are analyzed. Second, an SID of the AtN is defined, and a model of the SID is constructed based on the associations between the spatial attributes. Third, the importance of the location of the AtN is weighted based on the SID model. Fourth, an algorithm to automatically select AtNs based on the maximum coverage of the SIDS of preselected AtNs is developed. Finally, several AtNs are selected automatically using the algorithm. The experimental results demonstrate that (1) the proposed method can automatically select AtNs and the results comply with the requirements; (2) the automatic selection can eliminate the human-induced errors or the inconsistent results of manual selections from different operators; and (3) the efficiency of the proposed method is higher than that of current manual methods.     

海水养殖鱼类细菌性疾病研究概况

本文概述性地介绍了发生在海水养殖中的常见细菌性疾病包括弧菌现、链球菌症、卡诺氏菌症、细菌性肾病、巴斯德氏杀菌症、爱德华氏症、滑走细菌症和杀鲑气单胞菌症等的主要症状、流行情况及病原菌特征，通过遗传育种筛选有抗病力的养殖品种是人们正在努力的方向，目前的研究重点主要在鱼类抗病能力方差分析研究及抗病育种的方法研究，免疫接种是提高养殖鱼类对某一特定性病原菌的最有效的方法，目前已有三十多种菌苗上市，通过使用免疫刺激剂提高鱼类的非特异性免疫力已显示一定的应用前景，虽然抗菌素由于存在抗药菌株和鱼体残留，但它仍是目前对细菌性疾病疾病的常规疗法，而海洋藻类提取物对细菌的抑制作用已被人们所认识，它将会成为一种有潜力的抗菌药物。     

南极长城站附近企鹅栖息地粪土沉积地层放射性核素210Pb定年

用γ谱方法测定了南极长城站附近特有生物群落栖息地沉积物中放射性核素含量,地表沉积物中40K,137Cs,210Pb,226Ra,228Ra,228Th.和238U平均比活度,分别为143,7.56,24.1,3.65,5.36,4.15和6.5Bq/kg.同时测试了阿德雷岛企鹅栖息地粪土沉积地层中放射性核素含量,利用其中的210Pb比活度,210Pb_ex垂向变化特征,推演沉积物的沉积速率和地质历史年代:其中AD1-a柱样时间跨度约为74a(1928~2002年),据此计算了沉积速率为0.063mm/a(r=0.794),并讨论了在南极特定条件下,放射性核素示踪对定年影响以及与区域现代气候环境变化的内在联系.     

A System for Fast Cartographic Sounding Selection

Soundings on nautical charts provide information about the shape of the ocean bottom between chart depth curves. A single chart may have thousands of soundings posted on it. This article describes a system for automatic cartographic sounding selection. In this system, a new algorithm for automated selecting soundings is developed. The algorithm is employed in such a way that it can guarantee (a) fast and accurate selecting sounding, (b) processing the circular depth curves, (c) solving the overplotting problems between depth curves and soundings, and (d) keeping selected soundings “prettier.” The system was intensively tested using real data sets, and its superiority has been revealed by the testing results.     

中国民营企业对外直接投资区位选择研究--以温州民营企业为例

自上世纪90年代后半期开始,世界经济呈现一体化整合的态势,全球和中国对外直接投资的规模不断扩大,温州民营企业对外直接投资也快速发展,被商务部称为“走出去”的“温州模式”。本文基于全球生产网络理论,建立了民营企业对外直接投资区位选择的优势拓展因素、制度利用因素和优势寻求因素的分析框架。通过温州民营企业对外直接投资区位选择的实证分析结果来看,优势拓展因素仍起到重要作用,优势寻求型的技术创新资源、市场营销渠道、市场信息资源是区位选择新的重要因素,但制度因素的作用也不可忽视。     

MASCULINISM,POWER, AND DISCOURSES OF EXCLUSION IN BRIAN BERRY’S “SCIENTIFIC” GEOGRAPHY

This paper examines the impact of changes in industry mix, changes in technology, differential plant growth, plant entry and plant exit on metropolitan labor productivity growth between 1963 and 1997. Analysis is based upon unpublished plant-level data from the United States Bureau of the Census. We show that manufacturing productivity varies markedly between metropolitan areas in the United States. The most influential components of productivity growth are technological changes within incumbent plants and changes in industry mix. Significant differences in the relative contributions of these components of productivity change exist across metropolitan areas. Regional differences in rates of plant openings and plant closures also exert considerable impact on metropolitan productivity improvement.     

G-凸空间内的连续选择定理和重合点定理

在G-凸空间内得到了一些新的连续选择定理,并利用连续选择定理证明了非紧G-凸空间内及乘积G-凸空间内的重合点定理.这些定理改进并推广了最近文献中的一些已知结果.