A-FPN算法及其在遥感图像船舶检测中的应用

光学遥感图像船舶检测主要面临两个挑战:光学遥感图像背景复杂,船舶检测易受海浪、云雾及陆地建筑等多方面干扰;遥感图像分辨率低,船舶目标小,对于其分类与定位带来很大困难;针对上述问题,在FPN的基础上,提出一种融入显著性特征的卷积神经网络模型A-FPN (Attention-Based Feature Pyramid Networks)。首先,利用卷积提取图像特征金字塔;然后,利用顶层金字塔逐级构建显著特征层,抑制背景信息,通过金字塔顶层的细粒度特征提高浅层特征的表达能力,构建自上而下的多级显著特征映射结构;最后利用Softmax分类器进行多层级船舶检测。A-FPN模型利用显著性机制引导不同感受下的特征进行融合,提高了模型的分辨能力,对遥感图像处理领域具有重要应用价值。实验阶段,利用公开的遥感目标检测数据集NWPU VHR-10中的船舶样本进行测试,准确率为92.8%,表明A-FPN模型适用于遥感图像船舶检测。     

光照强度和光色对番红砗磲(Tridacna crocea)氨氮、活性磷酸盐及氧代谢的影响

采用室内实验生态学方法,以水体中氨氮、活性磷酸盐和溶氧变化为指标研究了光照强度和光色对番红砗磲代谢的影响。结果表明:(1)在0—10000lx光照时,番红砗磲排氨量逐渐减少,在光照12000lx时转为吸收海水中氨氮,并在16000lx时达到最高[1.89μg/(ind·h)];在2000lx光照条件下,番红砗磲可吸收海水中活性磷酸盐并释放氧气。其活性磷酸盐吸收量随光照强度增强而增加,在16000lx时达到最高峰。在实验光照范围内,番红砗磲产氧率随光照强度的增强逐渐增加。上述结果显示番红砗磲最佳光照强度约为16000lx。(2)当番红砗磲自黑暗转移至各光照组后,其氨氮、活性磷酸盐吸收率均随时间逐渐降低,而产氧率则逐渐升高。(3)光色显著影响番红砗磲合成代谢,蓝光最佳,红光次之,白光最差。     

非线性波消波及波场分布研究

本文基于雷诺平均N-S方程,并结合RNG k-e方程建立了粘性数值波浪水槽,对不同波陡、不同相对水深、不同相对波高的非线性规则波的阻尼消波问题和波场分布进行研究。文中提出了两种描述消波区内部阻尼变化的阻尼函数,分别适用于小波陡情形和高波陡情形。研究结果表明,小波陡组消波区可设为一个波长,阻尼系数取10~4～10~5即可满足消波要求,计算结果与实验结果及造波理论吻合良好;高波陡组消波区可设为两个波长,阻尼系数取10~4～10~5亦可满足消波要求,计算结果与实验结果吻合良好。此外,当波陡较小时,波场内反射情况的小幅改变即可对整个波场造成影响,特别是当水深较浅时这种影响极为明显,需谨慎考虑。当波陡较大时,水波能量较高,整个波场沿水波传播方向可观测到明显的衰减现象,在具体试验中需进行考虑。     

An atlas of the optical spectrum of supernova 1987A in the large magellanic cloud

Photographic spectra of SN1987A in the LMC have been obtained from 1987 February 25 to 1988 June 30. Microdensitometer tracings of these have been reduced to intensity and corrections for instrumental response have been applied to the spectra. This paper presents these data in an atlas format, discusses the reduction procedures in detail, and presents radial velocity measurements of selected lines in the spectra     

利用自回归模型检测地球极移重力效应

同时采用4个台站的国际超导重力仪长期连续重力观测资料和国际地球自转服务中心提供的同步地球自转参数，研究了极移引起的地球重力场变化特征。利用自回归模型估计了各序列的功率谱密度和积谱密度，结果表明极移导致的重力效应的主要能量集中在Chandler摆动和周年项附近，叠积后实际重力观测与极移重力信号理论值之间的差异分别为0．4％和3．9％，说明超导重力仪可有效监测极移导致的重力变化。     

Radio wave heating of astrophysical plasmas

The coherent plasma process such as parametric decay instability (PDI) has been applied to a homogeneous and unmagnetized plasma. These instabilities cause anomalous absorption of strong electromagnetic radiation under specific conditions of energy and momentum conservation and thus cause anomalous heating of the plasma. The maximum plasma temperatures reached are functions of luminosity of the radio radiation and plasma parameters. We believe that these processes may be taking place in many astrophysical objects. Here, the conditions in the sources 3C 273, 3C 48 and Crab Nebula are shown to be conducive to the excitation of PDI. These processes also contribute towards the absorption of 21cm radiation     

Near-Infrared Imaging Spectrophotometry Of Comet C/1995 O1 (Hale-Bopp) Near Perihelion

We present 1- to 5-μm broadband and CVF images of comet Hale-Bopp taken 1997 February 10.5 UT, 50 days before perihelion. All the images exhibit a nonspherical coma with a bright “ridge” in the direction of the dust tail approximately 10″ from the coma. Synthetic aperture spectrophotometry implies that the optically important grains are of a radius ≤0.4 μm; smallest radius for any comet seen to date. The variation of the integrated surface brightness with radial distance from the coma (ρ) in all the images closely follows the “steady state” ρ⁻¹ model for comet dust ablation (Gehrz and Ney, 1992). The near-infrared colors taken along the dust tail are not constant implying the dust grain properties vary with coma distance. This revised version was published online in July 2006 with corrections to the Cover Date.     

An efficient technique for pre-selecting low-redshift damped Lyα systems

The number of z ∼ 1 damped Lyα systems (DLAs, log  N (H  i ) ≥ 20.3) per unit redshift is approximately 0.1, making them relatively rare objects. Large, blind QSO surveys for low-redshift DLAs are therefore an expensive prospect for space-borne ultraviolet telescopes. Increasing the efficiency of these surveys by pre-selecting DLA candidates based on the equivalent widths (EWs) of metal absorption lines has previously been a successful strategy. However, the success rate of DLA identification is still only ∼35 per cent when simple EW cut-offs are applied, the majority of systems having 19.0 < log  N (H  i ) < 20.3. Here, we propose a new way to pre-select DLA candidates. Our technique requires high-to-moderate-resolution spectroscopy of the Mg  ii λ2796 transition, which is easily accessible from the ground for 0.2 ≲ z ≲ 2.4. We define the D -index, the ratio of the line equivalent width to velocity spread, and measure this quantity for 19 DLAs and eight sub-DLAs in archival spectra obtained with echelle spectrographs. For the majority of absorbers, there is a clear distinction between the D -index of DLAs compared with sub-DLAs (Kolmogorov–Smirnov probability = 0.8 per cent). Based on this pilot data sample, we find that the D -index can select DLAs with a success rate of up to 90 per cent, an increase in selection efficiency by a factor of 2.5 compared with a simple EW cut. We test the applicability of the D -index at lower resolution and find that it remains a good discriminant of DLAs for full width at half-maximum (FWHM) ≲ 1.5 Å. However, the recommended D -index cut-off between DLAs and sub-DLAs decreases with poorer resolution and we tabulate the appropriate D -index values that should be used with spectra of different resolutions.     

Non-Sobolev modelling of radiation-pressure-driven flows in active galactic nuclei

We present a new general scheme for calculating the structure and dynamics of radiation-pressure-driven photoionized flows. The new method goes one step beyond the Sobolev approximation. It involves a numerical solution of the radiative transfer in absorption lines, including the effects of differential expansion and line interactions such as line locking and blanketing. We also present a new scheme for calculating the radiation pressure due to trapped line photons in finite, differentially expanding flows. We compare our results for the radiation pressure force with those obtained using the Sobolev approximation and show the limitations of the latter. In particular, we demonstrate that the Sobolev method gives a poor approximation near discontinuity surfaces and its neglect of line blanketing can lead to erroneous results in high-velocity flows. We combine the newly calculated radiation pressure force with self-consistent photoionization and thermal calculations to study the dynamics and spectral features of broad absorption-line flows and highly ionized gas flows in active galactic nuclei (AGN). A comparison with Sobolev-type calculations shows that the latter overestimates the terminal velocity of the flow and, conversely, underestimates its opacity. We also show that line locking on broad emission lines can have a significant effect on the dynamics and spectral features of AGN flows.