Large-scale simulations of 2-D fully kinetic Farley-Buneman turbulence
Large-scale simulations of 2-D fully kinetic Farley-Buneman turbulence
复制标题
二维全动力学 Farley-Buneman 湍流的大规模模拟
DOI：
10.5194/angeo-26-543-2008
复制
发表时间：
2008
影响因子：
1.9
通讯作者：
L. Dyrud
中科院分区：
文献类型：
--
作者：
M. Oppenheim;Y. Dimant;L. Dyrud
在地球电离层电喷流中流动的电流经常发展成法利-布曼（FB）流装置并成为湍流。由此产生的电子密度不规则导致这些区域容易散射甚高频和超高频雷达信号。这些雷达测量的许多观测特征都是由等离子体的非线性行为引起的。本文描述了一组高分辨率、二维、全动力学的电场驱动的电射流湍流模拟。这些显示了波的饱和振幅；线性增长模态与阻尼模态的耦合；系统由较短（1 m-5 m）波长模式向较长（10 m-200 m）波长模式演化；在低于线性预测相速度和接近但略高于声速的相速度处，主要模式的传播。这些模拟重现了许多1型波的观测特征。它们为精确模拟FB湍流提供了有用的信息，并证明了我们在模拟电射流方面取得的重大进展。仅限于二维和网格分辨率只有4096 （64 × 64）模式。今天，利用现代的、大规模并行的超级计算机，我们可以在一个完全的动力学模拟中解析超过1600万个模式（4096乘4096）。在本文中，我们将描述从这些最近的高分辨率模拟中得到的1型波的频谱，以及这些结果与测量结果的关系。这些研究使用的模拟器模拟电子和离子动力学与动力学粒子在电池（PIC）方法。在50 mV/m的电场驱动下，对海拔约101 km的极光e区进行了模拟。我们将此运行与驱动场在较高和较低场（(44,70,100,140)mV/m）的运行进行比较。我们还比较了碰撞率降低的跑步，对应于103千米的高度。不稳定阈值约为40 mV/m。在所有情况下，我们发现最具能量模式的相速度低于线性预测的相速度，但略高于声速的最高估计。对于所有波长（<1 m），相位速度大致与波矢量与E0 - B0方向夹角的余弦成正比。最初，模拟产生了线性动力学理论预测将占主导地位的短波长（1- 5m）模式。随着模拟的发展，这些模式趋于饱和。后来，较长波长的模式在振幅上增长，直到它们主宰了系统，尽管它们的增长速度似乎比用线性理论预测的要快。对于160m × 160m的模拟，这些长波长模式在饱和系统中占主导地位。模拟不能精确地模拟波长与模拟大小相当的波的演化。在自然界中，要么是三维效应，要么是梯度或边界等外部效应会控制这些最长波的演变。本文首先概述了e区非正则性观测、理论和模拟。接下来是本文中使用的数值方法的描述
Currents flowing in the Earth's ionospheric elec- trojets often develop Farley-Buneman (FB) streaming insta- bilities and become turbulent. The resulting electron density irregularities cause these regions to readily scatter VHF and UHF radar signals. Many of the observed characteristics of these radar measurements result from the nonlinear behavior of this plasma. This paper describes a set of high-resolution, 2-D, fully kinetic simulations of electric field driven tur- bulence in the electrojet. These show the saturated ampli- tude of the waves; coupling between linearly growing modes and damped modes; the evolution of the system from dom- inance by shorter (1 m-5 m) to longer (10 m-200 m) wave- length modes; and the propagation of the dominant modes at phase velocities that lie below the linearly predicted phase velocity and close to but slightly above the acoustic velocity. These simulations reproduce many of the observational char- acteristics of type 1 waves. They provide information useful in accurately modeling FB turbulence and demonstrate the significant progress we have made in simulating the electro- jet. limited to 2-D and meshes resolving only 4096 (64 by 64) modes. Today, taking advantage of modern, massively par- allel, supercomputers, we can resolve over 16 million modes (4096 by 4096) in a fully kinetic simulation. In this paper, we will describe the spectra of type 1 waves from these re- cent high-resolution simulations and how these results relate to measurements. The simulator used for these studies models both elec- tron and ion dynamics with a kinetic particle-in-cell (PIC) method. We ran a set of simulations appropriate for the auroral E-region at approximately 101 km altitude driven by a 50 mV/m electric field driver. We compare this run to runs with driving fields at both higher and lower fields ((44, 70, 100, 140) mV/m). We also compare to runs with re- duced collision rates, corresponding to 103 km altitude. The instability threshold is approximately 40 mV/m. In all cases, we found that the phase velocity of the most energetic modes lies below the linearly predicted phase ve- locity but slightly above the highest estimates of the acoustic velocity. For all wavelengths (<1 m), the phase speed varies roughly in proportion to the cosine of the angle between the wavevector and E0◊B0 direction. Initially, the simulations generate the short wavelength (1- 5 m) modes which linear kinetic theory predicts will domi- nate. As the simulation evolves, these modes saturate. Later, longer wavelength modes grow in amplitude until they dom- inate the system, though they appear to grow faster than one would predict using linear theory. For simulations reach- ing 160 m by 160 m, these long wavelength modes dominate in the saturated system. The simulation cannot accurately model the evolution of waves having wavelengths compara- ble to the size of the simulation. In nature, either 3-D effects or external effects such as gradients or boundaries would control the evolution of these longest waves. This paper begins with a general review of E-region ir- regularity observations, theory, and simulations. It follows with a description of the numerical methods used in this