Solar Wind Control of Radiation Belt Electron Flux

我们越来越依赖天基技术(通信、定位等)，这意味着预测地球辐射带的“空间天气”是非常可取的。地球外辐射带内的高能电子通量在几个小时到几天的时间尺度上是高度可变的，但我们还不知道控制这种变化的主要因素。很明显，重要的太阳风瞬变特征，如日冕物质抛射和流相互作用区，可以改变外辐射带的大小和强度，但这种变化很难预测。航天器的异常，包括静电放电和单粒子翻转，与航天器环境中高能电子密度的增加有关。外辐射带从地球中心到地球同步轨道以外的距离约为2.5地球半径，为6.6地球半径，但直到最近，对高能电子环境进行科学采样的机会一直很少，主要是因为它对航天器和电子仪器非常危险。外辐射带的基于物理的数值模型还处于初级阶段。将无碰撞的相对论电子动力学的复杂等离子体物理简化为一个可以在必要的时间尺度上求解的数值系统是一个重大挑战。这个项目将对连接太阳风和外辐射带的最重要的物理过程提供宝贵的见解，这些物理过程可以用来在未来建立这样一个基于物理的模型。计划中的项目有两个关键的新颖性。首先，我们将使用NASA范艾伦探测器任务的数据来提供有关电子流量在离地球不同距离范围内的可变性的信息。这项任务(于2013年启动)将为该项目提供至少四年的空间等离子体数据，其详细程度前所未有，并同时在整个外辐射带的多个地点提供。由于缺乏来自其他地点的现场数据，以前调查太阳风可变性和辐射带通量之间的关系的尝试仅限于地球同步轨道附近的地点。来自太阳风和Van Allen探测器的航天器数据将被结合在一起，以调查太阳风的可变性如何与地球辐射带的可变性有关，以及两者中是否存在可以成功预测的可重复模式。第二，将使用尖端机器学习技术来研究太阳风可变性和电子通量之间的关系，以及哪些参数集最能预测未来的辐射带条件。机器学习技术可以用来在经验数据中找到可重复的模式，然后将它们构建成预测模型。太阳风参数，如速度、数密度和磁场方向，都有助于地球磁层的变化，特别是外辐射带的变化，但由于太阳风的物理特性，它们也表现出相互依赖的关系。我们将从研究太阳风变率的模式开始，在那里我们将开发出空间中单个点的时间序列的技术。随后，这些技术将被用来根据太阳风数据的输入，建立离地球一段距离内电子通量可变性的模型。经过仔细解释，机器学习技术允许我们确定那些对变化的电子通量影响最大的参数，并为它们施加这种影响的物理机制提供不可或缺的线索。通过在空间等离子体物理框架内明智地解释机器学习算法的结果，人们希望对太阳风如何控制整个外辐射带的可变性有新的见解。

Our increasing reliance upon space-based technologies (communications, location-finding, etc) means that predicting the "Space Weather" of Earth's Radiation Belts is very desirable. The high-energy electron flux within Earth's Outer Radiation Belt is highly variable on timescales of hours to days, but we do not yet know the major factors that control this variability. It is clear that important solar wind transient features, such as coronal mass ejections and stream interaction regions can alter the size and strength of the Outer Radiation Belt, but the changes are difficult to predict. Spacecraft anomalies including electrostatic discharges and single-event upsets are related to increases in the density of high-energy electrons in the spacecraft environment. The Outer Radiation Belt spans the distance from around 2.5 Earth radii from the centre of the Earth to beyond geosynchronous orbit, at 6.6 Earth radii, but until recently there have been few opportunities to scientifically sample the high-energy electron environment, principally because it is so hazardous to spacecraft and electronic instruments. Physics-based numerical models of the Outer Radiation Belt are in their infancy. It is a significant challenge to reduce the complex plasma physics of collisionless, relativistic electron dynamics to a numerical system that can be solved on the necessary timescales. This project will provide valuable insight into the most important physical processes linking the solar wind the Outer Radiation Belt that can be used to build such a physics-based model in the future.There are two key novelties in the planned project. First, we will use data from the NASA Van Allen Probes mission to provide information on the variability of electron flux across a range of different distances from the Earth. This mission (launched in 2013) will provide the project with at least four years of in-situ space plasma data in unprecedented detail and in multiple locations simultaneously throughout the Outer Radiation Belt. Previous attempts to investigate the relationship between solar wind variability and Radiation Belt fluxes have been restricted to locations near geosynchronous orbit due to a lack of in situ data from other locations. Spacecraft data from the solar wind and from the Van Allen Probes will be combined to investigate how variability in the solar wind relates to variability in the Earth's Radiation Belts and whether there are repeatable patterns in both that may be predicted successfully.Second, cutting-edge machine learning techniques will be used to investigate the relationship between solar wind variability and the electron flux and which sets of parameters best predict future radiation belt conditions. Machine-learning techniques can be used to find repeatable patterns in empirical data and then build them into predictive models. Solar wind parameters such as speed, number density and magnetic field orientation all contribute to changes in the Earth's magnetosphere, and especially in the Outer Radiation Belts, but they also exhibit inter-dependencies due to the physics of the solar wind. We will begin by studying patterns in the solar wind variability, where techniques will be developed for time series at a single point in space. Later, these techniques will be employed to build models of the variability of electron flux over a range of distances from the Earth, based upon inputs from the solar wind data. Carefully interpreted, machine-learning techniques allow us to determine those parameters that most influence the changing electron flux and provide indispensable clues for the physical mechanisms by which they exert that influence. By judiciously interpreting the results from machine-learning algorithms in the framework of space plasma physics, it is hoped to gain new insight into how the solar wind controls the variability of the whole Outer Radiation Belt.