EAGER: Machine Learning and Data Assimilation for Discovery of Generalized Fokker-Planck Equation for Radiation Belt Modeling

EAGER：用于发现辐射带建模的广义福克-普朗克方程的机器学习和数据同化

• 批准号: 2211345
• 负责人: Dmitri Kondrashov
• 金额: $ 25.01万
• 依托单位: University of California-Los Angeles
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-04-01 至 2024-03-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2211345&HistoricalAwards=false
• 关键词: EAGER; Machine; Learning; Data; Assimilation

This project supports a two-year investigation into a better understanding and prediction of the extreme dynamic regions of space known as the Van Allen radiation belts. The near-Earth radiation environment is filled with high-energy particles trapped by the Earth's magnetic field. The particle intensity and distribution change as part of a much larger space weather system driven by the Sun. The radiation belts can be hazardous for satellites and astronauts in space. Therefore, it is crucial to be able to predict these energetic particles better for our space exploration. The new investigation is expected to lead to a more accurate understanding of the radiation belts using a novel combination of data assimilation and machine learning methods to discover generalized Fokker-Planck equations for radiation belt modeling. The Principal Investigator (PI) will use University of California-Los Angeles (UCLA)'s Versatile Electron Radiation Belt (VERB) model and data assimilation to learn and incorporate additional dynamical terms that would describe nonlinear wave-particle kinetic effects. The project will promote and pave the way for novel use of machine learning and data assimilation tools to discover missing physics in high-dimensional and complex space physics models. The initial implementation and vetting of combined data assimilation and machine learning algorithms in the radiation belt diffusion model will not only lead to significant advances in space weather modeling capabilities. Still, they will also have substantial broader impacts on space plasma physics and other disciplines. The project will thus have the potential to drastically change our understanding of radiation belt modeling, and advanced tools can find broader use for Earth's climate studies and engineering.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

该项目支持一项为期两年的调查，旨在更好地理解和预测被称为范艾伦辐射带的太空极端动态区域。近地辐射环境充满了被地球磁场捕获的高能粒子。 粒子强度和分布的变化是由太阳驱动的更大的空间天气系统的一部分。 辐射带对太空中的卫星和宇航员来说可能是危险的。 因此，能够更好地预测这些高能粒子对于我们的太空探索至关重要。 新的研究预计将使用数据同化和机器学习方法的新颖组合来更准确地了解辐射带，以发现辐射带建模的广义福克-普朗克方程。首席研究员 (PI) 将使用加州大学洛杉矶分校 (UCLA) 的多功能电子辐射带 (VERB) 模型和数据同化来学习和合并描述非线性波粒动力学效应的附加动力学术语。 该项目将促进机器学习和数据同化工具的新颖使用，并为发现高维和复杂空间物理模型中缺失的物理现象铺平道路。辐射带扩散模型中数据同化和机器学习组合算法的初步实施和审查不仅将导致空间天气建模能力的重大进步。尽管如此，它们也将对空间等离子体物理学和其他学科产生更广泛的影响。因此，该项目将有可能彻底改变我们对辐射带建模的理解，先进的工具可以在地球气候研究和工程中找到更广泛的用途。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力评估进行评估，被认为值得支持。优点和更广泛的影响审查标准。