Mathematical Data Science in the Ionosphere

Abstract

The proposed work will produce algorithms to obtain lightweight, data-driven forecasting models of plasma density in the ionosphere given a time series of observations. The proposed effort will leverage recently published work by the PI and Co-PIs on methods to fit dynamical models from data measurements exhibiting complex, nonlinear, and sometimes multiscale behavior. We will achieve this by merging what we have learned using modal decomposition and deep learning along with information theory and nonparametric statistics to yield improved short-term forecasts over the current state-of-the-art. This effort will consist of two major aims. In Aim I, we will focus on generating machine learning forecasts for ionospheric specification via deep-learning-based dynamic mode decomposition. Aim II will then characterize the spatiotemporal distributions of fluctuations in the ionosphere not captured by our forecast using nonparametric statistical methods. Together, these two aims form a complete, fully data-driven characterization of the local ionospheric profile dynamics and the error signal

Document Details

Document Type
DoD Grant Award
Publication Date
Mar 03, 2023
Source ID
N000142312106

Entities

People

  • Christopher W. Curtis

Organizations

  • Office of Naval Research
  • Salk Institute for Biological Studies
  • United States Navy

Tags

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Research Science/Academic Research
  • Space/Atmospheric Physics.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks