Poster Session 4

Poster Gallery for Session 4 (February 11th): Space Science

Moderator: Ryan Mcgranaghan, ASTRA, LLC

📑 Combined PDF of posters for this session

Using an LSTM and Classification Methods to Determine Risk of dB/dt Threshold Crossings as Proxy for Geomagnetically Induced Currents

Authors: Michael Coughlan, Amy Keesee, Victor Pinto, Hyunju Connor, Jeremiah Johnson
Theme: Space Science
Poster ID: 53

Abstract: The interaction between the solar wind and the Magnetosphere can produce Geomagnetically Induced Currents (GIC’s) on the ground, which are capable of causing power outages and damage to crucial infrastructure. The ability to predict when and where these events may occur could allow us to avoid the worst of this damage.

Applying Machine Learning to MOMA Science Data for Scientific Autonomy

Authors: Victoria Da Poian, Eric I Lyness, Ryan M Danell, Melissa G Trainer, Xiang Li, William B Brinckerhoff, and the MOMA team
Theme: Space Science
Poster ID: 54

Abstract: (Not available.)

Using Machine Learning to Infer Pre-Entry Properties for Asteroid Threat Analysis

Authors: Jonathan Gee, Ana Maria Tarano
Theme: Space Science
Poster ID: 55

Abstract: Accurately assessing asteroid threats relies on knowledge of the asteroid’s pre-entry properties such as size, velocity, and mass. Directly measuring these properties can be infeasible due to the sparsity of events and the accuracy and fidelity of various sensors. Current analysis of an asteroid’s pre-entry properties involves modeling the asteroid’s entry into the Earth’s atmosphere. This process can be time consuming and can require manual adjustment of uncertain modeling specific parameters. NASA Ames has developed a genetic algorithm that can help automate asteroid modeling using the Fragment-Cloud Model (FCM). The algorithm generates realistic energy deposition curves based on actual energy deposition curves from real, observed asteroids. By using these synthetic, labeled energy deposition curves, we developed a one-dimensional convolutional neural network that can predict an asteroid’s pre-entry parameters.

Super-resolution of MDI Solar Magnetograms: Performance Metrics and Error Estimation

Authors: Paul J Wright, Xavier Gitiaux, Anna Jungbluth, Shane A Maloney, Carl Shneider, Alfredo Kalaitzis, Michel Deudon, Atılım Günes Baydin, Yarin Gal, Andrés Muñoz-Jaramillo
Theme: Space Science
Poster ID: 56

Abstract: Aim: Develop an approach to convert and upscale line-of-sight magnetic field data to a reference survey in order to under- stand long-term variability of the mag- netic field on time-scales larger than the lifespan of a single instrument.

Comparison of Time Series Techniques to Model Connections Between Solar Wind Input and Geomagnetically Induced Currents

Authors: Amy Keesee, Victor Pinto, Michael Coughlan, Connor Lennox, Md Shaad Mahmud, Hyunju Connor
Theme: Space Science
Poster ID: 57

Abstract: Geomagnetically induced currents (GIC) can drive power outages and damage power grid components while also affecting pipelines and train systems. Developing the ability to predict local GICs is important to protecting infrastructure and limiting the impact of geomagnetic storms on public safety and the economy. While GIC data is not readily available, variations in the magnetic field, dB/dt, measured by ground magnetometers can be used as a proxy for GICs. We are developing a set of neural networks to predict the east and north components of the magnetic field, BE and BN, from which the horizontal component, BH, and its variation in time, dBH/dt, are calculated. We apply two techniques for time series analysis to study the connection of solar wind and interplanetary magnetic field properties obtained from the OMNI dataset to the ground magnetic field perturbations. The analysis techniques include a feed-forward artificial neural network (ANN) and a long-short term memory (LSTM) neural network. Here we present a comparison of both models’ performance when predicting the BH component of the Ottawa (OTT) ground magnetometer for the year 2011 and 2015 and then when attempting to reconstruct the time series of BH for two geomagnetic storms that occurred on 5 August 2011 and 17 March 2015.

Developing Deep Learning for Solar Feature Recognition in Satellite Images

Authors: Michael S Kirk, Raphael Attie, James Stockston, Matt Penn, David Hall, Barbara Thompson
Theme: Space Science
Poster ID: 58

Abstract: Abstract not availalble.

Correlation of Auroral Dynamics and GNSS Scintillation with an Autoencoder

Authors: Kara D Lamb, Garima Malhorta, Athanasios Vlontzos, Edward Wagstaff, Atilim Günes Baydin, Anahita Biwandiwalla, Yarin Gal, Alfredo Kalaitzis, Anthony Reina, Asti Bhatt
Theme: Space Science
Poster ID: 59

Abstract: Abstract not availalble.

Complex Data Explorer (CODEX) – A multi-use Machine Learning Powered Tool for Rapid Data Exploration

Authors: Jack Lightholder, Lukas Mandrake, Josh Rodriguez, Rob Tapella, Patrick Kage
Theme: Space Science
Poster ID: 60

Abstract: Modern science datasets from missions like OCO-2 and telemetry records in Ops may have 500+ simultaneous measurements at each of millions of time samples. Scientists would often like to look through the record and discover not only expected trends but ones they did not initially guess, while Ops personnel perform the same task under serious time pressure should an anomaly occur. In both cases, the optimal environment for this rapid exploration large data would be one where visualizations were clear, interactive, and responsive, permitting the investigator to “play” with the data and gain rapid insight, falsify hypothesis, and make discoveries. Machine Learning (ML) has proven invaluable in providing some of these key data insights, but to do so in a statistically robust and reliable manner requires a data science professional and a lot of custom Python code, losing any sense of interaction and play. CODEX will address these concerns by providing a desktop-like environment with standard scientific graph types that are robust to rapid, powerful exploration.

COSMIC: Content-based Onboard Summarization to Monitor Infrequent Change

Authors: Lukas Mandrake, Gary Doran, Masha Liukis, Steven Lu, Umaa Rebbapragada, Jimmie Young, Kiri Wagstaff
Theme: Space Science
Poster ID: 62

Abstract:(Not available.)

A Deep Learning Approach to GNSS-R: Predicting Soil Moisture with Delay-Doppler Maps

Authors: T Maximillian Roberts, Ian Colwell, Rashmi Shah, Stephen Lowe, Clara Chew
Theme: Space Science
Poster ID: 63

Abstract: GNSS reflection measurements can be calibrated with data from SMAP to yield estimates of soil moisture with increased spatiotemporal resolution, useful to certain hydrological/meteorological studies. Current approaches which use simple models of the relation between the DDM (delay-Doppler map) and soil moisture which can fail in certain regions. Complex information contained in the complete 2D DDM could help in these regions, and can be extracted through the application of deep learning based techniques. This approach simultaneously provides the ability to incorporate additional contextual information from external datasets. Our work explores the data-driven approach of convolutional neural networks (CNNs) to determine complex relationships between the reflection measurement and surface parameters. CYGNSS DDMs were aligned with SMAP soil moisture values and ancillary datasets, a network was developed and trained with these measurements, the results of which are analyzed and compared to existing global soil moisture products.