1 of 6

Save the earth from another Carrington event

The Solar Palm Readers

2 of 6

Geomagnetic storm

The Carrington event was the largest geomagnetic storm recorded on earth

1859 - minimal damages

  • harmed telegraph transmission lines.
  • caused sparking and fires in telegraph stations.
  • glowing night sky auroras were recorded globally.

2022 - life-threatening global disaster event with severe long term consequences.

  • damage telecommunications satellites.
  • damage power lines.
  • damage electronic and electrical systems.
  • The high voltage power lines and substations could potentially be put out of commission for months or years.

3 of 6

The objective

Prediction and Preparation are the key!

The model used today can warn us about 45 minutes in advance of an upcoming storm. The main goal is to improve the data and get more accurate and reliable alert.

In the project, we try to predict the solar wind from DSCOVR spaceship and compare the results to the measurements from WIND satellite.

Wind spacecraft

DSCVOR satellite

4 of 6

Dynamic time warping

The first challenge of the project was to calibrate the measurements from DSCVOR and WIND, the two are located in L1 point. The vast majority of time they record a similar signal, offset from one another a bit in space and time.

We used Dynamic Time Warping process in order to compare the magnetic field measurements of the two spacecrafts.

In the graph on the right we show an example result of the DTW process we aligned and plotted.

WIND

DSCVOR

5 of 6

Neural network model | RNN Model

We chose the RNN model because RNN takes into account previous states in addition to the current state, and this is necessary when working with sequential data.

We wanted to train the collected data from DSCOVR on an RNN in order to predict the solar wind ion parameters - density, temperature and speed. We were supposed to compare our predictions to the “ground truth” data from WIND.

Due to a lack of time and computing resources, we couldn’t run the model itself in the scope of the Hackathon

Input - Faraday Cup measurements

Output - Solar Wind parameters: density, temperature, velocity

Time

6 of 6

Suggestions for the future

  • Test the architectures and compare them to simple baselines such as Linear regression and a simple artificial neural network.
  • Use the RNNs to predict the future solar winds to provide extra time for a warning signal.