Helios

Medellín | Save the Earth from Another Carrington Event!

Helios

High-Level Project Summary

We design an algorithm wich is represented in a intuitive and undestendable dashboard, wich is created thanks to the etconsists of a random forest classification method which was trained with historical data. After this training, an accuracy of 60% was reached. This algorithm can be improved with optimization methods in the presentation of the variables.

Link to Final Project

https://github.com/JD314/Proyecto-solar

Link to Project "Demo"

https://docs.google.com/presentation/d/1lHmlQFah_QEgCKyWdqtGvbq-SMJXXrTdukqNiE0VjmI/edit?usp=sharing

Detailed Project Description

Since Galileo's observations with his telescope to the Sun, no one imagined the of our star in modern times and beyond.

The development of high technology from the 1950’s and 1960’s (sinde the 20th and 21st centuries )has made it possible to advance science, such as programming language, to be more effective in predictive models. The event of 1859 Carrington event) passing through the events of the 20th and 21st century in 1989, 2013, have make possible the need to develop instruments to analyze and prevent energetically dangerous solar events for our high-tech era, is undoubtedly a necessity for the human species, both: on Earth as in space, a verifiable reality in our daily routines, in our time and in the development of Solar science -heliophysics- has allowed us to create high-tech instruments, to detect solar events with a quite remarkable precision; space weather has entered, with other variables that put our planet at risk, in addition to terrestrial weather. To observe the Sun we have ground and space telescopes. At the level of astrophysics, ground-based telescopes are the most preferred to study the sun, but space probes play an extremely important role, since they allow us to observe events in our star, for which our eyes cannot perceive them with the naked eye. To analyze these events we use the electromagnetic spectrum, that consists of the energy distribution of the set of electromagnetic waves, it allows us to see in the spectrum the electromagnetic radiation - which emits an emission spectrum or absorbs an absorption spectrum - The electromagnetic spectrum extends from the radiation of shorter wavelength such as gamma rays, X-rays passing through ultraviolet radiation, visible light and infrared radiation till electromagnetic waves of longer wavelength, radio waves. Within the most important solar events we find the flares: these are a sudden and intense release of electromagnetic radiation, in the chromosphere of the Sun with an energy equivalent to hydrogen bombs of up to 6x10^25 joules, which accelerate particles near the speed of light. They are associated as precursors of coronal mass ejections. To observe these events are we use the following observatories: The AIA, the AIA take images of the Sun in 7 UV/visible channels, graphs can be created at solar corona temperature from 1Mk to more than 20Mk through the six channels. The presence of ion emissions of iron shows that the corona is really hot and other parts of it get much hotter during flares.

AIA 131 angstroms of SDO telescope, is designed to study solar flares. It measures extremely high temperatures around 10 million K (18 million F), as well as cold plasmas around 400,000 K.

Both the flares, the prominences and the coronal holes, generate flow of the solar wind through and all propitiate geomagnetic storms, being the event of the flares the most destructive with respect to the previous events, because they are not only formed by particles: they are also formed by radiation, as have been recorded with the DSCOVR.

The DSCOVR (Deep Space Climate Observatory) is the best anticipation system for solar wind alerts: it give alerts of coronal mass ejections and solar wind, since it not only shows the behavior of the solar wind as such, but also the impact of ejections of coronal mass in real time and in the same way, the intensity of a geomagnetic storm, which is what shows the possibility of a Carrington event in terms of the kp index simulated in the project.

About HELIOS

Initially, it was considered to take the route proposed by NASA for development of the challenge, which consisted of the use an ML algorithm in the "black box" form that consisted of using the data from the Wind probe to validate the satellite data.

DSCOVR in order to re-calibrate data produced by noise in the Faraday cup; later this route was discarded because the method was complex and inefficient since the resources of the challenge were available. It was notable that the already existing algorithm and saturated with conditionals would not be worth fixing, and the data used from the DSCOVR was too dirty, so an alternative method was chosen using other satellites in which cleaner and more optimal data could be available to use in a neural network, with the added benefit that it would add more reaction time to the alert because These satellites receive data from the Sun through photons and not through plasma (slower) as the DSCOVR does. This method consists that when an EMC occurs through the analysis of the image from the LASCO coronograph, we can discover particular properties before DSCOVR does.

Additionally, the Goes 16 satellite detects the X-ray flux produced while the EMC is being given, so using an ML algorithm and adjusting the inputs with the Lasco analysis and the GOES X-ray flux, an output categorized into 6 levels can be given.

According to the intensity index (Kp) proposed by NOAA (from G0 to G5) from which we can represent the danger of an EMC.

The algorithm used consists of a random forest classification method which was trained with historical data. After this training, an accuracy of 60% was reached. This algorithm can be improved with optimization methods in the presentation of the variables.