High-Level Project Summary
Our solution provides the prediction of the next solar flare as well as its magnitude. First, using a dynamic API, the AI collects and manipulates the data to transform it into a pandas.dataframe. After that, the AI performs the prediction of the next relevant event and generates a list of the other predictions. The main benefit provided by the solution is to frequently predict events even before they are detected by the missions that provide the database.In general lines, we make the task of visualizing the data easier, showing them in a simple, but complete, way. Besides that, the predictions facilitate the decision making on how to counter the damages that this event can lead to on Earth
Link to Final Project
Link to Project "Demo"
Detailed Project Description
We were presented to a problem that we've never even heard about before, the high-magnitude solar storms, especially those that are classified by the “Carrington” class. Initially, we focused on having a great understanding of the problem, so we searched about it and found out that a Carrington event is a solar storm intense enough to affect the Earth’s magnetic field, and that it has already happened, more precisely in 1859, leading to damage on some telegraphs.
Bringing that situation to our current society, we would have problems with our communication systems, GPS satellites, internet signals, electrical power distribution, and many more things that, nowadays, society relies on completely, causing a loss of billions or even trillions of dollars.
Given the seriousness of the problem, we understood that our solution would be extremely helpful to our society, because, even if the probability of a Carrington event occurring is 1 in 500 years, if that happens humanity would be in serious problems.
To help prevent those damages, we decided to develop an IA that, using data collected from a sensor in some NASA’s satellites, would predict the intensity of the recent solar storms and when exactly they happened in the sun. A differential is that our model syncs automatically with the most recent database uploaded by NASA, transforming the archive that is .cdf into a pandas.dataframe, enabling us to manipulate these data more easily. Since the velocity of those solar flares isn’t as fast as the velocity of the transfer of data collected by the satellites, which is as fast as light speed, these information would arrive fast enough to enable the model to do the predictions and help society counter the damages that the solar event would cause on Earth.
Finally, to make the application useful for NASA engineers, we created a website where all the important information, as well as the predictions made by the model, will be presented in a simple, but at the same time complete, way, making it easier to work with these information and make decisions, thus solving the given problem.
Space Agency Data
We used the newest dataset available at the time we made the application. The dataset was reached by one of NASA's websites, where all information about solar storms collected in 2022 were available. We transformed the original dataset format (.cdf) into a pandas.dataframe.
https://2022.spaceappschallenge.org/challenges/2022-challenges/carrington-event/resources
Hackathon Journey
During the whole process of idealization and creation of our solution, the entire team stayed motivated and dedicated to making the best assing we could, always searching and learning new topics and tools that could help us achieve our goals, such as machine learning and deep learning techniques, team management, data manipulation and business and design techniques, making it a great experience for us to work together.
Our team chose the “Save the earth from another Carrington event” challenge because we saw an opportunity to prevent the earth from being affected by a cosmic catastrophe that could make the technological progress that we made as a society fall apart, leading us to dark times - literally and figuratively.
We found out that making the best possible data process was the most important thing to do, so our initial approach was to define how we would do that. Once we did it, we focused on making the predictions and handing them in the best and simplest way possible to our main client.
To help us achieve our objectives and overcome the difficulties, we made intense benchmarks and concept validations with the mentors.
We would like to thank all mentors that helped us conclude this project with excellence, with special regards to Guilherme Ciuffi and Emmanuelle Richard.
References
Project code’s Upload: GitHub
- Design
Slides: Canva
Royalty free images and GIFs: Pixabay
Wireframes and prototypes: Figma
- Business:
Idea’s validation: Google forms
Workframes: Miro
- Coding:
Code writing: Visual Studio Code
Programming language: Python, TypeScript
¬ libraries:
Numpy: NumPy
Pandas: Pandas
Spacepy: SpacePy
SkLearn: SkLearn
Pickle: Pickle
Glob: Glob
TypeScript Framework: Next.js by Vercel
- Project Hosting:
Front-end: Vercel
Back-end: Heroku
Tags
#CarringtonEvent #AI #Python #Data #Website #Prediction #sun #Brazil