High-Level Project Summary
In the first instance we developed an algorithmic code in Matlab, because it was easy and designed for scientific-based purposes, but then we transfered our code to Python because of its versatility and compiling speed. There, we developed a neural network-based algorithm and other related techniques to remove the tropospheric signal from various radar datasets and discern the ground motion as real as possible. So basicaly, we took the data that we had in databases provided by NASA, to visualize them through code in a programming software and then plot the changes over time that occur in the behavior of the events.
Link to Final Project
Link to Project "Demo"
Detailed Project Description
Our solution is mainly intended to visualize the problems and the great disadvantage in which we are involved when we do not know or ignore data that, studied and analyzed in a correct way, would allow us to obtain good predictions about future events, the procedures carried out to reach these solutions taken as methods and the results being useful to reduce as much as possible the negative effects of events such as those that occur in earthquakes in cities, among different populated areas.
We take data that we have in databases provided by NASA, to visualize them through code in a programming software and then plot the changes over time that occur in the behavior of the events. Once analyzed, we have a better approximation of their nature and can detect factors such as noise, thus having the opportunity to learn more about how these events are carried out.
The benefits of our project is to have an approximation towards the behavior of seismic events that allows us to prevent negative effects in the areas to which these events correspond, this because the predictions will be good since we have followed methods based on mathematics and physics.
The possibility of having a code that allows us to enter images corresponding to the same event but at different times, thus knowing how a seismic event proceeds, analyze different of them to better understand them and support in the predictions of these, as well as in the areas where the event will occur. Some of the tools we use are programming languages such as Matlab, Python, Neural Networks, Mathworks, Jupyter notebook, etc.
Space Agency Data
The data we used was found in the Earth data from NASA. We used the information from two earthquakes that shook California in the summer of 2019. We saw the motion of the ground due to both of these earthquakes using InSAR. With the information provided by all the images that were taken before and after both earthquakes, we tried to identify and separate the real ground motion from the signals you see due to the delay of the radar beam through the atmosphere. We also used the data from the Kilauea volcanic eruption in Hawaii that took place during the spring and summer of 2018. It was a magnitude 6.9 earthquake that struck the southern flank of Kilauea near Leilani Estates. We obtained an interferogram that allowed us to see the earthquake and some of the other signals related to the volcanic eruption. Finally, we used the information obtained from some faults that lay hidden and quiet right until an earthquake breaks them apart, but others move little by little, continuously or in small bursts. In Central California, part of the San Andreas Fault is always moving at a slow rate of a few millimeters per year. These movements are visible in many InSAR images and those were the images we analized. This dataset was perfect for the idea that we had to solve the challenge and also brought us deep inspiration as we love to analyze data in our bachelor of science in Physical Engineering.
Hackathon Journey
This Space Apps experience was unique and challenging. It gave us the opportunity to put into practice the knowledge acquired in the two years we have been studying the bachelor of science in Physical Engineering such as the numerical analysis and modeling of stochastic and nature-based systems. We learned about the signal detection from earth ground by Synthetic Aperture Radar Interferometry (InSAR), a technique which is largely affected by the fluctuations of the atmosphere and also to easily clean large amount of data from different resources, as well as explaining the metodology and development of a complex project. It is obvious that the challenge we had in the subject Numerical Modeling of Stochastic and Nature-Based Systems in which we developed a neural network to clean data from the SIR model, inspired us to choose this challenge. So, we focused on obtaining and managing data and images of the seismic movements of the Earth provided by NASA to later do a deep debugging of them using neural networks in programming languages such as Matlab and Python. It is important to say that the best way to solve the setbacks we had, such as the duration in downloading the NASA files and obtaining matrices from the images, was by adequately distributing the work of the project according to our individual strengths as team members. Finally, we would like to thank PhD. Servando López Aguayo for being a constant inspiration and for giving us his support in solving problems from all the subjects we have studied with him.
References
Richard Bamler and Philipp Hartl. (1998). Synthetic aperture radar interferometry. Retrieved from: https://iopscience.iop.org/article/10.1088/0266-5611/14/4/001
]
Interferometric Synthetic Aperture Radar (InSAR) | U.S. Geological Survey. (2018). Retrieved from: https://www.usgs.gov/centers/land-subsidence-in-california/science/interferometric-synthetic-aperture-radar-insar
Image, G. (2021). What is SAR and InSAR? Geo. Retrieved from: https://www.geoimage.com.au/blog/what-is-sar-and-insar/
Tags
#NASA #Matlab #Python #Science #Technology #InSAR #Image