 Okay, our next presenter, please welcome to the stage Paulina Soto-Robles from the University of Arizona. Her presentation is titled Evaluating the Multiscale Atmosphere Geospace Environment Model Through Comparisons with Rural Observations. Is the mic working? Yeah. Okay. Hello, everyone. My name is Paulina Soto-Robles. I'm from Mexico. I study at the University of Arizona. I'm a junior undergrad majoring in astronomy. My mentors this summer were Dr. Michael Wilterberger and Dr. Wen-Wing Wen. And what I did this summer was evaluate the performance of the MAGE model through a comparative analysis with the mesural observations. So I would like to start this presentation by defining the concept of what is space weather. So space weather refers to the dynamic and ever-changing conditions in the space environment surrounding the Earth driven by interactions by the solar wind and the Earth's magnetosphere, ionosphere, and atmosphere. Space weather can occur anywhere from the surface of the sun to the surface of the Earth. As space weather storms leave the sun, it passes through the corona and through the solar wind. When it reaches Earth, it energizes the Earth's magnetic sphere and accelerates the particles on the Earth's magnetic field lines, where they collide with the atmosphere and produce all sorts of phenomena. Each component of space weather can have different effects on our technology. So this is why we care about it. So it can impact our satellite communications, navigation systems, such as GPS, or power grids, pipelines, astronaut safety, and can increase atmospheric drag. It can affect climate and weather and our polar regions. For an example, in 1981, there was this situation in Quebec where there was a major breakout of the power, and it affected 6 million people. So now, what is this big name that I gave at the beginning of the presentation? So what is a multi-skill atmosphere geospace environment model? So it's a physics-based predictive model of stormtime geospace, which means the system of systems representing interconnected physical domains of the Earth's environment. The MAGE model will spend the domains of geospace from the lower atmosphere to the thermosphere ionosphere to different regions in the magnetosphere. It will reserve global dynamics and critical mesoscopic processes with highly-precised technical techniques. I talk in a future tense because this is an ongoing model. Right now, they're in phase one, so they have fully two-way couple, the Gamera model, the REE-CM, their incurred model, and they have done the TCM model that will be substituted by the WACCMX model. So in simple terms, it's a big model that will predict space weather in lots of different regions on Earth and near Earth environment. So how are auroras related to space weather? So auroras are directly related to space weather because they're one of the results of their geostorms and are the most visible and fascinating manifestations of it on Earth. So auroras, also known as the Northern and the Southern lines, or Borealis and Australis, respectively, are natural light displays that occur in the polar regions near the Earth magnetic poles that are caused by the interaction between charged particles from the solar wind and the Earth's atmosphere. So on the right part of the slide, I'm showing a really simplistic scheme of what steps need to happen for an aurora to form. So first, the solar wind approaches the Earth. It interacts with the magnetosphere, the charged particles that come from the solar wind enter the atmosphere where they interact with the atoms, excite them, and emit visible light. So now to the research part of my project. So as you might remember, my purpose was to evaluate the performance of the model using aural observations. So for this purpose, I had two data sources. I had the run results from the simulation, the Mesh model. And we used them as on-ground all-sky imagers that are mostly located in the United States and Canada. So on these two movies that I already played and I hope you saw are just evolution of the aurora on November 4th in 2021. And you also can see the names of the instruments that were used in the Mesh all-sky imagers. So why did we choose these data? So I'm going to give a little bit of background on solar physics. So solar cycles refer to the periodic variations in solar activity that are approximately 11 geocycles. These cycles are characterized by changes in the number of sunspots, solar flares, and coronaviruses mass ejections, which are pretty much just a mass of plasma flying through space and into Earth. During solar maxima that we were actually reaching is where most of these geomagnetic storms become more frequent and strong. And it should be here around 2025. So geostorms are becoming more frequent right now. So in November 4th, there was this big geostorm. There was actually two recent ones that happened in March and April. But due to limited data, we couldn't work on them for the purposes of the research. On the first plot, you're looking at the disturbance term time index, which is just a fancy word for how we're able to measure the perturbance on the Earth's magnetic field. And so a more negative number means the stronger perturbance on the magnetic field and a closer to zero positive means almost no perturbance. So as you can see, at 9.15 UT time, it's one of the pigs for the DST index. And it matches with the activity, the rural activity from the simulation results at 9.15 too. So in the plot that you're looking in the downside of this slide, I integrated the total energy flux from the results of the MATCH model and over the whole day. So as you can see, the time matches. And the yellow line only is there to show you the second peak that happened on the DST index. But by that time, for chemical and physical processes, it was already on their recovery time. So that's why you don't see another peak on the rural activity. Here, I'm showing a little more background on auroras. Auroras have three phases. They have growth, expansion, and recovery phases. For the purpose of this research, we only were able to concentrate on the expansion phase. But for a fun fact, for aurora hunters, this is the stage of the auroras that they're usually looking for, because it's when it's the brightest. So yeah. So now on to the data analysis. So as you might have realized on the first two movies, the Themisolsky majors gave us the data set in terms of intensity, while the Mage model results gave us in terms of energy flux. So there's a discrepancy on the terms of units that we're working with. And also, I will go more in depth a little bit later, but we had to go into some grid resolutions with this. But for the first part, the resolution for both was the same. It was one degree per, well, allowed it to learn to do per grid box for both of them. So we didn't have to worry about that then. We normalized the data on a scale from 0 to 1. And on the first two plots, you're just looking at the data sets, normalized values, and magnetic local time. So if you remember the polar grid, the midnight in magnetic local time represents 180 degrees. So we're looking at midnight and before midnight around 135 degrees. What you're looking at is where our latitude, the rural activity, happen. So as you can see at a given moment in time, which is 1235 UT, does not happen in the same latitude. So they don't happen at the same place. And on the plot on the right, it's just a more generalized view. It's a full width half maximum, so the width of that peak. And you can see that on the mage model, the full width half maximum is more constant. It doesn't show as much movement as it shows in the famous Oscar imagers. OK, so here. The times that you're looking at is 0, 7 UT and 10 UT. They were random times. They don't have a specific purpose. But what's important to look at here, it's pretty much the same thing as the last part, is looking where the activity, like at what latitude happened, activity peak throughout the polar grid. So you're looking at from 5 magnetic local time through 1 PM. So yeah, you can see there's not a lot of correlation between the two data sets. So now, the important part of the research. So about a week ago, we realized that the data that we had gotten from the semi-soul sky imagers captured cloudy data, which was giving rise to a decreased quality of the data sets overall. And you can also tell from the movie that the sky coverage was not extensive. So we're comparing the observations with the model, but the model has these data sets from the whole polar grid. Well, the semi-soul sky imagers only pretty much concentrated at the second quadrant. And there's blank spots of data. So last week, I rerun the code. And this is from the new data set, which doesn't show as much discrepancies as the other ones, but there's still a lot to do. Same thing, we're looking at a specific region of the polar grid right before midnight, so right before 180 degrees, so still the second quadrant. And we're looking where a lot of collatitude the activity pick happened. So for this, you run in different times in different regions in the polar grid. But for this specific timing at the peak, at 9.15 UT, we got a root mean square error of 445. So since we normalize the data from a scale to 0 to 1, this is a pretty significant error, which tells us there's a big discrepancy. So key takeaways. So we found discrepancies between the two models, both Mage model and the observations, indicating variations in a rural peak locations at a specific moment in time. The limitations of the predictive model and limited spatial coverage contributed to these differences. And we got a significant error. When we run on different times, the error was between 0.3 and 0.5, so the mean was 0.4. So future work. So we know that there's a lot of work still to do on the Mage model, as I told you, we're still on the first phase and there's still a lot of physical processes that need to be added to the model for it to give an accurate prediction of what space weather is. One of them, I know they haven't implemented the magnetosphere ionosphere coupling, so that should have a really big impact on the data sets that we have. Also, for having very comparisons, it would be nice to improve the models with resolutions because the University of Calgary, who are the people who provided us with the Themisolsky images data sets, they're moving into a bigger or better resolution. It would be about a third of 32 per longitude per grid box, so we have to match it to be able to compare it with the Mage model results. Also, it would be nice to have a more extensive data coverage covering the whole polar grid, the 360 degrees and multi-time, and excluding cloud obscura sites from the files. I would like to thank my mentors, Dr. Michael Wilberger, who was working remotely with me, Dr. Wenming Wan, Dr. Long Lee, which was not my mentor, but for his patients, Jerry Sikhan, Oya, Ben, and my fellow Nessie Cypersters interns, and N-Kart and NSF for this opportunity. Questions from the audience, Jill? So yeah, my question is, have you ever seen Aurora Borealis in person? I have not, and I want to. Like, I will for sure add some fun in my life, but I haven't yet. So congrats, it was really good. So like you say, you got your data, like you recorded data one week ago, right? So just like want to know, like, how do you feel like all this summer working with not a good data and at the end just like getting, so how was like the experience? Because research is no linear. And this is an example of how you're like spending 10 weeks and then you have, oh, this is not what I was expecting. This is like cloudy data. So I just want to know your experience about that. So it was really stressful at the beginning because that's not what I had planned. So I completely changed my plans. But at the end of the day, it was not such a huge change because all the code that I had used for the past data applied to the new one. So in terms of that, it was fine at the end. Hi, Paulina, great presentation. I'm curious, what causes like the different colors in the auroras? So I'm just going to go back because I have a little picture on the pretty much hidden on the corner. I don't know if you saw it. So the colors depend on the wavelength of the light emitted. So it's a termine the specific atmospheric elements and its electric state and the energy in the particle that hits it. So for example, at higher latitudes, the green and red colors are given by oxygen particles. While at lower altitudes, it's given by nitrogen. And if you go even lower, it could be hydrogen and helium, but they're not seen as much. Thank you, Paulina. That was really cool. I have more of a technical question regarding code. In fact, what was your experience in working with data? Did you have to pick up new skills? You just mentioned some code, your old code applied to the new data. Like, what do you even do this analysis in? Because I have like no idea. Your polar plots looked like they were Python. So I didn't use Python. I know that for the model they use Fortran. But for terms of my research, I drew Python. And what was the other part of your question? So I had previous knowledge, not as much though, because I took my programming class at the beginning of my college career. And I'm already almost three years in, so it was a while ago. But I don't think it was as bad, mostly because my mentor provided me with these tutorials from the model. So like pretty much a rewrite of the Fortran code into Python so that I could like read it and understand it. Perfect. Any other questions for Paulina? Thanks, Ben. Awesome job, Paulina. Thank you so much.