The Quiet Planet challenge asks teams to use space data provided to document environmental changes and the subsequent societal changes. Our team decided to address this challenge by focusing on the issue of light pollution. Our goal is to raise awareness about light pollution and pique interest in our most beautiful natural feature, the night sky. We used data from NASA Worldview in order to assess differences in light pollution before and during the COVID-19 crisis. We found a statistically significant relationship between the onset of the crisis and a decrease in light pollution. Once we verified this conclusion, we created a game that would use the beautiful constellations of the night sky to quickly build interest in space. Though we certainly don’t want the pandemic to persist, we’d love it if our project could influence others to preserve and protect our right to appreciate the stars at night.
For as long as we can remember, each member of our group has loved stargazing. We can all remember seeing our first shooting star and the first time we got far enough away from New York City to see the Milky Way. We love seeing stars and planets and galaxies each night, and it’s inspired all of us to care for space exploration. We also know that the nighttime sky is an inspiration to so many others. That’s why we’ve always hated light pollution: it cuts us off from the starry nights we love so much.
One night in May, we noticed that the stars had seemed just a little bit brighter since we started socially distancing from each other, and, when we saw the Quiet Planet challenge, we realized that it gave us the perfect opportunity to do some in-depth research into one of our passions, all while participating in a Hackathon that would be a great way to spend our weekends.
Once we decided on a plan for our project, we had to actually verify that light pollution had decreased as a result of quarantine and social distancing guidelines--our hunches from before weren’t enough. In order to do this, we used NASA Worldview, a data source that allows for a day-to-day comparison of the Earth’s surface. Using the Nighttime Imagery overlay, we were able to obtain black and white images of the Earth’s surface in specific regions. Comparing similar dates from 2019 and 2020 for the same regions allowed for somewhat certain visual confirmation of the decrease in light pollution between the two years. However, we were still faced with two challenges: the quantification of this data, and the certain statistical confirmation of its significance.
To quantify the change, we used ImageJ, a Java-based image processing software. Once we had two same-size images of the same region, one from 2019 and one from 2020, we opened both in ImageJ. Using a standardized light intensity (160/255), we binarized both images to contain only black and white pixels (example of the binarized images here). ImageJ allows the quantification of the specific number of pixels of each intensity, so we imported that data into Excel (full spreadsheet here). Using a simple proportion of white pixels : total pixels, we created a number which represented the average brightness of a region at night. At the end, we were left with average intensities for 2019 and average intensities for 2020, with 2020 consistently having the lower light intensity in our data. However, this still wasn’t enough for us to be certain of the significance of our data. After all, there are several factors at play and data collection is never perfect. Thankfully, some members of our group have a background in statistics, and they were able to do an analysis of the differences for 16 international regions to ascertain whether or not we could accept our conclusion that light pollution decreased.
We decided to determine if the mean difference in light intensity change from 2019 to 2020 was statistically significant by running a one sample t-test for a mean difference. This statistical test means that we will be calculating the difference in light intensity scores for each region that we generated and from the set of differences we will determine if that value is statistically significant by determining its p-value.
The methodology and results of that test can be found here. In summary, the test confirmed that the data was significant.
The purpose of carrying out this statistical test was to prove that there has been a significant decrease in light intensity from 2019 to 2020. Although this test does not allow us to claim that the change in light intensity is caused by COVID-19 (this was not an experiment but an observational study), we can conclude that there is a strong association between COVID-19 and light intensity. Using this data we can help establish that COVID-19 and its associated societal changes have helped to lower the amount of light pollution and has increased visibility of the stars at night. If we continue to follow in this trend it will lead to less light pollution and a more beautiful night sky long-term.
Verifying that light pollution did in fact decrease because of the stay-at-home orders of the COVID-19 pandemic was a relief, as it meant that we had found a valid foundation for our project. Now, we could begin the product side of our project, the part that would actually solve the problem that we’d identified.
Our idea was relatively simple: a game called STAR, similar to Bingo. Our plan was to host live games of STAR at night on a regular schedule in order to garner interest in the night sky. The object of the game is simple: as we announce the constellation to look for, you find it in the sky, take a picture, and mark it down on your board. Once you have a full row or column, you indicate that you’ve won, and get a nominal prize for winning that day’s game. Obviously, this game gives an advantage to people who are familiar with the night sky, so we’re incentivizing participants to stargaze in their free time, increasing awareness of our goal. Other games like this (eg. HQ Trivia and The X), which give away prizes to young people, have proven to be successful at gaining huge amounts of interest in the past. So successful, in fact, that over 1.2 million people competed on the HQ platform on a single day in January 2018. As high school students ourselves, our group is acutely aware of these sorts of trends, and hoped to capitalize on that knowledge for this project.
After doing some online research, we found that there are 16 constellations visible in the Northern Hemisphere during the month of June. We focused on June based on the assumption that social distancing guidelines will be lessened as the summer progresses, bringing light pollution back up to its previous levels, which is to say that our window of opportunity to raise awareness with the game is limited. After obtaining this list, we wanted to make sure that it would be viable to make enough unique boards using only 16 constellations. 16P16, which represents the number of permutations of 16 items, confirms that the number of unique boards would be 20,922,789,888,000. That’s plenty.
We used a few simple Java algorithms, found in our public Github repo, to process the list of constellations and produce a random board. We realized that a 4x4 grid of names isn’t as useful or attractive as a well-designed board, so we created a mockup in photoshop. If we were given the opportunity to fully implement this product, we would add a way to process the output file of the Java program (a .txt file with a 4x4 grid) into a graphically designed board. It would also be interesting to create a version of the game that is applicable for people in all regions of the world. Unfortunately due to our limited experience in computer science and the time limit on this project, we were forced to stick with what we know.
Despite these limitations, we’re quite happy with both the data analysis and the ultimate product of this project, and we genuinely believe that it could impact a wide range of people to care more about the beauty of the night sky. We’re very thankful to NASA and its partnered agencies for the opportunity to demonstrate our knowledge and learn new things during the course of this project.