Mapping UFO Sightings in 2015 with CartoDB


Note: If you’re looking for Part 2 of the GeoDuino post, I’m… uh… working on it.

I Want to Believe

Let me start by saying that I am the Scully in the Mulder-Scully dichotomy. I can be coldly rational, scientific and rarely whimsical. And while I absolutely love science fiction and fantasy, it is because it sparks scientific imagination. That said, I kind of secretly love UFO conspiracy theories and videos of alleged sightings and Ancient Aliens. I grew up watching the X-Files with my parents. My biggest irrational fear when I was a child was that I was going to be abducted by aliens (thanks, Mom and Dad). As an adult, I’ve mostly overcome this fear and replaced it with adult fears, like my student debt – the real scary monster (thanks, society). But you could say…
I want to believe

I was partly inspired by a student’s GIS project that mapped UFO sightings in Georgia. She attempted to correlate sighting density with various demographic and economic data. This was posted to Reddit last month. She showed where to find the data. To be honest, I was also partly inspired by the return of the X-Files, because of course I was.

Paranormal Data for All

The National UFO Reporting Center (NUFORC) maintains a fantastic database of UFO reports worldwide, with the most extensive reporting in the United States. Users submit reports through NUFORC’s web portal and 24-hour hotline. The database includes date, time, locations and descriptions of the sighting. These reports are reviewed and cleaned by NUFORC, with the most obvious hoaxes and wild misinterpretations flagged or outright deleted. It’s a pretty sweet database, all things considered, even if the actual report summaries are truncated to fit the field’s character limit. And while the locations aren’t as specific and beautiful as latlng , the city and states give enough information to geocode and generate some pretty interesting maps on a national or international scale.

The best part of the dataset is definitely the summaries. The reports range from essentially “I saw a weird light and I don’t know what it is” to “I saw aliens entering an interdimensional portal.” Here are some of my favorites from a quick scan:

1/6/15 – Phoenix, AZ – Please take seriously.

4/3/15 – Las Vegas, NV – I saw a green dark green spotted alien being hovering over me not a ufo but an extraterrestrial.

6/3/15 – Richland, MS – I, along with 6 witnesses saw Fleets of UFOs disappearing into portal I have 4 1/2 min of video.

6/5/15 – Mount Pleasant, MI – I was taken to an alien laboratory from my bedroom.

6/25/15 – Philadelphia, PA – At first I thought it was a blimp but when I looked closer I saw that it was flat on the bottom. Blimps aren’t flat.

6/29/15 – Omaha, NE – Omaha aliens?

7/4/15 – Elkhart, IN – UFO observing 4th of July activity.

7/15/15 – Houston, TX – Two aliens were in my backyard and I recorded the entire event.

8/20/15 – Commerce City, CO – Dancing stars in the sky. ((NUFORC Note: We suspect “twinkling” stars, but cannot be certain. PD))

10/1/15 – Barstow, CA – White, cigar shaped, large, high, no noise, fast, no chemtrail to say jetliner.

11/14/15 – Bernalilo, NM – Loud horns/trumpets flying the sky, Nothing visible. REALLY REALLY LOUD! Sounded like a frieght train, or a dragon.

This data must be shared for nerds (and UFO conspiracy theorists) everywhere. I’ve been poking around CartoDB and futzing with D3js for a while now. I chose to use CartoDB for these maps for three reasons: it’s super easy to make pretty maps, it makes for fun time series visualization, and I really wanted to produce something with it. I also drew up some okay charts with D3js.

The reports are from all over the world, but I did filter them to the United States and Canada. It seems clear to me that this site is predominantly American. International reports were so few and far between that I wouldn’t have been able to produce any meaningful visualizations.

Mapping the Weird

Do yourself a favor when looking at these maps and put them in fullscreen mode for a few minutes.

The above is an intensity map of all (geolocate-able) reports made through the website in 2015. It’s essentially just a point map with overlapping points increasing the intensity values. That means it’s clickable! I’ve created tooltips giving readable data on each report. Unfortunately, stock CartoDB maps do not allow for “cycling” tooltips on points that share¬†a single geography. That means when you click one it returns the first report in the query, and you can’t scroll through the entire set of reports. This can be accomplished using their Javascript library, CartoDB.js, but after playing with it for a bit I didn’t find a method that I liked. I could create a huge scrolling list of reports, not a page for each one. This may be an area where ArcGIS Online shines a little brighter than CartoDB–these kinds of tooltips are standard and easily created.

CartoDB seems to have exploded in popularity in the last year or two, and I think that’s largely due to the addition of their Torque Maps feature. The popularity of these grew rapidly after it was used to show how the news of the shooting death of Michael Brown spread across the world, and later the Twitter reaction to the grand jury decision not to indict Darren Wilson. Both of these maps were shared widely on social media. This was many people’s first exposure to this type of map.

This particular dataset works very well with this kind of map. We do have a date_time field. Using it, we can see some interesting things.

There appear to be days in which the map lights up, indicating that more reports were made that day. By breaking down the data in chart form, we see it clearly.

If we look at the above bar chart, displaying number of sightings by day, we see that on most days the number of sightings vary between about 20 and 35 sightings. There are, however, some pretty obvious spikes. The first spike is the very first day, 1/1/2015 (53 sightings). Maybe aliens like to visit us on New Year’s Day, but I think I’ll attribute that spike in strange-lights-in-the-sky sightings to the tiny little warheads we tend to blast into the sky called ‘fireworks.’ Further, a large number of reports mention watching fireworks during the sighting. Maybe I’m being dismissive, but the next spike in the year, a huge one, comes on 7/4/2015 (170 sightings). The Fourth of July, another day we (Americans) tend to blow up the sky with lights. Boring, boring fireworks. The remaining spikes I do find interesting. These are found on 9/27/15 (74 sightings) and most definitely on 11/7/2015 (284 sightings).

Wild speculation

So what exactly happened on these dates? Swamp gas? Was it visitors from another world? Was it part of a shadow government secret operation to breed human-alien hybrids?

I'm not saying it was aliens...

While it probably wasn’t aliens, it wasn’t anything mundane either. Even going with the Occam’s Razor approach, it’s really quite interesting.

Starting with September 27th, some quick Googling led me to the most probable answer. It turns out that September 27th, 2015 was quite a rare day in astronomy, as it was the night of both a supermoon and a total lunar eclipse. A “super blood moon.” Events like these tend to draw people out of there homes to look up at space, and people who don’t typically look up at space can often see things that they can’t readily explain. Some might call that a UFO. This may be particularly true for those who’s first thought is to file a report with an internet UFO database, but that’s not very scientific of me.

Blood Red Supermoon

You gotta admit, that is an impressive super blood moon.

So what about the huge spike on November 7th? Surely it must have been something normal and boring. I’m a smart guy, I’ll know what’s going on. Like a meteor shower or an airplane or a music festival or…

Oh. Uhh… well okay then. That sure is a thing right there, isn’t it? Ummmm…

Turns out that there may have been some government secrets involved after all. A U.S. Navy submarine was test firing a Trident II ballistic missile in the Pacific Test Range off the coast of Southern California. Apparently this was a ‘secret’ test in the sense that the Navy neglected to tell anyone about it. Obviously it couldn’t stay a secret for long, as it was launched right off the coast of the second most populated city in the country.

So what did the map look like on that day? Using CartoDB’s built-in PostgreSQL, it’s easy to filter the reports by date and build a map. Below is a heat map of reports based on that filter. I’ve also done my best to trace the Pacific Missile Range; the missile was launched somewhere within this area, right off the coast.

Clearly, the most dense cluster of reports was in Southern California, followed by the Bay Area, Las Vegas and Phoenix. These are all within viewing range of the Trident launch. In fact, if we break the data down into states, we can see that California had far more sightings on November 7th than other states.

A surprisingly heartfelt reflection

I started this project thinking it would be fun to find correlations between real events and sightings, debunking reports with maps. I was right; It was fun, but after reading a lot of the summaries I found myself sympathizing with the writers quite a lot. As I said earlier, I love science fiction and fantasy because it inspires scientific imagination. What we have here are inspired minds looking to explain the unexplained. While I quoted some of my favorite at the start, the vast majority of the posts read like a police report might. For example:

5/15/15 – Akron, NY – Elongated cyndrical shape bright red lights all around descending from night sky north east of akron ny at 9:30pm duration 1min.

8/23/15 – Paradise Valley, AZ – Bright flash, then round ball moves slowly across sky, does quick button hook, followed by 180 degree turn, heads NE, disappears.

9/5/15 – Clermont, FL – 24 bright lights over Clermont fla.headed in southerly direction at approximately 9:30 at night on sept. 5th 2015

Cold, unemotional reporting in 135 characters or less. Whether or not it accurately describes what they saw is irrelevant. Forgiving that many of these are likely hoaxes, for most it’s what they believe that they saw.

I’ve also developed a deep respect for the National UFO Reporting Center. They maintain a well policed, largely unbiased report database for the entire county. That’s some pretty important data and could make for some really cool science. The fact that idiots like me can pull the data for analysis, free of charge, is even better. I’ll admit: what I did here is peanuts. If you’re ever going to find a pattern in the UFO reports for something not easily explained, you’re going to need big datasets like this.

I’ve made the data used to generate these maps and the D3js charts available on my GitHub profile here. The maps and related data are publicly available on my CartoDB account here.

I thoroughly enjoyed working in CartoDB. It’s easy, quick, and–dare I say it–fun to make these maps. Further, I’ve only just barely scratched the surface here of what is possible.

This is super meta.

One more thought: I posted the above picture (SUPER META!) on Instagram/Facebook/Twitter when I was pulling all of this together and a friend/former instructor of mine pointed out that someone very recently did a report using years of data. It’s a pretty interesting read: Summer Spikes of UFO Sightings. I disagree with the point that “it is very clear that UFO sightings have been exponentially increasing since 1995,” though my issue may be with phrasing, and it’s not even the author’s main point. We are dealing in reports, not sightings. I think the reason NUFORC is obtaining more reports is most likely due to it being a largely internet platform, and the internet has become much more ubiquitous since 1995. The author hints at this by noting it may be due to how the “evolution of technology made it easier for people to report such things, be it by phone, emails or dedicated online platforms like the one made by NUFORC.” While I can’t necessarily dispute the argument that the “rise of science fiction movies has made people more interested in UFOs and thus make them tend to look at the sky more often, wondering if they might have their own sighting,” I can’t help but feel like the reason there are more reports is because more people are using the internet.

Going Forward

If I continue to work on this (and I may) I’d like to get better tooltips on the maps for starters. I think it would greatly improve the interface if users could cycle through reports from the same city. I’d also like to dive deeper into CartoDB.js and D3js and put together a Torque time series map that also traces a D3 bar or line chart at the same time. Eventually, my goal is to setup a full page “story map” type thing with all relevant data, and update it regularly with new reports.

Lastly, I’d like to refresh some of my spatial data analysis skills by looking for significant spatial-temporal clustering and look for more correlation to documented events and get more hard science-y numbers. I’ll probably remove the dates above (1/1, 7/4, 9/27 and 11/7) from the analysis as outliers and see what else I can find. This post was meant to show fun data visualizations that are possible in CartoDB, and the conclusions are pure speculation. I believe that more interesting insights can be found.

Update 2/10/2016: A lot of this looked terrible on mobile. Sorry about that. I’ve updated the maps and made the D3 charts responsive, so it should look a little better now.

Leave a Reply

Be the First to Comment!

Notify of