UAP Duration and Timing Study

UAP Data Analysis Do the over 100,000 UAP experiences reported over the past 30 years have a common thread? Are they all related, or are these experiences completely independent of each other? I've spent the past few weeks trying to answer this question starting one key observable: time. When are UAPs observed and for how long do they appear? What follows is a statistical analysis where I look at the time and data information we have on UAPs.

I found that most UAP encounters last around 3 minutes and they are more likely to happen in the evening toward the end of the week and in the summer time. However, there is a lot of variation with plenty of cases extending into hours at all hours and times of the year. Check out my UAP Data Science project to see how I created the data visualizations below that support these assertions.

How Long Do UAP Encounters Last?

In order to get a clear picture of the duration of UAP encounters, I focused on the observation length in minutes. It was hard to show this in a plot because a small number of reports have extremely long durations. For instance, case 31427 describes lights hovering near an airport and lists the time as 2,400 hours. This would be 100 days and seems more likely to be a data entry error. In any case, I took out extreme outliers that lie outside the 1st and 99th percentiles for the sake of the plot below.

UAP Duration Plot

This plot shows that the majority of sightings happen within seconds and minutes. Furthermore, all but the most extreme data points fall below just over two hours.

In order to get a better picture of the durations in the lower end I decided to look at the 50% of cases that fall in the middle of the distribution we are seeing. To do that I just found the durations at the 25th percentile and 75th percentile.

shape count minutes_p25 minutes_p50 minutes_p75
All Reports 121832 0.500 3.00 10.00

This paints a pretty clear picture for me. Essentially most UAP observations happen within this 10 minute window with a great majority being near instantaneous.

Do Different Shaped UAPs Have Different Durations?

Duration and Shape are the two strongest factors that we have to systematically distinguish UAP reports. Do they interact? Are some shapes more likely to be seen for longer times than others?

In this analysis as well as the one above, I'm looking at the "Median" statistic. This serves the same purpose as an "Average" but this statistic is less affected by outliers which is why it's useful here. What I did below is plot the median duration by shape.

UAP Duration By Shape Plot

As you can see, there is some variation here. Most notable are the "changing" shape which has a median duration of 10 minutes! This is intriguing. A shape of "changing" does suggest something more interesting than other simple geometric shapes. This category of shape could warrent a closer look.

Here are the detailed stats including the count for each type of shape. FYI you should be skeptical of the stats associated with very low counts.

Shape Count P25 Median P75
All Reports 121832 0.50 3.00 10.00
light 24883 0.63 4.00 15.00
triangle 11112 0.50 3.00 10.00
unknown 8369 0.75 4.00 15.00
disk 7025 1.00 4.00 12.00
formation 4084 1.00 4.00 10.00
teardrop 1039 0.17 2.00 8.50
sphere 7906 0.58 3.00 10.00
rectangle 2092 0.50 3.00 7.50
fireball 8702 0.20 2.00 5.00
oval 5257 0.50 3.00 10.00
chevron 1431 0.20 1.00 5.00
circle 12811 0.75 3.50 10.00
other 8258 0.50 3.00 10.00
changing 3004 2.00 10.00 30.00
diamond 1729 1.00 5.00 15.00
cylinder 1905 0.63 3.00 10.00
egg 1010 0.50 3.00 10.00
flash 2134 0.07 0.58 7.50
cigar 2953 0.50 2.50 10.00
cone 458 0.79 4.00 14.00
cross 392 1.00 4.00 12.00
crescent 2 0.13 0.13 0.13
flare 1 0.50 0.50 0.50
delta 9 2.00 5.50 12.75
dome 1 0.03 0.03 0.03
pyramid 1 2.00 2.00 2.00
hexagon 1 3.50 3.50 3.50
round 2 3.81 7.54 11.27
changed 1 60.00 60.00 60.00

I would say that overall, shapes are fairly consistent in terms of duration. There are notable exceptions though: Changing, Flash, Chevron, and Fireball are all a little different. When you think about these attributes it makes sense. Flashes and fireballs imply things that pass quickly from sight. Chevrons don't seem to me to have an obvious reason why they would appear for less time and this may warrant some further analysis. The Changing shape type is most interesting still.

Has Duration Changed Over The Years?

We have thirty years of UAP reports so naturally I'm curious to see if things have changed over the years. So I plotted the median duration times over the past thirty years.

Duration Over Years

As you can see there does seem to be a shift in the early Nineties to slightly lower duration times. While I can't know for sure what the cause of this change is I believe it's related to the change in data collection that NUFORC implemented in 1994. See this explanation from their About Page:

The principal means used by the Center to receive sighting reports is this website, which has operated continuously since 1994. Prior to that period, the telephone hotline and the U.S. mail were the primary means of taking reports.

What Time Of Day Do Most UAP Sightings Take Place?

Are you more likely to see a UAP in the morning, afternoon, or evening? Or does it not matter what time of day it is? When I first thought about this question, it seemed obvious that the answer would be "nighttime" and yes it does turn out that there are more sightings at night than day. However, I read about many notable sightings that take place in the day. Here is what the data shows:

UAP Sightings By Hour of Day

In the plot above, I've reverted back to using average as the statistic since we no longer have to worry about outliers. As you can see there is a real pattern here. Sightings stay low until they gradually increase starting at 5PM and peaking at 9PM.

So does this mean that UAPs keep Banker's Hours and come out only after Happy Hour? LOL. It seems more likely that this pattern is showing us when people have time to look up and the sky is dark enough to show the most common UAP - Lights.

NOTE It would be interesting to see this plot with only weekends to see if the pattern changes. Maybe there are more daytime UAPs that we think?

What Days Of The Week Do People See UAPs?

Do you think that more people see UAPs on Wednesday or Monday? Saturday or Sunday? I expected that just like in the section above, UAP sightings likely follow the workweek pattern. Indeed they do, but it's notable that even though we see more UAPs on weekends we are still seeing many throughout the entire week.

UAP Sightings By Day of Week

What Days Of The Year Do UAPs Appear?

Are you more likely to see a UAP in February or July? Do we see any patterns throughout the year or are their certain days when UAPs seem to appear more often than others? It turns out that there is a pattern throughout the year and that there are a handful of days that year after year have dramatically more sightings than other days.

This was another analysis that surprised me, check out the plot here:

UAP Sightings By Day of Year

Woh, see that spike in the middle? Can you guess what day that is? So that is the 185th day of the year, also it's July 4th. Here are the statistics for the days around the 4th of July and the day itself so that you can see the differences.

Day Of Year Count Min Max Mean Median P25 P75
180 298 1 31 9.03 5.00 3.00 13.00
181 673 2 34 14.32 13.00 9.00 19.50
182 724 2 50 13.66 12.00 7.00 18.00
183 396 1 31 10.42 8.00 6.00 14.00
184 495 1 69 13.03 7.00 1.00 17.00
185 1698 1 269 36.91 10.50 2.00 38.50
186 863 1 195 17.61 6.00 1.00 19.00
187 409 1 79 11.05 7.00 2.00 15.00
188 359 1 29 8.55 7.00 2.00 12.75
189 376 1 43 10.16 9.00 2.00 16.00
190 317 1 26 9.06 8.00 2.00 13.50

So what gives here? Are UAPs particularly patriotic? Perhaps. Or is there a more reasonable explanation?

Another big day throughout the year is January 1st which is New Years Day. In case you were not getting there yet, it's important to note that these days have a lot of fireworks and people staying out late at night. This is also true of summertime, which I found out when I looked at this data using a smoothed version of this data. This version of the data basically averages the observations to give you a smooth line with error bars and has the effect of taking out those outliers so that we can see the overall pattern of sightings throughout the year.

Smoothed UAP Sightings By Day of Year

This is a pretty dramatic plot which essentially shows more sightings in the summer months.


So what does all this mean? I really take two things from these findings so far. The first is the beginnings of a profile for the most common UAP sightings and the second really reflects the behavior of the people reporting UAPs.

Most Common UAP Profile

You could take from this study that the most common type of UAP sightings are described as lights in the sky observed for about 1 to 10 minutes at night and more often in summer. In addition to this very common case, you can start to see other profiles emerging. For instance the Changing type of UAP which is observed for a longer period of time.

Observers Behavior

The second big takeaway is when people are telling us that they see UAPs. While I can't prove that this is largely an artifact of our behavior it is telling that reports seem to track with the work week and vacation time in the summer. It would be interesting to test this agaisnt similar data from the summer hemisphere or from a culture that follows a different pattern.

However, there is also the datapoint that shows UAPs appearing more often at night. This one may hint at a deeper aspect of UAPs. Considering that the most common UAP shape is a light which would require darkness to appear most of the time this is a telling feature. But, it would be interesting to see if different shapes tend to appear at different times of the day.

UAP Profile Attributes

Now that I'm at the end of this project, a method of categorizing UAP is starting to emerge. So far I think that these three attributes would be a good starting point for the task of developing a profile of UAP events:

  • Shape
  • Duration in Minutes
  • Time of Day

Another potential attribute could be geographic location, but since I haven't fully explored that yet this will have to wait until I can take a deeper look at that.

Future Analysis Ideas

This feels like a good start and a picture is starting to form in my mind. The NUFORC dataset still has more that can be mined. I haven't really touched the text accounts themselves nor have I looked at the geographic data. And finally, more work can be done in developing profiles and sighting types.