Hacker News new | past | comments | ask | show | jobs | submit login
Is this plane landing or departing? (stackexchange.com)
256 points by bluedino on Jan 15, 2017 | hide | past | favorite | 80 comments

There is motion blur on the traffic cone suggesting that camera was moving right and slightly down. There is no blur on the plane itself since the camera was locked to the plane's motion. So I deduce a landing photo.

I 100% agree, and used this method. :P

A consideration not listed on the OP is that Air Force pilots are required to conduct partial-flap landings on a continuing basis to maintain qualification. This could explain why the flaps do not appear to be in landing configuration. Normally, this training is conducted in the simulator, but if the president is not onboard and and the aircraft is inbound for maintenance at Boeing Field this would be plausible.

Partial flap landing distance is dependent on the aircraft, but 4,000+ feet as the position suggests would be very reasonable. The pilot could have also intentionally landed long to expedite taxi to parking.

All that said I still have no idea if the aircraft is arriving or departing.

See the comments for the answer: http://aviation.stackexchange.com/a/34609

> So...abelensky found the series of photo this was taken from, see for yourself: https://drive.google.com/file/d/0BxQOwrukppp5WTlfaEFLTVNzTnc... ...it's clearly landing.

Highly doubtful it would be going to Boeing Field for maintenance. BFI is a 737 delivery center. I'd expect it to go to Paine Field if it was returning to the factory, as that's where the 747/67/77/87 are manufactured.

The approach I would have taken is to check the EXIF data in the photo to see what time it was taken. You could then look at historical records or news records to see about when the President left or landed at that airport on that day. If the time is closer to when he left its a take off, if it is before it is a landing.

These planes can fly without the president, so you're not guaranteed an answer this way.

I was thinking this too, but was guessing that there are databases of all flights in the US - I know there are those flight tracker websites out there and someone is probably recording that data.

Yep, flightaware advertises flight history "dating back to 1998".

Example: https://flightaware.com/live/flight/DLH458/history

First thing I thought of. The hosted pic [1] does not seem to be the original, though.

1. https://i.stack.imgur.com/IWD2X.jpg

Edit: the person asking the question rejected a request for the EXIF data in a comment. Seems to want to limit the puzzle to direct analysis of the plane.

Just for comparison, look at the standard "landing" sign in an airport https://www.google.com/search?q=airplane+landing+sign&tbm=is... I always chuckle

Oh cool, now we know why it looks so stupid: it's because it's almost impossible to distinguish takeoff from landing if you don't exaggerate the angle.

Exactly, that's why I think this is a great pictogram: it makes it clear what it's supposed to represent, even though it has nothing to do with what a healthy landing looks like. A pictogram's job is to fit most people's conceptual model of what it represents while being simple and distinguishable, not to be an accurate representation of whatever it stands for.

This pictogram is simple (just a few shapes), it's easy to distinguish (e.g. from the pictogram for "departure"), and it fits most people's conceptual model of a landing (plane comes down from sky to earth).

Now that you point it out, that icon is terrifying

Huh, I have a question about that: do planes actually point downward at any point during a normal descent, or do they just adjust velocity and control surfaces so as to produce less lift?

I understand that planes are pitched upward at the moment of landing, but I don't know if that's true during all stages of normal flight or not!

> do planes actually point downward at any point during a normal descent

For a "Sarajevo Approach" the nose of the plane actually points down until the very last moment [0]. I saw once live and it looked quite sketchy but they routinely did this routinely during the Bosnian war in order to stay out of the range of rockets for as long as possible.

[0] https://www.youtube.com/watch?v=hNQRGOgHrZU and https://www.youtube.com/watch?v=gc6s9fopyg8

Previously known as the "Khe Sanh approach" [0]

[0] http://h2g2.com/approved_entry/A4187621

From your link:

>One of the weirdest things about this is the fact that you get to see the upper side of the wing. With planes this big, this rarely happens. If you see them on the ground, you walk under and around them, always looking upwards. When you see them in flight, you obviously very rarely get to see them from any quarter other than below. Seeing the top half of the Hercules like this therefore feels inherently wrong, especially since the plane is pointing directly at the ground. Like seeing a picture of sinking ship perpendicular in the water or a crashed car upside-down with wheels still gently turning, seeing the unfamiliar upper side of the wing induces a sort of subconscious panic.

This perfectly explains what I was feeling when I was watching GP's youtube links.

Generally speaking, during the approach the nose of the plane will actually be pointed slightly downward to control the airspeed (pitch for airspeed, power for altitude). Once you cross the threshold, the pilot generally removes any remaining power and begins to pitch the airplane upward (flare) to reduce the airspeed to stall the airplane. Ideally, the airplane basically stalls at the exact same time the wheels touch down on the runway.

General aviation airplanes typically fly nose-down during final approach, then pitch nose-up for the flare. Jets typically fly slightly nose-up during final approach and maintain the same attitude during the flare (or you could say they don't flare, they just fly straight down to the runway).

Just as an aside: if you have too much energy on approach in a small single engine piston plane (e.g. a Cessna 172), you can pull the throttle to idle, pull the nose up until you're below full flap extension speed (85 knots), then put in full flaps, and then point the nose down and keep descending (in a steep nose down attitude) at 85 knots. It's quite fun, and brings you down fast (i.e. steeply).

They do point slightly downwards during approach, before levelling off then pitching upwards slightly before touchdown.

I'm impressed that this question is so difficult. I presumed that, for an aviation expert, a quick glance at any photo of a similarly situated aircraft was enough to make a determination.


my first impression was landing but I can't quite explain why. I used to be an air traffic controller so I have some amount of "machine learning" in this area though not a lot from this angle or this kind of aircraft.

The angle didn't quite look right for a takeoff though after looking at a few takeoff videos of similar airplanes it seems they mostly try for steeper takeoffs but there's a lot of variability. There are conflicting requirements, you want to get as much altitude as possible as fast as possible so you have more options if there's any problem but you also want to gain speed and your initial angle might be limited by stall speed or other factors ( http://www.boeing.com/commercial/aeromagazine/aero_12/attack... ).

Even if I had 50% of being right it's always nice to be able to explain to oneself how good you are and when you get it wrong you can blame something else ;)

For a human, it's just called "learning". :)

I assumed the parent meant something slightly different from just learning - something along the lines of: observation of instances of planes taking off and landing but without conscious consideration of the features that differ between the two sets of "training data". Not really learning per se as that would require some form of study of the planes and their differences??

I'd say the most appropriate term here is intuition (which I'd say is definitely regular ol' learning, but that's a matter of definitions and semantics).

"If you wanted to take a picture of a plane taking off, I'm sure you would have done it while it stil had some of its wheels on the ground. But if you wanted to take a picture of a plane landing, you would probably do it while it was still in the air."

The plane is in the air, so I say it is landing.

amazing! How did such a discussion inducing question survive on modern stack overflow? I guess the question-closing-mavens do not stalk the aviation section as they do the programming/IT ones.

My guess would be that the aviation SE doesn't see variants of the same questions posted every day.

I think it could also be because there's clearly a single correct answer, even though it generates a lot of discussion.

The smaller SE sites tend to be more tolerant of discussion. I've asked a couple blatantly subjective questions on math.SE that were answered by moderators and didn't get closed.

Simple: because it's not Stack Overflow ;) It's a different site of Stack Exchange, with slightly different goals, and several orders of magnitude smaller (low tens of thousands vs. high millions of users) - which is IMNSHO the crucial difference. On the scale of SO, this question would get asked by ten different users a day, and consequently it would get closed.

Can we train a convnet to recognize departure vs landing and then analyze some of the layers to see what is being activated?

Although this may have been posted in jest, it would probably work quite well or at least be an interesting experiment.

There's a large corpus of images in both categories, and it's a binary classifier which keeps things simple. There are numerous image cues that people have identified such as the flap angle relative to the wings, absence of smoke from the tyres, absence of heat haze from the jet wash, wheel-spin and so on. Depending on angle you could also get the landing markers on the runway. It seems reasonable that a convnet would identify these features.

My only concern is that features like heat haze and wheel-spin might require very high resolution inputs to the net. Also it's quite hard to search explicitly for "planes taking off" as you get images from aircraft about to take off, aircraft taxiing, etc. Finally it's quite hard to find publicly available images, e.g. most photos on Aviation.net are copyrighted.

nope! not in jest at all. i think the only possibly difficult thing would be finding the solution classification.

The angle of attack looks very low for an aircraft taking off with it's wheels maybe 30' above the ground. If you look at this takeoff video they tend to rotate more than that. https://www.youtube.com/watch?v=5ptrsx-A3H0&feature=youtu.be...

This is a perfect example of nerd sniping. (https://xkcd.com/356/)

As it relates to human nature, my non-scientific guess is that most people want to witness the president arrive, so I bet the OP took the photo when the plane was landing ;)

It's interesting that a lot of answers say that it should be possible to figure out if it's landing or taking off by figuring out the direction of the landing strip. Do most airports have only one-way runways? I know my local airport, in Krakow, Poland, can operate either way, depending on the wind on that particular day. I've both landed and taken off in both directions, many times.

It's not so much the direction as it is the location of the aircraft along the runway. Boeing Field (KBFI) has a number of recognizable landmarks along the runway (as seen in the discussion), so some of the theories advanced had to do with whether or not the aircraft could be in a given phase at the recorded location.

Furthermore, both landing and departing aircraft will move in the same direction (into the wind).

You are correct. However, at many airports the winds are almost always favoring one direction. At LAX, for example, the wind is almost always coming from the West.

Just to make this about me: I used to work across the street from the Museum of Flight, shown in the aerial view of the field in question.

Landing. It"s not nose up enough for takeoff. This is a partial-flaps landing at a relatively higher speed than is normal. If this was a takeoff the nose would be higher. Shalow landings are optional, you can abort, but nobody does shallow takeoffs. You want altitude asap (within reason) because there is no abort once rubber leaves runway.

I immediately thought landing because the entire plane doesn't look like it has nice "straight" angles aligning with the runway. But I guess it is hard to tell from the angle of the photo itself.

On take off, the back wheels would've been nicely aligned with the runway.

Oh well, I deduced right?.

I don't see any spin blur on the wheels.. if the wheels are not rotating , it would be a landing.

I can just hear the Dread Pirate Roberts say "Truly, you have a dizzying intellect."

"I'm just getting started!"

OP has the picture and the file should have the date-time stamp of the flight. Couldn't they look up visits by Air Force One or travel by the president around that point in time? It may not give an answer 100%, but it should help.

Do we know the president was on it? These planes fly without the president for maintenance, photo ops, etc.

Do those big planes take off with flaps? I didn't know that, so went with landing.

Yes, the 747 (which this aircraft is based on) generally takes off with flaps 10° or 20° depending on load, runway conditions, temperature, etc. Landing flaps are 25° or 30°.

Yes, flaps are routinely partially deployed for normal takeoffs. On the 747, there are leading edge flaps in addition to the "normal" trailing edge flaps.

On the leading edge, they're called "slats" [1]. Most passenger jets have them, and they're rather important for take-off: The first fatal 747 crash, a Lufthansa machine departing Nairobi, came down because the slats were not deployed [2].

[1] https://en.wikipedia.org/wiki/Leading-edge_slats

[2] https://en.wikipedia.org/wiki/Lufthansa_Flight_540

You're correct on most airliners, but not the 747 (on which the VC-25 is based).

A 747 has Kreuger flaps [1] between fuselage and inner engines and 'variable camber leading edge flaps' on the rest of the wing.

[0] - https://en.wikipedia.org/wiki/High-lift_device [1] - https://en.wikipedia.org/wiki/Krueger_flap

Awesome, thanks. Nice illustration as to the differences in the Wikipedia article. The LH crash Wikipedia article seems inaccurate, then.

Edit: put a note in the Wikipedia article.

Yeah, flaps decrease speed but also increase lift, so they're good for both takeoff and landing.

I'd suggest to try and teach a neural net to decide upon airplane pictures whether they're landing or taking off. That way the stackexchange debate could be settled (with a chance to be wrong here, haha xD )

the flaps is in down-ward angle ,so i believe the plane is taking off


By the position of the flaps its most likely departing the airport.

Landing, there is hardly any disturbed air behind the engines. While modern engines don't produce all that much smoke they do produce lots of heat when running at full power, leaving a tell-tale trail of visibly disturbed air and exhaust gases behind them. No such trail is visible here so the engines are at a low power setting (idle or close to it).

In addition to that, the engines of the Air Force One aren't really modern. The disturbed air should be even more visible there.


There's nothing in the site guidelines that states all content will conform to a narrow definition, in fact this particular post is relevant because the question and subsequent answers on SE are interesting to engineers and aviation enthusiasts. Here is the relevant section of the guidelines:

> On-Topic: Anything that good hackers would find interesting. That includes more than hacking and startups. If you had to reduce it to a sentence, the answer might be: anything that gratifies one's intellectual curiosity.

By contrast, your entire comment is in violation of the same guidelines:

> Please don't submit comments complaining that a submission is inappropriate for the site. If you think a story is spam or off-topic, flag it by clicking on its 'flag' link.

The full text of the guidelines can be found via the link at the bottom of the HN main page.

FYI: Most people here are familiar with stackoverflow.com and related 'Stack Exchange' sites, earlier this week this question was a 'featured question from the Stack Exchange network' and, if you are on Stack Overflow looking up something difficult, it is very tempting to click on the other questions listed in the sidebar, including this one.

Therefore, this story is on HN in part because it was a promoted link on Stack Overflow this week.

this is so much better than talking about politics, isn't it?

why does this matter?

The self-censorship in this thread is insane. I've noticed at least two (valid) comments now that have been deleted due to downvotes.

That cat is both dead and alive.

Clearly a departure. Went wheels up right after rotate. No way they'd be that low with the gear position the way it is without a failure. No landing markers in sight. This is either a take off or a disaster.

Asker found the full set of photos confirming that it's a landing: https://drive.google.com/file/d/0BxQOwrukppp5WTlfaEFLTVNzTnc...

Theory vs experiment. More data always wins.

The gear aren't in the process of being retracted in that photo, they have that backward tilt on landing too.


Gear are retracted once a positive rate of climb has been achieved, but not that early.

To add to your comment, the FO watches the speed tape and VSI and calls out events to the captain (V1, V2, rotate, positive rate-gear up).

I looked around on Youtube for some good examples, and this was the best I could find:


The captain keeps a hand on the throttles until V1, and then it is several seconds after takeoff that FO retracts the gear.

That gear is down. The trucks are designed to hang at an angle like that.

In which the internet pretends to know things.

Isn't it a false choice? Why can't the answer be "both"? Explain in detail.

> Isn't it a false choice? Why can't the answer be "both"?

How does a plane both take off and land at the same time?

Unrelated to the picture, but touch-and-go probably qualify as "both", although not exactly same time either.

I'm very surprised this maneuver wasn't assessed in any of the answers. It actually does look like they're practicing a touch and go maneuver. The tires aren't producing any smoke and the flaps are positioned at a T20 so its both a landing and takeoff configuration. This is also a very late position for a 747 to be landing on an airstrip of this size. All evidence, for me at least, suggests a touch and go.

The fact that its airforce one aswell and they'd likely be practicing different techniques and training pilots.

Taking the lack of air distortion from the behind the engines, it could be because the photo is extremely grained/blurry to begin with and the flaps foiling the air to the ground rather than directly behind the engine.

My only other guess is they were practicing late landings.

By combining these internet answers with the airplane on treadmill absurdity from years ago?

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact