Tracking the Contrail of ANA173 with GOES-16 Images

Mick West

Administrator
Staff member
20171009-161325-dzdjh.jpg
http://rammb-slider.cira.colostate.edu/

The GOES-16 Satellite takes an 121 Megapixel image of the Earth every 15 minutes. This makes it idealy suited for viewing the contrails of planes. With the frequent imagery you can often see the contrail in the process of forming, like this one which just passed over Seattle (and must have looked quite impressive from the ground)
20171009-161530-brzhg.jpg
Here we see the contrail narrow to a point, indicating where the plane is at that time. Since the time is in UTC, it's a pretty easy task to find it in FlightRadar24
20171009-161737-rt7mz.jpg

Flight ANA173 from Houston to Tokyo (still in the air over the Pacific as I type this). So it's cool to see the scale of the contrail, but another interesting thing is seeing it evolve and move.
20171009-162055-scfwv.gif

That animation goes from 20:30 to 22:45, 2 hours and 15 minutes of the contrail persisting. Clearly it's not dissipating, and it's quite obvious why - it's a cirrus cloud, in the middle of a bunch of other cirrus clouds. If they don't dissipate then why would it?

By 22:00 it's spread out about as much as it's going to and still be recognizable as a contrail.
20171009-162922-nb6zx.jpg

It's about as wide as the Puget Sound, around three or four miles across.

GOES-16 starts a new era of contrail observations in the US. Contrail season is just starting in Northern California, and I'm looking forward to using GOES-16 to track down a bunch of interesting contrails.
 
There are two points to bear in mind when matching the time of the satellite image with the time from the flight track.

One is that the GOES-16 image resolution is only about 1000 metres per pixel, therefore it can capture only those contrail parts, which are sufficiently spread out, that is, at least a few minutes old.

The other is that it probably takes several minutes to collect the full Earth image, therefore there will be some offset between the reported time of the image and the actual time when the image of a particular area was taken.
 
Last edited:
The other is that it probably takes several minutes to collect the full Earth image, therefore there will be some offset between the reported time of the image and the actual time when the image of a particular area was taken.

Here's a good description:
http://spaceflight101.com/goes-r/goes-r-instruments/
ABI employs a scanning mechanism that sweeps out an East-West image swath and steps in the North-South direction after each successive sweep, capable of assembling a global image every five minutes.
Content from External Source
upload_2017-10-10_9-24-3.png
I've emailed the Slider team to ask what the timestamp corresponds to.
 
One is that the GOES-16 image resolution is only about 1000 metres per pixel, therefore it can capture only those contrail parts, which are sufficiently spread out, that is, at least a few minutes old.

I don't think it would necessarily take a "few minutes". A white line that's less than a pixel wide can still be visible, it's just dimmer as it gradually blends in with the background color.
20171010-093633-r1qam.jpg

This flight was a 777, wingspan of 200 feet. The width of the contrail exceeds this in a few second. Even when it's only 1/10th of a pixel, if the conditions are right for contrail growth it's likely going to be optically dense, so bright white, and hence visible as a faint grey line.

Still, it's a good point that we can't indicate a spot and say "here is the plane", especially with the uncertain timestamp.
 
I don't think it would necessarily take a "few minutes". A white line that's less than a pixel wide can still be visible, it's just dimmer as it gradually blends in with the background color.
I have been experimenting previously with Himawari-8 images and got lag times of up to ten minutes for a contrail to become visible in the satellite image. Even for the MODIS satellites with the resolution of 250 m/px, it is not always possible to detect a contrail that has not spread out enough.
 
I have been experimenting previously with Himawari-8 images and got lag times of up to ten minutes for a contrail to become visible in the satellite image. Even for the MODIS satellites with the resolution of 250 m/px, it is not always possible to detect a contrail that has not spread out enough.

On the 20:30:41 image I'd say the visible trail is just over the central valley of Olympic national park when it either ends, or is lost in the cloud
20171010-105916-7pxhg.jpg


This is before ANA173 makes a distinctive turn, putting it around here on the FlightAware track, when the plane itself is about 30 miles further along.
20171010-110012-zxg2u.jpg
That's about at the 8:26 point on the track. Only four minutes behind the GOES-16 time - although we don't know exactly how that relates to the actual image (start of capture, end of capture, transmit time?)
 
Here's flight AA60 during a distinctive west-east portion of its flight at 17:45 GOES-16 capture time.
20171010-114524-xdg3a.jpg

It's a bit hard to see there, here's contrast enhanced.
20171010-114808-w37lo.jpg
 
Regarding timestamps:

Hi Mick,

My name is Kevin and I'm the web developer for RAMMB. Thanks a lot for the kind words about SLIDER---it's been a fun (and challenging) project.

The times that are displayed in SLIDER are the start times of the scan for any given image. So for full disk imagery, the scan starts roughly 37-38 seconds into the full 15-minute-long Mode 3 cycle, which lines up with the image you attached depicting the scan mode. That said, you're right that knowing the relative position of the point of interest in a given swath could give you a more accurate time for that location.
Content from External Source
 
@Mick West
I see that you previously selected Full Disk sector. It appears, however, that for CONUS (continental USA) sector, there are shorter (5-minute) intervals between the images and probably a higher resolution too.
 
@Mick West
I see that you previously selected Full Disk sector. It appears, however, that for CONUS (continental USA) sector, there are shorter (5-minute) intervals between the images and probably a higher resolution too.

Hmm,yes that shorter interval might make a significant difference.

CONUS does seem a tiny bit better quality, not a lot though, I don't think the resolution used in SLIDER is different.
20171010-162517-sol4b.jpg
 
CONUS does seem a tiny bit better quality, not a lot though, I don't think the resolution used in SLIDER is different.
20171010-162517-sol4b.jpg
Your link above (http://spaceflight101.com/goes-r/goes-r-instruments/) says that the resolution is 500m:
The resolution is band dependent and ranges from 0.5km for the visible and 1.0km for short-wave IR to 2.0km for mid-wave and thermal IR.
Content from External Source
However, dividing the area of Full Disk (~125M sq. km) to the number of pixels (121M) in its image gives about 1 sq. km per pixel, that is, the Full Disk resolution is only about 1000m.
 
Last edited:
Yeah, it should be better, but does not seem to be. It's possible that SLIDER simply sets the same zoom level regardless of source.
 
Back
Top