Starlink Satellites, 04 January -
04 November 2020
Starlink is a group of thousands of satellites being deployed by SpaceX into low Earth orbit to provide broadband Internet access. A test flight took place in February 2018 and, in May 2019, deployment began, in groups of 60, referred to as constellations. Due to the test flight, the launch number is one greater than the constellation number, thus the second launch group deployed constellation Starlink 1, and so on. Below, observations are grouped by constellation.
The first constellation of 60 Starlink satellites, launched on 11 November 2019, was well-placed for observation in early January 2020. The night of 04 January was clear with a first quarter Moon, but not particularly transparent. Most satellites of the constellation passed virtually overhead. I was able to record 41 of them, including three slightly out-of-plane with the main group, along with several other satellites that passed through the field-of-view (and two aircraft).
In the animation below, the Starlink satellites pass from 2 o'clock to 8 o'clock. Most were invisible to the naked eye in the moonlit sky, but two were very very prominent, with brightness similar to the stars in Cassiopeia.
By mid-February 2020, satellites from the third Starlink constellation, launched on 29 January 2020, had come to visibility in the mornings. The morning of 12 February, although very cold, presented an opportunity to image the objects. The satellites came out of the Earth's shadow at an altitude of about 20°, close to the Moon. The main group consisted of 31 satellites, all in a line with a typical spacing of about 12 seconds. It was possible to see at least 10 satellites at the same time - the sight of them moving as one relative to the stellar background was quite surreal! Towards the end of the pass, one of the satellites was moving visibly faster than the others and overtook another. Currently, most of the satellites are in a circular orbit at an altitude of about 340 km, orbiting the Earth 15.75 times a day, climbing towards an operational altitude of 500-550 km. Starlink 1179 is in a lower orbit of 279x291 km, circling the Earth 15.95 times a day, so appearing to travel 1.3% faster.
I videod the passage of the satellites through the border of Corona Borealis - Serpens. Equipment used: Sony A7S camera, driven, 100 mm telephoto lens at f/2, ISO 16,000. Some acceleration of the event was necessary to reduce the size of the video file. The satellites are nominally in the same orbit, but from the rotating Earth they appear to sweep through a band of the sky. The impact of the satellites on astronomy, flashing through the fields-of-view of telescopes around the word, remains to be seen.
On the morning of 19 February it was again clear and I took the opportunity to record the passage virtually overhead of the constellation. I used a fish-eye lens. Although the satellites are launched in groups of 60 into a low earth orbit, they start to spread out as they propel themselves into their working orbits at an altitude of 550 km. On 19 February, the main group contained 28 satellites that passed in about 8 minutes. The movie below captures most of the satellites from approximately 05:40:30 to 05:49:50 UT. The still frame, taken at 05:47:11 UT shows 14 satellites. Equipment used: Canon 60Da camera, 8 mm fish-eye lens f/4, ISO 3200, 2 s exposures.
On 16 March, there was an opportunity to catch a train of satellites from constellation 3 passing directly overhead. Forty of the 60 satellites passed over in a group, one every 25-27 seconds or so, with eight visible at any one time. At this time the objects were mainly in an intermediate circular orbit at an altitude of 380 km.
The video below was taken using a driven mount. The evening sky was not particularly clear, with cirrus cloud present. Again I used a fish-eye lens, which meant that aircraft were a major problem, and removing them from the video was tedious!
The morning of 12 March was not very clear, the Moon was up and it was very, very windy, but it was the first opportunity to observe constellation 4, launched on 17 February. The main pack (about 44 satellites out of 60) passed by in 18 minutes. They were mostly 25-27 seconds apart, not as close as some earlier launch groups have been, but it was still possible to see six or more at once. All bar one could be identified using predictions from heavens-above.com: the exceptional one moved like all the others but there was no associated prediction. All the satellites passed close to the star Pherkad (γ Ursae Minoris, magnitude 3.05) which most of them matched in brightness.
I normally use a driven mount when recording Starlink satellites, but I forgot to switch it on initially! As a result, the stacked composite near the beginning of the video shows star trails.
On 07 April, I finally saw the Starlink satellites belonging to constellation 5. Conditions were mixed: the atmosphere was cool, calm, silent and still, but the sky was hazy and moonlit with cirrus clouds. I set a camera running, taking images to make a time-lapse movie. I had initially intended to set up a second camera in video mode, but concluded that conditions did not warrant it.
Usually, with the camera running, I watch the satellites pass overhead, but this time I could not see them. Nevertheless, the camera recorded them together with some additional "features" including a large brightness gradient across the field of view and moving cirrus clouds. (I have removed the gradient in the second part of the video, to give a much more pleasing appearance.)
I don't know why the satellites were so faint. Towards the end of the video, Starlink 80 (from constellation 1) can be seen passing above the path of the satellites from constellation 5, and appearing much brighter. Curiously, heavens-above.com does not include magnitude estimates for constellation 5, whereas it does for all the others.
By 19 April, heavens-above.com was providing magnitude estimates for constellation 5, indicating that the satellites would be exceptionally bright. I recorded the passage that night, half expecting another disappointment. Unfortunately, I was late in starting, so missed the first few satellites. I was stunned by how bright the satellites appeared, some appearing roughly as bright as Regulus.
The video below has not been polished - I have not removed the aircraft, nor labelled the satellites.
The passage of constellation 5 on 20 April was much different; in fact it was a disappointment! I recorded the passage, using exposures of two seconds, rather than four seconds used on 19 April, other details being the same. In the 20 minute interval from 22:00 - 22:20 BST, some 42 satellites passed through Ursa Major but a video shows only seven and a multiple exposure only 10.
The passage on 21 April occurred during evening twilight. This made it tricky to record because of the variation in background light level as night fell, necessitating more sophisticated processing of the images than had previously proved necessary! In the video below, the left hand pane is a view of the sky as night falls: it is difficult to see the satellites as the Sun is only 9° below the horizon. (However, dust on the sensor is very clear!) The image was captured using a fish-eye lens with a very wide field of view, showing a wide variation in brightness across the sky. The right hand pane shows the sequence of differences between each frame and a running nine-frame median, highlighting moving objects against a much flatter background.
My video of the passage of constellation 5 on 21 April is below. The two brightest stars in the lower left are Castor and Pollux. The video starts at 20:00 UT. It is cropped and resampled from a sequence of full frame images taken with a Canon 550D and Samyang 14 mm lens at f/4, 2 s exposures.
The following time-lapse sequence shows the passage of Starlink constellation 5 above stellar constellation Leo on 21 April. It spans the period 19:58-20:14 UT. It was recorded at 6240x4160 pixels then cropped to 1280x720 pixels. Taken with a Canon 6D camera and 24-105 mm zoom lens at 24 mm, f/5.6. ISO 4000.
The launch of constellation 6 was unexpectedly moved forward to 20:30:30 BST on 22 April 2020. This provided an excellent opportunity to witness the first pass of the carrier rocket in the evening twilight, some 20 minutes after lift-off. As the satellites are released some 15 minutes after launch, they were expected not to have separated sufficiently from the rocket to be individually visible.
I imaged the passage with a 600 mm telephoto lens on a tripod with a video head. Predictions for Ipswich showed that the rocket would pass close to Venus, providing a bright marker in the evening sky. Having found the rocket next to Venus, it came closer and rose to higher and higher altitude making it harder and harder to keep in the field-of-view. As it passed nearly overhead at an altitude of 74°, I could no longer follow it.
It's difficult to obtain good results tracking a moving object with a long telephoto lens and, indeed, the rocket appears to dance all over the frame. So I have picked through the video, keeping all the frames where the image is reasonably still, then centred them manually, and turned them into a video. Unfortunately, the time difference between frames is not fixed!
The video shows initially two Starlink objects, a long one and a point below and to the right and, much further to the right, Venus. The point is the carrier rocket. The long streak is a stack of 60 satellites, not individually resolved, after being released by the launch rocket. As they come closer the odd stars fly by from the top of the frame. Near the satellites some stray objects start to flash into view - they are retention rods that held the satellites in place on the carrier. The stack of satellites appears to rotate, but this is an effect of perspective as they pass overhead. Then I lose them before finding them again. By now the stack of satellites and the carrier are back-lit and have become one object again due to foreshortening, but four retention rods are clearly seen as separate items.
Below the video is a composite of selected frames, showing the evolution of the Starlink train as it passed overhead.
Opportunities were initially limited to see the constellation 6 once it had been placed into orbit. However, an opportunity arose in the evening twilight of 24 April. Visually, I could not see anything, but a video showed a train of satellites low down in the south-west. The video below is a 7 s clip of the original recording, cropped 50% both horizontally and vertically. The lower pane is a contrast-stretched version of the original (cropped): the satellites are still somewhat hard to see. The upper pane is a sequence of differences between individual frames and a running median of nine frames.
I then applied further processing to individual frames of the video, all 1500 of them, to produce the still images below the video. The upper still is a sum of individual frames, showing a trail effect, with the satellites following one another through the sky. The lower still is a stack of 25 frames (one second's worth) aligned on the satellites. (The fact that they are moving at slightly different rates makes it not possible to achieve good results stacking a larger number of frames.) It shows 36 distinct satellites, with one or two looking like unresolved multiples. This aligns with the fact that the 60 satellites of the constellation have separated into a group of approximately 40 followed about a minute later by another of approximately 20.
There was another opportunity to observe constellation 6 on 15 June. Some 40 satellites appeared in a regular procession with an interval of almost exactly 20 seconds. They were visible to the naked eye; this contrasted with constellation 7 which was visible earlier in the evening, but required use of binoculars (see below). In part, this was because the satellites of constellation 6 were at an orbital altitude of ~500 km compared to ~700 km for constellation 7. While most of the satellites become visible within the frame, there are a few times (26, 28 and 30 seconds into the video) where the satellites flare at the right-hand edge of the frame. (Unfortunately, the right-hand edge is marred by an ugly green colouration, an artefact of compressing the video from 1080 to 720.)
Constellation 7 was launched early on 04 June and there were four passages of the satellites over the UK during the hours of darkness on the night of 05-06 June. Unfortunately, conditions were not ideal, with a full Moon, at low altitude in Scorpius, causing considerable glare in the south.
The first passage was in the early evening with the Sun only 8° below the horizon. I could see no satellites, either by naked-eye or in binoculars. I captured with a video camera a series of moving faint dots, as well as a couple of satellites that brightened to magnitude 3 or so. The spectacle was unimpressive.
The second passage was at much higher altitude, up to 80°, in a dark sky. There was not a lot to see, until one satellite appeared at magnitude approximately 0.5. Binoculars then revealed a very impressive train of faint moving dots; however, they were not visible to the naked eye. Light pollution from Ipswich and sky conditions combined to produce a limiting magnitude of approximately four. The resulting video is below: it shows 54 distinct bodies, as a well as a couple that look like unresolved multiples. By way of comparison, Alcor is magnitude 3.96.
The third passage was early in the morning of 06 June, again at high altitude, up to 75°. The sky did not look as clear as before. Again there was a prominent bright satellite, but the rest were binocular objects.
The fourth and final passage was in the morning twilight, with the added complication that the view was towards the full Moon. Several satellites were visible to the naked eye at magnitude one or thereabouts, but most could be seen only with binoculars.
In response to concerns from the astronomical community, SpaceX has introduced a policy of orienting the satellites en route to their operational orbits so as to minimise their impact on the night sky. See https://www.spacex.com/updates/starlink-update-04-28-2020/.
By 15 June, the satellites had spread out and appeared low down in the south-west after sunset. They were visible only with binoculars, not to the naked eye. I recorded 52 in less than 10 minutes, sometimes two or three appearing in the field of view at a time. Later that evening, there was was a passage of the constellation 6 train - details above.
I generally use orbital elements, loaded into a planetarium program, to identify individual satellites. However as the satellites are climbing to working altitude, the elements rapidly become out of date. I found that two sets of data gave a different order to the satellites. Indeed, if the predictions show satellites to be, say, only one second apart, confidence is low that the order is correct. So in labelling the satellites in the videos I follow Eric Morecambe's stance on his interpretation of Grieg's piano concerto for Andre Previn "These are all the right satellites - but not necessarily in the right order!" (Morecambe and Wise Christmas Show, 1971).
I captured the second passage of constellation 7 on 05 June starting at approximately 22:45 UT. I used a 135 mm telephoto lens at f4 on a Canon 550D in video mode under pretty clear skies. I recorded the video as 1920x1080 px then rotated and cropped it to 1080x720 px. heavens-above.com predicted that all satellites would appear in the range of magnitudes 1.8 to 2.2. However, in practice they appeared fainter than this so I increased the brightness and contrast of the video to make them more apparent; in consequence, it is rather grainy.
The video is below. The two stars on the left are Mizar and Alcor in Ursa Major (magnitudes 2.23 and 3.99 respectively) and the video gives a good impression of the naked eye brightnesses that I observed - all but one satellite were rather dimmer than Alcor and the bright one was considerably brighter than Mizar. I could barely see the satellites by naked eye but they looked very impressive (almost a space armada!) in binoculars. I tried matching the timings on heavens-above.com with my video but couldn't obtain a good correspondence, so can only identify the bright satellite as around number 40 in the passage. (Satellite 20 in the sequence is tagged on heavens-above as a Visorsat model, i.e. it has a special sun-shield to help avoid reflections from the antenna.)
I expect that the majority of satellites were dimmer than predicted due to efforts by SpaceX to reduce visibility of the craft by angling them during the ascent phase to reduce reflection. Perhaps something went wrong with the bright one?
On 05 July, I finally recorded the Starlink 8 satellite train, launched in daylight on 13 June 2020. However, with all the interest in comet C/2020F3 (NEOWISE), my report on the observation had to wait!
Fifty-eight Starlink satellites were launched for Starlink 8, along with three Skysats. I recorded 46 satellites from the launch, together with several from the Starlink 3 launch on 29 January 2020. The majority were in one long group, with, in most cases, a satellite being about 20 s from its neighbour. Those not at 20 s intervals were in slightly lower, faster orbits. In the video is a short clip at the live rate, but the major part is speeded up and processed to make the satellites more visible. I also left the aircraft in - they are relatively easy to identify as there are few of them about!
I did not see the majority of the Starlink 8 satellites with the naked eye from my suburban location, as they must be magnitude 4 or 5, but binoculars easily showed them. So it seems that the "string of pearls" effect seen previously is likely to be a thing of the past, as Starlink is taking steps to reduce the impact of the satellites on astronomy.
The morning of 12 August provided the first chance of imaging the Starlink satellites from the latest launch on 07 August. Previous nights during the heatwave of early August all suffered from high level, obscuring cloud.
Only 57 satellites were on board the latest launch - the remaining capacity being taken by a pair of earth observation satellites from BlackSky. All Starlink satellites launched now are fitted with sun visors to reduce their visibility to astronomers.
By the morning of 12 August, many of the satellites had separated from the main train, but 39 were scheduled to pass through Cassiopeia within 40 seconds. Unfortunately, I set too long an exposure time, ¼ s instead of 1/30 s and, with the satellites appearing so close together, in some cases, streaks caused by different satellites overlapped. In the below video I counted 36 satellites, including the straggler at the end. Some of the longer streaks are obviously caused by a pair of satellites, so I counted them as such. The remainder of the group are some 8 minutes behind and spread out.
The typical summer haze meant that the satellites were not visible to the naked eye. The first I saw of them was on the back of the camera viewfinder, then in binoculars.
While I was pleased with my observations on 12 August, I was annoyed that I had used the wrong exposure. On the morning of 13 August, there were two more opportunities to see Constellation 9: the first at 02:36 UT had better weather prospects than the second at 04:13 UT. I awoke at 02:15 UT and a quick check out of the window showed stars. I went outside to find that much of the sky was covered in cloud - lots of little high, dark, almost stationary clouds. Even the Moon was obscured, but some bright stars were visible. I could discern Cassiopeia and, knowing that my cameras can record more than I can see, I decided to proceed.
The clouds were shifting and I was rather surprised to see a moving object in the electronic viewfinder; I looked up and could see that it was a satellite and that there were several more moving in the same direction. They were Starlink satellites! I was surprised to be able to follow a handful of them in the direction of the Moon.
The following video, taken using a Sony A7S with 50 mm lens, shows the unmistakable pattern of Cassiopeia and indicates how cloudy it was. Eleven Starlink satellites are easily visible, most continuing to be visible through advancing clouds. They were of comparable brightness to the stars in Cassiopeia, magnitude 2. The remaining satellites of Constellation 9 don't really show in the video.
I used a second A7s with a longer, 100 mm, lens to produce the video below. It shows the bright satellites and the advancing cloud-front. After the last bright satellite passes, faint satellites are visible in the gap above the top of the advancing cloud bank, until the gap closes.
The forecast for the earlier observation had been fair, but it turned out cloudy. The forecast for the later observation was grim. But I decided to stay up and have another go. I went outside periodically to check the weather, to find at times that the only things visible were the Moon, Mars, Venus and Capella. As it approached 04:00 UT, Cygnus was quite clear, but it was hard to discern other constellations, as the limiting magnitude seemed to be about 2.
When photographing Starlink satellites (other than with a fisheye lens) it is necessary to aim the camera in advance at where the satellites will appear. I chose an area in Andromeda, approximately due south. Just before the arrival of the satellites, I started scanning Cygnus, in binoculars, for moving objects. Sure enough, the satellites appeared and were in fact visible to the naked eye. I watched as they came into the field-of-view of the first camera, but not the second: it was not aimed correctly. A quick adjustment rectified the situation, but the second camera missed the first part of the passage. My mistaken aiming came about as follows. With the camera set to high sensitivity (ISO) and with a narrow field of view, I had found a set of stars that looked similar in the viewfinder to what I could see with the naked eye. In fact, the camera was pointing several degrees too low. In future I may use a green laser pointer to highlight the desired area of the sky.
The following video, taken with the camera with 50 mm lens, shows 14 bright satellites and a few fainter ones. The 14 bright satellites look comparable in brightness, if not brighter, than the labelled stars, which are all around magnitude 2.0-2.1. The limiting magnitude was around 5.
The initial six bright satellites are missing from the following video because of the pointing error. The clip starts with a faint satellite in the extreme bottom left (before the bright pair); thereafter I counted a total 23 faint satellites, making 37 in all. This includes 2 pairs that are very close together, one at 10 seconds, the other at 17 seconds. Stars are recorded to magnitude 7. www.heavens-above.com predicted a passage of 39 satellites.
The morning of 06 September 2020 was the first opportunity to observe the Starlink 10 satellites launched nearly three weeks previously. When I got up there was a long bank of cirrus cloud drifting by, well-illuminated by the 18-day old Moon. The satellites had started to spread out but the main group of some 35 was predicted to pass by in less than eight minutes. The bank of cloud moved away, only to be replaced by another one just at the critical time - how frustrating!
In the short clip below shown at x1 speed, I show only three satellites from the earlier part of the pass, one at approximately magnitude 4 and a pair at approximately magnitude 6. The stars at the top are the Hyades, including Aldebaran. Overall the spectacle was unimpressive, made worse by the back-lit clouds obscuring the sky.
It will be another week before satellites from the latest launch (Starlink 11, launched 03 September 2020) become visible in the morning.
On 09 September, I finally managed to record the Starlink 10 constellation without clouds, passing virtually overhead by the Perseus Double Cluster. The spectacle was singularly unimpressive: of the 32 satellites that passed in some seven minutes, only three were visible to the naked eye, the remaining 29 being binocular objects, perhaps magnitude 5.
An incidental highlight was the capture in the video of a small meteor - it flashes by about eight seconds into the clip. I replay that section at the end of the clip at much slower speed.
In the early hours of 20 September 2020, I got round to recording the latest (as I write) set of Starlink satellites, launched on 03 September. The constellation had spread out and, while there was some clumping, it was predicted to take 18 minutes for 58 out of the 60 satellites launched to pass overhead.
I expected to see faint dots visible only in binoculars with a handful that might be visible to the naked eye. I waited for the first one to emerge from the Earth's shadow and was pleasantly surprised that it was visible to the naked-eye. Then several more appeared, visible by eye. It started to look as if all the satellites would be bright, but a quick check in binoculars revealed in addition fainter satellites, invisible to the naked eye. The brighter satellites matched nearby Mirfak in brightness, at magnitude 1.8, possibly brightening as they moved towards sunrise.
A driven camera recorded many satellites but the field of view was too narrow to capture their lateral spread. A stationary camera with a wider field of view recorded all 58 satellites, 37 bright and 21 faint and, in addition, four satellites from previous launches, appearing coplanar with those of Starlink 11, various other Starlink satellites and other satellites and two meteors. The four satellites coplanar with Starlink 11 are orbiting at an operational height of 550 km, with an orbital period of 95.64 minutes, and move more slowly than the Starlink 11 bodies.
Within Starlink 11, there are two sub-divisions of satellites, those currently at a height of about 380 km with a period of 92.15 minutes, and a second, smaller group at an altitude of 360 km with a period of 91.80 minutes. The two groups have started to separate due to the differing orbital velocities and now follow two slightly different tracks in the sky. In the video below, the former group appears first, then members of the second group appear, at slightly higher altitude. The first part of the video is speeded up x20 relative to live. This is repeated in the second segment at x100. The final segment is at live rate (x1), true to what I observed, showing five bright and one faint satellite.
The Starlink 12 constellation was launched on 06 October 2020. Unfortunately, dismal weather prevented me from imaging it until the morning of 25 October. The date was almost the last opportunity to capture the morning passage of the constellation, after which it became visible in the evening, with individual satellites spread out much further.
On the morning of 25 October, clouds had cleared but the sky was not transparent: it looked miserable and felt very damp. Fainter naked-eye objects such as Praesepe were not visible. Fifty-six of the 60 satellites emerged from the shadow of the Earth in Gemini and Taurus during a period of some 12 minutes. Most were bright, some rivalling Castor (magnitude 1.6) in brightness, and I could see perhaps six at any one time.
The first part of the video below shows all 56 satellites, speeded up by a factor of 20. It includes several other satellites, including other Starlink satellites from previous launches, but I was not able to identify all of them. Then there is a clip speeded up even faster: x100. Lastly is a short clip showing the motion of the satellites in real time. The overall view shows much more detail than could be seen with the naked eye.
The morning of 04 November brought the first frost of the winter. It also provided an opportunity to capture the Starlink 13 constellation, launched on 18 October 2020. The presence of the 18 day-old Moon in Taurus resulted in a bright sky bright.
The passage of the constellation was underwhelming. Most satellites required binoculars to be seen. The leading satellites were binocular objects but, after a while, some were visible to the naked eye at magnitude 2, rivalling the stars in Ursa Major.
The video below shows 58 satellites from Constellation 13, several from Space X constellations launched previously, and other satellites, not all of which could be identified. In the video, first there is first section showing the passage at x20, then a frenzied version at x100, then finally a clip in real time. The video shows more detail than was visible to the naked eye.