Model of Starlink flares - some progress

From: Richard Cole via Seesat-l <>
Date: Mon, 11 May 2020 09:50:24 +0000
I have been working on a model of Starlink visibility based on a 
Sun-pointing model used in the low-drag configuration below 550km. This 
is discussed at the link below and is termed here Model-A.

I have used the same model to analyse various flare events gleaned from 
personal contacts, reddit posts and SeeSat-L posts. The analysis looked 
at simple modifications of Model A to try to explain the flares.

Model-A was modified (to Model-B) to allow larger variations in 
roll-angle round the velocity vector and also deal properly with 
Starlink locations at large distance from the observer. The Sun and view 
angles to the panel was analysed to indicate when specular reflections 
might be expected, at multiple positions across the sky.

Model-A does not predict flares at high observer altitudes, since a 
Sun-pointing panel (even one aligned to the velocity vector) will 
reflect sunlight in the Sun-orbit plane, which is at low altitudes 
(though under study to see if there are observable events at those 
altitudes). Model-B uses large roll-angle to allow reflecting surfaces 
at orientations that might give flares at higher altitudes (where the 
angle of ray deviation has to be larger to reach the observer).

I deal with four reported flare cases which are charted here (fig 1):

Case 1: Italy 17th April - (unknown observer)

The observation is shown in a nice video here:

This shows multiple Starlinks crossing the virtual image of the Sun 
created by some reflective structure on each Starlink, all aligned in 
the same direction by the same attitude control law. I have used the 
brightest flare for the analysis in fig 1.

A rough match to the flare is given by a roll-angle deviation of 30 
degrees (the same deviation is used for cases 2 and 3).

Case 2: Sydney 5th May (Reddit user ChodaGreg)

At lower altitude.

Case 3: Leiden 21st April (Marco Langbroek)

The same 30 degree roll-angle deviation give a reasonable fit to the 
flares near Pollux.

I am reasonably confident that the basic model is correct, that is the 
panel is sun-facing as described in the article. The detailed analysis 
of the Leiden data using Model-A would break down if a large roll-angle 
applied to the whole spacecraft.

Therefore, I am suggesting there is some sub-structure that is tilted 30 
degrees from the main structure around the long axis of the whole panel, 
that is the short axis of the spacecraft block itself. Possibly this is 
the sub-panel holding the dish antennas at ends of the panel (used for 
s/c control and telemetry, not the user links). These panels are 
required (in this scenario) to articulate the dishes in deployment, to 
point the dish towards the ground. The images of ejection of the 
spacecraft from the Falcon show the dishes are pointing in the other 
direction at that time.

I note the published sunshade design for Visorsat may shade the area of 
these dishes, suggesting SpaceX is seeking to mitigate any effect of 
these panels in the final orbit configuration. A polished panel only 
15x15cm acting as a mirror can give a bright flare on the ground. 
Published images of a Starlink stack prior to launch show a lot of shiny 
aluminium panels.

Case 4: Arizona 4th May (Austin Godber)

Austin reported a number of Starlinks showing bright flares in sequence, 
with the Starlinks visible before and after the flares.

This one is different. Model-A predicts this pass should be totally 
invisible from Austin's location as the 'open book' panel would be 
pointed away from the observer due to high Sun azimuth. Model-B needs a 
roll-angle of 67 degrees (downwards) to get a flare at the right sky 
flare direction.

This pass was over southern California. I suggest that in accordance to 
recent information from Space-X, the roll-angle has to be changed to 
allow communications with the ground and so in case 4 the spacecraft 
baseplate is pointing nearly to the nadir and these flares are from the 
main baseplate, not the dish support panels (if that is what is flaring 
in Cases 1-3)

More work is needed to understand the deviations of observations from 
Model-B, predict some flares (and try to observe them) and apply the 
model to Starlinks at 550km.


Richard Cole

mob:  0771 858 8940
Seesat-l mailing list
Received on Mon May 11 2020 - 04:52:07 UTC

This archive was generated by hypermail 2.3.0 : Mon May 11 2020 - 09:52:07 UTC