The MH17 discussion has become extremely messy. It is a highly politicized topic, with active disinformation campaigns and trolling from both the Russian and Ukrainian sides. It is sometimes hard to discern who is honestly seeking the truth, who is honestly seeking the truth but very naive with regard to sources, and who is actively involved in peddling propaganda and disinformation.
Ever since
my expert participation in the Dutch Parliament committee hearing of Jan 22, and ever since I expressed caution on this blog about certain satellite images and factually clearly incorrect statements on satellite positions released by the Russian dept. of Defense, some of the trolling has been directed at me.
I have learned not to bother too much with trolls, but when they actively and tenaciously disseminate
disinformation and seriously flawed counter-arguments around one of my analysis, I reserve myself the right to a
rebuttal, just to set the record straight.
In a
previous post I showed that a satellite image released by Russia, purported to be a Resurs P1 image from 17 July 2014 which claimed to show Ukrainian BUK's in a field near Zaroshchens’ke, is problematic. The viewing angles of the "BUK's" in the image do not appear to fit the satellite-to-location geometry, which only allows clearly oblique viewing angles between 45 and 57.5 degrees with the horizontal, from directions ranging from northwest via west to southwest (for more details,
see my earlier post in question). I therefore urged caution with regard to these images.
My post has drawn fire on twitter, notably from a twitter user nicked "
Masami Kuramoto". While the nick sounds Japanese, and the twitter account claims to be located in Germany, I have a strong suspicion that the entity behind it is Russian.
Kuramoto's chosen line of attack is by questioning the accuracy of the Resurs P1 orbital information which I used. That orbital information came straight from
JSpOC (formerly known as "NORAD"), argueably this world's most reliable source of orbital elements. Kuramoto basically tries to advance a claim that the JSpOC tle's for Resurs P1 are either highly inaccurate or even deliberately doctored, and that the satellite in reality passed along a somewhat different trajectory (but with a similar pass time, to match the time listed in the images), thus advocating for the existence of an imaginary trajectory that allows to reconcile the imaging angles with the published images. In order for this to be possible, it is necessary to argue that the real orbit amounts to a significantly shifted orbital plane, i.e. a shifted
RAAN value, compared to the JSpOC published orbit.
In an attempt to argue this position, Kuramoto suggestively quoted from the Space-Track terms of use, taking text out of context to insinuate that JSpOC tle's were not accurate for the task:
Kuramoto kept insisting on his perceived "unreliability"of JSpOC tle's, even after I had set him straight on this:
Let me first elaborate on what I pointed out in the tweets above. Kuramoto tried to capitalize on this warning in the Space-Track User agreement:
"A TLE AVAILABLE TO THE PUBLIC SHOULD NOT BE USED FOR CONJUNCTION ASSESSMENT PREDICTION"
This statement is relevant to close encounters (with the risk of collision) of two objects in space. What this statement simply means is that it is unwise to base decisions on debris avoidance manoeuvres solely on published tle's. In such cases, very small uncertainties matter. If one uses tle's produced for the epoch of today to make a prediction on a future position of two objects (say: 3 days from now), that prediction for a moment days from now will have a small uncertainty. SGP 4 after all is only a model.
These uncertainties are negligible for other purposes, but for close encounter mitigation they matter. Satellites in Low Earth Orbit move some 7 km/s, so a 0.1 second uncertainty in the time of passing a particular point in orbit, hardly something to bother about under normal circumstances, amounts to a positional uncertainty along the orbit of 700 meter. This might not seem much and for other purposes 0.1 seconds and 700 meter is negligible, but for collision avoidance it matters: it might be the difference between a miss or a hit, certainly because the other object introduces a similar uncertainty (i.e., if both objects have 0.1 second uncertainty, the uncertainty in relative distance is 2 x 700 meter = 1.4 km. So if your analysis says they will safely pass 1.4 km apart, they might in reality collide instead. Or conversely, if your analysis says they will collide, the reality might be that they pass each other at a km distance rather than colliding).
In other words: the warning by JSpOC
is only relevant to a very specific situation, and concerns uncertainties that are
completely negligible for the subject at hand: the position of a satellite with respect to the viewing geometry of a location on earth. The more so because the latter assessment actually uses a tle with epoch very close to the the time of interest, unlike a collision avoidance assessment of a moment more removed in future. The uncertainty pointed out,
in no way can change the viewing angles to the extend that it would solve the discrepancies I pointed out in my earlier post.
Now, this could have been a simple misunderstanding, based on a lack of knowledge and insight in the matter on Kuramoto's side.
Kuramoto however next took it to a new level and suggested that JSpOC might have deliberately altered the orbital elements for Resurs P1 post-fact:
The point is: if JSpOC would have done that, the simple reality is that many people working with these data would notice it. Satellites suddenly would be at different positions than where the JSpOC orbital data would put them. Our tracking network for example, frequently catches Russian satellites as byproduct of our tracking of classified objects. On these occasions we would suddenly note large positional errors in that case, and we would even start to see UNIDS (unidentified satellites, which always have our immediate attention) that next turn out to be Russian satellites in orbits not matching their JSpOC orbit. No way that would go unnoticed.
As for the suggestion that the elements were only retrospectively altered, Kuramoto was a bit shocked to learn next that several of us (including me) actually regularly archive the full JSpOC database of orbital elements. I do so several times each month (for July 2014, I for example have archived elements from July 14 and can compare these with elements for that date retrieved from the JSpOC archive today: they are the same, they have not been restrospectively altered). In a retrospective analysis, altering the elements starting at some given date (or only altering them around a given date) would show up as a sudden change in the elements as well.So no: such a plot is simply not realistic.
After this, Kuramoto nevertheless still wished to cast doubt on the JSpOC tle's:
Notwithstanding my earlier rebuttal, he at first simply restated his already rebutted argument:
That would not do of course, and Kuramoto seems to have realized that. In order to maintain his position, Kuramoto had to grasp the next straw. He next brought up a paper by
Kelso et al.:
This paper discusses what factors might introduce predictions that do deviate considerably from reality (with the focus again on the accuracy of data needed for orbital debris avoidance manoeuvres). One such case is when for example a position is based on a tle issued 4 hours ago, but the satellite in question meanwhile has actively manoeuvered to a new orbit. In that case, the predicted position indeed would be incorrect. Kuramoto (of course) tries to seize on that, but in doing so again shows a lack of insight in the matter. Whether a satellite (Resurs P1 in this case) had just manoeuvered can easily be checked: by looking at a series of tle's issued around the time of interest (17 July 2014, 8:32 UT in this case), a manoeuvre around the time of interest would be visible by a sudden change in elements.
For Resurs P1 around 17 july 2014, I did this check (Kuramoto obviously didn't). There is no such change, i.e. the satellite did not make a significant manoeuvre. This can be seen in the diagrams below which depict the evolution of the orbit (from JSPOC data over July 2014). A manoeuvre would show up as a clear discontinuity (a clear sudden change) in either perigee and/or apogee altitude, argument of perigee, inclination, Mean Motion and /or RAAN (and notably in RAAN for Kuramoto's argument to hold). Tampering with the orbital elements around 17 July by JSpOC would show up similarly, by the way. But none of this happens, as you can see below. So again, Kuramoto's next grasp at a straw, is futile, and by now his attempts to argue my analysis away are bordering the pathetic.
post-edit 24 Feb 2014, 15:05 UT:
Kuramoto is still trying to advance his ill-fated argument:
Again, his argument is largely irrelevant. The kind of deviations pointed out are very minor: a maximum error of 9.3 km in position at a given time really will not significantly change the viewing angles. That would need cross-track errors an order of a magnitude larger.
post-edit 24 Feb 2014, 15:40 UT:
Well now, somebody has seen the light it appears:
