# Issue with propagating TLE sets using SGP4

I am trying to make some analysis on conjunction events using the TLEs available from Celestrak's SOCRATES page. Take for example the following predicted event:

Here's my issue: when I try to propagate the two TLEs of the objects (accessible by clicking on the button TLE Data) up to the time predicted for the closest approach (the field Days Since Epoch) I get different results from those predicted by SOCRATES. In particular, the relative speed comes out exact (equal to the value of Relative Velocity (km/sec)) whereas the relative distance comes out completely different from the value of Min Range (km). In particular, for the example in the figure I get a relative velocity of 14.9166 km/s (correct!) but a minimum range of 128.1554 km, completely off from the 0.040 km predicted by SOCRATES.

For the propagation I use the MATLAB code for SGP4 provided directly by Celestrak (link). When I first downloaded the code I quickly validated it by trying to propagate an example of TLE available from Vallado's book (Fundamentals of Astrodynamics and Applications, 4th ed.) and the results matched. There is something I am missing and I hope you can help me.

• Just for clarity, I calculate the relative distance as: norm(r_1 - r_2), where r_1 and r_2 are the computed position vectors at the time of closest approach. Apr 17, 2022 at 10:17
• To find the minimum, how often are you computing |r1-r2|? The first thing I'd check is the time resolution of your calculation. At 14.9 km/s, the distance 0.04 km represents only 0.0027 seconds, so to get that close in your calculation you have to try a bunch of times separated at the millisecond level. 128 km at that speed is 8.6 seconds, which sounds like you're checking the difference every 10 seconds, and your resolution needs to be much finer (not that a TLE should be trusted at sub-kilometer scales). Apr 17, 2022 at 18:23