Negative Drag Coefficient

Hello Orekit Community,

I have been using BLS Observer to get information about each iteration and one of the values it returns is drag coefficient. My initial drag coefficient is 2.2, but it keeps going down and returns negative values going towards the end.
This is a non-realistic value for drag coefficient. What is happening here?


The limits for drag and SRP intentionally allow negative coefficients
because sometimes other forces end up in the drag. This was for example the case in the W3B LEOP as an on-board leak did have the same signature as a negative pressure.
Do you have also another force like parametric acceleration acting at the same time? This could introduce observability problems and diverging estimated parameters.