Issue: Air Drag possibly inconsistent and tolerance problems

WILL EDIT ISSUE LATER Comparison of poliastro Cowell, Python sgp4 and TLE from Tiangong 1.

What can be seen is that for a torelance of 1e-5 the orbit decays, but for 1e-7 the orbit does not. Of course one needs to take into account the less accurate SGP4 than Cowell, however, based on the TLE’s Cowell seems to be inaccurate.

Relative Tolerance = 1e-5

Relative Tolerance = 1e-7

🐞 Problem

The relative Cowell tolerance has an influence on the outcome of the propagation. SGP4 is known to be inconsistent for long propagation times, however Cowell/Dopri853 should be rather accurate/similar to TLE data.

🖥 Please paste the output of following commands

• conda info -a (only if you have conda)
• conda list (only if you have conda)
• pip freeze

# Paste your output here:

• python -c "import poliastro.testing; poliastro.testing.test()"

# Paste your output here:

🎯 Goal

💡 Possible solutions

📋 Steps to solve the problem

• Comment below about what you’ve started working on.
• Add, commit, push your changes
• Submit a pull request and add this in comments - Addresses #<put issue number here>
• Ask for a review in comments section of pull request
• Celebrate your contribution to this project 🎉

[jorgepiloto]

The [link]http://www.braeunig.us/space/atmmodel.htm) that @WolfsSky shared is going to be really useful to fully implement pressure and density in COESA76 up to 1000km. After implementing locally the equation they use to compute previous state variables it seems to work fine. Pull request is coming soon.

On the other hand, they also include the martian atmosphere, really cool 😍

[applysomesky]

Finding 1 What I find the most peculiar is the too long Linear decay, compared to TLE and SGP4. The atmospheric model uses a scale height of 8 km with a rho_0=1.3 kg/m^3. This is true for an atmosphere till 100 km, but above that it is enormous erroneous compared to for example the data found in Space Mission Analysis and Design by Wertz for higher altitudes: For example, at 300 km the current code will find a density of 6.73e-17 kg/m^3 while Wertz gives an average value of 1.95e-11. poliastro thus computes a factor 1e-6 lower atmospheric drag.

Finding 2 There is something going on with the rtol value compared to the stepsizes of the time of flights (tofs). Same propagations with different rtols lead to significant different results, to the point where for rtol=1e-7 no decay happens, where in TLE data the satellite has already re-entered. Reducing the rtol in my situation to 1e-5 lets the satellite re-enter more close to the TLE’s.

It seems a high rtol makes the solution too stiff? Although I would expect the propagator to make smaller steps to account for this. Maybe a minimal step size in the settings lets the propagator use that with the allowed rtol, and thus non-accurate results? (as in rtol is adhered to, but a minimal step size is used for that, which should actually be lower than that minimal stepsize) Not sure how this propagator works exactly.

I think this normally is not an issue, but decaying satellites change their orbit significantly.

Also I had to implement an own termination condition for below 250km attitude, otherwise the propagator gets stuck.