Question about the NRLMSISE00 Model

I am working on code to perform long-term (5-10 days) high-fidelity propagation of satellites, and when I use the NRLMSISE00 atmospheric model I get excellent agreement with my known truth results (I ran 4 separate cases and repeatedly was off by only approx. 3-4 km over 6 days of propagation). However, I only get those results when I set some of the 24 switches to non-default settings:

  • 1 = 0 - F10.7 effect on mean
  • 9 = -1 - Daily AP. If you set this field to -1 , the function uses the entire matrix of magnetic index information (APH) instead of APH(:,1) .
  • 2 = 0 - Independent of time

While I’m glad to repeatedly get good results, I would like to understand why and I haven’t really been able to find more than the most basic description of what these switches control (descriptions for relevant switches included above). I understand why setting 9 to -1 would improve things because you’re using more detailed/accuarte AP information, and if I understand the description of 2 correctly setting it to zero means I am making my calcuations dependent on time of day, so I can also see why that would improve things. But my understanding of switch 1 is that I am turning off the daily variation in the F10.7 cm solar radio flux value and telling the solver to use an average daily value instead. My general assumption for numeric integrators is that if I use an average value instead of including the time variations, then the accuracy of my integration would decrease, but instead I repeatedly see the opposite. Can someone who understands the NRLMSISE00 model better than me explain to me if I made a mistake in what these values mean, or if I haven’t why turning the F10.7 flag off would improve the propagator results? Thank you.