Measurement model simulating more observed cases than total cases

[erinclancey]

I am using a negative binomial measurement model to simulate observed case reports from an SEIR process model. My accumulator variable, H, tallies incidence over the course of a week. When I simulate the model I often get more case reports in a given week than the number of actual cases accumulating in H. I am simulating a small population. Does this issue come up with other models? What does pomp recommend when this happens?

[kingaa]

The negative binomial has support on the whole set of non-negative integers, so this is expected behavior. What were you expecting?