Hi,
You should specify the satellite angles that apply to the observation you are simulating. If you carry out a nadir simulation, the simulated radiances will be quite different in general to the actual observations. I’m not an expert on satellite observation datasets, but I think it is common that they include the satellite zenith and azimuth angles corresponding to the observed radiances.
If you are carrying out experiments/investigations purely involving simulated radiances (no real obs), then you can set the angles to zero. But just be aware that they won’t be representative all observations across the swath of the sensor.
Best wishes,
James