When using the AngularAzElBuilder for azimuth and elevation I noticed that the input is azimuth standard deviation. Azimuth uncertainty is a direct function of elevation, and usually a standard deviation of cross elevation is utilized (see section III.A for reference: https://ipnpr.jpl.nasa.gov/progress_report/42-88/88F.PDF). Is this azimuth uncertainty considered to be constant through elevation or is this already accounted for?
We don’t model anything special. There are two sigma values as we pass a two-elements array. The first element in the array corresponds tor azimuth and the second element corresponds to elevation.
These values are just copied as is in the generated measurements, so all measurements will share the same values, regardless of the current elevation.
If we want to set up a model with a dependency between elevation value and azimuth uncertainty, we should set up an interface that computes the uncertainties and let the user pass a custom implementation of this interface to be used in the build method.
Contributions welcome!
Hi,
Actually I’ve been thinking about adding a sort of covariance model provider to Position and PV builders. The same thing would benefit AzEl and RaDec as indeed their noise is correlated in reality.
Cheers,
Romain.