pyrocko
Quadnode phi is Nan
Despite the num.nanmedian function phi maybe nan if all the pixels in the quad are nan. Why the quad does exists then is still unclear as it should have been removed if the nanallowed threshold is crossed.
Must be a problem in stamps2kite, phi/theta values at the left image end are not always valid although there is s valid displacement. I guess somewhere in the interpolate_look_angles
Yes the LinearNDInterpolator returns Nans for values outside of the interpolation range- at the border of the gridded angles this is the case ...Consequently there are sometimes Nans in the phis and thetas although there are displacement values vor the respective pixels. How to fix ...?
Added warning for number of pixels that have no look angle info here: https://github.com/pyrocko/kite/commit/f3e68d1d0ff3cd37d8fbbfb746f1ea7fbf271caa
Tested that on two different scenes, one has 1236 pixels without that info, the other 0 ...