The issue is that sometimes the lines and pixels are not monotonous. This case is not handled by the Scipy function interpolate.RectBivariateSpline. So for these cases (caught with a ValueError exception), a more robust Scipy function is used: interpolate.interp2d.
The more robust function is only used for those cases as it is much more costly than the one already in use.
I have checked on the two S1 reference products (IW and EW) that the results from both functions are very close.
I have not checked why sometimes the input data is not monotonous.
Fix for #5
The issue is that sometimes the lines and pixels are not monotonous. This case is not handled by the Scipy function interpolate.RectBivariateSpline. So for these cases (caught with a ValueError exception), a more robust Scipy function is used: interpolate.interp2d.
The more robust function is only used for those cases as it is much more costly than the one already in use.
I have checked on the two S1 reference products (IW and EW) that the results from both functions are very close. I have not checked why sometimes the input data is not monotonous.