I like more this latest results, as now are not ramps there anymore. You could see what you have removed using the command ps_plot(‘o’,1) .
Answering to your question, well, you could try to use TRAIN, the toolbox for atmospheric noise removal developed by David Bekaert, and there you can use weather models, and other data to reduce the APS of the scene.
In fact, the trends you have removed using the ‘o’ option, other people say that are tropospheric ramps, and not orbit ramps, so you could also use this toolbox to help you removing noise.
Additionally, you could use GPS data or similar to reduce bias and fit your output to that, but it is something more difficult that one forum thread can handle.
Still, I invite you to look for this techniques in literature and see what you could think to use/apply.
Another option is to use the step 8 integrated in StaMPS, which does a spatio-temporal filtering to estimate APS signal. Then, for removing it the option is ‘s’ (from slaves) and should be used as follows: ps_plot(‘u-dmos’,1) but I cannot recall if this option is not supported. Just check it out