I ran the site processing steps using a single star graph. Then, I changed the graph from the ‘select dataset’ module and start the site processing again. In the site processing module, only the “interferogram processing” button turns into yellow and other buttons remain green. However, I expect that after changing the graph the coherence generation and APS estimation should also be repeated (their corresponding buttons should also turn into yellow).
first of all, the concept behind: changing the graph does not correspond to a fully new analysis (like re-starting from the beginning). It’s just a different way to combine your data. Soon we will allow to change it also from within some modules to make it easier and more convenient.
If interferograms are missing it’s easy to detect, so, that is straightforward. For the spatial coherence, it’s less easy. We will anyway include a txt note just to keep track how a spatial coherence file was processed. But I don’t think it is so useful to re-process it everytime you change the images graph. It’s up to you.
(a brief note: I hope you realized that you do not need to create the STAR graph interferograms. In the APS and MISP modules the sw automatically generates the interferometric phase. Process interferograms only when you want to filter them)
The APS is always saved in STAR format: the APS is the atmospheric delay for each image (referred to the Master one). Also the estimated parameters (as height and velocity) are independent from the images graph.
Changing the images graph could help you to estimate better APS and parameters, and you will choose what you think is the best. But the buttons simply tell you if the APS has been estimated or not…
This is the current philosophy.
We take your notes into consideration, but we would need a broader discussion and hear many opinions before deciding to change the philosophy…
Thanks for the comments in any case!
For the coregistration the same applies:
the final aim is to have all images coregistered on the same grid.
If you do this by processing the coregistration via a STAR graph or via an other graph, the accuracy/precision may be different, but the final result will be the same: N images coregistered on one common grid.
By the way, up to this moment we did not fully implement yet the coregistration via a non-star graph. We only did some experiments.
In 98% of cases, it is not needed.
We have it in the todos list, with a low priority.