I have a question about changing a master image for an already coregistered dataset. The reason can be e.g. the prolongation of temporal baselines when updating a dataset, to make the master more “central”.
If I change it in the DATASET SELECTION window, there is no change in DATASET_INFO/DataSet.txt. Is it possible to save it somewhere so I do not forget to change it again next and next time when I get back to this dataset?
And, when can I find the actual baselines? Just in small area .mat file?
baselines are saved in the DataSet.txt file. However, it’s just an indicative number, since the baselines changes within the analyzed area. In the processing, such value is calculated from the orbital information.
If you use Matlab you can also load the baselines values using the command “fast”. They are stored together with other infos in the DataSet structure.
About changing master, we may think to save your settings for future use. However, it does not really make any difference to change the master image without re-coregistering your data. From the phase point of view, changing the master is like adding an offset, and when you estimate your parameters it does not have any impacts. Of course interferograms change, but you don’t filter the interferograms generated w.r.t the master, right? And for other interferograms the only impact (which may be significant) is given by the coregistration process.
So, here the key point is understanding when it is necessary to reprocess all images w.r.t a new master. Or otherwise finding a different strategy, that is gradually changing the master, processing only a subset of images, and connecting them with the old time series after processing.
Anyway, the matter is open, but changing the master without re-coregistering is not very useful.
Thank you. My aCcess to matlab is now more complicated, but i will try to solve it somehow.
I think it is sometimes useful to change the master. There are two reasons why I consider it:
– the temperatures. The master temperature should also be in the middle of the temperatures measured, especially in case where ktemp is neglected (forced to zero), in areas where small values of ktemp are expected, comparable to their accuracy. I think in such a case it is better to neglect the ktemp in processing, but some dilation residuals are expected and I think it is useful to let them be (close to) zero mean.
– accuracy evaluation. For standard deviation evaluation (according to Colesanti03) one needs both temporal and perpendicular baselines to have as high variance as possible, and if one starts with 30 images and updates the dataset progressively, the master is better to be changed due to the fact that it is for sure “out of center” in the time dimension.
InSAR is by definition a differential operation, so, constant terms have no impacts.
What counts in the temperature is its variation, not its average.
Also standard deviation and variance are independent from the mean value.