search

yumorishita / LiCSBAS2

GNU General Public License v3.0
27 stars 5 forks source link

LiCSBAS_cum2vel.py #37

Closed HamedAmirkhani1378 closed 10 months ago

HamedAmirkhani1378 commented 1 year ago

Hello When I use LiCSBAS_cum2vel.py and put the start and end dates of the whole network, the results should be the same as results/vel.filt.mskd and results/vstd? To me they are not exactly the same and slightly different.

HamedAmirkhani1378 commented 1 year ago

Also, if I set a short interval (for example 12 days), the vstd becomes very large. So we should not use it for short intervals?

yumorishita commented 1 year ago

I am not sure if the vel should be exactly the same because np.linalg.lstsq might include some random element. The vstd should not be exactly the same because a random element (i.e., bootstrap) is included in the calculation of vstd.

Also, if I set a short interval (for example 12 days), the vstd becomes very large.

It is natural that the vstd gets large within a short period.

HamedAmirkhani1378 commented 10 months ago

Hello

I compared LiCSBAS_cum2vel.py results from LiCSBAS and LiCSBAS2 and the results are very different. Also I think LiCSBAS has more accurate results than LiCSBAS2.

Is it possible for this to happen in the calculation of velocity and displacement in step 13?

yumorishita commented 10 months ago

The results should be the same between LiCSBAS and LiCSBAS2 because I did not change any algorithms as far as I remember.

HamedAmirkhani1378 commented 10 months ago

That's right. But great if you can check it out.