[Help]: standard deviation of the los velocity

[Smitrgeo15]

Hi Alexey, how to get the standard deviation of the los velocity from PyGMTSAR?

[AlexeyPechnikov]

Do you mean spatial distribution? If so, you can calculate it like this: da.std(['y', 'x']), where da is your xarray DataArray.

[Smitrgeo15]

Along with the mean los velocity, we also obtain the standard deviation of the velocity for every single pixel. I want that values for every pixels and also if It is converted to lat long.
Although I hope once I get the stand deviation of the velocity using sbas.ra2ll it will help me to convert that into latitude, longitude. Am I right ?

[AlexeyPechnikov]

Since trend velocity per pixel is a constant, its deviation is always zero. Perhaps you meant the deviation of displacements instead? That can be calculated using disp.std('date'), where disp is your data array. However, this approach might not be entirely insightful, as the standard deviation isn’t necessarily zero even when displacements precisely match the velocity trend. A better measure might be (disp - vel).std('date'), which calculates the standard deviation of the differences between the displacements and the modeled velocity over time.

[Smitrgeo15]

i was talking about this standard deviation which i have highlighted. Below i have attached the whole paper.
remotesensing-12-01305.pdf

[AlexeyPechnikov]

The common way is to estimate RMSE error as shown in the 'PyGMTSAR SBAS and PSI Analyses: Lake Sarez Landslides, Tajikistan' notebook:

rmse_ps = sbas.rmse(disp_ps_pairs, disp_ps, corr_ps)
rmse_ps

Pay attention, correlation-aware RMSE is more accurate while you can omit the correlation:

rmse_ps = sbas.rmse(disp_ps_pairs, disp_ps)
rmse_ps

[Smitrgeo15]

I want it for SBAS method, how can i do it?

[AlexeyPechnikov]

Simply replace PSI phases and displacements with SBAS ones.

[Smitrgeo15]

how to get disp_sbas_pairs?