Discrepancy in velocity estimation between velocity.h5, timeseries_ERA5_demErr.h5, and tsview.py output

[DD-ZHAO]

Hello everyone,
After completing the MintPy time series processing, I understand that the velocities in velocity.h5 are obtained by performing a first-order least-squares linear fit on the deformation time series in timeseries_ERA5_demErr.h5. However, when I compare the results for a given pixel using three approaches: visualizing the time series with tsview.py timeseries_ERA5_demErr.h5, manually extracting the time series from timeseries_ERA5_demErr.h5 and performing my own least-squares fit, and directly reading the velocity from velocity.h5—I obtain noticeably different values. For example, tsview.py shows -5.83 cm/yr (as shown in the figure), my manual fit from timeseries_ERA5_demErr.h5 gives -5.48 cm/yr, and velocity.h5 reports -6.08 cm/yr. While these differences may seem small, they become significant when validating against GNSS velocities at the millimeter-per-year level. I also noticed that the time series displayed in the tsview.py command-line output differs from the time series extracted directly from timeseries_ERA5_demErr.h5. Could you please explain the reason for this discrepancy? Thank you for your help.

[ditafaith]

I also watch this issue to get a better understanding.

[DD-ZHAO]

Thank you for your interest. Let's look forward to a clearer explanation.

[scottstanie]

Could you post more about exactly how you got the three values, especially how you perform your own least squares fit?

[DD-ZHAO]

Thank you for your reply. The value of -5.83 cm/yr was obtained using the tsview.py timeseries_ERA5_demErr.h5 command as shown in the figure. The value of -6.08 cm/yr was extracted from the corresponding pixel in velocity.h5. The value of -5.48 cm/yr was obtained by performing a least squares linear fit (y = Ax + B), where the decimal year was used as the x-variable and the deformation time series in timeseries_ERA5_demErr.h5 was used as the y-variable; the coefficient A represents the estimated rate.