[Help]: Inquiry Regarding Differences Between Cumulative and STL LOS Displacement

[ncbc-04]

Dear Alexey,

I am writing to seek clarification regarding the difference between cumulative LOS and STL LOS displacements in synthetic aperture radar interferometry (InSAR) data analysis.

As in the attached pictures, I got the results of the SBAS analysis for my area of interest for the year 2023, Could you kindly provide insights into the key distinctions between these methodologies and offer guidance on which of these two is appropriate to utilize/focus on for analyzing LOS displacement data?

[AlexeyPechnikov]

Cumulative LOS displacements is a result of least-squares solution on a set of interferogram phases. It can be affected by atmospheric effects, seasonal changes and more. Trend calculations are targeted to remove all the factors and detect a long-time trend and STL is the one of the best algorithms for that. For STL algorithm explanation and links see https://www.statsmodels.org/dev/examples/notebooks/generated/stl_decomposition.html

[ncbc-04]

Dear Alexey,

I’ve used your work extensively as a basis for my research, I have come across findings that indicate a significant level of subsidence using SBAS analysis, with measurements suggesting a displacement of approximately 50mm in the area under study for a period from 2017 to 2022 as shown in the figure below. However, I have come across another contribution within the field that presents a contrasting view, suggesting a different interpretation of the subsidence dynamics in the same area. For your reference, I have attached the relevant contribution to this comment. This difference has led me to question whether I might be approaching the analysis or interpretation of the data incorrectly. I am reaching out to you seeking your expert opinion and insights on this matter. Given the differences in findings between my research (SBAS analysis through PyGMTSAR) and this other contribution, I am eager to hear your expert insights and perspective on the potential factors contributing to these discrepancies. The study by[Kakar et al.] provided subsidence values ranging from 3.2 cm/year to 16 cm/year, while my research[SBAS using PyGMTSAR] yielded values of 5 cm/year. This difference has raised questions to inquire about the factors that might contribute to such variations in reported subsidence rates. Your expertise and guidance on on this matters would greatly assist me in gaining a clearer understanding of the subsidence dynamics and in ensuring the accuracy and reliability of my research outcomes.

remotesensing-16-01521 .pdf

[AlexeyPechnikov]

Hi, the red area around the blue on the plots above is a subsidence zone, according to the paper. If the red area has a subsidence of -40 mm/year instead of an uplift of +40 mm/year, then the center blue zone has a velocity of about 150 mm/year, which is comparable to the paper's results (15 cm/year). This means you selected too small an area for the processing, and your velocities are biased. InSAR velocities are relative and calculated from an average for your processing area. When the area, on average, is not stable, you have a bias. You should select the same area as in Figure 7 of the paper because it is stable on average and should not produce the bias.

[AlexeyPechnikov]

FYI, I've recently updated the velocity calculation function to use a least-squares fit (instead of the previous fast solution based on average displacement):

velocity_ps = sbas.velocity(disp_ps)
# optionally, materialize to disk and open
velocity_ps = sbas.sync_cube(velocity_ps, 'velocity_ps')

You can call it on displacements as well as on STL trend, but be careful as this trend is not constant.

[ncbc-04]

Hi,

I wanted to update you on my recent SBAS analysis. Following your insight from the paper to increase the area of interest, I expanded the scope accordingly. However, despite this adjustment, I found that the analysis produced the same results as before. I'm reaching out to seek your guidance on this matter.

Analysis with Initial AOI

Analysis with Expanded AOI

[AlexeyPechnikov]

Show your POI displacements plot with trends and RMSE estimation. And how do you calculate the velocity? Try the sbas.velocity function.

[ncbc-04]

Dear Alexey,

I have used the updated notebook for sbas.velocity calculation but i come up with the following error. Could you please provide some guidance or assistance in resolving this error?

[AlexeyPechnikov]

It seems you need to upgrade your PyGMTSAR library.

[ncbc-04]

Hello sir, the issue is still there even after upgrading the library and it is not showing the updated notebook. you can see below in the figure which show that its 23 days ago.

[AlexeyPechnikov]

The function is used in 'PyGMTSAR SBAS and PSI Analyses: Golden Valley, CA.' and 'PyGMTSAR SBAS and PSI Analyses: Lake Sarez Landslides, Tajikistan.' Google Colab notebooks.