Visible ramp in the velocity map

[Sayyed Mohammad Javad Mirzadeh]

Hi all,

Using the HyP3 products and MintPy, I got the velocity map of a stack of Sentinel-1 over my study area (attached here) but when I checked the velocity map in more detail, I found that it suffers from a northeast-southwest ramp.

I tried to change the type of ramp from linear to quadratic and got new results. The velocity map is better but I am worried about using the quadratic ramp removal, which is uncommon in some cases.

I will be glad to see your comments.

Sincerely,
Sayyed

[Eric Fielding]

Hi Sayyed,

Did you apply a tropospheric correction to your time-series? Sometimes a ramp can appear if there was a strong tropospheric effect in a few of the dates, but performing the tropospheric correction can reduce this.

The question about whether you should use a linear or quadratic ramp depends on what kind of signal you are studying. If you are studying subsidence due to groundwater extraction, then you probably don't need to worry about what long-wavelength ramp has been estimated and removed. This is because the groundwater signal (which I assume is what is in the southwest corner of your figure) is limited to a small part of the area and not affected by the ramps much.

If you are looking at interseismic deformation, then you would have to be much more careful about the long spatial wavelengths and ramp removal.

All the best,

   ++Eric

[Sayyed Mohammad Javad Mirzadeh]

Hi Eric,

Thank you for your comments.

Yes, I applied a tropospheric correction on the time series in both cases using the ERA5 data and PyAPS package. Since the study area is Tehran City, surrounded by mountains, I should check the tropospheric signal again as you suggested.

But as you said, it is a case related to subsidence due to groundwater extraction. BTW, I think there is not a considerable difference between the results of ERA5 and GACOS-based tropospheric corrections. 

Sincerely,

Sayyed

[Sayyed Mohammad Javad Mirzadeh]

Hi Eric,

As you know, the MintPy will generate the last H5 file named "timeseries_ERA5_ramp_demErr.h5", after the steps of ramp removal, tropospheric and topographic corrections. When I checked the plots of this file and compared them with other cases I already studied, I found that they have different numbers of ramps in different epochs (in the attachment). I got confused that we should see the ramps of subsidence in the southwest of the area (the corner) not across the center to the northeast of the area! Even, when I applied the quadratic type of ramp removal, I can still see these ramps in these plots.

Do you know how it can be possible?!

Sincerely,

Sayyed

[Eric Fielding]

I don’t know what is happening, but that ramp is very large. Maybe the subsidence in the corner is confusing the ramp estimate. What does the time series look like before the ramp removal? 

[Sayyed Mohammad Javad Mirzadeh]

Before the ramp removal, it seems to be very normal as you can see in the attachment. You can compare before and after the ramp removal here.

Do you think it is better to extend the study area and have the subsidence frame in the central part to let ramp removal do another reaction?

It was an interesting challenge I see here :)

Sincerely,

Sayyed

[Zhang Yunjun]

Hi Sayyed,

It looks like ramp removal step went wrong. My guess is that your subsidence signal is strong in many pixels, which significantly biased the global ramp estimation. You could either use a custom mask file (using generate_mask.py --roipoly to generate one) to exclude the subsidence region while estimating the phase ramp via the mintpy.deramp.maskFile option; or processing a larger area so that your ROI is small relatively.

I hope it helps,

Yunjun

[Sayyed Mohammad Javad Mirzadeh]

Hi Yunjun,

Thank you for your point.

Exactly, as you and Eric said, I thought that the location of subsidence in the corner forced the ramp removal toward the subsidence and made it wrong. So, I decided to consider the whole frame, instead of cropping in the first step of MintPy, and let the program do it correctly. Then, for the post-processing and decomposition, I can crop it to my AOI.

Sincerely,

Sayyed

[Eric Fielding]

Sayyed,

You can also turn off the ramp estimation and removal if it is not working well for your problem.

All the best,

    ++Eric

[Sayyed Mohammad Javad Mirzadeh]

Hi Eric,

Thank you.

Yes, it can also be another solution, but will it not impose the overestimation in the subsidence rate in the zone of deformation?

Sincerely,

Sayyed

[Sayyed Mohammad Javad Mirzadeh]

Hi all,

An update is that I was successful in fixing this ramp issue by extending the area to a larger region and letting the program have more non-subsidence pixels for ramp estimation.

It was a good lesson that we should take care of cropping the input files into the MintPy for other following steps of processing.

Sincerely,

Sayyed