Re: Fail in step 9.1 residual_RMS: ValueError: Axis limits cannot be NaN or Inf

[hike L]

Later I changed my DEM file covering the whole scene and rerun isce2. I had an another problem in correct_unwrap_error as follows:

Traceback (most recent call last):

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 1268, in <module>

    main(sys.argv[1:])

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 1250, in main

    app.run(steps=inps.runSteps)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 1042, in run

    self.run_unwrap_error_correction(sname)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 527, in run_unwrap_error_correction

    mintpy.unwrap_error_phase_closure.main(iargs_closure)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/unwrap_error_phase_closure.py", line 556, in main

    run_unwrap_error_phase_closure(inps.ifgram_file, common_regions,

  File "/public/home/zzusaa02/tools/MintPy/mintpy/unwrap_error_phase_closure.py", line 512, in run_unwrap_error_phase_closure

    cc_obj.label()

  File "/public/home/zzusaa02/tools/MintPy/mintpy/objects/conncomp.py", line 177, in label

    raise ValueError('input reference point is NOT included in the connectComponent.')

ValueError: input reference point is NOT included in the connectComponent.

Then I selected a reference point located in a bridge through Google Earth, but an error came out as follows:

Traceback (most recent call last):

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 1268, in <module>

    main(sys.argv[1:])

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 1250, in main

    app.run(steps=inps.runSteps)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 1045, in run

    self.run_network_inversion(sname)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 559, in run_network_inversion

    self.generate_temporal_coherence_mask()

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 614, in generate_temporal_coherence_mask

    raise RuntimeError(msg)

RuntimeError: Not enough reliable pixels (minimum of 100). Try the following:

1) Check the reference pixel and make sure it's not in areas with unwrapping errors

2) Check the network and make sure it's fully connected without subsets

I check my network and it was fully connected, but I fail to solve this problem. I skip this step and run the left. However, I come across the same error that I posted originally at the beginning. I find that  RMSs of all acquisitions are inf in rms_timeseriesResidual_ramp.txt. Here are some figures: .

Any advice on how to solve these problems would be appreciated.

Best regards,

Liang

在2021年10月4日星期一 UTC+8 下午10:44:07<hike L> 写道：

Hi everyone,

I have a  problem when runing step residual_RMS and the version of MintPy is 1.3.1. My platform is supercomputer and I installed isce2 by conda.

 Thanks advanced if any possible help.

 My cfg file is attached and the error message follows as:

/public/home/zzusaa02/miniconda3/envs/mintpy/lib/python3.8/site-packages/scipy/linalg/basic.py:1220: RuntimeWarning: overflow encountered in square

  resids = np.sum(np.abs(x[n:])**2, axis=0)

/public/home/zzusaa02/miniconda3/envs/mintpy/lib/python3.8/site-packages/numpy/core/fromnumeric.py:86: RuntimeWarning: overflow encountered in reduce

  return ufunc.reduce(obj, axis, dtype, out, **passkwargs)

/public/home/zzusaa02/tools/MintPy/mintpy/objects/stack.py:396: RuntimeWarning: overflow encountered in square

  self.rms[i] = np.sqrt(np.nanmean(np.square(data), axis=(0, 1)))

Traceback (most recent call last):

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 1268, in <module>

    main(sys.argv[1:])

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 1250, in main

    app.run(steps=inps.runSteps)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 1063, in run

    self.run_residual_phase_rms(sname)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/smallbaselineApp.py", line 866, in run_residual_phase_rms

    mintpy.timeseries_rms.main(iargs)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/timeseries_rms.py", line 228, in main

    analyze_rms(inps.date_list, inps.rms_list, inps)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/timeseries_rms.py", line 129, in analyze_rms

    ax = plot_rms_bar(ax, date_list, np.array(rms_list)*1000., cutoff=inps.cutoff)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/timeseries_rms.py", line 175, in plot_rms_bar

    ax = pp.auto_adjust_yaxis(ax, np.append(rms, rms_threshold), font_size, ymin=0.0)

  File "/public/home/zzusaa02/tools/MintPy/mintpy/utils/plot.py", line 464, in auto_adjust_yaxis

    ax.set_ylim([ymin, ymax])

  File "/public/home/zzusaa02/miniconda3/envs/mintpy/lib/python3.8/site-packages/matplotlib/axes/_base.py", line 3849, in set_ylim

    top = self._validate_converted_limits(top, self.convert_yunits)

  File "/public/home/zzusaa02/miniconda3/envs/mintpy/lib/python3.8/site-packages/matplotlib/axes/_base.py", line 3439, in _validate_converted_limits

    raise ValueError("Axis limits cannot be NaN or Inf")

ValueError: Axis limits cannot be NaN or Inf

Here is my command of isce2 :

stackSentinel.py -s /public/home/zzusaa02/S1 -d /public/home/zzusaa02/S1/DEM/demLat_N34_N36_Lon_E113_E114.dem.wgs84  -o /public/home/zzusaa02/S1/orbits  -a /public/home/zzusaa02/LCM/AUX -b '35.10 36.00 113.25 114.10' -c 3 -u snaphu -n '1' -f 0.9 -z 5 -r 20 --reference_date 20190103

[Zhang Yunjun]

Hi Liang,

This seems like a detailed issue related to the quality of the dataset. It's difficult to guess. Could you share the entire "pic" folder here for me to take a look at?

Regards,

Yunjun

[hike L]

Hi Yunjun,

Thank you very much for your reply. My study area is in a urban region and the dataset does look not good. My 'pic' folder that only include configuration files without any figure because of the failure of step 9, is attached below.

Best regards,

Liang

[Zhang Yunjun]

Hi Liang,

You could still run smallbaselineApp.py --plot to plot the existing files.

Yunjun

[hike L]

Hi Yunjun,

 

Following your advice, I got some figures in the zip file from Google drive. I hope it work. Please have a look. Thanks a lot.

 

Kind regards,

Liang

[Zhang Yunjun]

Hi Liang,

1. The phase closure correction result is not completed based on your pic folder content. You might want to either turn it off or re-run that step (by removing ifgramStack.h5 and re-run from the beginning of mintpy processing). 

2. Your DEM does not cover the entire area. You might want to re-run stack processing with a larger DEM to fill the lower part of your 2nd swath.

Regards,

Yunjun

[hike L]

Hi Yunjun,

Thanks for your patience. Following your advice, I re-run isce2 and mintpy processing with a lager DEM and less SAR images acquired in 2018. At this time, the phase closure correction is completed, but I find that RMSs of all images are Inf. Does it mean that the number of reliable pixel is 0 according to the RMS formula in your paper?

 Unluckily, I have the same problem as my post title mentioned. Could you please take a look at my "pic" folder(pic.zip)?

Best regards,

Liang

[Zhang Yunjun]

Hi Liang,

The google drive link you posted is restricted by access, so I can not see it.

There are detailed explanations in the jupyter notebooks for smallbaselineApp (https://github.com/insarlab/mintpy-tutorial), with recordings from previous years' UNAVCO short course. They are helpful to understand the figures in the pic folder, in case you have not gone over them.

Yunjun

[hike L]

Hi Yunjun,

Sorry for the google drive problem. I've learned a lot after watching the UNAVCO short course given in 2021. Then I dropped some interferograms with obvious errors and re-run the network inversion step but I still got the same problem with RMSs equalling Inf. I also found that the average velocity looked like a constant and the velocity map of ERA5 was empty (with tropospheric correction ending normally). I am confused by these results. 

Here is the new 'pic' folder files(pic.zip). Could you take a look and give me some tips? Thousands of thanks!

Best regards,

Liang  

[Zhang Yunjun]

Hi Liang,

Your stack of interferograms and the estimated time series look good. The issue happened during the ERA5 tropospheric correction. I would suggest you to:

1. make sure you use the latest pyaps3 on https://github.com/insarlab/PyAPS; and complete the testings there as described in the README to ensure it's working correctly.

2. make sure you use the latest mintpy on github; then delete "inputs/ERA5.h5" and "timeseries_SET_ERA5.h5" file and re-run smallbaselineApp.py.

It helps to read the printout message while running smallbaselineApp.py, [I know it's long, so do it ONCE at least] to get familiar with which files are generated at which steps. The steps in smallbaselineApp.py correspond to the workflow section in my 2019 paper. Thus, you will get much more comfortable if you could connect the files to the steps, and then to the descriptions in the paper.

Regards,

Yunjun