Documentation on the bachelor thesis: AN ASSESMENT OF SENTINEL-1 RADAR IMAGERY TIME-SERIES ANAYLIS FOR MONITORING SEASONAL SURFACE SUBSIDENCE DUE TO PERMAFROST DEGRADATION AFTER A TUNDRA FIRE.
This Project is part of the Course AB_1096 of the Free University Amsterdam departement Earth and Climate. Based on Procesing of Sentinel-1 data of 2016 and 2018 the project tries to describe seasonal change in permafrost degradation caused by wildfire induced dynamics. The processing is done using ESA’s SNAP [Veci etal., 2014] and the software StaMPS by the Universitoy of Leeds [Hooper et al., 2007].
You are watching: Search code, repositories, users, issues, pull requests…
Below instructions will get a copy of the software and data working on your machine to test and contribute to above project.
The processing Chain is dependant on
Linux is required for all processing after Downloading the Data and preprocessing using Snap and Snap2StamPS, compiling with cygwin on windows has not been succesfull for almost all users. for this project you can use a Debian linux distribution like Ubuntu 14.04
Downloading and preprocessing:
All commands can be run in the basic terminal directory except for the make and make install.
Install Matlab This step requires a full matlab license Download unix installer for Matlab from: https://nl.mathworks.com/downloads/ note: if the installer asks if you want to change path names to keep libraries from conflicting choose yes!
Install SNAP (not necessary if preprocessing is not done on the linux machine) Download Unix installer from: http://step.esa.int/main/download/
Install more dependencies
In this chapter we will discuss where to download Sentinel-1 data for the PS-InSAR processing. You can use either alaska satellite facility or the copernicus hub as platform to download the data but this instruction will be for the Alaska sattelite facility. for the copernicus hub you could use: https://forum.step.esa.int/t/python-data-downloader/14308
Surf to https://vertex.daac.asf.alaska.edu/#
After searching data that meets the requirements add all the data files to the download que. In the download que window choose download python script and enter your earthdata login credentials. Move the download python script to the folder where you want the data to be downloaded to. Open a terminal in the directory the python download file is located. Execute the command:
Example:
All Data will now be downloaded and if already downloaded controlled for corruption and redownloaded or skipped.
In this chapter the preprocessing using snap2stamps and SNAP will be covered. The Preprocessing can be done on both Windows and linux, this documentation is for use on a Windows machine. Snap2stamps are python wrappers for using the S1tbx of SNAP for interferometric preprocessing. To be able to use the snap2stamps package a working python 2.7 installation is required: check chapter installation
Before we can use snap2stamps the directory has to be set up. Snap2stamps expects the data to be sorted in 2 seperate folders: master and slaves. In the master directory the preprocessed master and original data for the master should be located. In the slaves folder all unprocessed and unzipped slave data files should be located
Step 1 – Master selection: Open all images in SNAP desktop Navigate to radar – interferometric – INSAR stack overview. Open all data in the window and click overview this will tell you which image is ideal to use as a master based on the temporal and baseline characteristics of the SLC products.
Step 2 – Master preprocessing: Use radar – interferometric – tops – tops split on the master to select the necessary subswath and bursts. Use radar – apply orbit file on the output of the last step (be sure to use precise orbits).
Step 3 – Setup project.conf file: Open the config file located in the snap2stamps folder with a text editor. Change all parameters accordingly to below example.
Note: The importance of the correct file structure The / after the paths or not The IW1= parameter needs for example IW2 – not 2! The complete path to the master.dim!
Move the config file(can have every name you want) to the Snap2stamps/bin folder and open a terminal in this location(tympe cmd in search bar).
In terminal type:
Read more : Wedding Invitation Postage: How to Mail Your Invites
For example:
This script will put the slaves in the right file structure to be processed by SNAP
After step 4 also execute the other 3 scripts in this order: splitting_slaves.py, coreg_ifg_topsar.py, stamps_export.py
To use an External Dem change all the mentions of DEMname (for example ASTRM) into External dem like this:
and change the path to the external dem to the path where the dem (in GeoTiff and wgs84 unprojected) is located. like this:
Another option could be changing the estimate coherence from False to True like this:
Or oversampling the signal by changing the coregistration parameters like this:
StaMPS is a software package developed for linux bash and the linux matlab environment, in order to complete all the steps we need todo both the preperation step in the terminal as the processing steps in matlab.
To start matlab processing we first have to setup an environment using the config.bash file that is delivered with the STaMPS installation. Navigate to your StaMPS folder and open the config.bash file in a text editor, file is located here:
Change all paths in this file to the correct paths to the software bin, if you can’t find a path use the whereis command to locate the install directories like this:
change the paths in the file like this:
And then upon opening the terminal add all paths to your ENV by sourcing the config file:
If you intend to use the TRAIN toolbox for atmoshperic correction aswelll you will have to source that config file aswell. Navigate to your TRAIN folder, open the file in a text editor.
Edit the file like this:
And source the file in the same terminal as StAMPS
Upon having setup the environment we can start processing, remember to source the config.bash file everytime a new terminal for processing is opened. Open the folder with the stamps_export data:
Then do the first processing step using this command:
for example:
This will start mt_prep_Snap with an amplitude dispersion treshold of 0.4 (between 0.4 and 0.42 is normal with 0.4 having less PS), 6 patches of 3 in azimuth and 2 in range and the patches overlap with 50 and 200m each. If processing in 1 patch is possible without memory issues this is recommended.
Carefully read to the log of mt_prep_snap upon finishing and look whether the mean_amplitude is never 0 and there are not to much ps points without amplitude, this will throw errors later in stamps processing. When done reading dont close the terminal but type:
This will open a matlab window in the right directory to do the further StaMPS processing.
In the opened matlab window all further processing will be done, by simply typing stamps(1,8)all steps of Stamps will be executed. Processing with all steps seperate may be better for parameterization though, in order to run 1 step or a certain ammount of steps type:
for step 2:
for step 3 and 4:
Read more : Best Buy Pre-Order Policy (All You Need To Know)
Except for running the program with default parameters we would like to be able to change parameter values. To get all parameters and their values and then change them:
Another usefull command is the information about interferograms and their temporal and baseline decorrelation
Converts the data into the formats required for PS processing and stores them in matlab workspaces. This step has no important parameters to change
Step 2 estimates the phase noise for every pixel in the interferograms via an iterative step. There are alot of parameters controlling this step but barely any documentation on how to use them correctly for noise reduction. Some sources note that changing the filter_grid_size , clap_alpha and gamma_max_iterations could make the filtering stronger and reduce the overall noise.
By buffering and using threshold values this step decides on letting PS with random phase fall out of the total PS. By lowering the possible density of these random phase pixels using the density/percent_rand parameter less noisy pixels will be kept.
at step 4 other bad PS pixels(too much noise or ground contributions) will be weeded from the PS pixels left from step 3. The parameters are easy to grasp so some adjustements can be easily made here. Reducing the weed_Standard_dev can easily remove the worse pixels from the group, any value between 0.65 and 1 is quite realistic.
If you have chosen to process in multiple patches from mt_prep_snap the patches will be merged here for step 6. It is also an option to resample the PS points to a lower resolution for noise reduction. Keep in mind that making merge_resample_size higher can lead to undersampling of the deformation that is being detected so should be used carefully.
Phase unwrapping creates the displacement values for the final results based on a stochastic process over all interferograms. Step 6 is really important and prone to error so reprocessing this step multiple times is necessary to get good results. The important parameters for step 6 are:
unwrap_prefilter_flag The recommendation for this parameter is to keep it on ‘y’ because this will extract the errors calculated at step 7 and adds them in after unwrapping to improve the accuracy of the unwrapping. If you want to process 1 run without the pre_filter you can use scla_reset to clear the values from step 7.
unwrap_grid_size This parameter determines to what grid the PS pixels will be resampled for the unwrapping. Higher values will reduce noise from unwrapping but also can undersample your deformation, use this parameter for noise reduction but with caution.
unwrap_gold_n_win , unwrap_time_win and unwrap_gold_alpha these parameters control the Goldstein filter that is used on the grid before unwrapping takes place and reduces noise on the fringes. Highering the alpha value will generally set the filter to filter more and can therefore undersample your deformation, the window and time size of the filter should be played with to obtain a good result. unwrap_gold_n_win can cause problems when not enough pixels fall inside of the window, it has to be highered than.
drop_ifg_index This parameter makes sure the interferograms listed (by number as shown in ps_info) are not used in the processing, if a interferogram is added the processing has to be rerun from step 3. Use this parameter if after multiple times this interferogram can still not be reliabely unwrapped.
Spatially-uncorrelated look angle (SULA) error was calculated in Step 3 and removed in Step 5. InStep 7, spatially-correlated look angle (SCLA) error is calculated which is due almost exclusivelyto spatially-correlated DEM error (this includes error in the DEM itself, and incorrect mapping ofthe DEM into radar co-ordinates). Master atmosphere and orbit error (AOE) phase is estimated simultaneously.
The errors calculated in step 7 are importtant to improve the unwrapping results so the final result reaches its optimum state.
The most important parameters for step 7 are scla_drop_index and scla_deramp. If a result is not reliabely unwrapped and you do not want to impact it the result of step 7 you can add the interferogram (number in ps_info) to the scla_drop_index and step 7 will skip this interferogram when calculating. Using scla_deramp may improve your results when high phase ramps are distorting your results, by setting parameter to ‘y’ it will calculate an extra error to be removed.
step 8 filters the results specifically for atmospheric disturbance, play with the parameters scn_wavelength and scn_time_win to see what works for your AOI.
keep doing this process until results are satisfying and command: ps_plot(‘v-do’. ‘ts’) to check if the total velocities are okay, if result is not satisfying try:
set your reference location for plotting using:
if all results are satisfying make a kml file with all data for visualization:
or make a csv file with all data vor visualizations:
if you want to use TRAIN check the train manual: http://davidbekaert.com/download/TRAIN_manual.pdf but using linear correction try this:
To convert the corrected LOS displacement from StaMPS into combined ascending and descending vertical displacement I have created a few python scripts to ease this process. The scripts can be downloaded from this git and meed to be adjusted to your folder system.
first script to run is vertical_displacement_calculator.py second script to run is asc_desc_combiner.py
Visualization of the data in this project is done using QGIS and Python. All scripts used can be found in this Git and shapefiles used can be requested by contacting me directly. Scripts will have to be adjusted to personal file structure and projections to be usable.
Source: https://antiquewolrd.com
Categories: Stamps
When it comes to buying postage online, the benefits are pretty straightforward: you save time,…
One of the most prominent hobbies in the world is collecting stamps, otherwise known as…
There are various types and options of rubber stamps, so it is quite natural to…
First-Class Mail Fact Sheet What is First-Class Mail Shape-Based Pricing? First-Class Mail shape-based pricing aligns…
When you’re sending someone a letter or a card, all you need is one Forever…
USP <51> is used to test preservative effectiveness. The number “<51>” refers to General Chapter…