Using InSAR LOS data in the inversion

InSAR line of sight data (LOS) are like static offsets except you only know the motion along the line of sight vector connecting some point of the Earth's surface to the radar. So, unlike GPS offsets where you have north, east and up offsets, here you only have a single number.

QuadTree resampling

InSAR data are very densely sampled so in order to use it you must first subsample it. The most inteligent way to do this is to use QuadTree resampling. This algorithm subdividies the InSAR scene into smaller blocks depending on the variance of the signal. There's a matlab script scripts/quadtree_los.m I have provided (sorry now Python equivalent of qtdecomp exists yet) that shows how to do this. The only tunable parameter is thresh which determines how often to subdivide a region. A lower value produces more resampled points than a higher values.

Preparing .los files

The matlab script generates an output file with lon,lat,los and los vectors. You need to convert each individual resampled point into a "station" with its own .los file, the stations will be numbered consecutively and a two letter prefix added. The module insartools contains the function quadtree2mudpy use it as follows

from mudpy.insartools import quadtree2mudpy
quadtree_file='/Users/dmelgar/Nepal2015/insar/t048_C02.txt'
out_file=u'/Users/dmelgar/Slip_inv/Nepal/data/station_info/nepal_alos_t048_C02.sta'
prefix='AU'
quadtree2mudpy(home,project_name,quadtree_file,out_file,prefix)

This will generate the .los files in the data\statics folder of your project folder and a .sta file in the station_info folder which you can use to add to the .gflist file. You can now use this data like in any other inversion. The last control flag int he .gflist file controls whether the InSAR data is used so don;t forget to switch that to 1.