The format of alos_asc.por.dat
1st column: y location (km)
2nd column: x location (km)
3rd column: LOS looking direction in x (negative means reverse to the x axis direction)
4th column: LOS looking direction in y (positive means in the y axis direction)
5th column: LOS looking direction in vertical (positive up)
6th column: LOS velocity (mm/yr, positive means ground moving away from radar, it also means reverse to the LOS looking direction)
7th column: standard deviation of LOS velocity (mm/yr)
the position and the velocity vectors are all in the PoD coordinates.                        

Note: the Y location are shift by 850 km and the X location are shifted by 400 km after trans_
pole.



**************************
Hi Xiaopeng,

I put a file on scratch called sample.tar.  Inside there is a file called sample.grd.  This is a grid of number 1-7 based on the second invariant of the strain rate.  

1 - sample every point
2 - sample every other point
.
.
5 - sample every 5th point
.
.

You can look at the ps file to see that there will be a 1-sample along the creeping faults.

So here is is a way to use this.  Run blockmedian on all your LOS tiles so there will be 5 sets of tiles.

1- downsample to -I0.005/0.004 (this should be about 400 m resolution)
2- downsample to -I.01/.008  (800 m resolution)
3- downsample to -I.02/.016  (1600 m resolution)
4- downsample to -I.04/.032  (3200 m resolution)
5-7 downsample  to -I.08/.064 (6400 m resolution)

I'm guessing at the numbers but you get the idea.  We will be running Bridget's model at 500 m.

Then your script will read the sample.grd and select the appropriate subsampled LOS grid.

I think this will keep all the LOS information.

David