Announcements

• Andria's UNAVCO one-pager, Denali high-rate GPS and sidereal/spatial filtering:
  text
  figure
  

• If your file contains 
  
  X   Y   Z
  Xd Yd Zd
  X   Y   Z
  Xd Yd Zd
  and so on...
  
  For odd number lines: awk '// {n += 1} {k = n%2} {if (k==1) {print $1,$2,$3}}' filename > odd_file
  For even number lines: awk '// {n += 1} {k = n%2} {if (k==0) {print $1}}' filename > even_file
  and then you can use "paste" to combine them together like "X Y Z Xd Yd Zd".
  

  I have /gipsy/bin/grace_odd.awk and /gipsy/bin/grace_even.awk that will grep out the times & positions and the velocities, respectively. you still have to paste them together, although if reading into matlab, no need. Just read the two files separately. You need to kill the header line though.

• Chairs:
  a company called Kare Products, here in Boulder 1-800-927-5273 www.kareproducts.com

  There's also a campus ergonomics office that has all the chairs available for you to try out, figure out which one is right for you. I think if you go consult with them first, the chair is 50% off. I think this is the guy we talked to: Marshall, Peter, Ergo Spec, ARCE A451 ... 2-6758

• Tip: make sure you APNAMS are listed in numerical order and don't skip a number.  If, just for example, you have 40 variables in your APNAMS list but fail to use APNAM(23), the filter will stop reading the information after APNAM(22).
• For runs longer than one day, solvekm only checks the /usr/local/gipsy/bin/bad.satellites file for the first day of the run.  I am using a modified version of solvekm.l, called solvekm.jp, which takes as a14th input the number of days in the run.  If you leave it blank it defaults to 1 of course.  This value is used to look at bad.satellites for all of the days in the run, any satellites it finds will be deleted from the temporary qm file in the /reg directory.  So instead of one day it can check for up to 5 days.  The deleting script is called multidaydel_qm and is in /gipsy/bin.
• TROPOSPHERE units, and how to use WETZTROP estimates for A PRIORI information: Very tricky.

• You can feed a priori data into your filter via the $APRIORI namelist
• How to set an a priori value in the filter namelist:  in the $APRIORI section of the filter, we normally assign a priori names(APNAMS) and a priori sigmas (APSIGS).  You can also assign APMNOM, which stands for A Priori Minus NOMinal.  If nominal values are assigned, they are in the qregres namelist*... but most things are not in that namelist, and thus the nominal value is zero.  APMNOM is therefore the same as the a priori value.   I have been investigating this some; if you pin the value with a small sigma, it will come out at the APMNOM value for your run.
•  The time associated with the APMNOM value is the first time in your data.  If you are doing a typical 24 hour run, the APMNOM value is assigned at 00:00.  However, you can assign the time of the a priori data as well with APTIME, but I haven't tried using that yet.

                       Here is an example from the $APRIORI filter namelist.

           APNAMS(1) =   'STABIAS AMC2',
           APSIGS( 1) = 3.172D-5,   APMNOM(1) = 3.994614254274801D-02

            Note: do not confuse this with the $APRIORINML for the smapper.  That remains empty.

 *Wet and dry troposphere nominal values are in the qregres namelist (but they are in meters and the filter namelist is in km)...

• Troposphere mapping:  default mapping for qregres namelists is 'LANYI' but apparently 'NIELL' is a better map.  The gipsy source code set_qr_tropmap is supposed to be used to change the mapping strategy, but it doesn't work due to the "replace" command.   I have fixed it, called the fix set_qr_tropmap.jp, and placed it in /gipsy/bin/.  Here is how to use it to modify a qregres namelist called 02sep23.qnml:

             set_qr_tropmap.jp  -i 02sep23.qnml  -m 'NIELL'
 
• Andria thinks these are useful sites:

http://www.3quarks.com/Backgrounds/
http://wp.netscape.com/assist/net_sites/bg/backgrounds.html
http://www.grsites.com/textures/
• From Andria: another matlab function, "intersect", should be easier to use than "ismember" for forming double differences. Here's an example, where A and B are time vectors with some times in common; the values you want to plot (y-axis) are Ay and By

[c, ia, ib] = intersect(A, B);
        c = common times
        ia = indicies of A
        ib = indicies of B
difference = Ay(ia)-By(ib);
plot(A(ia), Ay(ia), B(ib), By(ib))

• Remember I was complaining that Sean and Andria had set up unfriendly permissions on directories in /spc/valdemar3/test/summaries? This resulted in ppp crashing when another person tried to analyze data from the same station (a fairly friendly crash, but it wouldn't save your results all the same). The directories in question were created automatically by ppp. For reasons I don't understand, the ppp shellscript ignores the permissions you have set for yourself and does not allow group write. I checked though, and there is a workaround. You can specify the permissions when you use mkdir. I have made the necessary changes to ppp. If you see files and directories having the wrong permissions (I think ppp.plt has the same problem), here's what you do:

       change mkdir to mkdir -m g+rwx file.

I won't be offended if someone fixes ppp.plt.
• Say you have two time series x1,y1 and x2,y2. But, you have more x2's than x1's and gaps and so on. you would like to plot x vs. y2-y1, but need to synchronize the x's. This is particularly useful for making single and double differences. Here's an example:

  
  ijk1 = find(ismember(x1, x2));
  ijk2 = find(ismember(x2, x1));
  xdiff = x1(ijk1) ;
  ydiff = y1(ijk1) - y2(ijk2);
  plot(xdiff, ydiff)

• Change to solvekm.l : if you are using rapid orbits, and you are only using one day for your analysis (i.e. the orbit file is yymmmdd.reci)  then the qm file will be weeded based only on the rapid orbit for that day.  Reason: I have been having trouble with a satellite not being in the rapid orbit for the day in question, but existing in the previous or next day's orbit file.  Until now, the bad satellite would not be weeded from the file, and thousands of outliers would result.  Note I have committed solvekm.l to CVS so if this really bothers someone we can get the old version back.  --john 10/01/02
• another auto_newcin change. If the 5th argument is debug, then lots of print statements will go to the log. This will allow you to figure out where you screwed up either the stalocs or sitevecs file. You should still use the software program to keep a good copy of stalocs and sitevecs (and antenna).
• auto_newcin changes: test3 now goes in the 4th input, i.e.

 

auto_newcin 99jan01 transient.nml ITRF97 test3
auto_newcin 99jan01 transient.nml ITRF2000 test3
auto_newcin 99jan01 transient.nml ITRF2000
If you don't want the phase center variations (remember, it isn't the default because this particular version has crashed in the past) because you are using a choke ring, it is ok to ignore the test3 option.

• 08/26/02 Try this in your ~/.cshrc file:

limit coredumpsize 0M

I was running several years of data and didn't notice that I had filled up the disk - presumably with a lot of core dumps.

• 08/21/02: I have been able to kluge (sp?) a version of ambigon2 for the linux machines that is almost as good as the one I had running on the sun (see the web for more details). I will still try to port the sun code, but apparently I need access to a RedHat 6.2 machine (we are running 7.3). I will change solvekm.l to run the new code as follows: option A will be the new ambigon2; option a will be the old ambigon2. There are still issues associated with receiver mixing and ambiguity resolution which we will discuss next week at lunch.
• 08/19/02: Multi-day orbit files: if you want an orbit file from, say, 02jun03 until 02jun15, you can

type igs_clocks.jp 02jun03 02jun15  for final products, or type igr_clocks 02jun03 02jun15 for rapid orbits.  These will save the new orbit file as 02jun03A.eci or 02jun03A.reci, respectively.   These scripts create a .tpnml file and save it in the tpnml directory as 02jun03A.tpnml
•  08/19/02: ultrarapid orbit solutions:  to use ultrarapid orbits, option 6 on solvekm should be set to ultra and the orbit file for option 4 will require the suffix ".ueci"
• 08/14/02: Up-to-date user guides for most modules are in /gipsy/user_guides/source
• 08/12/02: I've come upon some inconsistency in how the linux version resolves ambiguities when compared with an older sun verison (really really old). I've informed JPL - but it is something we might have to fix here without their help. The short story is that the "new" code doesn't resolve widelanes well, and thus doesn't resolve as many ambiguities as it should. Stay tuned.
• 08/12/02: I've been working a new version of auto_newcin for a while now. I've been having problems with memory/core dumps, so if you see anything like that, please let me know (tell me which date and namelist produced it). This version includes the phase center variations correctly and uses ITRF97:

 

 
 
 
 
 
 
 
 
 
 
 
 
 

auto_newcin 02jan01 transient.nml test3

I have changed ppp as well so that you have the option of using the phase center corrections (pcv). These corrections are not important for almost all choke rings - as the pcv are assumed to be zero. type ppp to how to turn the pcv on. I would make this the default, but I want to be sure that we don't have core dumps going on all the time.

• 08/07/02: I've noticed that the NGS (MDO1, PIE1, TMGO) has started putting 5 second rinex files out - so downloads take longer. Clockprep is also crashing on them, which suggests clockprep doesn't expect something in the RINEX file. Am investigating. Installed a new version of clockprep, called clockprep_jeff. Have modified ppp to use this version instead of official GIPSY version.
• 08/07/02: Yousif provided me with a list of the data files that were lost or corrupted in the hard disk failure. Unfortunately, sometimes he put the corrupt files back on disk! This primarily impacts the qm files in the ppp directories, and as far as I can tell, primarily impacts me. But, you have been warned.