User Tools

Site Tools


analysis:solve

The full guide to using Calc/Solve is here. The following information describes the process used at Hobart and is aimed at routine processing of post-correlation data.

Prepare level 2/3 databases

Level 2 databases have an updated clock model applied (using calc11), replacing the model used by the correlator. To generate new databases, edit the up11.inputs file with the new database names (note the leading $):

$14FEB05XT
$14FEB05ST

and then run up11 up11.inputs AU. After completion, you should have 14FEB05XT_V002 and 14FEB05ST_V002 in /data/vlbi/mark3_dbh

Level 3 databases have cable cal and MET data added to provide an apriori atmosphere model. This is usually only applied to the X-band data. First, download the station log files with wget http://lupus.gsfc.nasa.gov/ivs/ivsdata/aux/2014/aust19/aust19hb.log, etc. Then run pwxcb aust19hb.log. You will be prompted to confirm some details of the experiment, and optionally edit the recorded MET data. When running pwxcb for the first log of an experiment, you will be prompted to give name of the database that this experiment is linked to - use the same syntax as in the up11 file (e.g $14FEB05XT). Repeat pwxcb for all stations in the experiment and then apply the calibrations by running dbcal /data/vlbi/wxcb/aust19.dbcal. After completion, you should now have a 14FEB05XT_V003 file in /data/vlbi/mark3_dbh

Start calc/solve & get a viable solution

First start Calc/Solve with the command enter AU. Once inside the Calc/Solve system, be extremely careful with your typing as almost any keypress can and will be quietly interpreted to change various settings. Also, default settings may not always be displayed, or displayed correctly on start-up. A last warning - Calc/Solve always starts up with the same data & settings loaded as when it was last run. To restart an analysis you should re-load the data, overwriting any previous databases in memory.

  • Load in the data, starting with X-band: Enter G to get some data through the SDBH program. Tap * to toggle the mode between between the append and replace modes. Use replace if starting afresh. Type K to enter the X-band database and version number ($14FEB05XT 03). Press enter to read the history file and a list of aprioris until you get back to the SDBH window. NB - you msy see some Warnings about Hart15 not being in the Flyby file - these are benign. Toggle the mode to append with*, and then use K to load in the S-band database. Again, review the information by pressing enter until you reach the SDBH window again & then return to the main window by pressing O
  • Set up appropriate calibrations by pressing +. Apply UT1Ortho, XpYpOrth, XpYpLib, UT1Libra, OPTLCont by entering 6, 7, 8, 9, #. Then press N to change to the S-band database and repeat. Return to the main menu with O
  • Set up initial solution (1^{st} order clock). Enter L. Press + and then 3 to use Group Delays only, and make sure that no EOPs are being estimated - you can move around the text menus with the arrow keys and toggle some variables by pressing the space bar. The screen should look like this:
Polar motion and UT1: Polynomial Parameterization         SETFL Ver. 2007.07.30
14/02/05 19:00 XWOB Coefficients   0 0 0 0
14/02/05 19:00 YWOB Coefficients   0 0 0 0
14/02/05 19:00 UT1  Coefficients   0 0 0 0
Select:(/)G.Rate & Segments (%)Only Segments (|)Sine Style (@)Reset Poly Epoch
Gamma, Precession rate             0 0    Nutation(.): Dpsi, Deps  0 0

Print residu(A)ls: OFF             Print corr. (M)atrix: OFF
Print (N)ormal Matrix: OFF         (Z)ero Normal Matrix: OFF

(^)Elev. cutoff: None              Pick parameters: (!)Sites OFF  (#)UT1/PM

Wea(K) Station Constraints: OFF    (R)Use rate: Yes

Use normally (W)eighted delays     Select: Baseline-(C)lock offsets

(:)Delay  Group  (;)Rates Off      Select: (B)aselines, (X)Data bases

Page: (E)Site       (S)ource       (O)ptions            (")Constraints
      (Q)Run least squares         (T)erminate SOLVE    (<)A priori clock
      (+)Change data type          Group delay only
      (')Change suppression method SUPMET__PRE98        (-) Singularity check

 Last page   Parms used / Max parms available:   20/ 2000

Then press E to set up the site parameters. For the initial solution, set the Clock Polynomials line to 1 1 1 0 0 * * * for every station bar one (the reference station - choose the same one as was used for a reference in fourfit). You can change between the stations with the N and P button. Make sure that all other parameter estimations are set to zero for all stations & beware of errant keypresses. Once ready, you can geenerate a solution by pressing Q.

The solution will be writtento the screen. Check that the Weighted RMS in delay is < 1 microsecond (if greater, suggests strong outliers or systematic problems). Also, check the second page where the clock solutions are listed:

  1. HART15M  CL 0 14/02/05 06:59                -7063.162 ns      1.95 ns
    2. HART15M  CL 1 14/02/05 06:59                   76.263 D-14    9.44 D-14
    3. HART15M  CL 2 14/02/05 06:59                    14.85-14/d    9.13-14/d
    4. HART15M  NG 14/02/05 06:59                  -44.17 mm         48.37 mm
    5. HART15M  EG 14/02/05 06:59                 -140.18 mm         41.28 mm
    6. HART15M  NG 14/02/06 06:59                  138.50 mm         46.24 mm
    7. HART15M  EG 14/02/06 06:59                 -151.57 mm         36.44 mm

If CL 0 is greater than 100000 ns (100 microsec) or CL 1 > 100000 D-14 in rate then you will need to apply an apriori clock model or, better yet, recorrelate the data. See the solve guide for instruction on applying a priori model. If the solution is ok, return to the main menu with O

  • Next, proceed to Ambiguity Resolution by pressing A. Change the minumum quality code to 5 by pressing Q and 5. Start the process by pressing A and check the printed results. You will sometimes encounter warnings like this:
 GAMB  S-band: r.m.s. of whole solution is OK
 GAMB  $b$ group ambiguities are resolved, but solution looks bad 
Are you sure, that you really need to save these results in scratch file ?

These are usually caused a few bad sources/observations - press S to save the results and return to the main window.

  • Now apply a per-baseline clock model. Press L and then C. Set the reference clock by entering the station number, and then press M. Generate a new solution by pressing Q. The weighted RMS of the solution should be in the range of 550-1500 ps and the magnitude of the baseline clocks should be < 1 ns. If now, refer to the guide on how to deal with permanmet ambiguities (this hasn't been necessary yet). Retun to the main page with O
  • Next inspect the residuals for any sign of clock breaks. NB - Don't try to run this on a machine with a display with < 1080 vertical pixels - the pgplot window is fixed at a very large size. Press P to start the plotting program. Press X in the pgplot window to return to the main window.
  • If you have to deal with clock breaks, you can do this by inserting a breakpoint in the clock solution. Press L and then E to get to the Sites menu. Use N to select the correct station and then press * to reveal and change the method until you get Insert (other options are Reset, Delete and Automatic which is the default). Then press C, and enter the time of the clock break. Press enter on an empty line when all the clock breaks have been inserted. You should now have additional entries for clock polynomials - use the arrow keys to highlight the first three zeroes and use the pace bar to toggle them to ones. The section below shows the set up for Katherine in aust19 which had two clock breaks at 035-133000 and 035-191200.
 Clock polynomials
 14/02/05 06:59                1 1 1 0 0 * * *
 14/02/05 13:30                1 1 1 0 0 * * *
 14/02/05 19:12                1 1 1 0 0 * * *
  • Next, set up for an intermiary solution.
    • Press L, + and 7 to set the solution type to G-Gxs combination.
    • Press ' and 3 to change the suppression method to SUPMET__COMB1.
    • Press -, A, 1, 2 (press enter), 2, 5 (press enter), 3 and 4 (press enter) to set up the singularity check with the recommended parameters. Press S to save & return.
    • Enter the sites menu with E and toggle mode to Automatic with *. Press C and set the time interval to 300 minutes (300 ), polynomial order to 2 (2), and use Batch mode (B). Set the reference station as before.
    • Press A to estimate the atmosphere. Use an interval of 300 minutes and batch mode
    • Return to the main menu with O and then go to the Weighting menu with H. Press C to add a noise floor to the weights, then save areturn to the main menu with O
    • Generate a solution with Q. If the weighted RMS is < 150 ps, things are likely ok. You should be able to produxe a version 4 database and NGS file at this stage if you want to work on the data in VieVS/OCCAM/etc. This avoids calc/solve flagging outliers automatically. To proceed on to obtain a final solution, read on.
  • Eliminate outliers with \. Set recommended parameters with the following commands: X, 1000 (press enter), U, 400 (press enter), C (press enter). When ready, the screen should look like this:
Automatic outliers elimination utility                    ELIM  Ver. 2007.08.01
                   �����������
$14FEB05XT <3>
'SUPMET__COMB1'
Information about residuals is not available yet

(X) Maximum uncertainty: 1000. psec        (A) Acceleration factor: 1

(U) Upper threshold for outlier detection: 400. psec      (E) EQM speed-up: No

(C) Cutoff limit for outlier detection:    not specified  (Y) Type: baseline

(Q) Quality code limit: 5                  (D) Update residuals

(-) Singularity check                      (') Change suppression method

(V) Verbosity level:    1                  (N) Confirm each action: no

(S) Return to Optin and save results       (O) Return to Optin without saving

(P) Proceed for outliers elimination       (T) Toggle elimination/restoration

(W) Weights update                         (H) On-line help
  • Press P to proceed with detection & elimination & press S to save the results & generate a new solution. If the wrms is between 15 and 100 psec with the used observations precetage > 80%, then it's regarded as good. Significantly worse wrms/low acceptance suggests mishanded clock breaks, un- or mis-applied calibration, or maybe a bad station.
  • Assuming that you have a good solution, or one with known and uncorrectable problems, you can then go on to generate a final solution.
    • Press L to enter the estimation menu. Press . to esimate nutation and press # to estimate EOPs. Use the arrow keys to change the UT1 coefficients line to UT1 Coefficients 0 1 0 0
    • Press E to enter the site menu. Toggle mode to Automatic with * and then set up both blocks (C) and atmosphere (A) using batch a 1 hour interval and batch mode. Set up gradient estimation with G and choose a 50 hour interval.
    • Generate the solution Q. Check the text listing makes sense and inspect the residuals (P) and estimated parameters (/).

Prepare level 4 database & convert to NGS format

You can do this at any stage after running GAMB which handles the ambiguity resolution & ionospheric correction process. It's a two-tage process of creating an updated database & the running a DOS script to convert it to NGS format.

  • Creating an updated database
    • From the main menu (O), press U. Make sure that only the X-band database is selected and then press N.
    • Press G, 5, 6 and 7 to set the options as shown below. Press N to proceed.
Database update                                           NEWDB Ver. 2007.06.05
-------------------------------------------------------------------------------


Database to be updated: $14FEB05XT

        Reweighting: (G)roup (P)hase (B)oth (#)None

 (1) Clk & atm parms, constraints, data configuration-->Yes No

 (2) Group delay editing and ambiguities--------------->Yes No

 (3) Group ionosphere calibration:--------------------->Yes No

 (4) Met., cable, phase cal status:-------------------->Yes No

 (5) Ocean, relativity, pole tide status:-------------->Yes No

 (6) Phase delay editing and ambiguities--------------->Yes No

 (7) Phase ionosphere calibration:--------------------->Yes No

(N)ext Menu             (O)PTION    (R)efresh Screen  (D)efault standard
(T)erminate SOLVE       Re(S)elect Databases
  • You'll be prompted to add a history entry via emacs at the next screen. Once emacs is started, add a line like “aust19 NGS conversion AU”. Close emacs and then press enter to update the file.
  • If it fails complaining about “Input experiment $14FEB05XT Ver. 3 not in catalog

” or “key not found”, etc, this is usually due to the database having already been updated. If this fails, you'll need to exit calc/solve , delete the offending file from /data/vlbi/mark3_dbh/ and from the catalogue. You can do this with the catlg program. Use the de command (with -1 as the “password”), select e for entry and then confirm the deletion of the most recent version. Once deleted, you should be able to run the update as per normal.

  • Once you have the 14FEB05XT_V004 file in /data/vlbi/mark3_dbh/, you can now convert it with MK3NGS2.EXE. First, start a DOS command prompt with wine cmd.exe. Then run MK3NGS2.EXE /data/vlbi/mark3_dbh/14FEB05XT_V004. The output file will be written into the current directory as 140205XT.NGS and is now ready. Note that VieVS expects a filename like 14FEB05XT_N004 though.
/home/www/auscope/opswiki/data/pages/analysis/solve.txt · Last modified: 2014/08/12 04:09 by Jamie McCallum