How to analyze AG campaigns at the drop level, and calibrate the SCG

Analysis and reduction of microseismic perturbations is subject of another document
 
Recipes under HOW TO are for a tcsh environment. In particular we use at command prompt level...
foreach loops (for loops in bash)
setenv parameters (export in bash)
arrays of parameters and strings

Activate shell settings for AG:

cd ~
. .4ag

Utilities in ~/bin

Date conversion:
Utility
converts
example
procag2kal
day-of-year to month-day procag ... | procag2kal
procag2jd
day-of-year to Julian procag ... | procag2jd
procag2h 
dates to hours after epoch
procag ... | procag2h -D 2014,05,28
tsf-kal2dy dates to decimal years cat *.tsf | tsf-kal2dy

Directory structure

TD/a/
TD/a/bin/  contains symlinks to everything needed that is not in the ~/bin

Table 1 - Directories and content. Process flow proceeds from left (acquired zip-files) to right (final results)
Important lst-files are shown in bold.
Top level =
TD/a/
IfE-2009-11/  
IfE-2014-05/ 
Recalc2014/
   ...     Original data
Alldrops/

Drop files
Allsets/

Set and project files
Allsmp/

Processing files
Allcamps/

               Analysis and Result files
Directories
result of unzip
per campaign as needed



PS/        plot
d/         data
tmp/
o/          output 
tsfo/    final
msets/  own sets.txt per project
msets/d/  raw for concatenation
msets/b/  bundled
t/    
urtapm dumps
Symlinks

symlinks to original
data sources,
*.drop.txt
symlinks to original
data sources,
*.set.txt,
*.project.txt

prl2tse       -> ~/Ttide/SCG/+
campaigns.lst -> ../Allsmp/+
Data files
.tar, .zip 


*.drop.smp
*.*-drift.tsf

Control files



*.drop.smp.tse
remdc.tse
wmean.tse
urtap-*.ins
agreduct.tse
allties.tse
autoset.tse

diff-*.tse
synth-*.tse
weed.tse
msets/msets-<campaign>.tse
File catalogs
and lists


project-files.lst

files.lst
bundles.lst
1)
campaigns.lst

dropfiles.lst
cdumps.lst (temporary)
files.lst (temporary)
campaigns.lst -> ../Allsmp/+
features-MJD.lst 2)
meters.lst
mc-files.lst
orientations.lst
project-dates.lst
equivalent-sets.lst 3)

drop-symlinks.lst 4)
smp-symlinks.lst 4)
Programs

All programs should
either reside in
~/TD/a/bin
or should have
symlinks there.

At least their path
should be in the
$path shell variable.

There are a few
exceptions:

scgts4ag
add-campaign

procag.pl
  (symlink in ~/bin)
agsmp2mc*)
make-drift-tsf*)
make-s-and-j-tsf

mc4campaigns
pf-of-project

*)
please no symlink
       in  bin/
analyse-mc
bundle
cal-wmean
calplot
calplot-sop
evaluate-b+s
evaluate-tp-by-project
final-analysis
final-results4plot
list-features

list-project-dates
MJD-tp
make-mc-files-list
make-outlier-mc
make-own-sets
make-scg-rms
make-urtap-ins
pf-of-project
plot-cal-results
plot-final-results
plot-project
plot-all-projects
plot-all-sets
plot-merged-residuals
plot-tp
prl2tse -> ~/Ttide/SCG/+
procag.pl
  (symlink in ~/bin)
produce-ag
project-from-to
project-times
project-times2tse
ps-add-platform
redtp
slope-bars4plot
the-campaign-of
the-drop.smp-files
the-meter-of
the-orientation-of

the-platform-of
the-project-files
tp-index-of-project
tplist
tplist-project
tp2tsf
vary-bias-and-slopes

1) bundles.lst
Bundles are concatenated own-set files, ready with headers

2)
awk '/^[/=]/{print $0}' urtap-*-mrg.ins

3)
equivalent.lst
Equivalences between a bundled set file and the covered AG original set-files.

4)
drop-symlinks.lst and smp-symlinks.lst
Used in tp-index-of-project to find aliasses of project file names given a project name .

------------------
www/4me/
www/4me/ag2014
png-collections


Definitions

Table 2 - Definition of terms. Contents in directories and files
Part 1
What is / are...?






drop files
.drop.txt
They result from analysis of each single mass drop,
comprise all drops within a project.

set files
.set.txt
time and measurements. One set comprises / averages
over a series of drops, typically 50

project files
.project.txt
Parameters and information

a project
a succession of sets
No move of instrument within a project

a campaign
several projects within a time period of days to weeks. The original data should be put into a directory under ~/TD/a , e.g. ~/TD/a/IfE-2014-05
Campaign names should be a common substring in all file names, e.g.
201502b
The association with drop, set and project files is defined in a file named campaigns.lst
If the original .txt-file names run month names instead if numbers, it is advisable to create symlinks with dates in the form YYYYMMDD.  There is some flexibility with the rest of the file name, b ut not too much should be expected.  A functioning file name in campaign  200907a  may e.g. be  Onsala_AC_20090706b_1.drop.txt

platform and
orientation codes
Due to limitations in urtap, the least-squares program, the platform cannot have more than two characters. In order to signify bias terms with a four-letter word, one character is needed for orientation and one to enumerate the list,  AAN3 . The slope terms are not depending on platform nor orientation, so that the are named SLOn




Part 2
File name extensions
...


What's in the files?


ASCII
Binary

txt
after  project. , set. , drop. , original result files from g-software
A
tse
instructions for tsf-edit, a subroutine e.g. in tslist
A
tsi
instructions for tsf-edit without a leader and a trailing END, biding their time for import into a complete tse-file. Use wraptse  if the instructions form a complete set.
A
ts
mc
binary time series files, one component (column)
multi-column
B
tsf
time series files, irregular sampling intervals possible
yyyy mm dd  hhh mm ss fff data data ... :: comment
A
prl
Result sheet from tide least-squares program urtapt
A
ins instruction files for (almost all of my) Fortran main programs
A
drop.smp
tsf-files from drop files, regularized sampling interval
yyyy mm dd  hhh mm ss fff jul : SCG-obs :: AG-obs AG-sigma
A


Frequently used programs and utilities:
GMT for plotting
tslist (tsl) with tsfedit
tslq
tslql
tsf-kal2dy
tslg
jdc
urtapt
urtap-outlit
prl2tse

In scripts:  ~/bin  beak fromto ecq sb helptext calc decopt
In plotting scripts: ~/bin  axlim axlim4ts addlegend pslegend ps-add-platform ps2png

Languages
Fortran-77 (Absoft 11.5)
perl
awk
sed
csh
sh


HOW TO

Most programs show usage information upon option -h

The name model for mc-files containing AG,VAL AG,SIG and SCG,VAL columns can be given with templates:

             template     meaning     ( Search string in project file )
               %D    - start date
               %S    - sub-Campaign, the last character of <campaign>
               %C    - Campaign name
               %I    - instrument ID        ( Meter S/N: )
               %P    - platform ID          ( ^Name: )
               %O    - Orientation first
                       character, N or S    ( toupper() ~ "ORIENT" )
               %N    - number of concatenated projects + `P'
               %M    - input file name substring between last `_' and `.drop.smp'
               %L    - input file name last char before `.drop.smp'
               %Z    - last character of the project file name (less .project.txt)

For extracting data from drop and set files, use
procag.pl  (alias  procag)

The purpose of the processing operations in this chapter is
  1.   Extracting SCG-readings synchronous with the AG
  2.   Reducing the SCG drift
  3.   Preparing the input (data and instructions) for least-squares analysis (using urtap); includes setting up bias+slope events
  4.   Examining the output of the least-squares fit
  5.   The results comprise: SCG-calibration factor; AG-biases and drift-slopes for platforms and orientations
Begin here: Copy project, set and drop files into a new subdirectory under   ~/TD/a/
mkdir ~/TD/a/LM-201606
cp <files> ~/TD/a/LM-201606
cd ~/TD/a
add-campaign 201606a FG5-233 LM-201606
!!! You'll get a number of lst-files in   a/Allcamps/  marked with the campaign name.
!!! They are to be added to the respective campaigns lists, see the pink box below.

Error:
File   project-dates-<campaign>.lst   won't be made unless there are .smp-files
Instead, after
scgts4ag ,
cd /TD/a
project-times <yyyymm> >! Allcamps/project-dates-<campaign>.lst
project-times -J <yyyymn> >> Allcamps/project-dates-<campaign>.lst


First a short way to sample SCG data and output SCG + AG ascii files
For one project only
cd ~/TD/a/Allsmp
cp files.lst files.lst-bup#n
For creation of .drop.smp and .drop.smp.tse -files,
cd ~/TD
scgts4ag -new <campaign> -O a/Allsmp -sigma -n -S %.smp.tse a/Alldrops/*A?_201304*.drop.txt
and this is for GGP-filter corrected files:
lpf-unbessel -N -GO -C 201707a
scgts4ag -new 201707a -O a/Allsmp-s -SCG lpfu/G1_g_%%%%%%-1s.ts -E NONE -L NONE
-f -sigma -n -S %.smp.tse \
         `the-dropfiles -D 201707a`
OBS! There seem to be offsets from one G1_g_ file to the next. We must adjust them.
lpf-unbessel has been built out to this end (2017-07-17). Test the result using
check-overlap-dc lpfu/G1_g_201707*-1s.ts
 

and binary mc-files containing the data from SCG and AG
agsmp2mc -O a/Allsmp -o AGSG_201304a_%N-%P-%I-%O   a/Allsmp/*A?_2013042*.drop.smp
(if there are multiple campaigns that match the filename expression, add the campaign's id-letter,  a  or  b  or ... )  like here:
agsmp2mc -O a/Allsmp -o AGSG_201405a_%N-%P-%I-%O   a/Allsmp/*A?_201405*a.drop.smp
If you have created lpf-unbessel'ed files, use instead
agsmp2mc -O a/Allsmp-s -C 201502b -DY -o AGSG_201502b_s_%N-%P-%I-%O \
         a/Allsmp-s/A?_20150204*.drop.smp \
         a/Allsmp-s/A?_20150205*.drop.smp \
         a/Allsmp-s/A?_20150206*.drop.smp \
         a/Allsmp-s/A?_20150207*.drop.smp \
         a/Allsmp-s/A?_20150209*.drop.smp \
         a/Allsmp-s/A?_2015021*.drop.smp

     # obs! drop.smp-files must be in temporal order
     # You may use `the-dropfiles -S 201502b`
If the drift series has been created in a subdir, like here DSYN , use the -DD option:
agsmp2mc -O a/Allsmp-s -C 201707a -DY -DD a/Allcamps/d/DSYN/ -dt LIST -o AGSG_201707a_%N-%P-%I-%O `the-drop-files -S 201707a`
and you're ready for calibration with urtapt. The instruction file  urtap-cal201304a.ins would contain
  21 ^ a/Allsmp/AGSG_201304a_2P-AA-FG5-233-N-5s.mc

You might like to create symlinks in Allcamps/d
ln -s ../../Allsmp-s/AGSG_201707a_2P-AA-FG5-233-S-5s.mc AGSG_201707a_FG5-233-5s.mc
If you want to explore the impact of low-pass filtering the SCG, generate parallel mc-files, like here for 201106a:
agsmp2mc -O a/Allsmp-s -C 201106a -DY -DD a/Allcamps/d/DSYN/ -dt 1 -o AGSG_201106a_s_%N-%P-%I-%O `the-dropfiles -S 201106a`

agsmp2mc -O a/Allsmp -C 201106a -DY -DD a/Allcamps/d/DSYN/ -dt 1 -o AGSG_201106a_s_%N-%P-%I-%O `the-dropfiles -S -d a/Allsmp 201106a`


but wait!
Below we will use a procedure that carries out this step for one or more campaigns in one stroke,
mc4campaigns
and warrant consistent file names using the .lst -catalogues.

      OBS! FILE NAME CONVENTIONS
      In the .mc-file names, the campaign is surrounded by underscores

      If there are file names with month names instead of numbers, then create symlinks in a/Allsmp/
cd a/Allsmp/ 
ln -s AA_2009Nov03a.drop.smp AA_20091103a.drop.smp
       etc.
Since we love to go automatic, we should concatenate  drop.smp  files by campaigns.
So we need a recipe to combine files and get the respective campaign parameters.

If you need a new catalogue, use scgts4ag to go through all files
(you can delete the catalogues except files-new.lst in  a/Allcamps/d/ afterwards)
cd ~/TD
scgts4ag -O a/Allcamps/d/ -fl files-new.lst -sigma -t -n -S - a/Alldrops/*.drop.txt
collects SCG 1s-data in synch with the AG drop files.; and a file catalogue, files.lst
-t  will prohibit another sampling.


Cataloging files and projects

This box explains the hand-work to be done for cataloging. If you have used the add-campaign procedure
check that the lst-files you have obtained are o.k.

cd ~/TD/a/Allsmp
Make a copy of the files.lst (a/Allsmp)
Update  campaigns.lst  (a/Allsmp) and edit in lines of the following kind

#C 201004a     AGSG_201004a_%I   AS_2010Apr17a.drop.smp AS_2010Apr17b.drop.smp ...
        ^-- Campaign      ^--- mc-filename stub      ^----input files...

all those files you want to combine for a campaign.
You may have help from the file  a/Allcamps/files.lst

Once the campaign's drop files are processed, add the information to
a/Allsmp/dropfiles.lst
in temporal order; example:
201606a  Onsala_AC_20160627a.drop
201606a  Onsala_AC_20160628a.drop
201606a  Onsala_AA_20160629a.drop
201606a  Onsala_AA_20160701a.drop
201606a  Onsala_AA_20160702a.drop
201606a  Onsala_AA_20160703a.drop

For all campaigns:
set campaigns = ( `awk '/^#C /{print $2}' ../Allcamps/campaigns.lst | sort | uniq` ) 
Edit this as required.

Then in  a/Allcamps
add orientations to the file  orientations.lst
Sometimes a one-liner can help:
cd ~/TD/a/Allsets
awk '/Nord/{print "%O",FILENAME,"north"}' A?_2011061*.project.txt
awk '/Sued/{print "%O",FILENAME,"south"}' A?_2011061*.project.txt
and in  meters.lst (a/Allcamps) #M records and  %P records

Finally, update   project-dates.lst  (a/Allcamps/ , manually editing in the output of
cd ~/TD/a/Allsmp
project-times 2011061
project-times -J 2011061

Then, heading for mc4campaigns, again in  a/Allsmp we are going to produce drift files.
Doing that on the basis of parameters solved in our croned SG jobs ( ~/TD/ts4openend ) can be risky. First check that the tsf-edit segments and commands in  make-drift-tsf  are o.k., and check the result with urtap's ph04 files.
Where the CONT command is used, check that the file has been renewed recently, and that the urtap solution was successful. 
The models for drift synthesis that are relevant here are DSYNTH and DWWFSYNTH . There are other solutions (OCHY  using hydrology and nontidal ocean in the analysis;  EXPFIT solving exponential decay parameter) that are either lagging in time or need to be re-executed manually.
make-drift-tsf [-DY] -O dest-dir/ *<yyyymm>*.drop.smp.tse
alt.
make-drift-tsf [-DY] -CY -O dest-dir/ *<yyyymm>*.drop.smp.tse
alt.
make-drift-tsf -DY -CY -O ../Allcamps/d/ `the-dropfiles -x .smp.tse <campaign>`
most actual (three drift solutions in one  AGSG*.mc -file):
tslist _63000 -B2009,06,15 -r1 -Emake-drift-tsf,DWWFSYNTH -I -o ~/TD/a/Allcamps/d/synth-drift-extd.ts
tslist _63000 -B2009,06,15 -r1 -Emake-drift-tsf,DSYNTH    -I -o ~/TD/a/Allcamps/d/synth-drift-simp.ts
tslist _63000 -B2009,06,15 -r1 -Emake-drift-tsf,EXPF      -I -o ~/TD/a/Allcamps/d/synth-drift-expfit.ts
make-drift-tsf -DY -from synth-drift-extd.ts   -CY -O ../Allcamps/d/DWWF/ `the-dropfiles -d ../Allsmp-s -x .smp.tse <campaign>`
make-drift-tsf -DY -from synth-drift-simp.ts   -CY -O ../Allcamps/d/DSYN/ `the-dropfiles -d ../Allsmp-s -x .smp.tse <campaign>`
make-drift-tsf -DY -from synth-drift-expfit.ts -CY -O ../Allcamps/d/EXPF/ `the-dropfiles -d ../Allsmp-s
-x .smp.tse <campaign>`
which will create samples of the SCG drift model in synch with the drop.smp-files
one file for every drop.txt -file.  In the case above both AG's of the parallel campaigns are included.
The directory for the drift.ts -files is  ~/TD/a/Allcamps/d  (hard-wired).
Destination directory can be  ./   but is usually  ../Allcamps/d/  Files produced:  *-drift.tsf
To get a list of drop.smp.tse files from ./dropfiles.lst , issue (or exec within ` ` )
the-dropfiles -x .smp.tse <campaign>
Prefer to add the check-up option -CY whether the old syn-drift.ts file is long enough.
If not,
cd ~/Ttide/SCG
tslist _#n -B2009,6,15 -r1 -Emake-drift-tsf,SYNTH -I -o ~/TD/a/Allcamps/d/synth-drift-new.ts
and copy to ~/TD/a/Allcamps/d/synth-drift.ts if the file is o.k.
Replace SYNTH with  DSYNTH  or  DWWFSYNTH  to base drift on the most recent open-end solution of the weekly SCG analyses, standard or extended solution, respectively.
All of that and much of the following  can be carried out with a super-script, urtap-merge-pp

Then, again in  a/Allsmp
mc4campaigns -DY
which will run a full series of agsmp2mc , one for each #C line in  campaigns.lst .
The filename expressions determine which drop sets will be combined for each individual calibration.
The mc-files are (as long as the hard-wired parameters of 2014-08-11 are intact) stored in ../Allcamps/d/ .
To reprocess one campaign,
mc4campaigns -DY <DVAR> [-DD ../Allcamps/d/DSYN/ ] -C <campaign>
where the -DY option asks to use the "synthetic" drift files (from TSF EDIT-recipes inside the  make-drift-tsf script).
 
And this is for GGP-filter corrected files:
scgts4ag -GO -new 201405a -f -O a/Allsmp-s -ilp -sigma -n -S %.smp.tse a/Alldrops/*A?_201405*a.drop.txt
cd a/Allsmp
mc4campaigns -DY -smpdir ../Allsmp-s -dt 5 -C 201405a
Most actual:
mc4campaigns    -DY DWWF -smpdir ../Allsmp-s -dt 5 -DD ../Allcamps/d/DWWF/ -C <campaign>
mc4campaigns -U -DY EXPF -smpdir ../Allsmp-s -dt 5 -DD ../Allcamps/d/EXPF/ -C <campaign>
mc4campaigns -U -DY DSYN -smpdir ../Allsmp-s -dt 5 -DD ../Allcamps/d/DSYN/ -C <campaign>
The latter one is most probable to extend through a campaign, since its creation is most independent of acquired model data.

OBS! In campaign 201006a, see urtap-$c.ins, there is only one "="-command spanning the whole

project period (there is only one project!).
cd ~/TD/a/Allcamps
cp mc-files.lst mc-files.lst-bup#n
make-mc-files-list -o mc-files.lst $campaigns
and manually update  mc-files.lst

There is a potential (and common) problem with combining projects:
The exact epoch of each drop file. The help text of mc4campaigns says:
A file ./agsmp2mc.problems is created with information on
eventually low retrieval quotas in the sampling process.
In such a case, analyse whether sampling schemes make jumps
(worst case if e.g. a 10 - 20 - 30 seconds scheme jumps to a 05 - 15 - 25 scheme).
The doubt margin of readfuf/readfmd cannot cope with this.
The best way is to force 1/2 of the auto-detected sampling interval upon the output mc-file:
In a/Allcamps/campaigns.lst add a line
#D <campaign> <dt>
decide on a common divisor of the sampling times (in 200907a this would be 2 seconds), and
mc4campaigns -dt p -DY -C <campaign>
or, if  Allcamps/campaigns.lst holds the information in a #D -record
mc4campaigns -dt LIST -DY -C <campaign>
However,  if the mc-data is to be combined with seismometer records where missing-records and precision to the second matter,
a series of mc-files by-project should be generated. The -s option in mc4campaigns can be employed.
More on this theme, in application the the campaign in February 2015 is here.
Check that  Allcamps/campaigns.lst  is consistent with the imposed sampling intervals!

Then we assemble urtap.ins files for each campaign, using the <campaign>-jump.dat files and a template inside the script
make-urtap-ins -C <campaign> -with-drift -outliers-from-raw
or, to fetch the mc-file name from mc-files.lst  rather than from the jump-file(s), like in campaign 201502b,
make-urtap-ins -C <campaign> -mc-from-list -with-drift-var SYNT -outliers-from-raw -include =
where we also specify a drift signal (SYNT) and allow only bias parameters in the signal model.
Manually update the list files meters.lst  and  orientations.lst (this may help:)
ls *$campaign[n]*.drop.smp | awk '{print "%O",$1}'
Then you can run
urtapt @ urtap-<campaign>.ins
or if you dare
foreach i ( urtap*.ins )
   urtap-outlit $i
end

     Branch here to urtapm.html for a multi-campaign analysis

    The sequel here continues with single campaigns

Now you can analyse the bias-and-slopes results, and the calibration result.
The weighted mean calibration factor from all campaigns:
cal-wmean [-d]
where the -d option makes cal-wmean more talkative.

Bias and slope results:
evaluate-b+s <campaign>
evaluate-b+s -b <campaign>
Then you can collect the outlier data (for showing them on the calplot)
make-outlier-mc `awk '/^#C /{print $2}' campaigns.lst`
make-outlier-mc $campaigns
producing e.g. o/AGSG_200907a_FG5-233-10s-del.mc  for campaign 200907a

CUTOUT:
Next step is plotting with calplot .
There is a simpler utility too:
plot-cal-results [-scg-rms] [-with-outliers] <campaign>
 But let's continue with   calplot . The bias-and-slope solution that
was produced by urtap will be plotted at all times of the campaign.
i.e. also through the pauses. In order to suppress it there,
project-times2tse -do-cutout -o o/cutout $campaigns
Have a look at the tse-files produced; use setenv to set variables to appropriate
values before you continue:
setenv CUTOUT o/cutout
foreach c ( $campaigns )
  tslist o/scg-cal-$c.ph01.ts -E o/cutout-$c.tse,CUTOUT -I
end
Finally, you can plot
setenv PNGOUT ~/www/4me/ag2014/
calplot -sj -cal -del d/AGSG_200907a_FG5-233-10s.mc
Extra settings are needed for e.g. 201004a
setenv WEEDOUT 3000
set c=201004a
calplot +LEG C:8.8,4.0 -sj REMDC -pfs -cal -del d/`the-mc-file-for $c`
Examining the bias-and-slope files:
Explore the uncertainty range of biases and slopes:
prl2tse -XB  1 -S o/scg-cal-201405a.prl BPS0 >! synth.tse
prl2tse -XB -1 -S o/scg-cal-201405a.prl BMS0 >> synth.tse
prl2tse -XS  1 -S o/scg-cal-201405a.prl B0SP >> synth.tse
prl2tse -XS -1 -S o/scg-cal-201405a.prl B0SM >> synth.tse
prl2tse        -S o/scg-cal-201405a.prl B0S0 >> synth.tse
or by a script:
vary-bias-and-slopes $campaigns
Everything from the above starting with urtap in a one-size-fits-all approach can be done with
foreach c ( $campaigns )
   analyse-mc -S -C -PP $c
end
(sorry, can  analyse-mc  with plotting only one campaign at a time)
____________________________________________________________________________________

Plotting with   plot-cal-results
Plot includes:
    AG samples with error bars minus SCG
    optionally, outliers can be included (option -with-outliers)
    bias+slopes
    SCG long-term residual
    microseismic RMS from SNSN Seismograph
    ... from SCG (optional)
If seismograph data is not available, use the option -scg-rms
or try produce with ~/Seismo/gcf/long-time-movrms
long-time-movrms -h
shows all necessary steps.

Next project:
CREATE ANCILLARY DATA FOR DIRECT CORRECTION OF AG-DATA WITH SCG DATA
Concept: AG = AG_raw - AG_jumps&slopes - ( Filter{SCG_G|V, deci.tse,F30} - SCG-drift ) * Cal_new/Cal_old

Cal_new = weighted mean of all cal campaigns
Cal_old = value in ~/TD/t/urtap.trs
Subdirectory for the example: tsfo/ under Allcamps/

Reducing Drop-data to Set-data:
   Example: As a demonstration of principle, do one campaign,
                   simply the measurements, don't care what, just as an exercise
tslist d/AGSG_200907a_FG5-233-10s.mc -LA -LA -E autoset.tse,A -C3
with autoset.tse
 TSF EDIT A
 OPEN 31 < set.ts
 AUTOSLMEAN U=6 ->W G=5.
 ; = use weights, print results ASCII to protocol, gap width=5*dt,
 ; then copy the result into the W-array and output:
 DUMPW U=31
 END
Another example, subtracting the SCG:
(that this works o.k. is owned to the fact that subtracting a series involves work storage W while the AG-series and its weights are imported into X)
setenv FUNIT 11
setenv OPEN22 '; '
setenv FNSUBLABEL 'S+V'
setenv FACSUB 774.16
tslist d/AGSG_201405b_FG5-220-4P-05s.mc -LA -LA -keep-open \
       -E subts.tse,ADDVL -E autoset.tse,A -C3 -qqq -w tmp/A-S-201405b-autoset.tsf

and converting dates to hours from midnight of day 1:
awk '{print ($7-56805)*24,substr($0,32)}' tmp/A-S-201405b-autoset.tsf

Again, we have a useful script for going through all campaigns:
produce-ag -results -weighted-means -with-slopes $campaigns
prodcue-ag -results $campaigns
which amounts to the same using -results defaults.
using autoset.tse (by default).
ask  produce-ag -h   what else it can produce

Depending upon the task for produce-ag, a specific block
in autoset.tse is loaded. The commands specify a.o. the
output subdir for the results. autoset.tse is responsible for
creating the columns in the final  ...$campaign.set.mc files
foreach c ( $campaigns )
  final-analysis -add 380. -S -I tsfo $c
end
final-results4plot -I tsfo > tsfo/final-results.tbl
plot-final-results

Produce own set-files, reduced as far as we can go (no biases though)

make-mc-files-list $campaigns

setenv REPORT msets/msets
setenv MSETSADDG 380.
foreach c ( $campaigns )
   project-times2tse -S -C 'AUTOSLMEAN U=32' -C+ '${FFMT:[+TSF]} ->W ${WOPT} G=${GOPT:[1000.]} ' -o msets/msets $c
end
setenv DELCOMMAND RJC
setenv DELTHRESH 200.

foreach c ( $campaigns )
    produce-ag -own-sets msets/ -weighted-means -with-slopes -use-gaplength 5 $c
end
set c = 200911a
produce-ag -own-sets msets/ -weighted-means -with-slopes -use-gaplength 2 $c

foreach c ( $campaigns )
  foreach m ( AA AC AS )
    echo M=$m C=$c
    bundle -by-camp $c -by-monu $m
  end
end
bundle -by-monu AC -by-instr FG5-233
bundle -by-monu AA -by-instr FG5-233
bundle -by-monu AA -by-instr FG5-220
bundle -by-monu AC -by-instr FG5-220
  procag commands with their and our sets:
Their sets are usually more than one per campaign, while our covers all sets on the same monument with the same instrument during campaign
procag.pl -o -u -s -r -A -c Gravity -f "%5.2f " $our
procag.pl    -u -s -r -A -c G_ures  -f "%5.2f " $theirs
See plot-project how it's used
agsetwmean  commands, examples (results in uGal!)
own sets:
agsetwmean.pl -q -o -R 0.0 -w msets/b/own-AC-FG5-233.set.txt
agsetwmean.pl -q -o -R 0.0 -w msets/b/own-AA-FG5-220.set.txt
their sets:
agsetwmean.pl -q -u -r -w -A ../Allsets/Onsala_AC*.set.txt
agsetwmean.pl -q -u -r -w -A ../Allsets/AA*.set.txt
plot-project -g 2 msets/b/own-AA-201405a.set.txt
The project plots suggest that we need to account for orientation biases. The respective ADD commands have to be issued inside produce-ag
and for that to happen the orientations have to be quoted so that they appear in the urtap results. So back to the urtap-step. The responsible
script is agsmp2mc   See and revise the script code at "NEW Sep 5 2014:"

Take North as the reference orientation
Use = commands for monuments different from AA and for AA if orientation is South.
Use / commands for each project.
Mark the bias events with platform and orientation, not JMP
Mark the slope events with SLO1...

After bundle you can...
plot-all-sets -M 233 -P AA
__________________________________

.bye