[daip] Re: Another potential upgrade to DELZN

[Leonia Kogan]

Mark wrote:

Leonia,

   I ran into some interesting data analyzing BX005A.
It had some geodetic-like observations and I solved for
zenith atmospheric delays.  Attached is a plot (.ps format)
of data (green filled circles), model (blue open circles),
and residuals (red crosses).  I've plotted all baselines
to antenna 4 (KP).   The left plot contains delays and
the right plot contains fringe-rates.  There is some bad data
for the MK antenna at the start of the data.  

   My program uses both delays and fringe-rates to find
zenith atmospheric delays.  Usually the fringe-rate data
is not very important, since for short scans one has fairly
large (up to 3 mHz) atmospheric jitter.  However, for this data
the rates were crucial to determining the correct zenith
delays.  Since the time rate of change of the multi-band
delays (caused by a large clock drift) can be calculated
from the fringe-rates, my program was able to conclude
that the problem in the MK multiband delays was with the
first group of data, not the third group.  Between the
extra information in the rates and the automatic editing,
I'm pretty happy that a program could figure out which
data to discard.

   So, I recommend adding fring-rates to the data that
can be fit by DELZN.  Of course, there needs to be a way
to weight the relative importance of the delay and fring-rate
data.  (I've been assigning an uncertainty of 3 cm (0.1 nsec)
to delays and 3 mHz to the fringe-rates.)

Mark

-------------------------------------------------------------




Mark,

I have in turn the creation of an option for CLCOR to correct the
given source as a difference of the source and the reference
calibrator.

This is an option you asked me to add for 'ATMO' and other relevant OPCODEs.

We decided here to create the different task for this: DFCOR, and I
started working but was terminated to carry out the comparison of
E-EVLA and ATA configuration. I have finished this job (I hope) with
memo #65 at the EVLA memo series.

So now I can return to DFCOR. Having finished DFCOR, I'll start thinking about 
your new suggestion to enhance DELZN.

By the way DELZN and 'ATMO' at CLCOR calculates right now the
correction to the derivative of the delay which is fringe rate (being
muliplyed by frequency) caused by the zenith delay correction. You can
use it by some way (for example comparing the found fringe rate
correction with the measured values) now. Although I understand you
want to solve the inverse task.

So give me a time 

Leonia