Calibration Attributes of the DRSP
2004/08/26 email from Michiel Hogerheijde announcing calibration attributes report:
As mentioned during last Tuesday's telecon, a report is now available
on the calibration attributes of the projects in the Design Reference
Science Plan (abstract below).
The report can be found at
The calibration requirements of the individual DRSP projects are at
Supporting materials (plots, spreadsheets) can be found at
Comments are welcome !
The Design Reference Science Plan offers a valuable opportunity to
assess the calibration requirements for the Atacama Large Millimeter
Array, as well as the impact that relaxing these requirements will
have on the science that can be achieved. This document summarizes the
calibration requirements based on 67% of the projects included in the
Design Reference Science Plan for which input was received. The
majority of projects can be successfully executed with an absolute
calibration accuracy of 5%. Stronger requirements are placed on the
repeatability and the relative calibration accuracy (both within a
single band and between bands), for which most projects require either
1-3% or 5%. The document discusses a number of key arguments that are
offered in support of these requirements. Only a few projects in the
Design Reference Science Plan include polarization measurements, and
no conclusions can be based on their listed requirements of 10 degrees
on the angle and 1% on the amplitude.
Comment from Bryan Butler 2004/08/26:
nice work. a few comments.
i'm a bit confused by tables 1 & 2 in the writeup. why is there an
included "1%" column, when, in fact, there was no such response option
in the questionnaire (only "1-3%" was available)? if that were an
available selection, there are several of the solar system projects that
would have chosen "1%" instead of "1-3%". ditto the "NoRq" column.
this wasn't an option, and it appears that you've used the comments
rather than strict questionnaire answers to infer that (a similar
problem exists in the solar system entries - if that option had been
available, it would have been chosen for all of the astrometric
projects, whereas it looks like you've only used it for 4.4.2, where the
respondents noted in the comments that there was no requirement [this
was also noted in my responses for 4.2.7 & 4.4.1]).
i'm also confused by the tabulated responses at
of the solar system responses are repeated (seemingly unneccessarily, at
least to me).
let's be a bit
careful about this, since it will likely be used by the
project to justify loosening the amplitude calibration spec. as you all
probably know, i've argued for that for some time (perhaps to my own
scientific detriment), but let's be sure it doesn't get loosened by
much. for instance, putting the amplitude calibration spec at 10%
(which would, after all, satisfy more than half of the DRSP projects
covered in this document, according to table 1) would be a disaster -
that's no better (or at least not much better) than OVRO & BIMA do
currently (i'm not sure about PdB
...and Michiel's response to Bryan's comments...
I've included 1% because a few responses specifically included that.
However, in the interpretation I did not make a distinction between 1% and
1-3%. Many reponses included N/A for various calibration attributes. I
felt they should be counted in some way (and not grouped in the 10% box,
because that distorts the statistics). Perhaps I shoudl go back and make
sure that I have found all cases where 10% was used to mean NoRq
I have indeed used the provided comments to tally the calibration
requirements, trying to be consistent.
If you see any inconsistencies, let me know! (That's why I've sent the
As for the duplicated responses: I have counted each DRSP project
individually, since they were designed to cover all science topics. If
multiple projects have identical requirements, these should still be
counted for each occurance (I think).
I agree that we need to be careful about this (hence the posting). I have
tried to indicate which science arguments drive the requirements toward
high accuracy - perhaps that should be given stronger wording so that -
indeed - it is not decided to relax the requirements to 10%...
...and Bryan's response to Michiel's response to Bryan's comments...
On 8/26/04 9:28 AM, Michiel Hogerheijde wrote:
'> Hi Bryan,
'> I've included 1% because a few responses specifically included that.
'> However, in the interpretation I did not make a distinction between 1% and
'> 1-3%. Many reponses included N/A for various calibration attributes. I
'> felt they should be counted in some way (and not grouped in the 10% box,
'> because that distorts the statistics). Perhaps I shoudl go back and make
'> sure that I have found all cases where 10% was used to mean NoRq.
well, it still seems to me that we should limit the tabulated results to
the answers that were available in the questionnaire. it certainly
would have changed the results in the solar system theme - perhaps in
others. if somebody replied with an answer that was not one of the
provided possible answers, that reply gets rejected - otherwise you get
a biased sample. or you can ask the person that sent in the reply to
limit it to one of the supplied ones. otherwise we ask everybody if
they would have chosen either "1%" or "NoRq" if it had been supplied,
and retabulate the results.
'> As for the duplicated responses: I have counted each DRSP project
'> individually, since they were designed to cover all science topics. If
'> multiple projects have identical requirements, these should still be
'> counted for each occurance (I think).
yes, i agree that the accounting has to include them in this way
(counting each separate entry), but in the full listing, they can be
combined, i think.
- 07 Sep 2004