Notes by Mireille Louys

Hackhaton : Single dish 2022, 16 march 
Marco, Alessandra , François , Catherine , Mathieu, Vincenzo, Mireille
 
we look at Alessandra 's slides presented also at Radio IG running meeting recently.:
    
observing mode : 
    Marco: swipe over freq: also for solar data 

observing modes : 4 kinds : on/off , raster map, cross scan , on the fly scan 
data packaging in fits ? 
each file has a fits header , and one summary file for all data collected  is also present 

T5

data productype : measurement : too fuzzy for this radio type : usually used for a source and its photometric parameters  
need to clarify as data_product_subtype or extend the labels for data productype 

Mireille: 
seems there are different data products depending on the observing mode :
    intensity per source : on off mode
    cube ( few localized spectra ): raster image mode 
    image : mode= xcross scan 
    cube,  mode =  on the fly 

multi_feed example : Alessandra shows an image of the receiver geometry projected on the sky . 
the circular sky regions involved can be used to compute the  s_region
mireille comments / after session : 
    for instance : an estimated disc region corresponding to the area covered ? 
does this makes sense in terms of sensitivity, detection power , etc ??? 

in this case how to interprete s_fov , which estimates the coverage size 
how to evaluate s_resolution? 


Hackhaton : Single dish 2022, follow-up 21 march 
a group level for radio data 

?mode of observation 

recap on s_fov : a region dimension that estimates where you may find a source information, 
and outside which you cannot trust any detection .
you cannot resolve two sources that are in the same beam

beam_size = minimum resolution element 
s_resolution 

t_resolution : 
    this is tricky for single dish 
    this concept is not relevant for SD data : 
        Marco: better fill it with NULL
        the coverage is more important for query
        
em_resol_power
multiple spectral windows
Alessandra: there is a zoom mode where for one spectral window you have a better spectral resolution in some band 

Can have a table extension for describing support 

The pb is then to bind a support with a data chunk. 
VLBI can split the initial data set in more than one file 
Marco : it is not mandated to split a single data set , you can serve the full data set for all sub descriptions 

radioarchive.inaf.it (http://archivio-web.ira.inaf.it/tap is the TAP BASE_URL, currently missing the ivoa.obscore table, even if it's there)
single dish search : 
    select SD search choose project id 13-19 and see how the freq set up is shown 

em_resol : more meaningful than the em-resol power 
min resolution elt in freq : is useful 

for wide band receiver, the resol power varies a lot in the  band interval --> cannot be used for search

unit for spectral axis : give access to freq units in radio:  agreement in RadioIG .

FB suggests add f_resolution in freq units. 

Vincenzo : 
    Polarization : each single pol can be tuned at diff frequencies 
2 axes coupled : R L can be tuned independently , f. i 

Marco : you can show a document showing the dependencies 

MM: Consider how you can query in various query steps : 
    query in general Obs_core 
    then query the SD added table for fine grain selection 
    
François: Obscore allows to collect various data products under the same Observation obs_id  

Mireille : Proposal for going further :
can we list the metadata that are specific to SD data , and write down  a set of use cases where these metadata are used in practice  : 
 e.g.    
give me all observation where POL RR is present with fmin, fmax in [ 100 GHz, 300 GHz]. 
(do not stick to my units , it is just an interval example, and I m not a radio expert sorry ).  
    
Alessandra : examine uses-cases within the INAF Group 
May be to present then at next interop Radio IG session next month.