Proposal: Add use option also for three character observation type
VOMAY opened this issue · 7 comments
Dear Scivision,
I can confirm that your effort to speed up the parsing, by making the indicators optional, was successful.
Unfortunately, the parsing is still a bit slow. I have another proposal to speed it up:
The use argument is very useful to speed up the parsing, by selecting only the constellations/satellites you want. Especially ignoring the GEOs makes a big different, since they are recorded every single epoch. How do you feel about adding an option to allow a similar selection for the signal types?
Here is an example of what I would like to do:
#Only read Galileo E1 Code
gr.readrinex('test.rnx', use='E', signal='C1C')
#Read all code and phase observations of all constellations but GEOs
gr.readrinex('test.rnx', use=['E', 'G', 'C', 'R', 'J'], signal=['C', 'L'] )
I am convinced the potential time saving is substantial.
Implementing this, one may consider changing the use keyword, since it is not unambiguous anymore. But this is a detail.
Thanks a lot for your effort. Keep up the good work!
Best Regards
Volker
Thanks Volker, yes due to read buffering and the like, it is very speedy to just skip satellites.
Right, for the command-line ReadRinex
as well as the direct Python API, a new keyword is needed to include specific measurements.
Yes I think this will at least offer substantial memory benefits, as often we use only a few of the ~30 measurements available for each satellite
Try bf068c3
this enabled meas='L1C'
or meas=['L1C', 'C1C']
etc. as you have proposed.
Currently, I did it for RINEX OBS 3 only (if it works for you then I'll do same for RINEX OBS 2)
If you aren't currently working off the Git repo master
(in which case you could just git pull
) you can do:
git clone https://github.com/scivision/georinex/
pip install -e .
This works, but it is not yet perfectly user friendly. I would prefer wildcards, so I do not have to select the whole list of signals.
Look at the example I posted:
#Read all code and phase observations of all constellations but GEOs
gr.readrinex('test.rnx', use=['E', 'G', 'C', 'R', 'J'], signal=['C', 'L'] )
I would like to select all available Code and/or Phase observations with a simple command. This requires to check the length of the meas input (1, 2 or 3 characters) and compare those to the first characters of the signal names (Regex might be helpful here). This would simplify the use a lot. Alternatively you could also just discriminate between C, L, D and S. The RINEX files I am analyzing often have more than 20 different signals. I'd like to AVOID something like this:
#Read all code and phase observations of all constellations but GEOs (your current implementation) (the signal list is still not complete)
gr.readrinex('test.rnx', use=['E', 'G', 'C', 'R', 'J'], signal=['C1C', 'C1W', 'C2W' 'C2S', 'C2L', 'C2P', 'C5I', 'C5Q', 'C5X', 'C7I', 'C7Q', 'C7X' 'C8Q', 'L1C', 'C1C', 'L2S', 'L2L', 'C2P', 'L5I', 'L5Q', 'L5X', 'L7I', 'L7Q', 'L7X' 'L8Q'] )
The same could be done for the satellites (use), e.g.: 'E0?' for E01-E09
Note: bf068c3 fixes a previous error in the new meas=
feature you were testing.
I agree, wildcard based on simple regex is highly useful as well
Thanks for focusing on this issue.
On my side, I am in Germany and will call it day now, to get home before sunset. Talk to you tomorrow.
This is implemented for meas=
by checking the first character(s).
meas='C'
or meas=['C']
takes all measurements starting with C
like C1, C2
meas=C1
takes all measurements starting with C1
and so on.
This is implemented for OBS2 and OBS3 type files.
I just tested it for OBS3 and it works like a charm. Thanks a lot! This will save me a lot of time.