search

e-merlin / eMERLIN_CASA_pipeline

This is CASA eMERLIN pipeline to calibrate data from the e-MERLIN array. Please fork the repository before making any changes and read the Coding Practices page in the wiki. Please add issues with the pipeline in the issues tab.
GNU General Public License v3.0
14 stars 11 forks source link

Flux scale seems wrong by factor 10 in test obs #79

Closed varenius closed 6 years ago

varenius commented 6 years ago

I processed a test observation TSARP220_20171013 C-band through the pipeline 0.6.3 (df2396d) with the following input switches: Pre-processing (0=don't run, 1=run): run_importfits = 1 summary_weblog = 1 hanning = 0 ms2mms = 0 flag_0_aoflagger= 0 flag_1_apriori = 1 flag_2a_manual = 1 average_1 = 1 plot_data = 1 Calibration (0=don't run, 1=run; 2=run and applycal if calibration is produced): flag_2b_manual = 2 init_models = 2 bandpass_0 = 2 flag_3_tfcropBP = 0 delay = 2 gain_0_p_ap = 2 fluxscale = 2 bandpass_1_sp = 2 gain_1_amp_sp = 2 applycal_all = 2 flag_4_rflag = 0 plot_corrected = 2 weblog = 2

The fluxscale.txt-file in the calib directory ends with this summary: Fitted spectrum for 1516+1932 with fitorder=1: Flux density = 0.065178 +/- 0.000149966 Fitted spectrum for 1407+284 with fitorder=1: Flux density = 0.307479 +/- 0.000785612 Fitted spectrum for 0319+415 with fitorder=1: Flux density = 4.16598 +/- 0.0100069 Fitted spectrum for 1532+2344 with fitorder=1: Flux density = 0.0203492 +/- 5.96974e-05

On top of my head, what I would have expected is something like 0.6, 3, 0.2 for 1516+1932, 1407+284, 1532+2344. 4Jy for 3C84 seems a bit weak to I think. If all values were ten times higher, I would have accepted them without an extra check.

Did I make some inconsistent choice of input switches above?

jmoldon commented 6 years ago

Yes, those fluxes are not correct. I suspect what is the problem, but can you attach or send the eMCP.log and the casa_eMCP.log files? I think not all those steps were actually executed.

Also, just for efficiency, you don't need to set the values of all the calibration steps to 2 if you are running everything. If you use 2, every step that produces caltables will have to run applycal in CASA just to overwrite once and again the corrected_data column. If you set all those to 1 everything will run the same and the applycal_all will only run at the end to apply all the calibrations. (The value 2 is useful when you want to run one particular step and check what is the effect the calibration up to that step).

varenius commented 6 years ago

Here are the log files. casa_eMCP.log eMCP.log

jmoldon commented 6 years ago

The error was that init_models = 2 does not run the step (it requires a 1). I have updated the inputs file to make it more clear, and now any positive value is accepted. Fixed in eaac7db7f54fb0b17ee9e1b43f6f08edd9012a42