Analysis computers on the mainland are: Tromso: goppi and culebra, Kiruna: kana, Sodankyla: sana when using vizu verbose, sometimes the colour plot scales are not what you want. Try vizu VERBOSE but accepting the default scales. VERBOSE seems to force better default scales. Integration ----------- The "analysis_intfixed=0;" means that we integrate raw data even when the antenna is moving. It is often use it with antenna scanning observations. Satellite detection ------------------- In GUISDAP for Dummies for tau1_u: There are (were ?) sometimes problems with satellite detection in the calibration. This note was valid about 25 Oct 2003. Temporarily these profiles should be removed as follows: a_satch.plot=8; a_satch.lpg_skip=629; For tau8 : a_satch.lpg_skip=[977,1954]; For tau1: a_satch.lpg_skip=[511,1022]; For tau2pl: a_satch.lpg_skip=[511]; To avoid satellite rejection acting on clutter in scanning experiments like tau1u_cp3 on 24-26 June 2003, for example, set a_satch.clutter=10 which means do not look for satellites in the first 10 signal gates. Qute from Ingemar on 20060317: 'Except those dropouts where Tsys drops.=20 Can be avoided by putting 'a_satch.sigmac=3;' (or 4 maybe). That is normally set to Inf since it's a bit unstable for scanning/old exps. A modern cp1 should be OK. ' This enables checking of satellites in the calibration, since inf allows satellites. When a satellite is there the previous dump is used for cal. So there might be problems in scanning experiments. beata: "The range resolution is given in the r_ppw parameter. As the values are weighted means of points, the resulted resolution varies somewhat between points and dumps. The resolution is 2.3-2.5 km" manda ----- The RTGraph top left window shows three coloured lines. The blue one is the zero lag which only goes up to about 120 km. The red and green lines are pseudo zero lags where the red one has many more points than the green ??. The green line is very noisy, much oversampling, and shows the centred-on-zero lag. This seems to be the one used when Nr is plotted. The red line is the undecoded data so closer to a true zero lag. That is why, when you enable time-development, the Line plots 1 and 4 show about 1000 points from the red, about 500 points from the green and a about 120 points from the blue in that order from the bottom. If you want to limit the analysis range, for example to speed up the analysis, use analysis_range=[80:0.36:90]; (Analysis_altit does not work here) IPY data --------- in the analysis put: analysis_ppshortlags=1; in the special box. To then plot ipy data from the short-lag power profiles >> global max_ppw >> max_ppw=10 max_ppw = 10 --------------------------------------------- To avoid dumps being skipped through azimuth or elevation changes: analysis_sweep='az' or 'el' or 'elaz' ---------------------------------------------- To put an analysed data .png file on the web: -------------------------------------------- cp 2003-10-28_arcd@vhf.png /home/eiscat/tmp/Gup_vhf.png curl -s -F file=@/home/eiscat/tmp/Gup_vhf.png http://portal.eiscat.se:/raw/rtg/upload.cgi On 8 May 2004 I changed the time limit in ui_cal.m from 0.0007 to 0.005, in order to get overlapping data points in the pia experiment of 25 etc March 2003. To update the m_table*.txt files on the web (on palver, or webhost): ------------------------------------------------------- run the script update_web which is on goppi and culebra ===================================== To analyse at Kiruna : ssh eiscat@kana.eiscat.irf.se (alternatively, or previously ssh eiscat@eisdev.eiscat.irf.se ) guisdap -t -a (-t is text input) To analyse at Sodankyla: ssh eiscat@sana.eiscat.sgo.fi (alternatively, or previously ssh eiscat@hero.eiscat.sgo.fi) guisdap -t -a When analysing in text mode (-t) then Extra commands ('#'=line break)? analysis_txlimit=0;#f_ch_Pt=1.25e6; -------------------------------- Within matlab type format short g to see the numbers in the par_pl vizu verbose choose pf for plasma line freq a=vizu('verbose') -------------------------------------- To write the NCAR file again (for example when result files have been moved to a directory from somewhere else) then from matlab: do_NCAR('/analysis/results/2002-12-06_tau2_ant60@uhf',2,'uhf') for the binary only NCAR file ---------------------------------------- To do calibrations with cp7 one needs to use calib_ne(2,[180,580],1,70) instead of 500 for the upper height. Otherwise no peaks are found. --------------------------- To plot data using vizu from MADRIGAL, type vizu verbose 2000/madrigal_path_name ----------------------- To analyse separate beams with tau8 you need to run the analysis twice, once with analysis_code=1 and once with analysis_code=2 This produces two outputs one vhfa and one vhfb ------------------------------------------------ To analyse with the new, better power for UHF: f_ch_Pt= 'd_parbl(65)*1000'; or f_ch_Pt=1e6; forces the tx power to 1MW analysis_txlimit=0; f_ch_Pt=1.2e6" will integrate the data even if the transmitter is recorded as off and force the tx power to 1.2 MW. Needed sometimes at remote sites. -------------------------------------------------- To force the antenna elevation to some value: f_ch_el=90; ---------------------------------------------------- To plot the two power profiles in tau1a data: Load a result MATLAB file, plot(r_pprange(1:546,:),r_pp(1:546,:),'b', r_pprange(547:1092,:),r_pp(547:1092,:),'r' ) For tau8: plot(r_pprange(1:600,:),r_pp(1:600,:),'b', r_pprange(601:1200,:),r_pp(601:1200,:),'r' ) for cp7: plot(r_pprange(1:1011,:),r_pp(1:1011,:),'b', r_pprange(1012:2022,:),r_pp(1012:2022,:),'r' ) or thereabouts tau1a etc and 2 beams: Normaly the analysis combines the two halves (2 receivers) of the VHF. If the Tsys is noisy when analysing tau1 normally (not analysis_code=1 or 2) then it may be that there was no signal in one antenna half. This was the case for 4 Nov 2002 and a few other days. To intregrate only from guisdap: d_saveint.dir='/data/integrated/arcd' in special box. analysis_save=0; To avoid analysis have: analysis_altit=[]; To integrate only plasmaline data: analysis_plasmaline=1; ------------------------------------------------------------------- In batch mode: guisdap -B file where file = file.m containing go_on file2 and file2=file2.m containing the commands of a gfd_setup.m file. The output goes to /home/eiscat/tmp/file ------------------------------------------------------ New transmitter powers. analysis_txpower controls what parameter to use. Defaults to 8 for VHF and before 2005-01-01 for the others (or really [8 1] or [65 1000], but the factor is automatic) analysis_txpower=8; to force the UHF tx power to the old HV-derived value -------------------------------------------- To integrate old data with Nigel Wades integration program, on tana (matilda) : setenv matlab matlab6 then type guisdap -a This starts up matlab 6.5 instead of 7. and then guisdap knows to integrate with Nigel Wade's program. ------------------------------------------------ To force the systemp to a certain value, in the special box: analysis_Tsys=110; f_calTemp=[ ]; ---------------------------------------- In the UHF arcd analysis there appear to be two raw electron density profiles where there is signal. This is because there are two pulses that are analysed: 1) the decoded pulse where there is an extrapolated zero lag. This is closer to the correct density but is noisier than: 2) the undecoded pulse where there is another extrapolated zero lag but which is not so good because the correlation time becomes too short. The analysis rsult is an average which is not strictly correct. ------------------------------------------------------------ In the text output of the analysis there is a series if digits given showing the fit status of each gate. 0=OK 1=Max iterations 2=no fit 3=Fail Notes on calibrations using foE. ------------------------------- The results of using the automatic foE scaled values are somewhat variable, more so than foF2. Studying foE in more detail for the arc_dlayer experiments of July 2004, I found that the real foE, at least that which gives consistent results with the peak density between 90 and about 130 km from the EISCAT data both UHF and VHF, is a bit less than the scaled values. It looks like a a sometimes weak point of inflexion akin to foF1 in the F layer, is often present in the E region trace but not picked up by the analysis. The data from 12 July 2004 is a good example of this. I manually picked out foE this way. I was really calibrating foE from the EISCAT data instead of the other way around. Notes on calibrations using plasma lines (mainly beata) ------------------------------------------------------- On UHF there are several frequency ranges used to get plasma lines. Some of them show spurious or interference signals. It is possible to select only one or fewer than the 3 frequency channels by using gates =0 to prompt for more questions like this: >> calib_pl_ne('../../integrated/2016-02-24_beata2_0@T','beata',0) Reading plasmaline data canon: ../../integrated/2016-02-24_beata2_0@T/04738815.mat Enter NaN to shortern the vector Ranges (km)? [ 47 128 209 290 221 302 383 464 ] No fft points? [ 0 ] No to integrate? [ 1 ] No of gates? [ 4 ] No of lags? [ 50 ] Max sigma? [ 2 ] No frequencies? [ 3 ]1 Skip interference? [ 0 ] Offset frequencies? [ -3.6e+06 -6e+06 -8.4e+06 ] Sampling frequency? [ 4e-07 ] Invert frequency scale? [ 1 ] Radar frequency? [ 9.3e+08 ] Approx elevation? [ 77.5 ] Up/Down shifted (0/1)? [ 0 ] Start address? [ 108612 190844 273076 ] Gates to display? [ 1 2 3 4 ] Start of upar? [ 46 ] Removing noise Data path? [/analysis/results/] . Start time? [ 2016 2 24 20 20 5 ]2016 2 24 20 20 0 End time? [ 2016 2 24 21 41 41 ] ... etc. -------------------------------------------------------- There are scripts on tserver called remove_EISXAT_data and flag_EISCAT_data. The flag_EISCAT_data script can be used to flag the hourley directories as Junk so that they are not copied to the HQ archive. This command must be given after the directory is finished, but before the next hour is up. If you give the command during the hour it gets labelled Junk for a short time until the next dump is written when it is 'OK' again.