Over the last ~7 days we’ve noticed that the sonic at LC has been intermittent during the overnight hours. During these periods we’re seeing an ldiag=1, although the sonic does still report values sporadically. As it turns out, the pattern of ldiag=1 from roughly 8 pm to 9 am has been consistent throughout the project, however the missing data started just within the last week or so. 

In looking at the DSM dashboard today, I found diagbits=0, 1 and 5 at various times while ldiag=1. I haven’t looked at past data to see if this pattern has been consistent. According to the EC-100 manual these represent “amplitude is too low”, “amplitude is too high”, and “sonic head calibration signature error”, respectively. 

It’s pretty clear that this is not an icing/obstruction issue, so the best course of action may be swapping out the sonic. Since we do not have a spare, Steve has suggested swapping out the EC-100 box to see if that helps. John and Jacquie will try this in the next couple of days.


  1. This is a post to note that LC has been experiencing significant data gaps with the sonic/irga since the end of Jan (~28th). We replaced the EC100 (See blog post) but that didn't solved the issue. As of Feb 11 20:32:30 no data is coming in.  This is the longest data gap I've seen. Steve mentioned that there is a sonic at the base trailer owned by Sebastian but the mounting stuff is different from ours.

    Stay tuned. 

  2. When you say "no data is coming in", I assume you mean the instrument is sending messages but the data are flagged as bad.  The dsp output you include shows that the sonic messages are being received, but it does not show how many messages have invalid measurements.  I can see in NCharts that ldiag is still not zero most of the time.  If those are the other images attached to this blog post, I can't see them for some reason.  I just see an error message "Ouch! We can't load the image."

    In case it helps, you can add the -D option to dsp or data_stats to see the average of the variables over the statistics period along with the number of NaNs.  Running data_stats over the last couple days of lc data, you can see that 3429894 / 4901076 = 70% of the wind samples are invalid.  That of course corresponds with ldiag=0.700:

    [isfs@barolo raw_data]$ data_stats -a -i 40,-1 -p -D -F lc_2022021*.dat
      dsm sampid    nsamps |------- start -------|  |------ end -----|    rate  minMaxDT(sec) minMaxLen
    [lc:/dev/ttyDSM3] u.lc,v.lc,w.lc,tc.lc,diagbits.lc,co2.lc,h2o.lc,irgadiag.lc,Tirga.lc,Pirga.lc,SSco2.lc,SSh2o.lc,dPirga.lc,ldiag.lc,spd.lc,dir.lc
       40     41   4901076 2022 02 09 23:59:59.227  02 12 20:04:14.687   20.00  0.000   1.874   64   64
            u=0.363   (*3429894 NaN*)        v=-0.544  (*3429894 NaN*)        w=-0.027  (*3429894 NaN*)       tc=4.531   (*3429885 NaN*)
     diagbits=2.870         co2=0.681         h2o=3.038    irgadiag=0.000   
        Tirga=0.413       Pirga=783.912     SSco2=0.958       SSh2o=0.910   
       dPirga=690.571     ldiag=0.700         spd=1.554   (*3429894 NaN*)      dir=212.617 (*3429894 NaN*)

    Below is what dsp -D output looks like when there are no flagged measurements from the sonic in a 10-second period.  No NaNs reported, and ldiag=0.000:

    daq@lc:~ $ dsp -D
    sensor               dsm sampid    nsamps |------- start -------|  |------ end -----|    rate minMaxDT(sec) minMaxLen
    u.lc,...              40     41       200 2022 02 12 21:21:19.834  02 12 21:21:29.783   20.00  0.048  0.061   64   64
            u=-1.121          v=0.391           w=0.082          tc=5.237   
     diagbits=0.000         co2=657.279       h2o=2.246    irgadiag=0.000   
        Tirga=6.080       Pirga=820.848     SSco2=0.938       SSh2o=0.882   
       dPirga=659.336     ldiag=0.000         spd=1.231         dir=107.068