These are the news starting 01 Jan 2017
Click here for the news of the previous years.
|12 Aug 17||The backup station computer is frozen, and the screen will not be updated (this is the link labeled "current"). I worked on this problem quite a lot of time, with a 12 hour test run; after reinstalling this morning everything ran smoothly...but only for a couple of hours. Thank you for your patience, as remedial action will have to wait for several weeks.|
|08 Aug 17|| 1. Maintenance and calibration work
2. The Thinclient computer serving the Davis backup station died; has been replaced by the spare.
|02 Aug 17|| 1. The rain sensor was blocked by an impetuous algae
growth; cleaned, new filter, ok.
2. I bought a second HT2000 CO2 sensor and started a new comparison with both HT2000 and the Vaisala GMP343; data capture will run for about 6 weeks.
|01 Aug 17|| 1. Upload of jul17.dat
file into data archive
2. Rain sensor is not working since yesterday (31 July); missing readings have been replaced in the jul17.dat file by those of the VantagePro2 backup station. Problem will be checked tomorrow.
|13 Jul 17|| The intercomparison mentioned the 8th July gave some
interesting results: the cheap HT2000 CO2 sensor shows a huge
offset of about 117 ppm, but reproduces quite well the changing
patterns of CO2 mixing ratios. The instrument is manufactured by
DONGGUAN Xintai Instrument CO. Ltd. in Shenzhen (PRC) under the
brandname "Xintest". It uses a NDIR sensor from the Swedish company
SenseAir, probably the
The big rain pour from the 10th July caused an anomalous peak, but otherwise I am astounded how good the HT2000 CO2 sensors reproduces the overall pattern. Look here for a plot.The intercomparison tests will continue over the next months.
|10 Jul 17||Maintenance and recalibration on ozone and NOx sensors.|
|08 Jul 17|| 1. I started a new year-long comparison series with two
silicon band-gap temperature loggers developed by the Geological
Survey of Belgium (and meant to be displayed in caves: see
here the link to paper at Researchgate). The bare loggers have
installed in a ventilated plastic Stevenson-type enclosure. This is
the third intercomparison with these loggers.
2. I privately bought an "el-cheapio" logging CO2 sensor HT-2000 from Professional Instruments; the logger has been installed in the weather-hut where the Vaisala GMP343 is mounted. A week-long intercomparison will hopefully show if this is a useless gadget or a possible low-cost analyzer.
|06 Jul 17|| 1. Some thunderstorms approaching allowed to check the
working of the new combo HP-ThinClient & Astrogenic Nexstorm: it
does'nt work :-((( The problem is excessive noise generated by the
thin client (the Boltek PCI card is very close to the processor and
its cooler). Using a complete different thin client (an IGEL
UD5-420) with a similar geometry showed the same problem. So I
reverted to the former old FSC E600 computer, installed Nexstorm...
and it runs! Seems we have to stick with the FSC computer as being
the last one with moving parts...
2. WBGT sensor bad (Wetbulb is dry), will be repaired tomorrow,
3. Upload of jun17.dat file into data archive. Please read metadata at the start of the file.
|The lightning software has been changed from Lightning2000 (which is alas no more supported due to the passing away of its creator Mark Mears) to Nexstorm from Astrogenic. It runs now on a HP T5720 thin client (which still has a traditional PCI slot mandatory for the Boltek Stormtracker). Some broken links on the lightning page have been corrected or removed.|
|27 Jun 17||Installed RS232 communications with the new CO2 sensor; was a bit complicated (if you did not read the manual carefully!) as the instrument was factory-programmed to RS485 mode. I had to switch off sensor power to reset to RS232 mode (this explains the down-spikes). Everything ok now.|
|26 Jun 17|| Raoul Tholl and I installed this afternoon
our new GMP343 CO2 sensor from Vaisala,
This is a NDIR sensor without any moving parts, working by
diffusion. The logged data are corrected for temperature, but not
for pressure (The CO2 data are in channel 11)). The next monthly data file will hold a message how to
do this. The plotted data will be corrected to a standard pressure
and thus show the CO2 mixing ratio at p=1013.25 hPa and T=25°C.
This replacement is one of our downsizing steps aimed to make meteoLCD more maintenance friendly.
We had to shut down several sensors during this work; this shows up as impossible peaks or readings in the plots. The process will be self-healing.
|22 Jun 17||The temperature regulation of the UVB sensor seems to be working again; I switched off the device for an hour, and apparently this was the medicine to apply :-)) Please disgard the readings of 07:00 and 07:30 UTC.|
|21 Jun 17|| The temperature stabilization of the UVB sensor broke
down 2 days ago; the sensor continues to measure UVBeff, but the
readings may be slightly different from those of a properly
temperature stabilized instrument.
If the fault is serious, we have to bear it, as our backup instrument is in Davos at the World Radiation Center, participating in a inter-comparison/calibration project, and will back only in September.
|14 Jun 17|| 1. I tried without success to replace the dead-sick API
CO2 sensor by the old MIR9000 predecessor, which went into
retirement in 2008; the Methuselah refused to make any meaningful
measurements (which does not come as a surprise). So for the next
time, there will no CO2 measurements available. We try hard to find
a definitive solution for this problem, but this will take some
2. The WBGT heat-stress sensor had a blocked tubing, so that the water which is pumped up two times a day did not reach the wick which keeps one of the 3 thermometers wet. The fault started the 12 June (look on the WBGT plots for the green line of the wet thermometer data which does not go lower as it should). This has been repaired at 14:00
3. I blocked the scheduler and some other service in the quest to find the offending process which stops the lightning computer at irregular intervals...for the moment the computer works ok.
|12 Jun 17|| 1. Since a couple of weeks the CO2 readings are
drifting lower and lower. I was unable to calibrate a correct zero
and span today. As a stop-cap the API CO2 sensor has been shut off
and the old MIR9000 (which was put to rest in 2008) powered up
again. But this sensor also has very big problems, so that we are
without good CO2 data for the moment. I am thinking to abandon the
big and complicated instruments with their many moving parts which
caused so big problems and demanded very heavy maintenance work. A
solution would be to revert to a passive low-maintenance
instrument even when the accuracy of the measurements will be a bit
2. The lightning sensor is online again.
|Upload of may17.dat file into data archive. Please read metadata at the start of the file.|
|21 May 17||Our Boltek lightning sensor is out of order: the problem is computer-related (an intermittent kernel error) and will not be solved during the next 3 weeks.Sorry!|
|14 May 17|| 1. Recalibrated CO2 sensor, no problems.
2. Recalibrated NOx sensor: probably zero-offset of about 4 ug/m3 for NO and still unknown (might be huge) for NOx. Seems like stabilization has not yet occurred (molybdenum heating chamber not yet in usual state). Look at these readings with a healthy suspicion.
3- O3 sensor ok.
|13 May 17|| Repairs by T.S.E.
1.The ground ozone sensor (O341M) is back from the Brussels calibration site. The graphs show some funny negative values for today; this is a self-healing process.
2. The AC31M NO/NOx sensor has been repaired (exchange of the chopper wheel assembly). It will go online tomorrow (hopefully!).
3. A problem with the CO2 sensor has been fixed. The instrument will be recalibrated tomorrow morning.
|12 May 17||Recalibration work on the CO2 sensor. Some problems with zero offset and span value.|
|10 May 17||WBGT (heat-stress) sensor is online again (even if we are still waiting for warmer days...)|
|03 May 17|| Davis backup station changed to a "new" Vantage
Pro2 Plus model. The old station had a bad humidity sensor and was
not upgradable anymore with the old Vantage Pro console. I found on
eBAY a used equipment (VantagePro2 console, ISS, Rain, T(H, windvane
and anemomete)r for a reasonable price, and so decided to upgrade.
This was a horror story: the bought station should have been a
wireless model, is was cabled (that was the least problem, as the
prior station also was cabled). The console VantagePro2 worked but:
1. the rain gauge was defect (someone had removed the tiny magnet which actions a reed switch)
2. the vane/anemometer cable was bad, and someone had removed the magnet from the anemometer cup assembly.
3. the console cable was bad
Luckily, I had a collection of spares from older Davis stations, so I replaced the anemometer, used different cables, put a magnet from an older rain-gauge in the "new" one (which has sexy spikes to scare away the birds and avoid them filling the gauge with their droppings). The solar and UV sensors are those of the preceding station. After quite a lot of working hours, the "new" station is now running fine, and its readings can be seen at the live-data page (here).
|01 May 17|| 1. Upload of apr17.dat
file into data archive.
2. Problems with the lightning computer; investigations into spurious malfunctions going on. No data until further notice.
|15 Apr 17|| I am switching the lightning computer (an old faithful
FSC E600) to a HP thin client T5735 which has a PCI extension module
(the Boltek Stormtracker hardware is on a PCI card). I replaced the
1 GB SSD with a 16 GB SSD, changed Ram to 2 GB and replaced the
original Debian with Win7e (this was a bit tricky).
A second important change is made to the web page which displays the radar screen and ancillary information: this page has been moved from the local meteo01 server to meteo.lcd.lu at Restena. This will simplify the somewhat baroque infrastructure. The "movie-making" feature has been disabled for the moment (it worked not well since a long time ago, for reasons I never was able to find...). I will try another solution, but this is not the most urgent work to do...so please be patient!
|03 Apr 17||Well, I was able to get that beasty NOx sensor working... but just for a couple of hours. It really seems that the bearings of the chopper wheel are heavily damaged. The sensor will be shut down, waiting for a reapair by T.S.E. at the end of month.|
|02 Apr 17||It seems like the chopper wheel of the AC31M NOx/NO sensor is stuck again...I will try a last "repair" tomorrow, and if nothing goes, will shut the sensor down.|
|01 Apr 17||Upload of mar17.dat file into data archive. Please read the heading of the data file, as this month had its share of problems.|
|31 Mar 17|| 1. The AC31M sensor (NOx, NO) is again on online. The
problem is a flaky bearing of a chopper wheel used to check the
motor rotation. We will see if that wheel continues to rotate freely
as it should...
2. The meteo06 computer which manages the radioactivity sensor has been replaced by a HP T610 thin client. From the 6 computers that keep meteoLCD running, 4 are now passive thin clients: 2 running XPe, one LUBUNTU and one Win7e. Only the lightning computer (must have an PCI slot for the Boltek Stormtracker card) and its server meteo01 are still computers with moving parts.
3. I made a recalibration of the CO2 sensor. Zero was ok, but span vastly too low...this may be a faulty indication...will redo this asap..
|29 Mar 17|| 1. The AC31M NOx sensor has a motor fault. It will
remain offline for the next weeks.
2. There was probably a software glitch with the making of the 7days data file (which was truncated to the one-day file). I made a correction (08:16 UTC).
|28 Mar 17|| 1. Maintenance by T.S.E. The ground ozone sensor has
been removed to be calibrated at Brussels. This will take 3 to
4 weeks. The first graph has been changed to temporarily show only
air temperature. The repaired NOx sensor is going through a lengthy
zero calibration phase (new molybdenum in heater); it will go
2. To avoid data corruption when the main Linux computer has to be exchanged with a spare, I made some modifications at the relevant scripts. No errors seem to be left :-))
|24 Mar 17||There was a bettering with the daily CO2 swing, but I am still not satisfied: there seems to be a problem with the inlet air. I checked all pumps and tubes, found nothing suspicious. As a kludgy test, the glass cover from the inlet filter has been removed and the filter fixed by a wire, so that the sample air will come through the open window, and not through the normal inlet and manifold. Let's see if there is a change...|
|19 Mar 17||There really seems to be a problem with the CO2 sensor, as the readings are more or less flat, without the usual peak during the morning hours. Solution hopefully during the TSE maintenance scheduled for 31 March.|
|18 Mar 17|| 1. The NOx/NO sensor has a fatal breakdown
(motor failure during the 13th March). Repair will be made at the
end of March during the next big maintenance by T.S.E. The live data
icon will be labelled with "No data".
2. CO2 sensor: no visible fault detected. Made new zero & span calibration (08:45). Will keep an eye on this sensor.
|17 Mar 17|| 1. There was a bug in the main command file making the
half-hour ftp transfers (one "bye" too many). The consequence ^was
that the meteolcd app (IOS, Android) did not receive its data, and
correctly reported "Out of Service". The problem has been corrected
today at 14:00. I apologize, and my face is red :-(((!
2. A second, independent problem seems to hamper the NOx measurements (since 13/03) and the CO2 measurements (since today 10:00 UTC). I will check this asap.
|15 Mar 17|| A major change is made at the computing
infrastructure: the old faithful Fujitsu-Siemens FSC E600
computer running OpenSuSE 11.2 is replaced by a slick refurbished HP
T610 thin client, running LUBUNTU 16.04 and having no moving parts
The T610 has been bought used at eBAY, and fitted with a 32 GB Sata SSD; it holds 4 GB Ram. New versions of all software have been installed, with a silly problem with GNUPLOT 5.0 crashing ImageMagick (an uncorrected change of a set command from version 4.2 to 5.0 created a header in the png file which was unacceptable to ImageMagick).
The installation was made from 08:30 to 09:10; please expect some disruption and possible some teething problems....
The picture shows upper left one of the older SUMICOM computers (retrieving the radioactivity readings), upper right a HP Thin client T5370 (with XPe) which handles the datalogger and lower right the new HP T610 with Lubuntu which makes all the graphs and the transfers to meteo.lcd.lu.
|06 Mar 17||Maintenance and calibration.|
|04 Mar 17||Upload of feb17.dat file into data archive.|
|02 Feb 17||Upload of jan17.dat file into data archive.|
|30 Jan 17|| Re-check of CO2 and NOx calibration (08:00-08:40).
Uncorrectible offsets: NOx = 18 ug/m3 NO = 3 ug/m3
|28 Jan 17||The trend analysis including the 2016 data is finished. There might be some future minor changes, as for instance the Hohenpeissenberg CO2 data for the full year 2016 are not yet available. Please tell me about eventual errors. The updated Excel file with the 2016 data is here.|
|27 Jan 17||Maintenance and calibration by T.S.E.|
|24 Jan 17||New intercomparison paper (pdf file) by F.Massen & M. Zimmer of the TOC Dobson measurements done at Diekirch and at the RMI of Uccle.|
|20 Jan 17||New additions to the "Trends" section: eff. UVBdose, UVAdose|
|09 Jan 17||New additions to the "Trends" section: ground ozone, rainfall, enthalpy, solar energy, sunshine hours.|
|02 Jan 17||In the "Trends" section, air temperature (including DTR and DJF winter temperatures) is updated.|
|01 Jan 17|| 1. The usual dogs-dinner in the 7days plot due to Gnuplot
problems with handling previous and new year (the logger files do
not contain the year number). This will be self
healing in the the next 7 days.
2. Added overplot of Uccle and Diekirch total ozone column measurements for 2016 (see Dobson page)
3. Upload dec16.dat and 2016.zip (= all 2016 .dat files concatenated with headers, and zipped) into data archive.
4. Made 2016_only.xls file with statistics, daily and monthly averages. Started work on "trends" page update (this will take some time...)