A ceilometer fires a laser straight upwards and times what comes back. It cannot report something that is not above it. Ours reports a layer at one hundred and twenty feet over the 08 threshold on nights when there is no cloud in the country.
I maintain the weather instruments at a medium sized airport. Twenty two years.
The instrument in question is a ceilometer and it is the one people find hardest to argue with once you have explained it.
It sits on the ground and fires a pulsed laser vertically. Cloud droplets scatter a fraction of it back down, and the time it takes to come back is the height, because light travels at a known speed. Your answer is a profile of backscatter against altitude, and the machine picks out the layers in it and reports up to three of them.
There is no interpretation in that, and no sensor quietly drifting away from a reference over a decade the way a pressure transducer will: a photon went up, something at a height sent it back, and the height is arithmetic on a number everybody agrees about. That is all it is.
Ours feed into the automatic system that writes the reports the crews read. There is one at each end of the runway and one mid field.
The reason a cloud base is a serious number: an approach has a minimum, and a reported base below it means you plan for not getting in. One hundred and twenty feet is below everything we are equipped for.
The 08 threshold unit reports a layer at one hundred and twenty feet.
On clear nights, which is the wrong way round. On cloudy ones it reports the cloud, at the right height, agreeing with the other two to within the ten feet you would expect of three instruments looking at the same sky from a mile apart.
Between about one in the morning and first light. Most nights from September to March, almost never in summer, and never once in July in eleven years of records.
The other two units, eight hundred and fourteen hundred metres away on the same airfield, report nothing at all on those nights.
That is the shape of this. A clear sky everywhere, and a layer four metres thick a hundred and twenty feet above one specific patch of grass and tarmac.
What has been done about it, in order, over eleven years.
The window cleaned. That is the commonest fault there is, because a film of dirt on the glass scatters the outgoing pulse straight back into the detector and the machine reports it as something just overhead, so you clean it, and then you check the blower that is supposed to keep it clean, and then you give up and replace the whole window assembly, which we did in 2018.
The unit swapped. A different ceilometer, a different serial number, off the mid field mast, bolted to the 08 mast. It does it. The unit that came off 08 sits mid field now and has never done it once in six years.
So it is the place and not the box, and I had to write that sentence in a report to a manufacturer's engineer who had flown in.
He went through the raw profile with me, which is what I should have done years before.
It is not noise. Noise in these looks like noise, and you can see it in the profile as a rough floor that rises when the sky is clear because of molecular scattering. This is a discrete return with a sharp bottom edge, an extinction coefficient in the range you get from water droplets, and a thickness of between three and five metres. He said that if somebody showed him that profile with no other information he would tell them it was a thin stratus deck.
The visibility sensor on the same mast reads ten kilometres or better on every one of those nights, and fog at the surface is the one thing that would explain a low return honestly.
I have asked crews, twice, through air traffic, on nights it was reporting. Both said clear above. One said he had the town's lights from thirty miles out and had been looking at the stars over the field while he was vectored round, which is not a meteorological observation and which I wrote down anyway.
What we do with the reports.
The automatic system is set to use the mid field unit as primary for the official report, which it already was, and the 08 unit's output goes into the archive and onto the engineering page and nowhere that a pilot sees.
That was signed off in 2016 on the grounds that the unit was unreliable and awaiting replacement. It has been replaced twice since and the sentence in the paperwork has not changed.
So officially there is a faulty instrument at the 08 end and I am the man who has not fixed it.
The thing I did last winter, which I had not thought to do in eleven years.
I plotted the height.
Not the nightly figure, which I had looked at a thousand times and which is always about a hundred and twenty and which I had let my eye round off for a decade. I pulled every night it has reported since 2016 out of the archive and took the median for each winter and put the eight numbers on a graph.
2016 is a hundred and thirty six feet. Last winter is a hundred and twenty.
Two feet a year, near enough, in a straight line, and the line does not care which box is on the mast.
I have not put that in a report and I am not going to, because the only thing anybody could do with it is the arithmetic I have already done, and the answer to that arithmetic is a year, and I work here.