chi and h site logo

Noctilucent cloud

Noctilucent cloud, 2009-06-17/18, Edinburgh
Noctilucent cloud, 2009-06-17/18, Edinburgh.

Contents

Noctilucent clouds (NLC) are clouds of water ice at the top of the mesosphere at an altitude of 84 km. At that height these clouds can be in sunlight after the Sun has set - or before it has risen - on the lower tropospheric clouds. Scotland lies in the range of geographic latitude where these clouds can be seen; at low latitudes the mesosphere is too warm to form water ice, at high latitudes the summer nights are too bright. The clouds themselves can form only during the season when the mesopause is coldest, i.e. the summer months. In the northern hemisphere the NLC season runs from late May to late August.

Geometry

The graphic shows a cut through the Earth with the observer on the left looking in the direction where the Sun is below her horizon. The observer is looking upwards into the sky. Near the observer the line of sight is in the Earth's shadow and any cloud will probably appear dark against the sky. But further away from the observer the line of sight crosses into the sunlit part of the atmosphere and any cloud there may appear bright against the background.

The geometry of illumination of NLC
The geometry of illumination of NLC.

Consider a cloud at height z above the surface of the Earth and along the surface a distance α away from the observer. If we were looking north the angle α would be the difference in geographic latitude of the cloud and the observer. We count α positive if the cloud is toward the Sun and negative if the observer sees the cloud at the opposite azimuth to the Sun. Given z, the altitude h seen by the observer and the distance angle α are related by

h = atan2[(R+z) cos(α) - R, (R+z) sin(α)]

α = acos[R cos(h) / (R+z)] - h

Note that the "altitude" h can be larger than 90°. This will happen when α is negative: Imagine the observer looking towards the horizon where the Sun is, then raising her head to look at the cloud. If h > 90° (if α < 0) she must bend over backwards to look beyond the zenith to see the cloud.

But the main question is whether the cloud is in sunlight. This depends on the angle hSun, which is a negative angle telling how far below the horizon the Sun is for the observer. We look for the intersection point of the layer with height z with the border of the shadow. The ground distance and observer's altitude of that point are

αsh = -hSun - acos[R/(R+z)]

hsh = atan2[(R+z) cos(αsh) - R, (R+z) sin(αsh)]

(Note that the maths is this simple only for the case of NLC in the same direction as the Sun or in the opposite direction of the Sun. The equations cannot be applied to NLC at azimuth significantly to the left or right of the Sun.)

NLC on 2006-07-12/13 from Edinburgh, Scotland
NLC on 2006-07-12/13 from Edinburgh, Scotland.

At sunset the light/shadow border is at hsh = 180°. This is on the horizon opposite the Sun, i.e. the whole sky is still sunlit (whatever height z we are considering). When the Sun has dropped 4° below the horizon the z = 15 km level is illuminated up to only hsh ~ 35°. Around this time normal cloud is passing into the shadow and will appear dark against the sky. When the Sun is 6° below the horizon (civil twilight) the 15 km level is illuminated only to 2° above the horizon. At the same time the 85 km level is still in sunlight to hsh ~ 168°, virtually all the sky is still sunlit for clouds at that level. This is why it is recommended to start NLC observations when the Sun has dropped 4 or 6° below the horizon.

When the Sun is 9° below, the shadow at the 85 km level is moving quickly across the observer's sky, and at 12° below NLC would be illuminated only if below an altitude of 12° and toward the Sun. The recommendation is to observe until the Sun is 16° below, because by then NLC would be confined to only 2° on the sunward horizon.

Automatic camera 2009-2012

camera on carboard recliner
The dSLR on its cardboard recliner, inside the cardboard light protection box.

camera lightbox and laptop
The NLC camera moved against the window pane. The laptop is in front.

camera view
The NLC camera and laptop on the window sill. The City of Edinburgh and Kingdom of Fife beyond.

For several years, I had set up an automatic camera to take images of the northern sky every 15 min during the night. These were then inspected and reports made to the BAA Aurora Section and to the NLC observers' home page. Reports of "no NLC" are as important as positive reports of seeing NLC. Of course, when there is tropospheric cloud or fog in the way, no report can be made.

The camera was a Canon EOS 300D digital SLR set to 400 ISO and f/3.5, connected by USB to a Linux laptop that had gphoto2 installed. This camera model is supported quite well by the software, so that the laptop could set the exposure time, release the shutter to take the image, and download the image to the laptop. The laptop was also networked, so that the images and associated logs could be copied elsewhere for reduction and analysis.

The exposure time (shutter speed) depends on the altitude of the Sun. The best exposure in seconds for a given altitude of the Sun in degrees seems to be (for 400 ISO, f/3.5)

t = 0.5 s · exp[-(hSun+7.3°)/1.2°]

Remember that hSun is a negative number. Images were taken whenever hSun < −4°. The resulting exposure times ranged from 1/4 s to 8, 15 or 30 s, depending on how dark it got at midnight. These values were good for the sky background and reasonably faint NLC. For bright NLC much shorter exposures were necessary, as short as 1 s for very bright NLC.

The camera drew power not from its battery, but from a power supply plugged into a wall socket. Nonetheless, it appeared to go into power save mode after a few minutes with nothing to do. The laptop ran 24/7 and its cron demon invoked a custom Perl script every five minutes at 4, 9, 14, ... minutes past the hour. The script used my Java application Sputnik (see my software page) to calculate the azimuth and altitude of the Sun for the next full minute (0, 5, 10, ... minutes past the hour). If the Sun was too high to look for NLC, the script would merely get a status report from the camera, wait two minutes, get another status report, and then quit. Three minutes later, the next cron job would run and repeat the action. This was how the camera was prevented from going to power save mode.

If the altitude of the Sun was appropriate to look for NLC, then the Perl script would also take a sequence of images. The longest exposure in the sequence was calculated from the altitude of the Sun by the formula above, but subject to the longest setting possible (30 s). Other exposures were shorter by factors 2, 4, 8 etc. Normally, no images were taken when the Sun was too deep for NLC to be seen, but images could be taken all night to look for Perseid meteors or aurora borealis.

Automatic camera 2013-2015

The Canon EOS 300D - both the original I had purchased in 2004 and the second-hand replacement - ended with mechanical failure. Since the 2013 season I use a cheap compact camera without controlling laptop. In his talk to the Astronomical Society of Edinburgh in 2012, David Small (who runs a similar camera project), introduced us to the Canon Hack Developer Kit, whereby certain Canon compact cameras can run scripts to take images autonomously.

I purchased a Canon PowerShot A810 for this project, which now runs a CHDK Lua script for the whole NLC season. At nightfall it switches to imaging mode and moves out the lens, at daybreak it returns to display mode and moves the lens back in. During the day (only!), I can plug a laptop into the USB connection to download new images, adjust the clock, etc. As before, I use Linux and gphoto2 for this. The imaging schedule is essentially the same as in previous seasons, a sequence of increasing exposures every 15 min with the longest exposure depending on the solar altitude as

t = 16.3 s · (S/ISO)−1 · (f/D)2 · exp[−(hSun+7.3°)/1.2°]

where S is the ISO setting and f/D the f ratio (focal length divided by aperture). This camera allows a maximum exposure of 60 s.

2017 season

The camera was active from mid May to mid August, based again at the Royal Observatory Edinburgh. Once again, Jose Sabater Montes has part of his view over the city blocked by the camera assembly.

These are the nights during which at least some useful observations were made from 2017-05-15/16 to -08-15/16.

2016 season

The camera was active from 19 May to mid August, based again at the Royal Observatory Edinburgh. Once again, Jose Sabater Montes had part of his view over the city blocked by the camera assembly.

These are the nights during which at least some useful observations were made from 2016-05-19/20 to -08-15/16.

2015 season

The camera was active from 19 May to mid August, based again at the Royal Observatory Edinburgh. Once again, Jose Sabater Montes had part of his view over the city blocked by the camera assembly.

These are the nights during which at least some useful observations were made from 2015-05-19/20 to -08-15/16.

The camera also recorded some aurora borealis.

2014 season

The camera was active from mid May to mid August, based again at the Royal Observatory Edinburgh. Once again, Jose Sabater Montes had part of his view over the city blocked by the camera assembly.

These are the nights during which at least some useful observations were made in 2014. The camera was active from 2013-05-14/15 to -08-17/18.

2013 season

The camera was active from mid May to mid August, based at the Royal Observatory Edinburgh. Once again, Jose Sabater Montes had part of his view over the city blocked by the camera assembly.

These are the nights during which at least some useful observations were made in 2013. The camera was active from 2013-05-14/15 to -08-20/21.

2012 season

In 2012, the camera was based at the Royal Observatory Edinburgh, which also provided an old laptop and network access. Thanks are due to Jose Sabater Montes, whose view from the office on the City of Edinburgh was spoilt a bit by the camera.

The camera did in fact run between the 2011 and 2012 seasons to look for aurora, but there had been several problems. While an NLC season requires about 10000 to 15000 frames, looking for aurora during the winter requires on the order of 50000 frames. This ultimately led to the death of the shutter, having performed about 100000 actions in total. Douglas Cooper from Doune was kind enough to sell me his old Canon EOS 300D so that automatic imaging was suspended only briefly during a period of bad weather.

These are the nights during which at least some useful observations were made in 2012. The camera was active from 2012-05-15/16 to -08-21/22.

The camera also recorded some aurora borealis.

2011 season

In 2011, the camera was back at the Royal Observatory Edinburgh, which also provides an old laptop and network access. Thanks are due to Jose Sabater Montes, whose view from the office on the City of Edinburgh is spoilt a bit by the camera.

These are the nights during which at least some useful observations were made in 2011. The camera was active from 2011-05-01/02. It failed on 2011-08-16/17.

2010 season

In 2010, the Royal Observatory Edinburgh could not be used to set up the camera: The copper domes dating from 1894 were being refurbished, and the whole building where the camera would have been sited was shrouded in tarpaulin for most of the year. Thanks are due to David Small, who lives in the Scottish Borders a few km north of Kelso and had offered his window for the camera to look out of. The camera was also using his domestic wireless network and his broadband connection to upload the images to the server at my home.

These are the nights during which at least some useful observations were made in 2010. The camera was active from 2010-05-13/14 to 2010-08-31/32.

2009 season

In 2009, the camera was set up at the Royal Observatory Edinburgh. Thanks are due to Michele Cirasuolo, who had his office view somewhat impaired by the equipment. The camera was then set to 200 ISO and exposures were twice as long as used from 2010 onwards, but still subect to a maximum of 30 s.

These are the nights during which at least some useful observations were made in 2009.

The dSLR had been in operation since the start of June 2009. During the second half of May I had tried to use a networked webcam, but the images were no match for a visual observer or a proper camera. These are some example data: