Map in the Sky - High Cloud Light Pillars in Xiamen, China

In the evening of May 10 2020, residents in Xiamen saw a strange, patterned cluster of light spots hovering over the city. Photographer YUAN Quan captured the phenomenon at its peak with his handphone.

© YUAN Quan, shown with permission. Equivalent focal length 39mm (from EXIF, unclear whether it's accurate).

Shortly after YUAN’s photo went public, LI Yahong and HUANG Tengyu from the China Sky Enthusiasts community found out that the light pattern, when flipped and rotated, perfectly matches downtown Xiamen’s map.

YUAN Quan's photo flipped and rotated 180°. Map from Apple Maps.

With the help of real night time images of Xiamen captured by Wuhan University’s Luojia-1A satellite (http://59.175.109.173:8888/index_en.html) and Chang Guang Satellite Technology's Jilin-1 satellite (https://mall.charmingglobe.com/Archive/), LI and HUANG’s finding was verified - we’re looking at a reflection of Xiamen’s city lights off the clouds.

*Special thanks to the Luojia-1A and Jilin-1 satellite teams for authorizing data usage.

Satellite images copyrighted to Jilin-1 and Luojia-1A satellite teams, shown with permission.

This phenomenon is most likely a high cloud light pillar event, similar to a previous case in Finland (http://www.thehalovault.blogspot.com/2017/03/map-in-sky.html) but on a much larger scale. The reflection captured in YUAN’s photo corresponds to an area of 200 square kilometers on the map. What’s even more interesting, the ground temperature in Xiamen during the event was 23°C -  a fairly warm night.

Due to the lack of background stars in YUAN and other people’s photos, it’s hard to do accurate cloud height calculations. A rough estimate by ZHANG Jiajie places the clouds between 6 and 7 kilometers above sea level. Sounding data from the night indicates the existence of a moist and wind-free layer in the 6 to 7 kilometer range so the height estimates may not be too far off.

Such warm night high cloud light pillar events are becoming more frequent in China - another two weaker ones were observed in the past few months in different cities. We believe the ubiquitous usage of decorative LED strips (which are very bright) on tall buildings, as well as the vast improvements in low-light photography on newer smartphones are two of the key factors leading to the surge in new observations.

Hidden Treasures in the Jutland Display

The Jutland display (http://thehalovault.blogspot.com/2020/04/new-halo-triple-cza-jensen-arcs-in.html) introduces us to the Jensen arcs, a brand new halo form and so far unexplained. While the new halo attracts all the plaudits, there are some hidden treasures in the display which are equally interesting and worthy of more attention.

Full circle 'pillar CZA'

During the peak of the display before the Jensen arcs appeared, another elusive, sub-visual halo lurked near the CZA and Kern arc. Anders’ Kern arc photo (likely the first high cloud Kern arc visible in a single frame), when enhanced, reveals a faint, full-circle arc outside the CZA + Kern combo.

Photo by Anders. Sun elevation 19.8°

First of all, it’s not a weaker version of the Jensen arcs. When compared to the actual Jensen arcs which emerged 10 minutes later, this one dips lower to around 57° elevation.

Left: ‘Pillar CZA’ (arrowed) , sun elevation 19.8°; Right: Jensen arcs, sun elevation 21.5°. Photos by Anders.

The 57° elevation reminds us of the multi-scattered 'pillar CZA' in the Siziwang Qi display (http://www.thehalovault.blogspot.com/2020/03/secondary-cza-from-sun-pillar.html). The overall appearance of the arc in Anders’ photo, albeit dimmer, greatly resembles the Siziwang Qi ‘pillar CZA’.  Chances of the two being the same halo are quite high.

Siziwang Qi photo by LI Tingfang. Jutland photo by Anders.

However, unlike Siziwang Qi where multi-scattering was strong, the absence of other multi-scattered halos (such as secondary UTA and PHC from the bright UTA) in Anders' photos clearly doesn't favor the conventional multi-scattered ‘pillar CZA’ solution for this display.

Nicolas Lefaudeux proposed an interesting theory which sounds highly viable in this case. He suggests that the sunlight scattered inside the cloud layer may have acted as an 'integrated glow', feeding light to each crystal from all azimuthal angles. In other words, the crystals are seeing from their viewpoint an infinite number of 'pillars' around the horizon, and each 'pillar' creates its own 'pillar CZA' via the crystals. The combination of these infinite numbers of 'pillar CZAs' will of course be a full circle with even intensity distribution.

This theory, when extended, may potentially explain the mismatch in the Siziwang Qi case too. In the Siziwang Qi display, available materials suggest the arc likely went full-ring as well (couldn’t be confirmed due to lack of DSLR photos). The sun pillar alone can't create a full-circle multi-scattered CZA without invoking unrealistically thick and triangular plates.

Simulations with ZHANG Jiajie’s program (https://github.com/LoveDaisy/ice_halo_sim/tree/dev/cpp). Sun elevation 20°.

Ji Yun from the Chinese halo watching community came up with the idea of facilitating the 360° glow from the snow covered terrain to increase the ‘pillar CZA’ azimuthal extent. His idea is in nature very similar to Nicolas’. If this ‘integrated glow’ theory holds, the name ‘pillar CZA’ should probably be changed to ‘integrated CZA’ instead. But for now, it’s still too early to make that move. More observations are needed to see if the arc always goes full-circle.

Skywatchers world-wide should start keeping an eye out for this elusive feature in future displays. Whenever there’s a bright CZA in diamond dust or high clouds, it’s highly recommended to always photograph the entire zenith area in RAW for in-depth analysis. It may also be worthwhile to go through past images and see if this arc shows up via heavy post processing.

Mismatching Lowitz arcs

Now let’s turn our attention to those arcs above the parhelia, which are seen in many photos of the Jutland display. Below is Anders’ photo for solar elevation of 14.8 degrees, and shown are four simulations of arcs, two from Lowitz oriented crystals (simulations 1 and 2) and two from alternate Lowitz oriented crystals (simulations 3 and 4). The alternate Lowitz orientation was included because recently it was realized that alternate Lowitz arcs actually exist. Ji Yun identified them in a display that Lasse Nurminen photographed in Raisio, Finland on 28 June 2019.

None of these scenarios really work. Only simulation 2 gives a match with the photo, and that’s just the outer arc – the inner arc is not right. And even the outer arc is on a rather short side for comparison. More length would have been preferred to see whether the apparent similarity in geometry really holds.

Photo by Anders. Sun elevation 14.8°. Simulations with HaloPoint 2

Below is another Anders’ photo, the sun elevation is seven degrees higher. Here no outer arcs are visible but we think the inner arcs are the same as in the earlier stage. This time comparison is made only with two Lowitz orientation scenarios. The circular Lowitz arc seems a better match here, but the arc in the photo doesn’t look quite as curved as in the simulation.

Photo by Anders. Sun elevation 20.4°. Simulation with HaloPoint 2.

Possibly these arcs in the Jutland display are like the subparhelic arcs (Schulthess arcs) of scenario 2, that is, arcs characterized by an added basal face reflection to the Lowitz arc raypath. The shape is not exactly what the simulations predict, but this is a known issue: several people have pointed out the disagreement of the observed subparhelic arc geometry with the theory for various displays. One such illuminating display was seen on 7 March 2017 in Rovaniemi, which happens to have solar elevation comparable to the first photo above.

And that’s just the geometry. Another contradiction, a glaring one, is that subparhelic arcs are in simulations accompanied by much stronger Lowitz arcs, yet usually Lowitz arcs are nowhere to be seen. Actually, the subparhelic arc and alternate subparhelic arc simulations above were tweaked for clarity: Lowitz arcs were removed in order to not let subparhelic arcs overwhelm them. Moreover, only the above parhelia parts of the subparhelic arc were plotted.

Until recently the subparhelic arcs have been exclusively seen in diamond dust, in which they are a common feature. The first high cloud case to show them appears to be the display that occurred on 7 February 2020 display in Finland.

2 and 10 o’clock spots

Many photos that Anders and others took of the 14 April display show an intensification in the subparhelic arcs at 2 or 10 o’clock positions, or both. Also one video has captured this feature nicely.

In some photos an outright spot is visible. Just to be sure, we tested with simulations for one of Anders' photos whether this could be a 24 plate arc. After all, the cloud contained pyramid crystals, as betrayed by the odd radius column arcs in some photos.

The position indeed corresponded to the 24 plate arc position, but the orientation isn’t quite right, as shown below. So, considering also the lack of other odd radius plate arcs in the display, these spots are probably better understood as parts of the subparhelic arcs.

Could they be just chalked up to irregularities in the high cloud? The fact that we haven’t seen such brightenings on subparhelic arcs in diamond dust displays – which tend to be more uniform – would seem to give appeal for such an explanation. However, considering that several photos from various people show more or less well defined spots in the Jutland display, it might be the case that there is something about the subparhelic arc crystals themselves that makes the spots.

The spot (on the right) in a photo by Richard Østerballe

The simulation has a 24 plate arc to compare with the spot on the left in Anders' photo. Simulation with HaloPoint2.

Another photo from Anders showing the spot. The 24 column arc is visible here, as well as in the photo above.

Odd radius column arcs

Anders' father, Ole Jensen also photographed the display and he took note of 9 column arcs on the sides of the sun. His photo is shown below. A selection of Anders’ photos, too, have 9 column arcs and additionally a very faint 24 column arc, which is visible in the two photos above.

9 column arcs. Photo by Ole Jensen. Sun elevation 13°

Seeing red in the blue spot?

There are some reports of people claiming to have seen red in the blue spot visually. But red has never been caught on photos to substantiate these claims and the theory is not forthcoming either.

Interestingly, some heavily enhanced Anders’ photos would seem to suggest that there is indeed red in the blue spot. However, Nicolas Lefaudeux demonstrated that it is an illusion that disappears when the blue color and its cyan end is covered. 

From the optical point of view, the colors in the blue spot are supposed to add to each other starting from the violet. It should be:

violet only = violet

violet + blue = blue

violet + blue + green = cyan

violet + blue + green + yellow = light cyan

violet + blue + green + red = white

So from this, any red should be an illusion (as there is never more red than any other color), and indeed, it seems to be the case when we look at the blinking grey patch on the blue and cyan of the blue spot (shown below).

Photo by Anders

The red on the blue spot more or less vanishes when blue-cyan is covered. Photo by Anders.

- Co-authored by Marko Riikonen, Jia Hao and Nicolas Lefaudeux