Tuesday, 14 March 2017

Trying to simulate arcs associated with parhelia in Rovaniemi solar display

Above is a display showing arcs extending outwards from parhelia and an attempt at simulating with Lowitz oriented plates using HaloPoint software. Sun elevation was 14.5 degrees.

In the lower right panel is a simulation with all rays. On the upper right is the same simulation (well, not exactly the same, ray numbers are a bit different) that shows from Lowitz orientation population only rays that experience 3 or 5 hits. This leaves the arcs from 325 and 315 type raypaths (the so called Schulthess arcs) visible while some others, such as Lowitz arcs are filtered out. The Lowitz arcs are seen in the lower left simulation, which has only rays with 2 and 4 hits from Lowitz oriented crystals.

The Lowitz arc scenario is clearly out of question to explain the arcs in the display, there is no match. Then how about the 3 and 5 hit scenario? Comparison is easiest with the flashing image below. The outer arcs fit together quite well but the inner ones don't. The inner arcs are angling too steeply inwards in the simulation.

Did I just do a botched job in matching the simulation to the photo? I admit it is not perfect work, but likely not so bad as to account for the large mismatch of the arcs.

Could it be because of the projection difference between the photo and simulation? I know that my 8 mm Sigma fisheye doesn't align perfectly with HaloPoint fisheye projections. For example, subanthelic arc is always narrower and taller in simulations than in photos. So maybe this too is just down to projection. Some day we might have a simulating tool that allows customization for individual lenses. That would settle the issue.

But even then we would still be long way off explaining these kind of displays. How do you get rid of Lowitz arcs in the simulation? I could not find a way. You have to really burn the Lowitz arcs to get 315 and 325 type arcs. This scorching inevitably gives also 46° contact arcs as is shown by the fuller version of the simulation below. In the fuller version of the photo next to it there are no signs of these arcs. Also, I was not able to simulate the correct length of the outer arcs. Lessening tilts seemed to make them disappear altogether. But I wasn't exhaustive in my attempts, it might have been possible to get them right.

What else? The display has no Kern arc even though cza was quite nice and diamond dust smooth. Clearly Kerns are not to be taken for granted. While observing, I thought there could be a chance here, but at the same time had a feeling that sun is probably too high, the plates too thin (you pretty much can say it without looking at a crystal sample) and tilting too much for the good old H.F.A.

Some words about the hunt itself  - the ryynäys as we say here. I was doing spotlight at the night and things got crappy as supersaturations increased. Diamond dust was everywhere but really poor halos. Expectedly it was no better in the morning. Close to the ski centre, diamond dust was so thick the sun did not shine through. At the edge of the cloud where the sun shone brightly there was quite some crystal glitter in the air, but, as it is with this glitter, it comes at the expense of halos: only pillar and subsun was visible. So I went shopping to a grocery store, hoping that in the meantime warming air brings a positive change.

It got better. However, the display was good only at the receding edge of diamond dust cloud. So I photographed in one place and when it weakened, hopped into car and followed the cloud to a new place to photograph again. I did six such manoeuvres, after which the swarm was soon done away completely by the heating sun. The photo shown here is the best set that I got at the third location.

Some technical info. The display was seen on 7 March 2017 in Rovaniemi. Temperature was probably around -16°C at the time I took the stack. It has 54 photos taken during 5m10s, starting at 10:13. Stacking was done with Registax 5.0. The artefacts seen in the image are a scourge of some series of Nikon cameras, they come in daytime photography and there is nothing you can do about them in post processing. 3000 series is one plagued with this issue (I have 3200 and 3300) and from what I have heard from other people's experiences 5000 series too. Definitely not cameras for halo photography.


  1. I don't think the projection difference is a culprit for mismatch between the inner arc's observation vs simulation. If You imagine rotating the view around zenith so that parhelion sets directly below zenith, ie. so, that a vertical line drawn down from zenith runs thru parhelion, the inner and outer arcs are in just about symmetrical angles relative to vertical line - in simulation. In real life, the inner arc is much closer to vertical line. If projection difference would cause the angle to differ between simulation and real life arc, it would do the same for the outer arc, too. Hence, no projection difference. The rotation procedure of course makes no difference, it’s just for making the symmetry (in simulation) easier to see. Actually, I made an animation comparing rotated halves of the third picture, I email it to You. (Picture tells more than thousand words). But what's the explanation, then? Beats me!

  2. Here is Sipinen's animation:
    Sipinen seems to be right. We then are dealing here with something that it truly different from the arcs that our Lowitz crystal model predicts. Actually, in the past I have superposed good Lowitz arc displays and simulations and noticed that the Lowitz arcs angles don't match either. The difference was not as striking as here, though. Just to be sure, I would like to see those comparisons done with lens customed simulations some day.

  3. Michael, it is very good, but could you be more specific on our question?