Rare Lowitz Middle Arc ~ Captured in diamond dust by freelance photographer Judith Howcroft (Wilderness Light Photography) in Northern New Hampshire on the morning of January 22, 2010. ©Judith Howcroft, shown with permission

About - Submit Optics Picture of the Day Galleries Previous Next Today Subscribe to Features on RSS Feed

Lowitz arcs, first recorded in 1790 and first photographed some 200 years later, are rarely seen clearly and they still guard their secrets.

The arcs are mostly described as extensions from a sundog towards the 22 degree halo but that is only one of their aspects.  Here we see a less usual manifestation – a bright ‘middle’ arc extending upward from the sundog and well separated from the 22 degree halo. In enhanced images it reaches the upper tangent and Parry arcs.

Lowitz arcs are thought to arise from hexagonal prism crystals (not necessarily thin plates) taking various tilts about a horizontal axis passing through two opposite prism edges. Rays passing between prism faces inclined 60° to each other have three routes relative to the axis. Two of them give the ‘upper’ and ‘lower’ arcs that extend from sundogs to meet the 22 degree halo. The third route where the ray crosses the crystal very roughly parallel the rotation axis generates the ‘middle’ arc.

In simulations we can tune the appearance of the three arcs by (1) changing the relative lengths of the crystal side faces, by (2) altering the extent of its tilt from horizontal and (3) by changing its thickness. Some aspects to look for are the way all these influence each arc's brightness, its angular extent, the variation in brightness along the length and the appearance near to the sundog.

The matching HaloSim ray tracing at right (sun 24° high) used quite thick plates, c/a=0.5, with their side faces next to the Lowitz axis twice the length of others. Tilts were restricted to a Gaussian distribution around zero with standard deviation 24°. The fit is not unique and other parameter combinations are possible. However, like other halos that we do not understand fully, the simulations are ‘OK’ but attempts to further refine the parameters do not greatly improve the fit. The arcs and how they are made are not fully understood and the model itself perhaps needs improvement.