Parallactic Angle: Calculate the Turning Vertical and Field Rotation

What is the parallactic angle?

Imagine two arrows in the target's tangent plane. One points toward increasing declination; the other points along the sky toward the point directly overhead. Their relative angle is q. The zenith arrow changes its relationship to celestial north as the target moves through the observer's sky.

Starlink's SLALIB documentation describes this as the position angle of the vertical. The examples below are constructed, unrefracted geometry. They are not a dated star catalogue, weather forecast or telescope-control prescription.

State latitude, hour angle and the direction convention

Let φ be the observer's latitude, positive north; δ the target's declination, positive north; and H its local hour angle, positive west. With matched coordinate conventions, H is local sidereal angle minus right ascension. A target east of the meridian has negative H in the examples, and one west has positive H.

Hour angle is often expressed in hours, but the tables use degrees. Convert with fifteen degrees per hour of hour angle. This angular unit must not be confused with elapsed civil-clock time. The sidereal-time guide explains why a sidereal clock and an ordinary solar-day clock are different.

Check the coordinate contract of any calculation routine. SLALIB's standalone PA description specifies geocentric apparent hour angle and declination, whereas its general altazimuth tracking routine specifies topocentric inputs. Precise work needs a coherent choice and appropriate corrections; copying labels between routines without checking them can mix distinct quantities.

A formula that keeps the quadrant

In the stated ideal geometry, define E = cos φ sin H and N = cos δ sin φ − sin δ cos φ cos H. Then q = atan2(E,N). Convert radians to degrees if necessary and retain a signed range such as −180° to +180°. The endpoints represent the same orientation.

The order of the arguments is important. The first component is east and the second north, so this measures from north through east. A one-argument arctangent of E/N loses quadrant information. For example, it cannot safely distinguish a zenith direction due north from one due south at a meridian crossing.

An equivalent nonpolar-latitude expression divides both components by cos φ: q=atan2(sin H, tan φ cos δ−sin δ cos H). The undivided form avoids introducing a tangent of latitude and is convenient for a vector interpretation. Neither form makes an undefined tangent direction meaningful.

See it as a position angle toward one special target

Choose the sky target's right ascension as zero for this derivation. Its unit vector is u=(cos δ,0,sin δ). The zenith direction is z=(cos φ cos H,cos φ sin H,sin φ). At the target, east is e=(0,1,0) and north is n=(−sin δ,0,cos δ).

Project z onto that tangent basis: z·e=E and z·n=N. This gives the formula directly and explains its relationship to the general position-angle calculation. The second direction is specifically the observer's zenith, rather than an arbitrary second celestial object.

The dot product u·z is sin a, where a is geometric altitude. Therefore sin a = sin φ sin δ + cos φ cos δ cos H. The two tangent components have combined length |cos a|. As the target approaches the zenith, that projected direction becomes short and its orientation can become highly sensitive.

Compare three complete tracks at latitude 35° north

For δ=+20° and H=−30°, q is approximately −54.111876°. At H=+30°, it is +54.111876°. The geometry is symmetric in this example: q changes sign under H→−H, while altitude remains unchanged. This does not imply that q grows at a constant rate between the two endpoints.

The +34° track passes only one degree from the zenith. Its q curve is steep near H=0, reaching about +74.950917° by H=+5°. Farther along, the curve can flatten and then decrease. Looking at the whole track prevents a single value or a simple constant-rotation assumption from being applied everywhere.

Read altitude alongside orientation

At meridian crossing, the three altitudes are 55°, 75° and 89°. Each altitude curve is smooth and has its maximum at H=0 for these selected tracks. Meanwhile their q slopes near that same crossing differ greatly. A well-centred target and a smoothly changing altitude do not establish that the field orientation is changing slowly.

All plotted directions remain above the ideal horizon over the displayed interval. The chart deliberately leaves out atmospheric refraction, terrain, obstructions, target brightness and daylight. A positive geometric altitude is not a guarantee of actual visibility.

The horizon-coordinate guide gives the companion altitude and azimuth transformation. Azimuth is a direction around the observer's horizon; q is an orientation within the tangent plane at the target. Substituting one angle for the other answers a different question.

Is the parallactic angle always zero on the meridian?

At H=0, E=0 and N=sin(φ−δ). If the zenith lies north of the target in declination, q=0°. If it lies south, q=180° or equivalently −180°. Thus “on the meridian” alone does not determine a zero angle. The relative latitude and declination also matter.

At φ=35°, δ=50°, the target crosses north of the zenith and the direction toward the zenith is southward in its celestial tangent plane. At latitude −35°, the same reasoning works with southern declinations; the sign of the latitude does not change the definition of celestial north.

The final row is different: φ=δ and H=0 put the target exactly at the zenith. There is then no unique direction from the target toward that same point. E=N=0, so q is undefined. A routine returning a conventional zero for a singular input must not be mistaken for a geometrically determined direction.

Why orientation can turn rapidly near the zenith

For a fixed declination away from the exact singularity, the local slope at H=0 is dq/dH = cos φ / sin(φ−δ). The ratio is dimensionless when q and H use the same angular unit. As δ approaches φ, the denominator becomes small and the slope becomes large.

The final column assumes H advances by fifteen degrees per sidereal hour, the ideal fixed-star tracking model. It is an instantaneous slope at the meridian, not an average rate for the whole plotted track. The +34° value therefore does not say that the field will rotate at 704° per hour throughout an observation.

A small finite example makes the distinction concrete. For φ=35° and δ=34°, H=−0.1° gives q≈−4.682988° and H=+0.1° gives q≈+4.682988°. The 0.2° hour-angle interval corresponds to 48 sidereal seconds under that model, while q changes by about 9.365976°. Broader intervals require the actual curve, not a repeated local slope.

Targets with changing declination or non-sidereal tracking need the corresponding time-dependent calculation. The SLALIB mount-angle specification explicitly states the fixed-declination assumptions of its derivatives and warns that real mount details can matter strongly near the zenith.

How does this relate to field rotation?

Keeping the image centre on a target constrains where the telescope points; it does not by itself fix the orientation of every direction around that centre. In an ideal altitude-azimuth arrangement, the local vertical and celestial grid change their relationship as q changes. That changing relationship is the geometric ingredient behind field rotation.

For two nearby times along a continuous track, Δq measures the change of the vertical's position angle in the celestial basis. If an image axis is tied to that vertical, the celestial grid turns by the opposite signed amount relative to that axis. A real detector can have a fixed rotation offset or reversed handedness, so its displayed or commanded angle requires the instrument's own convention.

An ideal aligned equatorial tracking arrangement can preserve the celestial field orientation while q still changes, because the local vertical is not being used as the fixed detector axis. The existence of a changing parallactic angle alone therefore does not diagnose a tracking fault or prescribe a particular motor command.

Handle wrapping and singularities before subtracting angles

A signed display can jump from +179° to −179° while the physical orientation changes by only two degrees. For a short well-sampled step, calculate the wrapped difference in a half-turn interval, or unwrap a continuous sequence with an explicitly stated method. A raw subtraction of −358° would misread that small change.

Unwrapping does not repair missing information at an exact zenith passage. The vertical direction there is undefined, and nearby samples may change very rapidly. Sampling must be fine enough for the intended task; otherwise a nearest-angle rule can choose the wrong continuous branch. At an exact celestial pole, the north/east reference basis also needs a stated convention.

Refraction, coordinate-frame transformations and actual detector orientation need separate treatment in precision observations. The graph demonstrates geometry under declared assumptions. It does not calculate a universal maximum exposure duration, promise sharp images or establish what a particular telescope can do.

A useful calculation record

Keep latitude, the target coordinates and frame, local sidereal angle or supplied hour angle, units, sign convention, altitude, q, and the treatment of wrapping. If comparing two times, state whether declination was held fixed and whether the reported change is geometric q or a detector angle.

With those details, another reader can reconstruct the example and identify why two tools disagree. Start with the general position-angle and horizon-coordinate guides linked above, then compare the same inputs. A drawing of a tilted field becomes much easier to interpret once “north,” “up,” and the reference point each have a precise meaning.

Frequently asked questions