Three Modifications for Azimuthal Projections
Aitoff's Projection
The equatorial aspect of the azimuthal
equidistant projection presents the whole world in the
familiar "horizontal" aspect; however, there is significant
areal exaggeration near the map boundaries.
|
| Original equatorial aspect
of azimuthal equidistant map,
inner hemisphere outlined |
|
|
|
| Inner hemisphere |
Doubled longitudes |
|
| Doubled
horizontal scale |
Noticing that an azimuthal equidistant map
encloses an "inner" hemisphere in a disc whose radius is half
that of the whole map, Aitoff proposed a very simple,
yet attractive modification:
- project the world with doubled longitudinal coordinates,
effectively cramming everything into the inner
hemisphere
- double the horizontal scale, stretching the disc into a 2:1
ellipse
The resulting projection,
no more azimuthal, is equidistant
only along the Equator and central meridian.
Projection equations follow directly from those for the equatorial
azimuthal equidistant, substituting λ / 2
for λ and multiplying a factor 2 in x coordinates:
α = arccos (cos φ cos (λ/2))
x = 2αR cos φ sin (λ/2) /
sin α
y = αR sin φ / sin α
Hammer's and
Eckert-Greifendorff's Projections
Aitoff's work was itself modified by Hammer, whose projection
applied the
same idea, but to Lambert's
azimuthal equal-area projection instead. As a
consequence:
- global scale is smaller than in Aitoff's
- the projected inner hemisphere is not half as wide as the
whole map, but encloses half its area
- the final doubling restores scaled proportions and the
final map is also equal-area
- scale is no more constant along the major axes
Again, formulas can be deduced replacing λ by
λ / 2, this time in Lambert's:

|
|
Aitoff (top) and Hammer (bottom) maps at identical scales
|
Scales are different but overall lines are fairly similar in
Aitoff and Hammer projections. Since differences in meridian
spacing are hardly visible in the inner hemisphere,
these two projections were frequently mislabeled.
Hammer's design was in turn modified by
Eckert-Greifendorff, in a
projection
applying a further 2 : 1 rescaling. Therefore
equations are identical, except for substituting
λ / 4 for λ / 2 and changing
the x factor from 2 to 4.
Winkel's Tripel Projection
Yet another modification of Aitoff's projection was
devised by Winkel. Much like in his first and second
hybrid maps, his tripel projection
averages the equidistant
cylindrical projection, this time with Aitoff's.
Again, φ0 = ±arccos 2/π are usually chosen as
the standard
parallels for the cylindrical base (although the final
projection has no standard parallels).
Equations follow directly from Aitoff's and the equidistant
cylindrical's:
 |
| Winkel tripel map with
conventional base parallels |
α = arccos (cos φ cos (λ/2))
w = 0 if sin α = 0, 1 / sin α
otherwise
x = R(λ cos φ0 +
2wα cos φ sin (λ/2)) / 2
y = R(φ + wα sin φ) / 2
 |  |  |  |  |  | | www.progonos.com/furuti August 22, 2005 |
Copyright © 2001 Carlos A. Furuti