○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Grids & Datums C O L U M N

THE REPUBLIC OF MADAGASCAR

The contents of this column reflect the views
of the author, who is responsible for the facts
measured for a length of 1,044.295 maps of Madagascar as well as for
and accuracy of the data presented herein. meters, and a repeat measurement dif- computation of the triangulation.
The contents do not necessarily reflect the
official views or policies of the American fered by a millimeter. This is typical The Antsirana Datum of 1906 was
Society for Photogrammetry and Remote
Sensing and/or Louisiana State University. quality for the French in the late 19th an update established by M. Lesage of
The Portuguese discovered the is- century. The Antsirana Datum of 1887 the French Navy at the Antsirana pil-
land of Madagascar in 1500, which was established at the Antsirana pillar lar where: Fo = -12° 16' 20.3" South
was in the same era as their explo- where: Fo = -12° 16' 25.5" South, (determined by Driencourt in 1904),
rations of the entire eastern coast of Lo = 46° 57' 36.2" East of Paris. The and Lo = 46° 57' 36.2" East of Paris (no
Africa. The French established defining azimuth was determined change from Favé in 1887). A later
some colonial stations temporarily from Antsirana to signal station report (date unknown) gives a local
in the 17 th century, and re-estab- Oronjia as: a o = 79° 12' 19". Antsir- Hatt Grid in the vicinity having an
lished their presence in the 18 th cen- ana 1887 Datum is referenced to the origin where: the Central Meridian,
tury. The local native Hovas group Clarke 1880 ellipsoid where a = 6,378, (l o) = 46° 55' 34.669" East of Paris,
signed a treaty with France in 1868, 249.145 meters, and 1/f = 293.465. the Latitude of Origin, fo = -12° 16'
and France declared that Madagas- The Hellville Datum of 1888 was 26.148" South, a False Easting of 80
car was a French protectorate in established at the meridian pillar km, and a False Northing of 30 km.
1882. The French reorganized the where: Fo = -13° 24' 20.7" South, Considering the coordinate preci-
administrative system of Madagas- Lo = 45° 57' 05.0" East of Paris. The sion reported, the International 1924
car in 1924 and independence was defining azimuth was determined ellipsoid was presumably used where:
achieved in 1960. from the observation tower in Hell- a = 6,378,388 m, and 1/f = 297.
Madagascar is the fourth largest ville to the fire beacon at station The Nosy Vé Datum of 1907 was
island in the world; its length is al- Tany-Kely as: a o = 197° 25' 18". The established for the hydrographic
most 1,600 km (995 miles) and its Hellville 1888 Datum is also refer- survey of the waters off the port of
maximum width is 576 km (360 enced to the Clarke 1880 ellipsoid. Tuléar (Toliara) in southwest Mada-
miles). Tsaratanana Massif in the The Mojanga Datum of 1888 was es- gascar. The origin is at the meridian pil-
northern part of Madagascar is the tablished at the pillar in Mojanga lar where: Fo = -23° 39' 00.20" South,
highest point at 2880 m (9449 ft). where: Fo = -15° 43' 24.2" South, Lo = 41° 16' 44.2" East of Paris. The Hell-
The island is obliquely oriented to Lo = 43° 58' 54.4" East of Paris. Al- ville 1888 Datum and the Hatt Azi-
the cardinal directions on a north- though the Clarke 1880 ellipsoid muthal Equidistant Grid also used the
east to southwest angle, and this was referenced, no record exists of quadrant system.
fact has had a whopping affect on the defining azimuth for this tiny In 1926, the Hellville 1888 Datum
the development of mathematical Datum. Note that all three of these was re-defined in terms of longitude
cartography for the island. The cen- local Datums in northwest Madagas- where Lo = 48° 16' 28.95" (“1.5”) East
trally located capitol of the Repub- car also defined Hatt Azimuthal of Greenwich. Presumably, this
lic of Madagascar is Antananarivo, Equidistant Grids for the respective change was identified as the Hell-
and other major cities include Toli- hydrographic surveys. The usual ville Datum of 1926. Although the
ara in the south, Mahajanga in the rectangular coordinate systems original meridian pillar was gone,
northwest, Antsiranana in the north, (SystPmes Usuel), were imple- the new origin was observed to have
and Toamasina in the east. The port mented in quadrant systems instead the coordinates of X = -28.1 m,
of Hell-Ville is on the island of of in false origin systems. Simulta- y = -2.6 m (note: X, y s.i.c.).
Nosy Bé off the northwest coast. neously, the French Army started In 1928 Laborde, a colonel in the
Original survey observations using the Bonne pseudo-conic equal French artillery, wrote that the
were performed by Owen of the Brit- area projection for topographic sur- Bonne projection chosen for the Ser-
ish Royal Navy in 1824-25, and veys and plane table mapping of vice Géographique de Madagascar
French Navy observations followed Madagascar. Note that the Bonne was as bad a choice as it had been
in 1830 and 1850. Geodetic triangu- was the projection “de rigueur” in for France. “Beginning in 1924, the
lation for a hydrographic survey of Europe at the time, and it was the defects of the Bonne Projection be-
the Bay of Diego Suarez was com- basis of the Depot de la Guerre map came totally inconvenient for use on
menced by Colonial Engineer M. series of France. The quadrant sys- the first part in geodetic work for
Grégoire in 1887. The baseline was tem was used for the topographic the calculation of geodetic coordi-

CLIFFORD J. MUGNIER, C.P., C.M.S.

142 Fe b r u a r y 20 0 0 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING


C O L U M N
nates on the ellipsoid, and on the
other part the rectangular topo-
graphic surveys accumulated a sub-
stantial error. In 1925, regions quite
removed from the Bonne Axis of
projection, notably the delta region
of Mangoky in the southwest part of
Madagascar, attained errors of 22'
for angles and ±1/580 for lengths. It
was decided to find another solu-
tion urgently so that the quality
would not be compromised for the
survey campaign of 1925. Pressed
for time, a local provisional projec-
tion was adopted in which the dis-
tortions were practically negligible
within the region of application.
Furthermore, difficulty was encoun-
tered in recruiting specialized per-
sonnel, and in 1926, the triangula-
tion operations were constrained for
lack of officers. Since training had
to be furnished to the personnel be-
cause of the isolation of the survey
parties do to the topography; the
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
○ ○
Grids & Datums
field calculations had to be simple
to employ and therefore rectangular
coordinates were used. Since the
Bonne Projection was useless, provi-
sional local coordinates systems
were improvised. Between 1925-26
four local projections were put into
service based on the Gauss Confor-
mal Projection. (Ed. note: The ac-
tual math model used for all four
was the Gauss-Schreiber Transverse
Mercator.) The Laborde Projection,
adopted in 1926, was put into full
use for the reduction of the calcula-
tions of the 1926 geodetic survey
campaign and for the topographic
survey campaign of 1927.”
The Laborde Projection is a triple
conformal projection, designed to
have two lines of zero scale error
oblique to the meridian. The pro-
jection of the ellipsoid onto the
plane is carried out in three steps;
the first is the projection of the el-
lipsoid on the sphere. The second
○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○
step is a Gauss-Schreiber Transverse
Mercator projection of the sphere
onto a cylinder that is symmetri-
cally secant about the central merid-
ian of the projection. The third step
in forming the Laborde Projection is
a conformal distortion of the plane
through a rotation in the secant
case. Beginning with the first step
(the conformal projection of the el-
lipsoid on the sphere), Laborde fol-
lowed Gauss. Gauss showed that
the ellipsoid could be mapped con-
formally on the sphere in such a
way that the distortions in the neigh-
borhood of some chosen point are of
the third order or higher. Also note
that the indirect or double projection
through the sphere is not exactly
the same as the direct one in that
they differ concerning the higher or-
der terms of the expanded transfor-
mation formulae. The conformal
projection of the ellipsoid on the
CONTINUED ON PAGE 144
Fe b r u a r y 20 0 0 143
 ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○

Grids & Datums C O L U M N

CONTINUED FROM PAGE 143 (on the complex plane) with the 16' 52.6884 East of Greenwich, East-
sphere was used by the Prussian UTM as a unifying tool. Laborde ing (Y) = 174896.590 m, Northing
Land Survey in 1876-1923 as the did this back in the 1920’s, and the (X) = 644433.000 m, (H = 31.421 m).
foundation for a double projection, Russians do it in an attempt to ob- Back in September of 1992, the late
and it is currently used in computa- fuscate the relation between their John P. Snyder asked for a copy of my
tions for cadastral retracement. “System 42 Datum” (Pulkovo origin) 1982 stuff on the Laborde Projection of
This triple projection of Laborde’s, and the WGS 84 Datum. The Rus- Madagascar. After John diddled a
using the Gaussian Sphere to sians still do it in the 21 st century while with trying to see how close the
achieve an Oblique Mercator, is fun- by using their screwy “System 1963” Hotine Rectified Skew Orthomor-
damentally different from the meth- series of local coordinates which are phic (RSO) Oblique Mercator would
od used some 20+ years later by merely translated and rotated fit, he concluded that he could
Brigadier Martin Hotine of the UK. Gauss-Krüger Grids. However, match my transformations to only
Most commercial software houses Laborde’s implementation for Mada- “about 0.08 meters for my part of
that I have noticed are abysmally ig- gascar was successful in simplifying Madagascar … worse of course away
norant on this point. (I’ll grouse on the task! from the island.” This approxima-
this later.) Laborde’s command of the Ser- tion is of interest for U.S. Geologi-
In reference to the four provi- vice Géographique de Madagascar cal Survey medium scale cartogra-
sional coordinate systems men- also resulted in the establishment of phic applications, but it is useless
tioned earlier by Laborde, he later the unifying coordinate system for for geodetic work and for large scale
commented on the relation of these the colony. After long observation, engineering mapping. I have seen
local systems to the Laborde Projec- the coordinates of the observatory in examples of much (apparent) hand
tion. His discussion referred to the the capital of Antananarivo were wringing done by software organiza-
1925-1926 local triangulation situ- used as the origin of the Madagas- tions (U.S. and elsewhere), that
ated in the region of Bemolanga- car Datum of 1925 where: have attempted to foist this sort of
Morafenobe. The work was carried F o = -18° 55' 02.10" South, “work-around” on government agen-
out over 4,200 km 2 , and was com- Lo = 47° 33' 06.45" cies in Madagascar. The Grid is
prised of 127 geodetic triangulation East of Greenwich. Of course, the commonly used at great distances
points calculated on a provisional ellipsoid of reference is the Interna- from the mainland because of the
Gauss-Schreiber Transverse Merca- tional 1924. The defining param- existence and official publication of
tor coordinate system. Two points eters were published by the retired a False Easting at False Origin!
of an old reconnaissance triangula- General Jean Laborde in the seminal The RSO is not a valid substitute for
tion were chosen as a base for a lo- work, “Traitè des Projections des the Laborde of Madagascar.
cal grid. This grid was “une projec- Cartes Géographiques a L’Usage des The U.S. National Imagery and
tion conforme de Gauss limitée,” a Cartographes et des Geodèsiens,” by Mapping Agency (NIMA) publishes
Gauss conformal projection of lim- L. Driencourt & J. Laborde, Paris, a parameter set for transforming
ited extent. Laborde further com- 1932. Remember Driencourt? He from the Madagascar Datum of 1925
mented that the transformation from performed the 1904 final astronomi- (“Tananarive 1925”) to WGS 84 Da-
the provisional system to the cal observations for the longitude of tum. However, they offer no infor-
Laborde Oblique Mercator was one Antsirana Datum of 1906. mation on accuracy of, or the origin
translation, one rotation, and a Driencourt later became the Hydrog- of, the parameters. I tried a nine
change of scale without any defor- rapher of the French Navy (long be- point solution and I obtained slight-
mation, “sans acune déformation.” fore the four volume book was pub- ly different results as:
This is an implementation of one of lished). Those parameters are: DX = -191.745 m, DY = -226.365 m,
the most valuable properties of con- Latitude of Origin (f o) = - 21 G South and DZ = -115.609 m. The resultant
formal projection: the ability to (-18° 54'), Central Meridian rms for Latitude = 0.215 m, for Lon-
maintain the undistorted shape of (lo) = 21G East of Paris (46° 26' 13.95" gitude = 0.323 m, and for ellipsoid
surveys and mappings through East of Greenwich), Scale Factor at height (h) = 9.958 m. The WGS 84
simple 3-parameter (translation in Origin (m o) = 0.9995, False Easting Datum observed coordinates for
X, translation in Y, and rotation) = 400 km, False Northing = 800 km, Morondava are: f = - 20° 17'
transformations! In my past column Azimuth of Axis of Strength (q) 39.079425" South, l = 44° 16'
on Belgium, I discussed how the = 21 G, and the False Easting at the 51.545430 East of Greenwich, and
U.S. Army Map Service imple- False Origin = 1,000 km. For exam- h = 31.421 meters.
mented the European Datum 1950 by ple, for geodetic station Morondava, ............
such 3-parameter transformations f = -20° 17' 34.2443" South, l = 44°

144 Februar y 20 0 0 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING