264
J. Populus et al.
Their attribution and management is run by Affaires Maritimes, a body within
the Ministry of Transport. Until recently, this job was still performed by hand, i.e.
graphical corrections on paper maps. The whole system was traditionally georeferenced to a local coordinate system centered on the spire of the church of the town
of Marennes. Although most sheets overlap correctly, there has never been any control
of the consistency of the whole structure. The system was
viable. With
the advent of its digitization and the prospects thereby raised, the question of an
accurate geo-referencing could no more be avoided.
The digitization of the 22000 parcels was contracted to a chartered surveyor,
who delivered the work under the dwg (Autocad) format containing the following
layers: parcels, channels,
(banks), coastline and miscellaneous. The sheet and
parcel numbers were attached to each parcel label point. Later, an ArcInfo routine had
to be written in order to transform the polylines and polygons into a topological
structure and to add to each parcel its unique number (a concatenation of the above
two). A comprehensive work of polygon cleaning, dangle correction and other trouble
shooting had to be carried out to end up with a reliable cadastre structure under
ArcView.
GEO-REFERENCING
In order to build a transform polynomial, some conspicuous points known in both
coordinates systems had to be found. The first operation consisted of assessing the
positional accuracy of various church spires, and conspicuous marks surrounding the
basin, which were likely to have been used for positioning. Discrepancies of up to 150
metres were found to exist between distances computed in local coordinates and in the
Lambert projection. As more points were necessary, a field survey was organized in
order to dGPS the few additional marks appearing on some sheets. An accuracy of 2
metres RMS was obtained, typical of a base station not farther than 20 km. The basin
was split in 4 zones where different polynomials were applied using surrounding tiepoints. Table 1 shows a set of five points in the NE zone which exhibits an average
shift of 18.5 metres. The dispersion of 5.4 metres shows that the positions within the
local referential are not consistent. This will obviously generate a distortion on the
final positions computed by the Helmert polynomial.
However, this is probably the best achievable result using global polynomials
based on only a few points. The alternative would be to work on a sheet basis, in which
case 3 points per sheet would be necessary, a great deal of work. The lack of
conspicuous points (the corner of a lease on the map having no physical existence in
the field) would in all cases limit the accuracy.
BANK MAPPING
Parcels in the field are grouped into entities referred to as
which are both
physical units and management units, i.e. under the supervision of a bank committee.
It is therefore quite a relevant geographical entity to be stored in the database. The way
J. Populus et al.
Their attribution and management is run by Affaires Maritimes, a body within
the Ministry of Transport. Until recently, this job was still performed by hand, i.e.
graphical corrections on paper maps. The whole system was traditionally georeferenced to a local coordinate system centered on the spire of the church of the town
of Marennes. Although most sheets overlap correctly, there has never been any control
of the consistency of the whole structure. The system was
viable. With
the advent of its digitization and the prospects thereby raised, the question of an
accurate geo-referencing could no more be avoided.
The digitization of the 22000 parcels was contracted to a chartered surveyor,
who delivered the work under the dwg (Autocad) format containing the following
layers: parcels, channels,
(banks), coastline and miscellaneous. The sheet and
parcel numbers were attached to each parcel label point. Later, an ArcInfo routine had
to be written in order to transform the polylines and polygons into a topological
structure and to add to each parcel its unique number (a concatenation of the above
two). A comprehensive work of polygon cleaning, dangle correction and other trouble
shooting had to be carried out to end up with a reliable cadastre structure under
ArcView.
GEO-REFERENCING
In order to build a transform polynomial, some conspicuous points known in both
coordinates systems had to be found. The first operation consisted of assessing the
positional accuracy of various church spires, and conspicuous marks surrounding the
basin, which were likely to have been used for positioning. Discrepancies of up to 150
metres were found to exist between distances computed in local coordinates and in the
Lambert projection. As more points were necessary, a field survey was organized in
order to dGPS the few additional marks appearing on some sheets. An accuracy of 2
metres RMS was obtained, typical of a base station not farther than 20 km. The basin
was split in 4 zones where different polynomials were applied using surrounding tiepoints. Table 1 shows a set of five points in the NE zone which exhibits an average
shift of 18.5 metres. The dispersion of 5.4 metres shows that the positions within the
local referential are not consistent. This will obviously generate a distortion on the
final positions computed by the Helmert polynomial.
However, this is probably the best achievable result using global polynomials
based on only a few points. The alternative would be to work on a sheet basis, in which
case 3 points per sheet would be necessary, a great deal of work. The lack of
conspicuous points (the corner of a lease on the map having no physical existence in
the field) would in all cases limit the accuracy.
BANK MAPPING
Parcels in the field are grouped into entities referred to as
which are both
physical units and management units, i.e. under the supervision of a bank committee.
It is therefore quite a relevant geographical entity to be stored in the database. The way