International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
7. DEFINING REALISTIC
REQUIREMENTS FOR THE FUTURE
It is important to set tolerances for absolute and relative accuracy
high enough to avoid that we should do over more than a small
fraction of the measurements for the sake of accuracy. This means
that we have to forbid the operators to replace data that are but
inaccurate by less than 3.290 (reliability 99.9%). oy and 6, are
still to be determined by the geodesy department. Having 22.7
millions of objects, even then there will be 22 700 objects to be
replaced for the sake of accuracy. Stereo-operators should
understand that the latest aerotriangulation is not necessarily
better than the previous one.
In practice the number of possible scenarios for updating is
limited. We can either quickly update a limited number of objects
or slowly update (nearly) all objects, or we can look for a
combination. Depending on this choice, the need for personnel
and means in the different services will take different proportions.
The present field teams need at least 15 years to completely revise
the whole country. It is not an option to adopt a 15 years’
updating cycle. Several studies show that users would like to
receive updates of roads every year or even every half year,
updates of buildings every 2 years, and updates of the other
objects every 5 years. In our situation, it would be more realistic
to adapt the updating cycle to the limitations in aerial
photography: due to the atmospheric conditions, we may be happy
if we succeed in covering the whole country with aerial
photographs every 3 years. It needs no study to know that with the
present production capacity we cannot update all objects within
3 years. On the other hand, 6 years is clearly not acceptable for a
first update of the roads. Therefore we shall examine whether
public roads, buildings and perhaps some other very important
objects may eventually be updated every 3 years, and the other
important objects every 6 years: Table 1 shows that apart from
roads and buildings, the presence and position of big rivers and
channels, railroads, high tension lines and woods are considered
as very important information. All this information can be updated
by stereo interpretation. Hence these objects are ‘natural
candidates’ for a quick revision. The only very important
information that needs to be gathered by other means is the
function of some buildings (factories, supermarkets, churches,
town halls, hospitals, schools). In this scenario we give priority to
the updating of very important and important objects. The field
teams are only available for updating the less important objects
when they are not needed elsewhere.
For assuring a constant map production level at long term, it
would be useful to maintain a balance between a (very) quick
revision of all (very) important objects in some areas and a slow
but complete revision in some other areas.
8. CONFRONTATION WITH THE
AVAILABLE MEANS FOR UPDATING
As mentioned in 3.1, it will not be possible to adapt all data
within a reasonable time to the specifications that are being used
at the moment. If we want to perform the first updating of the
important objects in a reasonable time we shall have to
temporarily accept a certain amount of heterogeneity in the data.
For the less important objects, that will be updated in a very long
cycle, we need to strive for homogeneity from the beginning, so
that at the end of the first complete revision all data may be
homogeneous, provided that we make the important objects
homogeneous during their second or third updating cycle.
Up to now, only very few arcas were partially updated for the
most important objects that are visible on aerial photographs
(about 700 km?). From this we learned that in practice a first
partial updating by stereoplotting of all major objects is about 4
times quicker than the original datacollection and, if restricted to
all roads and buildings it is even 6 times quicker. This means that
our stereoplotting capacity is big enough for updating the roads
and the buildings in 2'7 years: by the end of 2008 we can have
up-to-date data for all roads and buildings in the country, as
defined in the requirements. On the other hand, we estimate that
updating all other ‘important’ objects by stereoplotting would take
at least eight months for half of the country. This means that in
the first 3 years, we cannot succeed in maintaining a constant
map production covering half of the country along with the
updating of all buildings and roads, without using the additional
surveying capacity.
In the best case we may free a stereoplotters' capacity of 6 months
by updating all roads by surveying, after having detected the
areas to be measured by remote sensing. This would leave us a
whole year for updating the remaining ‘important’ objects in half
of the country. The remaining surveying capacity may even be
used for keeping the roads up-to-date in a shorter cycle,
depending on the availability of satellite images. This programme
may be a little bit too ambitious, but it is not very far away from
feasibility either.
If the surveyors are needed for other tasks so that we cannot count
on them for measuring roads, we may be forced to postpone the
updating of some'important" object types to the years 2009 —
2011. We expect the external data to become more important
during (and after) this period.
scen | Objects | 2006 | 2007 | 2008 | 2009 | 2010 | 2011
R
B
VI
A9SAInS XP]JA
]
O
R
B
VI
| Koamns ur
I
I
O
R
B
V
]
c KoAns utjq
O
Figure 2. R = roads, B = buildings, VI = very important objects,
I = important objects, O = other objects;
different colours represent different updating cycles.
Another possible scenario for 2006 — 2008 is of course
maintaining a constant map production by increasing the
‘important’ objects at the expense of making the first updating
cycle for roads and buildings a few months longer. However, this
would not reduce the long updating cycle of the remaining objects
that have to be collected in the field. The latter can only be
reduced by enlarging the expensive field teams and by finding
new sources for a part of the information. Which scenario NGI
will follow has not been decided yet. We do not recommend to
slightly increase the stereoplotting capacity for only a short period
due to the considerable training efforts that this would imply.
746
Intern
Thee
in fur
the u
pencc
classi
will |
techn
shoul
natur
Wem
deterı
photo
z-valı
espec
the fu
of wc
invest
capac
needs
of th
proce
Badai
geogr
In: J
Confe
Beyer
topog
Interr
Ather
Beyer
with /
In: 5
Sensi
Galett
struct
OEEF
pp. 8:
Galet
topog
Photo
30. P:
Heipk
Inforr
Photo
USA,
ISO 1
Kadri-
bases
des m
Franc
