maintain a drainage system per lot and he will decide per
growing season how to partition each lot into fields. These
lots might both belong to a superclass "farm lots” in a land
use data base and these again might belong to an even
higher superclass "lot" which also contains the classes
"forestlot" and "residencelot". The aggregation step from
level 1 to 2 and the next steps to the levels 3 and 4
where we have the farms and farm districts show that
after each step new objects are created. At farm level the
farmer will decide whether he will be a cattle farmer or
whether he will grow arable crops, in the latter case he
has to decide on a rotation scheme. At district level the
infrastructure and irrigation schemes will be developed.
The objects at each level have their own thematic
description expressed in an attribute structure that should
be defined in a class hierarchy according to section 2. In
this example each aggregation level requires its own
classification hierarchy. This should be structured so that
the generated attribute structures provide the information
to support the management operations defined at the
aggregation levels of the objects. The diagram of figure
8 represents the fact that a classification hierarchy should
be defined per aggregation level.
CONTEXT 1 CONTEXT 2 CONTEXT 3 CONTEXT 4
|
suot d
SE SUP.CL
AGGR. LEVEL 4
AGGR. LEVEL 3
M
fm im
fig. 8: Classification hierarchies related to aggregation
levels.
This is an example of a more general situation where
objects at each aggregation level are functional units with
respect to some process. In this case these were farm
management processes, but we could also take examples
like ecologic development, or demography and many more.
Each aggregation level within such a hierarchy will have
its own (sub) context within a thematic field, expressed
through a class hierarchy with related attribute structures.
The different (sub) contexts are related by the fact that
sets of objects at one level can be aggregated to form the
objects at the next higher level. There are often also
relationships between various classes ofthe different class
hierarchies related to the aggregation levels as was the
case for the cover classes for the farm lots, the farm types
and land use types of the farm districts at the levels 2,3
and 4 of figure 7.
There are bottom-up relationships between the objects
at different levels in the sense that the state information
of the lowest level objects, as contained in the attribute
data, can be transferred through a process like figure 6,
to givestate information about the objects at higher levels.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
There are top-down relationships in the sense that the
behaviour of lower level objects will be constrained by the
information contained in the higher level objects.
3.3. Structural Object Generalization
This strategy will be explained by means of an example
based on a database where the spatial description at a
1:50.000 scale, of a drainage system. The database has
been structured according the FDS as in section 3, see
(Martinez Casasnovas 1994). A generalization process will
be executed to derive the 1:100.000 representation, so
that wereduce complexity to stress spatial structure. Here
the spatial structure refers to the network structure of the
drainage system in relation to the subcatchments. The
generalization process will keep the area of the aggregated
subcatchments and the network structure of the system
invariant so that the computation of overland water flows
per node in the network will not be effected significantly.
The database contains geometric data and thematic data
of the elements of the system, as defined in the example
of section 2.1. Let the attribute ORDER contain the Strahler
order number of each drainage element. These numbers
in combination with the function Upstr[ ] make it possible
to analyze the stream network built by the drainage
elements. Through this network aggregation steps can be
defined for the catchment areas. The methodology for these
aggregation steps will follow to a great extend procedures
defined by (Richardson 1993 and 1994).
The process starts with the identification of the drainage
elements that are not mappable at the target scale, those
are the elements with Strahler number — 1 with an average
width aw < Thr(eshold). The minimum mapping width
of the drainage elements will be put at 0.75mm, that gives
a threshold 7hr = 0.75mmvscale at terrain scale, hence
Thr 2 75min this case. The average width for an drainage
element D; can be computed from the AREA, of the element
and the LENGTH, of its water line W, hence
aw; = AREA, /LENGTH.
The selection procedure applied to the drainage elements
is then
> select the drainage elements D, with ORDER, > 1
> select from the class with D, = 1 the elements
D; with aw; = Thr
The set of elements that should be eliminated is then
S =D, ORDER = 7, aw; € Thr],
their catchments should be combined with adjacent
catchments to form aggregates. The elimination of the
drainage elements D; € S should consist of the following
steps
> eliminate W,
> AGGRID,, C;) = Con
> find C, for which Upstr[C,, C;] = 1
2 AGGR(Cos; 0,1] 9 C;
where the notation Cy, means that the area of Di has been
merged into the area of its subcatchment, the notation
C,,, means that the area of Cy,; has been merged into the
area of C,. When water line W, joins the outlet point END(W,
) with only one water line W; of a drainage element that
550
will n
be me
> fin
/f
and
then
> AG
> AG
Then
and C
these
new S
eleme
Then !
until n
A Tes!
The dr
as an ¢
This fi
Draine
(Martii
age sy
mappe
water |
D; are
fig.
The tr
151090.
proced
with S
ated, t
stream
are recl
elemer
The re:
drainac
nine m
aggreg
scale. 7
catchrr
original
gates.
The Rc
