ul 2004
altern
)
Ince
)
Weibel
. These
oil map
per). In
ail any
il map
ased on
cs and
sses the
in the
Drocess,
aluation
tion
depend
n of the
factors
such as
graphic
map is
larity of
gion. In
arity of
rizontal
vertical
zonal
that are
d large
ribution
that is
;»hology,
human
n of soil
ities are
cale ar
ould be
1e could
middle-
ore than
2 should
map. If
for the
nning of
between
; used in
maller is
nce, and
)000 are
cribed in
with the
il maps
scale and
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
small-scale soil maps could describe the soil group or
soil subgroup (Huang et al. 1989).
- Constraint 4: the level of soil categories described in
the objective soil map should meet the requirements
of the purpose of the objective map. For example, a
soil map used in the planning of anthropogenic soil
should describe soil categories in detailed level such
as soil genus. The soil map used as reference of the
partition for the agricultural districts could just
describe coarse soil categories such as the soil group
or soil subgroup.
(2) Database Generalization
One of the main steps in database generalization is
attribute transformation. For soil data generalization, the
attribute transformation is to transform soil categories of
soil parcels from the lower level to the higher level. A
special representation, soil complex, is used to represent
soil distribution in some special cases. For example,
various soil genus intermix in a small geographical region
makes their boundaries impossible to be identified even in
large-scale map. Therefore, soil complex is used to
represent the complex components in one single soil parcel
with the associated soil types and their area percents are
recorded too. The premise of the soil complex
representation is the area percent of each soil type is not
more than 15% of the total area of the soil parcel.
- Constraint 5: emphasis the soil categories closely
relate to the purpose of the objective map. For
example, if the objective map will be used in the
planning of anthropogenic soil, the anthropogenic
soil parcels could be described in more detailed
categories than other kinds of soil parcels.
- . Constraint 6: one-way transformation. The category
of soil only can be transformed from the lower level
to the higher one in CST.
- Constraint 7: semantic consistency between the new
category and the old one of a soil parcel. That is to
say, the soil genera “Siallitic lateritic red earths”
should be converted into the soil group “Lateritic red
earths” rather than the soil group “latosols”.
- Constraint 8: category transformation of soil
complex. A soil complex involves more than one soil
category. Firstly, transform these soil categories to
their corresponding higher categories. After
transformation, the number of the soil categories in
the soil complex maybe reduced because some
categories may belong to a same higher category.
Secondly, recalculate the area percent for each
unique soil category. Thirdly, if the area percent of
one category is more than 75%, the soil complex
should be converted into a simple soil parcel with the
soil category.
(3) Graphic Generalization
Graphic generalization solves graphical conflicts caused by
the reduced space on a map or simplifies the graphic
information by eliminating the unimportant or unnecessary
details.
= Constraint 9: Preserve distribution pattern of soil
zonal discontinuity in large-scale or middle-scale
soil map.
- Constraint 10: Preserve the characteristics of
regularly continuous distribution in small-scale map.
207
- . Constraint 11: soil complex representation. In our
study, we employ soil complex representation to solve
some graphical conflicts caused by small-area soil
parcels which are adjacent each other and the sizes are
smaller than the minimal area. The soil types of these
soil parcels are associated in the soil complex and the
area percent of each soil type is recorded too.
- Constraint 12: different simplification operators
emploved on parcels of natural soil and
anthropogenic soil. Anthropogenic soil, such as
siltigic soil, athrostragic soil, paddy soil, is formed in
the process of long term utilization and reclamation
by human (Gong et al. 1997). Generally, the
boundaries of natural soil parcels are more complex
than that of anthropogenic soil. They should be
simplified with different simplification operators to
preserve the difference.
(4) Quality evaluation of the generalization results
Generally, constraints are involved in more than one phase in
the whole process of generalization. And some constraints (e.g.
minimum size, shape/angularity, alignment/pattern) are not only
used to measure conflicts and select operators, but also evaluate
the result after operators running. Most of the thematic
constraints and generic constraints are also used as the
standards to evaluate the validity of the generalization results
(Figure 2).
3. INTEGRATION OF THEMATIC CONSTRAINTS
WITH GENERIC CONSTRAINTS
Most constraints are contextually related and affect each
other, and they should be integrated into the whole
generalization process (Peter and Weibel 1999). Figure 1
demonstrates the integration of thematic constraints with
generic constraints in the different phrases of the generalization
process. The both kinds of constraints steer the generalization
operations in different levels and views. Thematic constraints
(e.g. Constrain 1, 2, 3, 4, 5, 9, 10) mainly portray the status or
characteristics of soil map which should be reached or
preserved during generalization, whereas generic constraints
focus on some concrete graphical characteristics. As a whole,
thematic constraint is the foundation and involved in every
phase of the generalization process. Some generic constraints
need to set the final value of parameters of relative operators
based on thematic constraints. Meanwhile, the generic
constraints support thematic constraints by adjusting
appropriate operators and parameters.
(1) Preparation and preprocess
The objective scale and the levels of categories represented in
the objective soil map are set according to Constraint 1, 2, 3, 4.
Then the distribution pattern of soil which should be
represented in the objective scale need to be explored and
designated based on Constraint 9, 10. After that, the minimal
size (e.g. minimal area), minimal distance can be defined and
the size ratio of each category, typical shape/angularity and
alignment/pattern need to be calculated and measured.
Generally, the minimal area of soil parcels belonging to detailed
categories is smaller than that belonging to coarse categories
(Weng 1997). In large-scale maps, the minimal area should be
smaller for the preservation of soil zonal discontinuity, but in
small-scale maps, the minimal area is larger to represent the
characteristics of regularly continuous distribution by
eliminating the zonal small parcels.
