nation, spatial data 
e underlying context 
ontext in the whole 
context. 
of a context. 
"a context. Both super 
ng the node/context 
ich are in the database 
ative names of classes 
nation about the data 
value which indicates 
e distribution options 
'y FGDC. A Boolean 
ita exist or not. 
ters of the database 
Joolean value which 
Ot. 
. types offered by the 
  
    
    
  
  
    
  
  
r quality 
zt. ; 
social Economic 
lysis Management 
onal National 
VSM 
zn 
ources 
y 
cates if the underlying 
her classes. 
oolean value which 
s the mechanism used 
the data set. It is a 
iere exist metadata 
atershed management, 
lves three types of 
is to prepare a concise 
itershed; 2) analysis: 
n order to assess their 
antify the causes of 
the intention is to 
cement plans. These 
'roups which can be 
responding to these 
gure 3. At the lowest 
y objects. Elementary 
1 abstraction hierarchy 
    
of a particular database. At the highest level of abstraction and 
corresponding to the three main activities of watershed 
management there are three different views. The views are 
further abstracted to other decision levels, e.g., local, regional 
and national 
e At the monitoring hierarchy elementary objects are 
interpreted as hydrologic response units in the context of 
land degradation analysis. They will be abstracted to 
subcatchments, catchments, and basins at the local, 
regional, and national levels respectively. 
e At the analysis hierarchy elementary objects will be 
abstracted to tessellation. These are the processing units for 
the simulation models which are used for analysing 
watersheds’ degradation and also the impact of the new 
management practices. The tessellation units can be cells, 
Management units Catchments and Processing 
(MANAGEMENT) subcatchments units 
(MONITORING) (ANALYSIS) 
«e. 
| Abstraction Phase 2 
Meteorology 
| Abstraction Phase1 | 
Real world 
Figure 3 Abstraction of elementary objects to the three hierarchies 
Abstraction to 
national, regional, 
andlocal 
  
  
  
  
   
  
  
  
  
  
polygons, or triangles. 
e At the management hierarchy elementary objects will be 
abstracted into management units, farms, districts, 
provinces, etc. 
Semantics of the analysis and management context hierarchies 
are defined respectively as follows: 
sem (analysis) = < earth resources, management, 
analysis.itc.nl, regional, none, (soil, land cover, hydrology, 
relief), none, True, True, True, (water quality, Best 
management, erosion, sediment, water level}, True, True, True» 
Following is an example of a rule that defines the relationship 
between analysis context and the supper context, earth 
resources. The rule is written in Pseudo language for 
illustration. 
data type = “soil erosion” and measurement unit = *kg/ m?" 
and target = “earth resources" — data type = “soil 
degradation” and measurement unit = *pounds/in^" and get 
function unitconv 
International Archives of P 
hotogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996 
The above rule means that if the context receives information 
for soil erosion and the requester is earth resources, then 
rename the data type to soil degradation and trigger the 
function unitconv in order to transform the units of 
measurements from kg/m^ to pounds/in?. The unit of 
measurements and other information are obtained from the 
metadata associated to the underlying context. At the super- 
context, earth resource in our example, a typical rule can be 
defined as follows: 
data type = “soil degradation” and target = “management” then 
data type = “upstream erosion” and get function project. 
The above rule means that if the data type to be transferred is 
soil degradation and the target context is management then 
rename to type into upstream erosion and apply function 
project in order to propagate the values stored per cell to that 
for the whole catchement. 
The association relationship ensures a consistency in the 
retrieved data. For example if a user requests a land cover and 
soil information which are functionally related, preference will 
be given to the context which has both types rather than 
retrieving them from two different contexts. 
4.3 Semantics of Hierarchies and Classes 
A context, or a database, contains one or more aggregation 
hierarchies. Hierarchies in turn are formed by classes. The 
reason for introducing semantics of hierarchies and classes as 
one unit is that the underlying hierarchies exist in one database 
and hence are introduced in one data model. Semantics of 
hierarchies and classes are defined by the 3-tuple. 
sem (hierarchy class) — «context, list of classes, class 
hierarchy, entity and attribute information? 
Context the name of the underlying database. 
list of classes is a set of class names which are in the database. 
entity and attribute information is about the information 
content of the data set, including the entity types, their 
attribute, accuracy and domains from which attribute values 
may be assigned., as defined by FGDC. 
Following the example in section 4.2, semantics of hierarchies 
and classes are shown as follows: 
Sem (analysis) 2 «analysis, (cells, chemical output, sediment 
output,), True» 
class = “chemical output” and nitrogen » XXX then change 
land cover and get function AGNPS 
This rule means that if the nitrogen value in the class chemical 
out is larger than a certain amount then change land cover type 
and run a simulation model, AGNPS, in order to calculate the 
new value of the nitrogen. 
5. Conclusions 
Information sharing has become an active area of research in 
the last decade. Exchanging information is currently achieved 
by providing files in a standards format. In the last five years 
several research activities were focused on providing a high 
level of semantic data sharing. This is achieved by providing 
mechanism for resolving the classic three aspects of 
63