ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001
160
oriented multi-value spatial data model, using multi-dimension
analytical approach, we analyze, compare and extract the hidden
information from spatial data, and to realize the integration of
data oriented and model oriented analytical approaches. Data
warehouse means the database can incessantly update data.
Data warehouse is a new technology used in data storage,
management and processing from 90’s. It is a management
supported, decision-making, theme oriented, changed by time
and long lasting data collection. The main task of data warehouse
is make procession of standardization, filter, matching, refining,
mark time, guaranteed quality for source data of different types,
different structures, different storage formats, different contents in
distributed databases which located in different places and
companies. Then integrates, divides, summarize, aggregate,
forecast, deduct, translate, transform, and image these data of
distributed database which have different features and different
format based on need of special task. At last, we make modeling,
aggregation, adjusting, affirmation for data warehouse and build
structure query function, and etc.
Spatial data warehouse includes the abstraction of real world, the
storage management and data reorganization of collected data,
and the tool of data-mining, extraction, and analysis based on
spatial data warehouse. In general, spatial data warehouse is
composed by four parts: data source, database, spatial data
warehouse data storage system and spatial data warehouse
analytical tools. Spatial data warehouse is a special form of data
warehouse; it is one of the key technologies of Web-GIS. This
paper mainly studies some key technologies to build infinitely
variable map scale oriented spatial data warehouse, and the
function of spatial data index in it.
2.3 Storage and Management of Internet-based Distributed
Geographic Data Warehouse
Mass, multi-source and multi-dimension spatial data brings
forward a new problem about the management and application of
spatial data. Traditional database has too many problems to deal
with integration spatial data in decision, theme, time variation and
permanence. Facing multiple, heterogeneous features and
spatio-temporal features of spatial data, spatial data warehouse
provides efficient approach to manage them.
Spatial data warehouse technology of Digital Earth provides firm
basis for building of macro-sophisticated decision support system.
We can simplify the design and development of system in logic
and application by using it to build subsystems or functional
system in Digital Earth, and make it very clear in logic. Therefore,
we can provide feasible data management scheme for system
application in Digital Earth.
2.4 Seamless Integration of Multi-source Data (SIMS)
The rapid development and extensive application of GIS
accumulate mass data source. They were stored into different
data format, and these differences make it very inconvenient to
use the data. Seamless Integration of Multi-source Data
technology realizes a special data access mechanism. It is not
only provides the ability to access different format data directly,
but also make GIS software has composite analytical function in
different data source.
SIMS is an advanced spatial data integrate technology that do
not need data format transform, and access multi-format data
directly. SIMS features are listed below:
• Direct accessing of multi-format data. This is the basic
function of SIMS. Because of avoiding transformation of
data format, it is so convenient to use different format data
source.
• Format independent data integration. When GIS user using
data, they may not care the storage format of data, realize
the format independent data integration.
• Position independent data integration. If we use large
relational database (e.p. Oracle and SQL Server) to store
spatial data, the data can be stored in Net Server, even
Web Server. If we use file to store spatial data, the data
usually located in local server. Accessing data through
SIMS, we may not care not only the storage format of data,
but also the storage position of data. User can operate
remote data conveniently as local data.
• Composite analysis of multi-source data. SIMS also allows
composite spatial analysis of multi-source data. For
example, user can overlap a land using data set of Arc/Info
Coverage format on an administrative divisions data set
stored in SDE.
2.5 Spatial Database Index
In digital Earth, mass geographic data change with time, and
appear diversity in spatio-temporal feature. The information
generalization of infinitely variable map scale spatial data also
make these mass data appear diversity with the continuous
change of scale. Especially in infinitely variable map scale
oriented spatial data warehouse, we much maintain the union of
all scales and all themes data set. Thus, there are mass data.
Traditional spatial data index is based on the static index of
hierarchical model, and these indexes are all derived from
traditional database index. This method can only improve the
efficiency of data searching. Thus, we must use new approach to
build spatial data index.
This new approach should have dynamic and complete features.
Dynamic means we can auto generate index for time dimension,
geographic dimension and theme dimension based on the
change of granularity in spatial data warehouse. Complete is to
build special spatial relations of spatial data based on spatial data
index, it mainly include topological relationship, geometrical
relationship and ordinal relation. Dynamic spatial data index
provides better support for online analysis and datamining,
especially provides a new approach to information generalization
of infinitely variable map scale.
2.6 Spatial Data Engine
Because Digital Earth mainly related to multi-dimension
spatio-temporal data, the traditional data warehouse and
datamining technology cannot satisfy this need. We should
organize and manage data unitedly through spatial data engine
based on spatial database, data warehouse and datamining to
provides high performed searching method and improves
analyze and aid decision ability.
Spatial data engine can support spatial data warehouse in spatial
data organization and data searching. Through research of
spatial data engine, we can improve searching speed to mass
spatial information, optimize searching result in infinitely variable
map scale information generalization, and return searching result
to end user through a friendly interface.
Concept model of infinitely variable map scale oriented spatial
data warehouse is described in figure 1.