ibul 2004
process is
tantiation
terialized
ts may be
gion and
:utable in
iced. The
tual geo-
o-object.
lemented
deriving
The geo-
)-tree can
'eospatial
> of a tiny
zeregated
ined, and
executed
'eused in
e enough
truct any
modules
vhere, for
action is
n. When
t series to
)-tree is a
ree is à
“chaining
fines and
an user
he chain,
es in the
a service
\wareness
ly being
b service
pen GIS
d access,
s. It also
nowledge
workflow
standard-
1 services
> required
t minutes
he multi-
available
front-end,
cts in the
that the
{dition to
easiness,
vork and
ndividual
lules and
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
those provided by GeoBrain for constructing complex web-
executable geospatial model. By executing the model on the
system on-line against any subsets of the petabytes of
geospatial data, end-users obtain not the raw data but a solution
to their scientific questions expressed in the model. Through the
proper peer review, user-developed modules and models can be
plugged into the system operational services. To other end-
users. the newly available web-executable models represent
types of geospatial products available although the products
will only be produced when users request them. These models
and modules can be reused to construct even more complicated
geospatial models. This accumulation of knowledge through
sharing and reuse of geospatial process models make the system
evolvable and increasingly capable with time. The architecture
also makes the idea of interoperable systems developed by a
community for the community implementable. The detail
description of GeoBrain architecture and components can be
found at Di, 2004. The same concepts are also being
implemented in the Grid environment in another project (Di, et
al., 2003; Di and McDonald, 2004; Di 2004b,c).
7. THE GEOSPATIAL SERVICE STANDARDS
It is envisioned that in the near future, there are many
independent geospatial data and service providers distributed
over the web. Services needed for creating a virtual geo-object
may be scattered in multiple service providers. In order for the
service modules to work together through the service chaining,
standards on service declaration, discovery, binding, and
execution have to be developed. The geo-tree concept requires
one service's output to be the input of another service.
Therefore, standards on service chaining are required. Those
standards and interfaces provide the common
environment for service providers to deploy standard-based
interoperable web services. In addition, a service may require
inputs from multiple data providers. A common data
environment, which provides standard interfaces to the data
provider's archives, is required.
service
The common data environment is a set of standard interfaces for
finding and access data in diverse data archives, ranging from
small data providers to multiple-petabyte NASA EOS data
archives. The environment allows geospatial services and
value-added applications to access diverse data provided by
different data providers in a standard way without worrying
about their internal differences in handling the data.
The interface standards for the common data environment are
OGC Web Data Services Specifications, including Web
Coverage Services (WCS) (Evans, 200, Web Feature Services
(WFS) (Vretanos, 2002), Web Map Services (WMS) (de La
Beaujardiére, 2002), and Web Registries Services (WRS)
(Reich, 2001). The specifications allow seamless access to
geospatial data in a distributed environment, regardless of the
format, projection, resolution, and the archive location. The
OGC WCS defines web interfaces for accessing on-line multi-
dimensional, multi-temporal geospatial data in an interoperable
way. Coverage data include gridded geospatial data and remote
sensing images. OGC WFS defines web interfaces for accessing
feature-based geospatial data. WCS and WFS together cover all
geospatial data. They form the foundation for OGC web-based
interoperable data access. OGC WMS defines interfaces for
assembling maps from multi-sources over Web. A WMS server
normally converts data to a visualized form (map) based on
requirements from the requestor. OGC WRS defines web
interfaces for finding data or services from the registries. The
191
OGC technology allows requestors to specify the requirements
for the data they want. An OGC compliant server has to
preprocess the data on-demand based on the requirements and
then returns the data back to requestors in the form specified by
them.
'3C and OASIS are the leading international organization for
developing general web service standards (W3C, 2004; OASIS,
2004). In the geospatial area, Open GIS Consortium (OGC) is
the only international organization dedicated to develop
geospatial web service implementation standards based on ISO,
FGDC, INCITS, and other standard-setting organizations?
abstract or content standards as well as technology standards
developed by W3C and OASIS. OGC standards are fully tested
at various geospatial environments. OGC specifications are
widely used by geospatial communities for sharing data and
resources and are becoming ISO standards. In the geospatial
web service area, OGC is modifying and extending W3C
standards for the geospatial web services through the OGC web
service initiatives (OWS2, 2004).
8. RESEARCH ISSUES
The current implementation of service-based distributed
geospatial system is mainly concentrated on enabling the user-
defined and workflow-managed service chains for geospatial
modeling and knowledge discovery. The technology for
enabling the opaque chaining, which requires artificial
intelligence for automatically construct the service chain based
on user's query, are not mature yet. More research in this area
is needed. The key technical questions to be answered through
research include:
l.How to automatically decompose user's query (user
object) to construct the geo-tree based on distributed data
and service catalogs?
2. How to represent the geo-trees in computer understandable
and executable workflows?
3. How to manage, share, and reuse geo-trees that represent
the geospatial knowledge of a specific domain?
4. How to execute the geo-tree at the distributed web service
environment automatically to derive the product that
exactly meets the user’s query.
Answering the first question requires domain knowledge to
understand relationships and constraints among the data objects
and the services that provide analysis, manipulation,
transformation, etc., of that data. In the geo-object concept, if a
user object is not at the archive or does not exactly match users?
requirements, the geospatial information system will
automatically create the object for users from the archived geo-
objects through a set of dynamically chained services.
In the geo-knowledge discovery process, the geo-data and
information assembly services involve in data reduction and
transformation services. Such services will not change the
meaning of the data they represent. Services falling into this
category include data subsetting, subsampling. reformatting,
geometric correction, radiometric corrections, etc. Those
services are common to most geospatial analysis, data mining,
and feature extraction processes. The rules for chaining such
services together to derive users products are quite simple and
universally accepted. Therefore, fully automating the geoquery
and geo-assembly steps of geo-knowledge process is quite
possible in the near term.
