International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
SOAP
SSL | WSDL | LE XML, BEPL
= rs = SER E - ET po ee]
HTTP | XML UDDI HTTP XML |
Security Service Service Service Modeling &
Stack Description Stack Discovery Stack Binding Workflow
: Stack Staci
(Publish) (Find) Stack
(Bind) (Chain)
Figure 2. The major web service standards
6. DISTRIBUTED GEOSPATIAL SERVICES
Geospatial services are the services that handle the geospatial
data and information (Di, 2004a). There are three major types
of geospatial services: The data services, the value-add
processing/client services, and the brokerage services. The data
services provide the data to requestors. In the service concept, a
requestor may be an application client that interacts with the
human users directly or may be a middleware service that
provides processing services to the data.
The distributed geospatial services under SOA have been
discussed by Di and McDonald 1999, Di 2004a, and OGC
(Lieberman, 2003). First, we consider a granule of
geoinformation (either a dataset, a query result, or
geocomputation output which describes some aspects of Earth)
to be a geo-object, which consists of data itself, a set of
attributes (metadata), and associations with a set of methods
(transformation and creation methods) that can operate on it. A
geo-object stored at a data center is an archived geo-object. All
geoinformation and knowledge products are derived from
archived geo-objects. Thus, from object point of view, all
processes for geo-information/knowledge discovery are the
processes of creating new geo-objects from existing ones.
If we consider a user request is a user-defined geo-object, or
called user geo-object, the object is either an archived geo-
object in a data archive or can be derived by executing geo-
processing algorithm (e.g., unsupervised classification) with a
set of input geo-objects. An input geo-object, if not exists in an
archive, can be further derived. by executing a geo-processing
algorithm with a set of input geo-objects and so on. The
decomposition process will construct a geospatial process
workflow tree, which we call a geo-tree. The construction of a
geo-tree is a geospatial modeling process; and the geo-tree
itself is a geospatial model that contains the knowledge of a
specific application domain. With the geo-tree, we know how to
produce the user-object although the object does not really exist
in any archives. We call such geo-object the virtual geo-object.
In fact, any sub-tree in the geo-tree is a virtual geo-object. Since
a geo-tree only captures the workflow, not a specific product, it
represents a type of geo-objects that it can produce, not an
instance (an individual dataset). The virtual geo-object can be
materialized on-demand for users when we have all required
methods and inputs available. When user requests such an
object, the user has to specify the geographic location, time,
format, etc. Those specifications will instantiate the virtual geo-
object. By propagating the specifications down to each node of
the geo-tree, the whole geo-tree is instantiated. This process is
called instantiation of geo-tree. Only by doing the instantiation
we can know if the virtual geo-object can be materialized
because in many cases the required archival geo-objects may be
not available for the users-specified geographic region and
conditions. After the instantiation, the geo-tree is executable in
the system and the virtual geo-object can be produced. The
production process is called the materialization of virtual geo-
object, which will produce an instance of the virtual geo-object.
The geo-object and geo-tree concepts can be implemented
under the web service framework for automatically deriving
user-requested geospatial information and knowledge. The geo-
processing algorithm (method) in each node of a geo-tree can
be implemented as a web service, which is called a geospatial
web service module. The algorithm may only take care of a tiny
step of overall geo-processing or may be a large aggregated
processing. However, the service should be well defined, and
has a clear input and output requirements, and can be executed
independently. All those service modules can be reused in
constructing different geospatial models. If we have enough
elementary service modules available, we can construct any
complex geospatial models by chaining those service modules
together.
A service chain is defined as a sequence of services where, for
each adjacent pair of services, occurrence of the first action is
necessary for the occurrence of the second action. When
services are chained, they are combined in a dependent series to
achieve larger tasks. From service point of view, a geo-tree is a
complex service chain. The construction of geo-tree is a
service-chaining processing. There are three types of chaining
defined in ISO 19119 (ISO, 2001):
o User-defined (transparent) — the Human user defines and
manages the chain.
o Workflow-managed (translucent)- the Human user
invokes a service that manages and controls the chain,
where the user is aware of the individual services in the
chain.
o Aggregate (opaque) — the Human user invokes a service
that carries out the chain, where the user has no awareness
of the individual services in the chain.
The geo-object and geo-tree concepts are currently being
implemented in the GeoBrain system under the web service
framework (Di, 2004a) The system uses the Open GIS
Consortium (OGC) standards for the data finding and access,
and OGC and W3C standards for the web services. It also
leverages the new and ongoing development in the knowledge
representation and management, especially the workflow
management.
GeoBrain is an open, interoperable, distributed, standard-
compliant, multi-tier web-based geospatial information services
and modeling system. This system can shorten the time required
for geospatial information extraction from weeks to just minutes
or seconds. At the back-end, the system accesses the multi-
petabytes of remote sensing and other geospatial data available
at NASA and other geospatial data centers. At the front-end,
end-users are able to find and obtain geospatial products in the
form that exactly matches their requirements so that the
products are ready for integration and analysis. In addition to
obtain the geospatial data and information with great easiness,
the system also provide an interoperability framework and
architecture, allowing end-users to develop individual
geospatial web-service modules and to use these modules and
190
Internation
those provi
executable .
system on-
geospatial d
to their sciet
proper peer
plugged int
users. the 1
types of ge
will only be
and modules
geospatial r
sharing and
evolvable ar
also makes
community
description
found at I
implementec
al., 2003; Di
7. THE GI
It is envisi
independent
over the wel
may be scatt
service mod
standards o
execution ha
one service
Therefore, s
standards a
environment
interoperable
inputs from
environment,
provider's ar
The common
finding and ;
small data
archives. Tt
value-added
different dat
about their in
The interface
OGC Web
Coverage Se:
(WFS) (Vret
Beaujardiére,
(Reich, 2001
geospatial da
format, projc
OGC WCS d
dimensional,
way. Coverag
sensing imag:
feature-based
geospatial da
interoperable
assembling m
normally con
requirements
interfaces for
