Mining Spatio-Temporal Data
Earth Systems Science
Dylan Keon,
Last modified: 25-AUG-04
Contact
Information:
Dylan Keon
541-737-6608
keon@nacse.org
NACSE's efforts have centered
around the development of interactive, Web-based visualization interfaces
capable of integrating both raster and vector GIS-based data with environmental
data derived from database queries (managed using relational or other database
management systems). We have made significant progress on integrating existing
open source code with our own custom code to create Web-based mapping
interfaces. The development of an easy-to-deploy set of dynamic mapping tools
is important, as this allows developers to easily integrate these tools within
an existing mapping application. It also allows users to take advantage of a
more intuitive set of dynamic tools within their browser, without incurring the
overhead of something like a Java applet.
NACSE's mapping products are integrated with the
Spatial Data Workbench (SDW), a meta-project that provides an integrated
collection of data management and analytical services for ecological data. The
goal of the SDW is to establish an integrated knowledge management system that
uses extensible and scalable infrastructure to provide data and analytical
services to the broad LTER Network community of field biologists and associated
collaborators working at these sites (over 1500 individuals). The SDW has
enabled access to complex datasets such as hyperspectral
remote sensing data to researchers who would not be able to use the data. The
current SDW consists of several components, including a SRB-based
collection management system for remote sensing imagery, analytical pipelines
(workflows) for high-throughput processing of remote sensing data, mining
algorithms for analysis, band selection and classification, and dynamic
Internet mapping of analysis results. These components are integrated through a
web services architecture that provides a scalable and extensible approach for
solving difficult application and database integration issues. This project
extends the impact of the evolving national cyberinfrastructure by providing a
tangible example of grid computing for the ecological community. The
architecture provides dynamic integration of resources across the distributed
LTER Network sites and the collections housed at SDSC.
This includes hyperspectral remote sensing imagery,
derived data products and both vector and raster-based GIS data layers for
spatial land-cover characteristics, and climate measurements.
Since this project's inception, we have accomplished our goal of integrating
existing open source mapping code with code we have developed to create a
dynamic mapping framework, capable of integrating not only local GIS datasets
and static remote imagery (via WMS - Web Mapping
Service) standards, but also dynamically created remote imagery, such as the
raster analysis output generated by the SDW. This real-time visualization of
analysis output imagery is provided via custom Web-based GIS applications,
which use RPC (Remote Procedure Calls) over HTTP to
communicate with the SDW. Once output imagery has been generated through the
SDW interface, the user is given the option of mapping the imagery against a
collection of base GIS layers representing various entities such as roads, site
boundaries, hydrography, and additional imagery such
as DRGs (Digital Raster Graphics) and DOQs (Digital Orthophoto
Quadrangles). This allows the user to view their custom output imagery in the
context of the corresponding LTER site. When the user chooses to map their
imagery a request is sent to the mapping application at NACSE,
which processes the request, retrieves the output imagery and georeferencing information from SDSC,
caches the image locally, integrates the image with the existing base layers in
the mapping application, and makes the modified mapping application available
to the SDW+ application at SDSC. While concepts such
as RPC over HTTP are not new, the combination of
methods used for mapping the analysis output imagery are novel within the realm
of Web-based mapping.
The mapping application was implemented in PHP
(www.php.net) and was created entirely using Open Source Software, including MapServer (mapserver.gis.umn.edu), PHP
MapScript (www2.dmsolutions.on.ca/mapserver/php_mapscript),
GDAL (www.remotesensing.org/gdal),
and additional tools. It runs on a Linux server at NACSE,
where a repository of spatial data (primarily base layers) is stored for most
LTER sites. NACSE is currently establishing a Web
Mapping Server (WMS) and will be implementing base
layers from all LTER sites as WMS layers, following OpenGIS (www.opengis.org) specifications. This will allow
anyone to remotely access these layers via web services for integration into
their own mapping applications.
The past 12 months have seen significant advances in both the scope and
functionality of the mapping interface. We developed a drop-in DHTML-based toolset that allows the user to zoom or query
the map dynamically by dragging a "rubber band" box across the image.
Previously, the user was required to simply click the map at the desired
location; the DHTML tools allow the user a much finer
degree of control over selection of desired location and extent (clicking a
point on the map still remains an option for the
user). Additionally, the DHTML tools allow the user
to query multiple features simultaneously, as defined by the bounding box they
create. This allows for more complex and meaningful spatial queries.
Concurrently, we spent a significant amount of time obtaining and processing
spatial data for a majority of LTER sites. The data were processed and reprojected, if necessary, then integrated into the
existing mapping application. Configuration files were written to represent the
spatial data available for each LTER site. The configuration files were then
"plugged in" to the existing application, which was written in an
extensible way so that it could easily accept multiple sites.
Image 1:
keon_npaci_01.png
Credit: Dylan Keon
Caption: Raster output from the Spatial Data Workbench can be seamlessly
mapped in the spatial context of the appropriate LTER site. The user can zoom,
pan, or query the map, as well as load/unload additional GIS layers such as
base imagery or vector data.
Image 2:
keon_npaci_02.png
Credit: Dylan Keon
Caption: The mapping interface now supports multiple LTER sites,
allowing the user to quickly switch among the various sites. Some of the layers
visible within selected LTER map windows are retrieved and made available as WMS (Web Mapping Service) layers, using Open GIS Consortium
specifications.
Image 3:
keon_npaci_03.png
Credit: Dylan Keon
Caption: This image demonstrates another LTER site available within the
interface. Additional DHTML functionality has been
added, which allows the user to drag a "rubber band" box across the
map image when zooming or querying, as opposed to simply clicking the map at
the desired location. When zooming, this allows finer control over the
selection of location and extent and, when querying, this allows multiple
features to be queried simultaneously.
