News

Information that has a spatial component is a core part of many scientific projects, from geology and climate to ecology and neuroscience. And the length scales of this data can range from the size of the Earth and larger down to cities, streets, and the human body, organs, cells, and even DNA. In helping build cyberinfrastructure to advance science, the Spatial Information Systems Lab at the San Diego Supercomputer Center (SDSC) at the University of California, San Diego (UCSD) is at the forefront of research and development of technologies that enable scientists to access, integrate, and manage spatial information in many forms.

The lab is extending spatial technologies beyond simply "maps in computers" to build a common canvas on which scientists can assemble an array of different kinds of information. The underlying capabilities involve integrating data from different spatial sources as well as across different scientific domains, different regulatory frameworks, and multiple scales; and then rendering the resulting information in a form that is understandable by both scientists and non-scientists for use in research, policy, and planning.

Part of SDSC's Data and Knowledge Systems (DAKS) program, the lab works in collaboration with other DAKS labs in projects such as the NSF Geosciences Network (GEON), as well as other UCSD programs including the National Center for Microscopy and Imaging Research (NCMIR), the Scripps Institution of Oceanography (SIO), and the Urban Studies and Planning (USP) program.

"The Spatial Information Systems lab is meeting a core need across many science domains, and with the other eight DAKS labs provides an unmatched set of IT capabilities," said Chaitan Baru, co-director of SDSC's Data and Knowledge Systems program. "As SDSC builds cyberinfrastructure to advance science, it's vital to bring SDSC's research and development expertise from groups like the Spatial Information Systems lab into our large-scale collaborations from BIRN in neuroscience to GEON in the geosciences."

Lab director Ilya Zaslavsky explains that "we have two interrelated roles, basic research on spatial information systems, and providing services to scientific projects." The main research focus of the lab is online visualization, analysis and integration of spatial data, geographic information systems (GIS), new technologies such as Web services using WSDL and SOAP standards, and XML-based Web mapping. "The scientific applications we work on are constantly giving us new research challenges," said Zaslavsky. "And as our research progresses, this in turn gives the scientific projects new capabilities they didn't have before." Through this process, the researchers are reformulating and extending traditional GIS and cartographic practices into more powerful tools for science.

Common Spatial Information Needs for Many Sciences

One of the most fascinating things about his lab's projects, says Zaslavsky, is just how many areas of science involve some spatial component in their data. "Everyone understands that Earth systems sciences like geology and oceanography require GIS, mapping, and georeferenced data," he explains. "But even fields like neuroscience, which you might not think would involve spatial data or mapping, have this need , and are already benefiting from applications of traditional geographic technologies."

Neuroscientists compile brain atlases, photographs of slices of the brain, which are stored as images that are often labeled or annotated. In order to advance their research, neuroscientists would like to be able to collaborate with colleagues across multiple institutions, combining different atlases and being able to search across them, for example looking for images in multiple collections that show a certain distribution of proteins in the cerebellum that may give new insights into diseases like Alzheimers.

Working with the NIH Biomedical Informatics Research Network (BIRN) project, which stores data in the SDSC Storage Resource Broker (SRB), Zaslavsky's team is helping build a system that enables them to search for brain images and display them along with other information, overlaying multiple kinds of data and images.

"These broad and powerful searches and data integration are key capabilities for today's larger collaborations," said Zaslavsky. "We're working to develop a common spatial data architecture that can be reused to support many different scientific applications." The basic elements include registering data into the collections, discovering relevant data through efficient spatial search mechanisms, and bringing multiple types of spatial data into a common environment to display, query, and process.

An example of the issues that the researchers face is finding ways to link the labels on a brain image to the image features, in order to enable neuroscientists to search the labels to find images that contain certain features. But often the image annotations are made up of drawing artifacts that do not connect into meaningful closed anatomic feature polygons, preventing any easy automated way to associate a label with its corresponding atlas feature. By developing ways to convert image markup into topologically correct spatial databases, Zaslavsky's lab is enabling scientists to search these databases of brain images for the particular images and fragments they need for their research - something that was not previously practical.

Another key area of research is spatial data integration, for example combining a map of rock distribution in one state with a map of rock distribution from another state that uses a different coordinate system. To join tabular data in a relational database, it is not difficult to relate values in one table's column to values in a column from a different table. In joining spatial data, however, in place of values in a table, researchers must deal with various shapes, which are presented in different coordinate systems. To be able to relate spatial data across different collections, the researchers have to develop ways to juxtapose the shapes from different data collections within the same coordinate space. In doing this they must address issues such as how to reconcile differences in the resolution of the different spatial data collections and differences among diverse coordinate systems.

Ontologies for Spatial Information Integration

As part of the GEON project, working with colleagues in the DAKS Knowledge-Based Information Systems (KBI) and Geoinformatics labs, Zaslavsky's lab has helped build an ontology-based geological map integration prototype that brings together in a single unified queriable map interface, the geologic maps from nine state geologic surveys in GEON's Rocky Mountain testbed region. Previously, those data sets and maps could only be accessed and queried separately, requiring far more labor and time.

Zaslavsky and lab member Ashraf Memon contributed technology to integrate map fragments retrieved from the multiple map servers. In response to user queries, the mediator transmits query fragments to the different ArcIMS servers that provide access to state geologic maps for the nine states. Because the map servers for each state use different database schemas and have different levels of detail, it is necessary to use query rewriting based on the relevant ontology to reconcile the differences. There are also semantic differences; for example, some states call the age of the rock samples the "period" while others call it the "age," and the ontology-based mediator reconciles those differences so that users can search seamlessly across all nine states.

To establish a common communication model, all components of the mediation system are exposed as Web services, that is, they have WSDL (Web Services Description Language) descriptions and respond to SOAP (Simple Object Access Protocol) messages. This W3C standards-based approach is ensuring that this infrastructure can interoperate with related efforts in GEON and beyond.

This GEON application is such an important demonstration of the value of ontology-based spatial information integration that it was highlighted in the keynote speech by Jack Dangermond, president of ESRI, the world's largest provider of GIS software and services, at the most recent annual ESRI users conference in San Diego. With more than 12,000 attendees, this was the largest GIS event ever held.

"Assembling maps like this is challenging," said Zaslavsky. "And to provide a full solution for scientists, we have to be able to assemble composite maps online from distributed spatial data that is of different types and in different coordinate systems." In accessing spatial data, scientists want to be able to navigate and query different maps. But they also need to interact with non-spatial data that is attached to map features, for example, point data that indicates the locations of earthquakes. The lab provides technologies that let scientists query by location the points corresponding to earthquakes, in order to find further information such as earthquake intensity, date, relation to nearby faults, and so on.

Online Mapping Services and Production GIS Support

In the Superfund Basic Research Project (SBRP), sponsored by the National Institute of Environmental Health Sciences (NIEHS), Zaslavsky's lab has collaborated with UCSD's Urban Studies and Planning program to develop an online mapping system that enables scientists and planners to combine existing publicly-available data sources, for example, using a map service from NASA with a map service from the USGS, and to overlay the maps with selected toxicant-emitting facilities retrieved from a data warehouse of the Environmental Protection Agency. This demonstration shows how planners and community groups can access powerful online spatial information services to analyze complex, multifaceted phenomena, without having to make a major investment in "in-house" spatial information capabilities.

Beyond research in spatial information systems, the lab provides production GIS services to other labs and projects, including making GIS data available to projects, answering questions about ESRI software, providing online mapping services, and managing site licenses for GIS and other spatial software. The team also helps evaluate and test new ESRI software, which keeps them abreast of the latest developments in GIS and mapping software. Zaslavsky is also a technical representative to the Open GIS Consortium (OGC), which develops GIS standards.

A Strong Lab Team

The Spatial Information Systems lab has a strong team, led by Ilya Zaslavsky, a specialist in GIS and spatial data analysis. His research focuses on spatial databases, Web-based GIS, and XML-based mediation between heterogeneous geographic data sources. Before coming to SDSC in 2000, Zaslavsky was at NPACI partner San Diego State University (SDSU) in the Education Center on Computational Science and Engineering, where he was lead developer of the Sociology Workbench, a collection of online tools for social scientists and data analysts. Before that he was on the faculty of Western Michigan University in the Department of Geography. Zaslavsky received his Ph.D. from the University of Washington for research on statistical analysis and reasoning models for geographic data in multi-layer GIS. He also received a Candidate of Sciences degree, the equivalent of a Ph.D., from the Soviet Academy of Sciences, Institute of Geography.

In addition to Zaslavsky, lab members include Ashraf Memon, Reza Wahadj, and Haiyun He. Reza Wahadj is a specialist in GIS who has worked for the City of Glendale and in UN projects that modeled environmental change in the Persian Gulf region. An expert in ESRI products, GIS and online mapping services, and the .NET framework, he participates in both the lab's research and support activities, and is working to establish a common GIS infrastructure that all SDSC projects can use. As technical point of contact for GIS databases, software, and applications at SDSC, he assists researchers by supporting spatial data and information infrastructure, coordinating site licenses, GIS, and remote sensing databases within SDSC, NPACI, and Cal-(IT)2, as well as conducting GIS training.

Lab member Ashraf Memon, who recently received his masters degree in Computer Science from SDSU, is a GIS programmer analyst and the lab's expert in Web services and map assembly middleware. Haiyun He is a programmer analyst with the lab who works in the BIRN project, and students in the lab include Joshua Tran, Wai Ho Wong and Dmitry Yudovsky.

"In addition to our own research on spatial data, GIS, geography, and related subjects," said Zaslavsky, "we welcome collaborations with other groups, and find it very rewarding to help solve the fascinating spatial data problems that they bring us." -Paul Tooby.