WEB SERVICES FOR GEOSCIENCE DATA: EXPERIENCES AND LESSONS

The lack of accepted standards for data interoperability is a continuing challenge in the geoscience community, and hampers our ability to make research results broadly available. The Open Geospatial Consortium, Inc. (OGC; www.opengeospatial.org) is attempting to address this challenge by developing Web service standards through a consensus process among industry, government, and academic partners. These standards include the Web Map Service (WMS) to compose and display map images from underlying data sources, as well as the Web Feature Service (WFS) and Web Coverage Service (WCS) to provide direct access to geospatial data.

WFS provides a simple standard for serving geolocated vector data such as points and polygons. The request is expressed entirely in the URL, and the response is delivered as a XML object using the Geography Markup Language (GML). Such a service can advertise a wide array of useful geoscience data including station locations, physical specimens, event catalogs, track lines, etc. WFS is currently supported by numerous GIS server products, both commercial – ArcIMS (www.esri.com), RedSpider (www.ionicsoft.com) – and open-source – GeoServer (www.geoserver.org), MapServer (mapserver.gis.umn.edu). It is supported in GIS clients such as uDig (udig.refractions.net) and GeoMapApp (www.geomapapp.org).

Deployment of WMS and WFS providers is underway throughout the geoscience community. UNAVCO (www.unavco.org), IRIS (www.iris.edu), and the Marine Geoscience Data System (MGDS; www.marine-geo.org) recently reported results from ongoing collaborative work. (Figure 1.) The National Geophysical Data Center (www.ngdc.noaa.gov), Petrological Database of the Ocean Floor (PetDB; www.petdb.org), and LDEO Borehole Research Group (www.ldeo.columbia.edu/BRG) have deployed WFS providers as well. The Marine Metadata Interoperability Project (MMI; www.marinemetadata.org) is pursuing activities including a formal OGC Interoperability Experiment for ocean observing data. Examples from these projects will be described in detail.

WFS can provide extensive information for each feature instance including identifier, location, time, elevation, URL (i.e. reference for further information), and any number of additional data attributes. For example, the PetDB WFS provides a Sample feature with an extensive listing of geochemical analyses at each instance. A WFS-enabled client such as GeoMapApp can load and display PetDB samples on a map, and allow the user to color, plot, and inter-compare different analytical values. (Figure 2.)

As WFS usage increases, several performance issues have become apparent. A request for a large number of feature instances can return a prohibitively large result, exhausting network or memory resources. The GeoMapApp client addresses this issue by restricting the user to a defined bounding box (world, ocean, or local study site as appropriate), thus keeping the server request to a manageable size. Further, there is no accepted standard for attribute names, units, or ordering. GeoMapApp addresses this issue by offering a generic interface for plotting data values, allowing the user to select and color attributes according to domain knowledge.  

(Figure 1. uDig GIS client displaying integrated results from UNAVCO, IRIS, MGDS providers at Gulf of California (Baja) local study site. Features and layers include GPS campaign surveys and continuous stations, seismic stations, earthquake locations, seafloor bathymetry, expedition tracks.)

 

(Figure 2. GeoMapApp GIS client displaying results from PetDB WFS provider. User selected Juan de Fuca mid-ocean ridge study site, plotted/colored distribution of K₂O versus SiO₂.)