Geographic information systems (GIS) are well-developed tools for query, analysis and display of point, vector, and polygon data and there are well-developed communities that use these data types and rely heavily on GIS tools. Many meteorological datasets consist of combinations of time series and grids. Tools for query, analysis and display of these data types are less well developed in the geospatial community. Developing approaches to integrating these datasets into geographic information systems is a critical step toward bridging the gap between science data and geographic information systems. We describe our experience with three approaches that cover the spectrum of geospatial systems from desktops to distributed geodatabases.

1) OPeNDAP GeoTIFF Server - OPeNDAP is a well known data access and http transfer protocol formerly known as DODS. The GeoTIFF format is a well known format for geolocated images. It involves extending the Tagged Image File Format by adding tags that include geolocation and projection information. The OPeNDAP GeoTIFF server takes an OPeNDAP URL, retrieves the data from that URL and provides a GeoTIFF file (or a set of files) to the user. This file contains the actual data values suitable for further analysis in the GIS. This approach is useful for a desktop GIS user (or ArcIMS server) that wants to view and/or analyze a science dataset in a GIS. They can create the GeoTIFFs they need once and download them to local disk to use as often as they need them. This approach also allows a science data provider to provide images for GIS users without translating all existing datasets or time slices into GeoTIFFs.

2) ArcIMS Data Source - In a partnership with OPeNDAP and the Environmental Systems Research Institute (ESRI) we have developed a custom data source for ArcIMS that allows map service providers to include gridded science data sources from OPeNDAP servers in ArcIMS sites directly. The data appears as a layer in the internet map along with traditional geospatial data from local or distributed geospatial databases. In this case the layer is an image of the data rather than the data themselves. This approach is useful for groups serving maps using ArcIMS that want to integrate data from local or distributed OPeNDAP servers into their internet maps.

3) Geospatial Database - The most powerful approach for integrating gridded science data into a geospatial environment involves the Gridded Datablade for Informix developed by Barrodale Computing and IBM. Using the datablade involves ingesting gridded data into an Informix database and accessing it using spatial extensions to Structured Query Language (SQL). This approach differs from the others because it changes the fundamental storage mechanism of the data themselves. It also provides the most power because of the capabilities already included in the datablade and Informix.

We will demonstrate all three of these approaches to science and geospatial data integration and describe our experience with each.