Integration of science gateways: a case study with CyberGIS and OpenTopography

Science gateways are collaborative software environments designed to enable community-driven development and use of cyberinfrastructure services, software tools, applications, and data through common interfaces, typically an online portal, customized to meet the needs of individual communities [1]. By abstracting the assemblage of cyberinfrastructure needed by the research communities and democratizing access to high-end computational resources, science gateways (e.g. those on XSEDE) provide a shared problem-solving environment and promote collaborations among community users. The integration work presented here represents a cutting-edge approach to coupling two independent geospatial software environments developed separately, namely CyberGIS [2] and OpenTopography [3]. CyberGIS -- defined as cyberinfrastructure-based geographic information systems (GIS) -- represents a new generation of GIS based on seamless synthesis of cyberinfrastructure, geographic information science, and spatial analysis and modeling [4]. Funded by the National Science Foundation (NSF), the CyberGIS project (http://cybergis.org) advances the science of CyberGIS, with a particular focus on enabling the analysis of big spatial datasets, computationally intensive spatial analysis and modeling, and collaborative geospatial problem solving and decision making [2]. The CyberGIS Gateway provides an online, high-performance and collaborative geospatial problem-solving environment to allow for the contribution, sharing of and access to CyberGIS services and tools by a broad community of geospatial scientists and GIS users. As a NSF cyberinfrastructure-based data facility, OpenTopography (http://opentopography.org) provides its community with access to high-resolution Earth science oriented topography data, related tools and geoprocessing applications published as a suite of Web services. All Web services are built and deployed by leveraging the open source Opal toolkit (http://www2.nbcr.net/data/docs/opal/), which provides a mechanism to streamline the process of encapsulating existing scientific applications as Web services. CyberGIS and OpenTopography represent two software environments that complement each other; with CyberGIS providing access to cyberinfrastructure-based spatial data and related analytics and OpenTopography serving as a data source of LiDAR-based high-resolution terrain data. Hence, integrating these software environments will extend their capabilities, improve their usability and bring benefits to the science communities by enabling large-scale geospatial problem solving through shared interoperable analytical and data services. The goal of this integration is to make seamless access to OpenTopography data when an analysis is planned on the CyberGIS Gateway and vice versa. More importantly, this work demonstrates multiple innovative aspects of software integration research and development that could serve as a template for integrating other such independently developed software environments. In particular, OpenTopography applications (exposed as Web services via the Opal toolkit) include: (a) the "Point Cloud Selection" service for access to LiDAR point cloud data; (b) the "Points2Grid" service for the generation of digital elevation models (DEMs) using a local gridding method; (c) the "Format Translation" service for converting DEMs from ArcInfo ASCII Grid to GeoTIFF and ERDAS formats using the Geospatial Data Abstraction Library (GDAL) toolkit (http://gdal.org); and (d) the "Derivative Products" service for the generation of the common geomorphic metrics of slope and hillshade grids in GeoTIFF and IMG formats. GISolve [5] -- the leading CyberGIS software toolkit -- together with its Open Service API, serves as the bridge to help resolve the major technical challenges posed by this integration through the following key integration strategies: (a) Security: To ensure that only authorized users can access related Web services, in OpenTopography, Opal was originally configured to restrict computing jobs based on IP addresses. However, this model was inefficient for service invocations from external problem-solving environments managed independently, such as CyberGIS. Therefore, a new security mechanism using the token-based authentication service in the GISolve Open Service API (http://gosapi.cybergis.org) was developed for Web service access between OpenTopography and CyberGIS. The security infrastructure (see Figure 1) that glues OpenTopography and CyberGIS together represents an innovative feature of our software integration and has been adopted for interoperable access to other service-oriented geospatial software environments and Web-based user environments from the CyberGIS Gateway. (b) Data Discovery: OpenTopography datasets are only useful to CyberGIS user communities if they can be understood, queried, and accessed transparently and coupled with analysis workflows in the CyberGIS software environment. This can be achieved by pairing data with relevant and useful metadata. To address this need, we developed a standard-based metadata service that allows for the discovery of and access to OpenTopography data sources from the CyberGIS Gateway. This OpenTopography metadata service is based on the OGC Catalog Service for the Web standard (www.opengeospatial.org/standards/cat). This service allows for programmatic access to OpenTopography metadata and related data processing capabilities. The integration of OpenTopography is demonstrated within the CyberGIS Gateway using a viewshed analysis application (Figure 2). The existing user interface for OpenTopography data discovery is reused within the CyberGIS gateway, and linked to backend metadata services and analytical workflows. Consequently, gateway users can select data sources from OpenTopography and use GISolve and OpenTopography services to transparently query OpenTopography LiDAR data and generate the DEM data needed by the viewshed application. Through this integration, CyberGIS users have access to a large, and continually growing catalog of high-resolution (sub-meter resolution) topography data. OpenTopography currently hosts 134 distinct LiDAR datasets, covering more than 91,283 km2 -- including, among others, important geologic regions and features such as the San Andreas Fault, and Yellowstone, Tetons, and Yosemite National Parks. The knowledge we have gained and the GISolve Open Services API approach to integration we have developed through this research has shed light on how to integrate other large geospatial data facilities and related software services to the CyberGIS software environment.