Gis in Building the Earth Information System

THE ROLES OF GIS IN BUILDING THE EARTH INFORMATION SYSTEM

12.1

May Yuan The University of Oklahoma, Norman, Oklahoma

1 INTRODUCTION The EIS envisioned here is a comprehensive and integrated system that Geographic Information Systems (GIS) monitors natural and built environments, emerge as an enabling technology for integrates, manages, and disseminates data geospatial data handling, integration, of physical and human environments, analysis, modeling and visualization. With a synthesizes data from multiple variables wide range of spatially enabling functions, across scales in space and time, formulates GIS technology and the science (GIScience) new knowledge, explores patterns of interest that underpins the technology ought to play and visualizes findings to facilitate a central role in building the earth understanding. Hence, a comprehensive information system (EIS). In practice, EIS consists of not only “a four-dimensional however, GIS technology has not yet been gridded set of quantitative, geo-referenced fully recognized in scientific information data” or “a digital earth” but tools and management and computing. Many procedures to reason, analyze, model, and technological and societal barriers constrain visualize the dynamics of the earth systems. the use of GIS technology in scientific inquiry. Societal barriers may be overcome To date, we have a suite of remote by organizational and institutional solutions, sensing and surface observing systems that such as setting data standards and open- regularly take measurements around the source means. Technical advances require world (Mikusch et al. 2000). Although many strengthening GIScience fundamentals and gaps exist, especially in ocean observations, technological innovations for direct and NOAA, NASA, and USGS have been robust information support. partnered with international organizations and the private industry to implement and This paper emphasizes the need for promote integrated earth observation scientific and technological innovations to systems across the world. Many specialized empower GIS roles in support for EIS programs have been developed by research development. While the significance of communities and the industry for data societal barriers cannot be over-stated, assimilation, information synthesis ad theses barriers will not be discussed in this visualization, and knowledge acquisition. paper for two reasons. First, they are well While GIS technology has much to offer in recognized by government organizations, these areas, the linkage of GIS to these developers and users, and therefore, much specially developed data analysis and resource has been allocated to develop computational models is, unfortunately, solutions for data standards, interoperability, weak. Adoption of GIS technology to and portals, such as major progresses and atmospheric research, for example, is still in accomplishments by the OpenGIS its infancy, and applications appear limited Consortium, USGS, FGDC, NASA and to data distribution, mapping, and simple NOAA. Secondly, removal of societal spatial query and analysis. There are many barriers facilitates data exchange and reasons contributing to limited GIS distributions but has very limited effects on applications in scientific computing and the relevance of GIS technology on scientific ultimately the marginal attention given to computing, which demands computational GIS in building the earth information system. support for data analysis, modeling, and One fundamental issue relates to the lack of visualization. GIS support for 4D spatiotemporal data and the use of GIS as a mapping system, rather Unidata Thematic Realtime Environmental than an information system. Distributed Data Services (THREDDS) (Domenico 2002). However, as specified by This paper calls for a multi-system the Federal Geographic Data Committee approach to integrate existing databases, (FGDC), the content standard for digital information systems, and modeling geospatial metadata (http: //www.fgdc.gov/ environments to achieve a comprehensive metadata/csdgm/) is considerably limited to and integrated EIS that can effectively information pertinent to maps and specific to facilitate understanding and reasoning both individual data sets. The lack of means to physical and human dimensions of the cataloging data that are spatially and dynamic world through monitoring, temporally correlated may have constrained integration, analysis, synthesis, modeling access to geospatial data. For example, the and visualization. Figure 1 illustrates the development of a severe weather event components of a comprehensive EIS. GIS (such as Hurricane Ivan) may be captured technology has much to contribute in by multiple sensors and observing networks management, dissemination, analysis, (radars, GEOS, MODIS, etc.) across space modeling, and synthesis. and over time. Geospatial meta-data should provide mechanisms to relate all kinds of data corresponding to a certain event. 2 MANAGEMENT Manual documentation of such meta-data seems impractical. Alternatively, GIS GIS technology is designed to manage functions of spatial overlays and buffering geospatial data. Its capabilities to can be used to automate the process. georeference data and associate attributes Spatial overlays of the path and outlines of at locations enable spatial integration of data Hurricane Ivan to other data sets can reveal from multiple sources. Data integration is related data sets from other sources that critical, since Earth should be considered as also capture the hurricane. an integrated system (Carr 2001). The integration allows users to query information Spatiotemporal query support can based on locations or spatial relationships be greatly enhanced with the additional embedded in geospatial data. GIS has been meta-data across multiple sources because recognized as a powerful tool for spatial the user is able to identify or retrieve data of data management (Kingsbury 2001). interest based on spatial and temporal Beyond basic spatial search and geospatial relationships among geographic phenomena data retrieval, GIS technology can advance and events (Yuan and McIntosh 2002). For the levels of information query, analysis, and example, we can submit a query to retrieve modeling by incorporating temporal and data at stream gages within the path of spatial data and meta data to represent Hurricane Ivan and 7 days before and after geographic dynamics. the arrival of the hurricane. GIS technology ought to provide the needed capabilities that Meta-data is central to geospatial spatially and temporally relate data sets by data management. GIS meta-data cataloging georeferenced features and document, in general, geo-referencing events and applying such catalogs for data information, attribute definitions, and data access. sources and credits. These kinds of information are useful to determine the proper use of a data set. The development 3 DISSEMINATION of digital libraries has pushed the needs of data catalogs that highlight the essential Data dissemination serves two purposes: (1) content of a data set. Significant efforts in distributing data to the user, now mostly via cataloging geospatial data contribute to the the internet; and (2) interfacing the data and implementation of Alexandria Digital Library, the user by visual means. These data can meta-data catalogs at Center for be observations or results from analytical or International Earth Science Information modeling. In a broad sense, GIS technology Network (CIESIN), NASA’s Earth Data is contributing to building the EIS. As the Information System (Schaefer 1995) and internet GIS technology becomes popular, A Comprehensive Earth Information System

Monitoring satellites remote sensing surface observing networks

Management geo-referencing Dissemination assimilation Distribution integration Visualization query

Analysis Modeling Synthesis spatial analysis geospatial models reasoning temporal analysis statistical models knowledge base Spatiotemporal analysis numerical models data mining simulation models

Figure 1: Components of a comprehensive earth information system. Arrows indicate directions of data flows. many web-based geospatial data clearing dissemination and, moreover, visualization, houses and data portals disseminate diverse analysis, and modeling. geospatial data worldwide. While there is no short of success stories on Internet GIS applications, GIS technology is currently not 4 ANALYSIS, MODELING, AND playing an active role in building geo- SYNTHESIS cyberinfruncture. Most GIS software packages are proprietary, which may be one GIS technology offers a suite of analytical of the key factors prohibiting the broad use and modeling functions for geospatial data. of GIS technology in science communities, Unfortunately, there is only limited set of who, like Meteorologists, have been spatial functions commonly used in developing their own data dissemination and geoscientific applications. As mentioned visualization packages that couple well with earlier, most GIS applications are on domain-specific data formats, quite different mapping. While mapping is critical to from the ones used in GIS. For example, visualization and revealing spatial netCDF files store meteorological data in 3D relationships, advanced analytical and space and 1D time. However, true 4D modeling functions can probe new patterns, spatiotemporal GIS data models are relationships, and thoughts into geoscientific unavailable, and therefore, current GIS investigation. technology cannot support adequately distribution or visualization of true Commonly used spatial analysis spatiotemporal data. functions include spatial join, buffering, routing, and spatial overlays, just to name a One solution to the problem is to few. These spatial functions are effective for enhance data interoperability in GIS so that map analysis or location-based analysis. domain specific data can be accessed, Clearly, the lack of capabilities to handle distributed, and visualized in a GIS temporal data results in the lack of support environment. Scripts have been developed for temporal analysis and spatiotemporal to import data in domain-specific formats analysis. Furthermore, the earth observing into proprietary GIS packages, such as systems acquire mountainous amounts of netCDF to ArcView GIS (Shipley et al. environmental data from every 5 minutes to 2000). These scripts have proven useful to a few months. The massive amounts of import non-GIS data into GIS environments. geospatial data prohibit exhaustive However, such a data transformation is examinations of all acquired data sets. conceptually and technically unsound Geospatial data mining and knowledge because it “degrades” 4D spatiotemporal discovery initiatives address the need to data to isolated 2D GIS data sets and strips extend GIS functions to enable exploration information that can only be represented and identification of novel spatial patterns and inferred in a 4D environment, such as embedded in massive geospatial data (Yuan the development of a convective storm. et al. 2004). Clearly, geospatial data mining should also seek patterns and associations An alternative solution is to couple a of interest in space and time, which depends GIS software package with programs (such upon GIS abilities to handle temporal data. as McIDAS) designed for data in domain Coupling of GIS and domain specific specific formats. Meta-data, including information systems holds a better promise georeferenced spatial features, and pointers to successfully develop tools for geospatial to data encoded in the native data formats data mining than GIS alone because we can will be stored in a GIS to support spatial take advantages of GIS spatial analytical query in these native data formats. functions and build temporal and Meanwhile, domain specific programs spatiotemporal functions upon true 4D data provide tools to distribute and visualize 4D in the domain-specific environment. spatiotemporal data. The alternative approach takes advantages of both GIS and As to modeling, GIS technology domain-specific information technology to enables a geo-processing environment to maximize support for earth information provide predictive models based on spatial variables. Map algebra is central to geo- processing, in which data representing capabilities for monitoring, management, different environmental, socioeconomic, and dissemination, analysis, modeling, and policy variables can be spatially combined synthesis. Each of the components affords using mathematic expressions, such as the EIS to ensure the integrity and use of linear algebra. Expanding from traditional, data in the system. Except for monitoring, nonspatial algebra, variables in the algebraic which is mainly supported by remote expression are gridded data. For example, sensing and observing networks, GIS grids of elevations (used to calculate grids of technology is well prepared to contribute slope and aspect) and Sun angles can be geospatial functions that are unavailable used to develop a grid of estimated solar from other kinds of information systems. As insulation. While temporal map algebra has earth-related data are by definition spatially not yet been fully developed, it promises a structured, GIS functions enable direct GIS that will greatly facilitate geoscientific manipulation and analysis of spatial data in interpretation and projection, especially support for understanding of the earth among multiple geospatial phenomena, environments. While the current GIS such as ENSO and droughts. technology is inadequate in handling true 3D spatial data and temporal data, research is Information synthesis emphasizes underway to extend GIS capabilities for 4D an accurate summary of patterns, representation and analysis as the next associations among variables, similarity, and generation of GIS technology. In the near information that can be deciphered as a future, GIS will be able to support general statement. Data Mining (DM) and complicated spatiotemporal computations knowledge discovery in databases (KDD) that dig into massive amounts of spatial and bring new directions to information temporal data to reveal the hidden patterns synthesis. Expanding upon the existing DM and correlations among features and and KDD techniques and a rich set of GIS processes on the Earth. analytical functions, the development of geospatial DM and KDD will add a new dimension to GIS applications that uncover hidden spatiotemporal correlates and Acknowledgement patterns and, furthermore, probe hypothesis This research was partially supported by the building for understanding the dynamic earth National Science Foundation (award no.: systems. A sample project in the research 0416208). Its contents are solely the direction is mining spatio-tempral data to responsibility of the authors and do not discover the behaviors of geographic necessarily represent the official view of phenomena (such as droughts and sea NSF. surface temperature change) and relationships among these phenomena. As geospatial data grow exponentially, References geospatial DM and KDD open a new avenue towards “informational science” that seeks Carr, T. R. (2001). "As we see Earth as an interpretation and understanding through integrated system, we must also information synthesis. integrate the information we gather about it." Geotimes 46(1): 16 (2 pages) 5 CONCLUSIONS Domenico, B., J. Caron, E. Davis, R. Kambic and S. Nativi (2002). "Thematic While there lacks a consensus of what Real-time Environmental Distributed constitutes the earth information system Data Services (THREDDS): (EIS), most cases refer the EIS as 4D Incorporating Interactive Analysis gridded earth data or the digital earth. Tools into NSDL." Journal of Digital However, an information system is more Information than just data; a suite of functions are 2(4).http://jodi.ecs.soton.ac.uk/Articl necessary to make the data useful and es/v02/i04/Domenico/ informative. To this end, a framework for a Kingsbury, P. (2001). The National comprehensive EIS is proposed to include Transition Database's GIS Module: A Management Information Tool for the National Weather Service, ESRI User Conference. Mikusch, E., E. Diedrich, et al. (2000). "Data Information and Management System for the Production, Archiving and Distribution of Earth Observation Products." 401-406 Schaefer, K. (1995). "NASA's Earth Observation System Data Information System." Bulletin of the American Society for Information Science / 21, no 4: 13 (3 pages) Shipley, S. T., I. A. Graffman, et al. (2000). GIS Applications in Climate and Meteorology. ESRI User Conference.http://gis.esri.com/librar y/userconf/proc00/professional/pape rs/PAP159/p159.htm Yuan, M., B. Buttenfield, et al. (2004). Geospatial Data Mining and Knowledge Discovery. A Research Agenda for Geographic Information Science. R. McMaster and L. Usery. Boca Raton, Florida, CRC Press: 365-388. Yuan, M. and J. McIntosh (2002). A Typology of Spatiotemporal Information Queries. Mining Spatiotemporal Information Systems. K. Shaw, R. Ladner and M. Abdelguerfi, Kuwer Academic Publishers: 63-82.