Priorities for GEOINT Research at the National Geospatial-Intelligence Agency (2006)

The National Geospatial-Intelligence Agency’s (NGA’s) vision for the future of geospatial intelligence (GEOINT) is ambitious and, if achieved, is likely to sustain the information dominance doctrine of the nation’s intelligence community (Joint Chiefs of Staff, 2000). Achievement of this goal, however, requires that numerous challenges be met, in both research and technology. These challenges involve problems of improving the existing GEOINT infrastructure, as well as designing and building the next infrastructure. These challenges also provide a basis for the specific research-oriented “hard problems” identified in the next chapter. A pivotal challenge to NGA is that it must remain operational during any disruptive transition or “paradigm shift” by exploiting quantum and incremental improvements in today’s operational systems and architectures. This challenge impacts NGA as it identifies areas for priority research. How can NGA build a targeted research program to create breakthroughs in geographical and information science that will lead to new systems, information advantages, and assets to match the needs of the next era?

In briefing materials given to the committee, NGA enumerates a set of problems of immediate concern: the top 10 list of challenges that must be met to prepare intelligence at the global, regional, and local levels:

1. Achieving persistent “tasking, processing, exploitation, and dissemination” (TPED);

2. Exploiting all forms of imagery, in the context of persistent surveillance, including the following challenges (3-6):

3. Detecting weapons of mass destruction;

4. Tracking moving targets;

5. Thwarting denial and deception; and

6. Targeting precisely;

7. Compressing time lines (for preparation and dissemination of intelligence);

8. Sharing with (coalition) forces, foreign partners, and communities at large;

9. Supporting homeland security; and

10. Promoting “horizontal integration.”

The immediacy of these challenges requires a research agenda that simultaneously addresses these short-term needs, while it pursues the evolution to a next-generation methodology for dealing with geospatial intelligence, GEOINT2. The vision for GEOINT2 involves bringing intelligence into a single operating environment that will allow analysts to draw from a variety of sources when making interpretations. Data for a specified location on Earth’s surface require both general knowledge of human and physical processes and specific knowledge of geography, culture, and tradition. Data on the physical environment may be directly measurable by sensors, or come from maps, but other data will come from the other intelligences (INTs), especially from open, public information about people and their lands. It is tempting to compare the necessary source integration to the multiple map-layer data model commonly encountered in NGA’s existing geographic information system (GIS) technology. Intelligence layers currently include weather information, strategic battle planning overlays, intelligence reports, and a layer attributing navigation safety. The underlying data foundation layers include georeferenced gravity data, vector features, bathymetry and elevation, intelligence baselines or reference data, and imagery.

A layered model, however, is inadequate. Multisource information does not easily resolve into layers; consequently, this information fits poorly with the existing GIS approach. For example, sensor webs provide data in real time that are spatially and temporally discontinuous and asynchronous, and often point-based. Human and signals intelligence may be in the form of textual reports with place name and other references. Data may also arrive as video, audio, web-based extensible markup language (XML), or any of a plethora of other media formats. The simple “data integration by spatial coregistration” standard that is the foundation of

GIS needs to be superceded by the placement of data into a time-space framework, with data elements being able to take the form of objects or features rather than components of a static map or image. In such a system, attributes should be attached to objects, not to artifacts of the systems architecture that created them: images, map sheets, reports, or networks.

Furthermore, these answers need to be delivered in near or actual real time, through distributed environments with multilevel security masking, and via context-sensitive interfaces on mobile, hand-held, or head-mounted displays. When working in near or actual real time, there is no time to wait for multiple forms of intelligence (MULTI-INT) to become integrated through conversion to GIS layers, nor will the intelligence demands of tomorrow fit into this simple model. Integration is indeed one of the central challenges. The National Intelligence Strategy dictates that “transformation of the intelligence community will be driven by the doctrinal principle of integration” (Negroponte, 2005). This will be true at NGA both organizationally and technically.

Not only is NGA responsible for solutions to its own operating and research challenges, but as functional manager for the National System for Geospatial Intelligence (NSG), NGA has the responsibility to set future directions for national geospatial activities, including an overall national vision of imagery, imagery intelligence, and geospatial investment. This vision is likely to have a profound influence on the future of most of the agencies involved in the collection and use of intelligence. Specific goals of the NSG Statement of Strategic Intent mandate specific actions from NGA (see Box 3.1) (NGA, 2004b). Research supporting the NSG Strategic Intent will have to transform the NSG infrastructure from a strategy for analysis based on data to analysis based on knowledge derived from interpretations.

The catch phrase adopted in the NSG community is “Now, Next, and After Next.” The phrase is designed to foster a transformation to GEOINT2 and to guide activities in specific NGA line directorates. Current activities, or Now, are accomplished by three line directorates: Source Operations, Enterprise Operations, and Analysis and Production. Next activities are carried out by the Acquisitions Line Directorate and refer to advancements in the immediate and short-term future that are based on results of research that are now moving into commercialization and production. After Next refers to the planned state of GEOINT that can be attained as challenges to existing and longstanding geospatial problems are met; it is carried out by InnoVision, NGA’s research and development directorate. This report covers both the “next” and the “after-next” stages and recognizes that it will take both shorter-term development of new technologies (next) and more basic research (after next) to reach the goal of GEOINT2.

The framework that can support research and bring about the transformation to GEOINT2 requires fundamental and high-risk research in both basic and applied science (NRC, 2001). Basic research builds on existing theory without regard to specific application, while applied research develops new theories and methods without regard to existing families of problems. In NGA’s case, high-risk research is both basic and applied. There can be a benefit in having diversity and an open-ended perspective for high-risk research. Given rapid changes in technology, there are many unknowns, and allowing sufficient flexibility to explore many research directions simultaneously may create benefits over a highly directed and coordinated research agenda. Therefore, four critical issues are (1) how much fundamental research should be basic and how much applied; (2) how much research in NGA’s portfolio should be high risk; (3) to what disciplines should the research be targeted; and (4) how can these disciplines collaborate to perform the interdisciplinary research necessary for GEOINT2?

The NGA chief scientist identifies the following disciplines as pre-eminent in pursuing basic research that is applicable to the NSG Strategic Intent:

• Geodesy and geophysics—including measurement and modeling of Earth’s shape and gravity, precision location, and photogrammetry

• Advanced geoprocessing—including architectures and design, special issues for geospatial-image computation, data mining, advanced synthetic aperture radar (SAR) processing, information technology (IT) for massive data files, mass storage, databases or structures, visualization, and high-performance computing

• Remote sensing—including sensor systems, phenomenology, analytical techniques, image processing, collection strategies or tasking, imagery science, polarimetry, and hyperspectral science

• Geospatial intelligence analytics—including spatiotemporal distribution, association, and behavior and interaction of natural phenomena on and near Earth’s surface

These disciplines will require attention from NGA, as will fostering interdisciplinary work among them, which will be more difficult. Attention is required because NGA has a vested interest in maintaining a strong U.S. presence in these fields and in ensuring a geographically distributed and representative body of expertise that it can draw upon for employees and leaders, as well as for research. Many U.S. programs—for example in surveying engineering and geodesy—have declined in size and quality over the last decades. Similarly, there is a growing need for skilled scientists and engineers who can work across these disciplines or facilitate exchanges among the specialized groups. Even fewer programs teach such interdisciplinary science, and the few that do have difficulty finding a niche to ensure continued funding and support. A model program to counteract the lack of interdisciplinarity is the National Science Foundation’s Integrative Graduate Education and Research Traineeship (IGERT) centers, which often target building human capital and centers of excellence in specific interdisciplinary matches.

The research needed to address NGA’s GEOINT challenges is not unique to NGA, but reflects overarching themes in GIScience (geographic information science) research. In other words, it reflects the need for advancements in data acquisition, target identification, integration of disparate types of data from many sources, data analysis to derive needed information, dissemination, and preservation for future use. Others in the GIS community have enumerated sets and supersets of longstanding research problems. One such enumeration originates with the University Consortium for Geographic Information Science (UCGIS), a consortium of more than 80 North American universities, professional organizations, and private vendors in which GIScience is taught and researched. UCGIS has established itself as the central network for the academic GIS research

community, especially in defining critical research areas (UCGIS, 1996) that can be addressed in the short, medium, and longer terms. The set of UCGIS enduring research challenges (or priorities) emphasizes basic research about unsolved problems associated with acquiring, processing, interpreting, disseminating, and preserving geospatial information (McMaster and Usery, 2004). The chronology of research priorities is available in white paper format.¹

The first group of UCGIS priorities concerns longstanding problems associated with the geospatial data infrastructure. These include spatial data acquisition and integration, distributed computing, interoperability of geographic information, and the future of the geospatial information infrastructure. All of these priorities match NGA’s research needs closely. A second group of priorities relates to data use and representation: extending geographic representation, cognition of geographic information, and GIS and society. A third group of priorities addresses specialized analytical methods required for processing geospatial data: geographic scale, spatial analysis in a GIS environment, and uncertainty. Since 2002, four additional research priorities have been enumerated by UCGIS, reflecting advances in information technology and in knowledge. These are geospatial data mining and knowledge discovery, ontological foundation, geographic visualization, and remotely acquired geospatial information. The 2002 UCGIS research agenda priorities are shown in Box 3.2.

It is interesting to note that the 10 longstanding problem topics originally identified in 1996 persist and, to a large extent, are closely reflected in the NGA top 10 list. This does not indicate a lack of advancing knowledge, but instead points to the complexity, breadth, and depth of the enduring challenges that arise in dealing with geospatial data in general and GEOINT in particular. It also highlights the current and potential synergy for innovative research in geospatial science between NGA and academia.

Missing from both agendas, however, is the need for research into new geospatial information systems architectures and associated software and standards that will facilitate flexibility in analysis tasks, as well as a greater degree of interaction between software components used for analysis, visualization, and archiving. Also needed are advances in IT research. Such advances are being addressed by the computational science and engineering community, but the IT and geospatial research communities will have to collaborate in order to achieve NGA’s research goals. Finally, within an organization such as NGA, research on the integration

of the complete system, from acquisition to preservation, including the human factors involved, both present and future, is relatively absent. This is in spite of important work on systems integration issues such as standards and interoperability.

Using knowledge about hard problems in geospatial information science as a starting point, in the next chapter this report focuses on the subset of these hard problems that are of most relevance to NGA in advancing GEOINT.

This chapter begins with a review of NGA’s own assessment of the short-term challenges it faces in current operations and an assessment of how these immediate challenges imply broader research needs. It then puts these challenges in the context of research in general, and of GIScience research in particular, using UCGIS’s past efforts to illuminate the research challenges for GIScience. While there is much overlap, NGA’s research must nevertheless be context specific to the challenge of creating GEOINT2 while supporting and improving existing systems. Traditional

GIS models that use layers and collocation as a means of data integration will be replaced by support for feature-level information that is independent of the collection systems used to acquire the data. In Chapter 4, the “top 10” list is used to structure the hard problems that NGA research faces and suggest promising solutions.