Geospatial Crystal Ball
MGT 2008 Volume: 6 Issue: 5 (September/October)
GEOSPATIAL CRYSTAL BALL
Experts are developing geospatial analytic predictive techniques that
seek to project where an event or individual can be found in the future.
by Karen E. Thuermer, MGT Correspondent
More than ever, U.S. warfighters, military planners, intelligence analysts, counterintelligence agents and policy makers today need to know their operating environment and who is in it. That’s because the United States has moved into an era in which stability and support operations, non-governmental organization activities, rescue operations and peacekeeping duties are arguably comparable in importance to manhunts for terrorists, counterintelligence operations and traditional military targeting.
“The new focus of counterinsurgency and information operations demands that we know the geospatial and temporal patterns of humans and their culture in the spaces in which we are operating,” said Swen Johnson, owner of a firm called Socio- Cultural Intelligence Analysis (SCIA). “As our sensitivity to collateral damage grows, the traditional kinetic options also demand an increased understanding of who is where and when in the operational environment.”
The U.S. intelligence community needs a better understanding of foreign societies and culture to bring and maintain peace and flush out adversaries hiding among populations. By placing the analysis of foreign, denied areas into a geospatial context, data and information becomes actionable intelligence, suggesting where and when operations should take place.
Some of the innovative methods now being used across the Department of Defense and intelligence community include geospatial analytic predictive techniques that seek to project where an event or individual can be found in the future.
“While still rudimentary, these techniques are proving to be valuable to a military intelligence community that will always be data-poor relative to their academic cousins working with clean, meticulously collected data sets,” said Johnson, a former Army counterintelligence special agent with a doctorate in sociology.
Johnson was honored last year by the U.S. Geospatial Intelligence Foundation with its Academic Achievement Award for developing an analytic approach of blending the social sciences with geospatial predictive technology. The analytic teams that Johnson built and installed in several elements of the DoD helped lead a transformation in the military understanding of the significance of the human terrain.
A cutting edge of GEOINT is the blending of the social sciences with recent advances in geospatial predictive technology to create models of human behavior that can narrow the analytic and operational search space in areas in which the DoD is interested.
“The creation of tribal polygons (the projection of where indigenous groups’ contours and edges are, based on using food or water shedding techniques to make up for the lack of good, factual data), status group mobility signatures (showing the ‘operating environment’ of certain groups and likely geospatial behavior for elements of those groups), geospatial pattern matching (the comparison of individual geospatial patterns to group patterns to assess membership in certain groups) and other similar methods are examples of how geospatial technology is being fused with the social sciences to improve the kind of analysis needed by today’s military,” he explained.
HUMAN TERRAIN
Johnson formed SCIA in 2005 to address what he saw as a critical intelligence gap in denied areas of foreign countries. SCIA not only founded the methods of “human terrain analysis” now being used across DoD and the IC, but also is the first to teach the topic to the academic community, with courses at George Mason University’s GIS department and in DoD.
SCIA’s mission is to help DoD and the IC understand the geospatial patterns of human groups and significant individuals, and how social structure and culture create the human terrain.
“Knowing where one’s adversaries are, as well as one’s friends, turns information into actionable intelligence at the tactical, operational and strategic levels,” Johnson said. “Previously, analysts working for the military doing ‘human factors’ research—a form of analysis that is heavily psychological and focused on individuals—and ‘cultural analysis’ were not meeting the needs of military operators and planners because they were not geospatial.” That’s because these are heavily influenced by anthropological approaches that some have criticized as subjective and anecdotal.
SCIA’s approach was inherently geospatial and took a sociological approach to understanding the people and culture of a given area. To do this, Johnson developed a method that was heavily sociological in its focus on groups and heavily empirical and quantitative. The result was a blend of all-source analysis that combined cutting edge techniques in social network analysis, a heavy dose of social-scientific theory, geospatial technology and a focus on culture.
Using this approach, SCIA developed a certification program and began training military veterans to serve as “socio-cultural analysts” and “human terrain analysts.” “This proved to be more practical than trying to take academic social scientists and putting them in the military environment,” Johnson said. “The requirements and pace at which military intelligence moves demands a pragmatism that often makes the academics blanch.”
SCIA now serves several different defense agencies and combatant commands. It is currently a subsidiary of Earl Industries, a ship-repair company that has formed a Technology Group featuring several defense and intelligence technology companies.
One of the most critical gaps in GEOINT systems, Johnson argues, is the conceptualization and measurement of the different types of empirical data that is needed in geospatial predictive models.
“Not to get overly complicated, but kernel density with spatial preference techniques— those that use sophisticated algorithms to look for geographic environments that are like the environments in which a set of events has been located to ‘project’ or ‘predict’ where a follow-on event might occur—demand geospatially referenced data that goes beyond the geophysical terrain to include what we started calling the human terrain back in 2005,” he said.
“We saw that the crude hot-spotting models that were being used in DoD to predict where events, individuals or interesting phenomena might be found were heavily reliant on non-moving, stable, geophysical data,” he continued. “As social-scientific-minded analysts, this didn’t make sense to us. We understood that people move in relation to other people, social structures, institutions and culture. Those were the things that needed to be mapped—not the mountains, rivers, and things like slope and altitude.”
Consequently, when SCIA began to “map” culture using its methodology, it found previously unknown and invisible geospatial patterns of tribal, ethnic, religious, occupational, language and other groups that provided the critical “intelligence preparation of the environment” analysis that it said were needed by military planners and operators.
“We see at least four levels of human terrain data,” Johnson said. “First, there are the physical manifestations of culture, such as schools, mosques or roads, that clue us in on what type of people inhabit a certain area. Second, there’s data on where human groups actually are. This is the geospatial referencing of attribute groups to discover their patterns. Third, there’s geospatially referenced ‘nonmaterial culture,’ such as ideas, attitudes, values and norms. These are the exciting maps that are not yet being made. And fourth, geospatially referenced data on human behaviors and actions, the kind of data that can reveal patterns and clues as to what type of groups live in the area of interest.”
CRITICAL INTELLIGENCE SOFTWARE
A number of other companies are also involved in offering geospatially related predictive technology. One is SPADAC, which has a lengthy track record of providing government and defense agencies with mission-critical and mission-specific predictive analytic and geospatial solutions.
The Office of Naval Research (ONR), for example, was in need of a way to predict and analyze tactics, techniques and procedures used by enemy individuals and groups, said Mark Dumas, SPADAC chairman and chief executive officer.
SPADAC answered this need by offering its enhanced, patented predictive analytic technology, Signature Analyst, which helps the ONR and other SPADAC customers determine the cause of certain behaviors, and identify features of individuals, events and organizations that pose threats to specific missions and determine where sensor resources should be allocated for maximum efficiency.
Signature Analyst is able to fuse data from multiple sources, including spatial information as well as human terrain and social networking elements.
“By synthesizing all of these inputs and applying high-level analytics and enhanced objectivity, Signature Analyst software is able to identify the causal factors that contribute to a specific situation or event in question,” Dumas said.
Identifying potential launch pad sites, locations and networks of criminal or terrorist elements are among the multiple applications for this software, although not specific to ONR use. For one, Signature Analyst can be used to efficiently and effectively respond to potential man-portable air defense system threats, which target aircraft within take-off and landing vectors.
Predictive analytics takes into account the environment, similar past events and the nature of the threat to identify the areas of highest risk, providing authorities the opportunity to prepare accordingly and take responsive actions. It can also be used to identify threats posed by organized crime groups or terrorist networks, whether in U.S. metropolitan areas or in hostile environments around the globe.
“SPADAC analyzes numerous data layers, including physical terrain, cultural, economic, human and social networking elements, and determines key characteristics and factors that contribute to the crime rate,” Dumas said. “By using previously identified criminal safe houses, SPADAC’s predictive analytics validates the existence of other safe houses in the area.”
According to Dumas, integrated intelligence systems will always have room for improvements to adapt to technological advances and changes in the military’s defense and intelligence needs.
“SPADAC recognizes this as one of the main characteristics of its industry and is driven to constantly provide its customers with the most innovative technologies and solutions for their geospatial, predictive analytic and imagery needs,” he said.
Amid these constant changes and industry challenges, the company consistently is developing and offering new and unique approaches to ensure that expert analysts, proven methods and innovative technology come together to minimize risk, maximize opportunities and significantly increase the likelihood of mission and business success for its diverse client base—both within and outside the government arena.
As Dumas pointed out, the gaps are wherever the enemy has made advancements either in technology or methodology. “It’s impossible to describe the gaps without revealing information relevant to national security,” he commented. “But by integrating more information from various data sources, thereby fusing multi-variable sources of data to identify patterns within that data, SPADAC can help clients fill any gaps and stay ahead of the enemy by identifying trends in tactics, techniques and procedures.”
SPADAC currently has three active contracts with the ONR. In the first contract, SPADAC was hired to research and provide ONR with a solution to predict and analyze tactics, techniques and procedures used by enemy individuals and groups.
“This solution will help analysts determine the causes of certain behaviors and to plan accordingly to exploit those behaviors,” Dumas said. “The solution will also provide a means of measuring changes in behavior.” In the second contract, SPADAC was hired to develop simulation and modeling tools to identify the features of individuals, events and organizations who pose a threat to specific missions and to determine where sensor resources should be allocated for maximum efficiency.
Under a third contract, SPADAC will research and develop a system to provide a valuable analytic component that accounts for the importance of cultural and organizational factors. The system will integrate cultural data with existing geospatial and temporal analysis algorithms, while simultaneously allowing human terrain analysts to form hypotheses about the structure and intent of organizations.
SPACE-BASED INTELLIGENCE
A different kind of use of predictive analysis answers the need of intelligence analysts to know when and how airborne or spacebased platforms can be used for intelligence missions. They also need to know platform position and attitude; image or communication sensor capabilities; weather conditions that may affect the ability to collect; and various constraints of the asset such as geometric, environmental and time of day. In most instances, the analysis will need to integrate with geospatial data to model and identify ground targets and assess collection opportunities.
“Our software satisfies these needs, by ingesting the data directly from geospatial databases and combining it with simulated or real-time models of collection platforms, their parameters and constraints,” said Todd Smith, business development director, new markets, for Analytical Graphics Inc. (AGI).
Furthermore, target acquisition operations may involve locating and providing persistent surveillance over a specified area, so that targets of interest can be continuously tracked and monitored for real-time situational awareness. This capability will enable analysts to develop course of action plans to optimally utilize critical assets.
“Our software can also be employed for time-dynamic target-to-asset analysis, and visualize the dynamic relationships of multiple assets using real-world environmental constraints,” Smith added.
AGI provides interfaces between GIS software packages, such as ESRI’s ArcGIS, so users can import, analyze and export geospatial data.
“Our STK desktop and STK Engine software products, used in conjunction with our GIS Analyst Extension, which integrates ESRI’s ArcGIS engine runtime, enables users to connect to any GIS format that ESRI supports,” he said.
STK and STK Engine leverage the ESRI map document, which permits all of the properties of the document to be preserved in the user’s scenario. Users can also select features within the geospatial data to be applied as input parameters to the analysis within the scenario. Intelligence products developed within the scenario, such as sensor data and coverage areas, can then be exported into any GIS platform to bring AGI’s analytical data to various environments.
“As such, these AGI-derived data products can easily be implemented within any display environment, whether it is Google Earth, Microsoft Virtual Earth or a Web application,” Smith explained.
Other software tools lack the ability to handle high-fidelity, high-velocity objects, he argued. Traditional GIS views time as a snapshot, while many GEOINT professionals require time to be viewed as a continuum. Analysts need to propagate to a specific time and location and perform analysis. Many times a time step, or snapshot in time, is insufficient.
“GIS does a tremendous job of modeling, analyzing and distributing geography, but can be challenged to manage temporal data at the required fidelity,” Smith said. “By creating a `geodynamic’ layer to traditional GIS, geospatial analysts can calculate precise locations and the performance of dynamic platforms and assets.”
Examples could be modeled sensor motion, fields of view for target acquisition and collection planning. The interoperability between GIS and temporal GEOINT analysis is critical for mission success. “AGI software bridges the gap between the traditional GIS and GEOINT analysis through our time-dynamic analysis and visualization capabilities,” Smith said. “Furthermore, AGI software is customizable and modular, allowing for integration or adoption within disparate systems, enterprise architectures and federated systems.”
For example, AGI is a member of the Commercial Joint Mapping Toolkit team, providing dynamic geospatial and 3-D software components for this National Geospatial-Intelligence Agency program. AGI products also are embedded in a number of C4ISR systems. Gaps exist in the systems that need to be improved, however. The main gap, according to Smith, is the lack of a true temporal data model within the GIS community that could accurately represent temporal geospatial products derived from high-velocity vehicles, sensors, communication systems and radars created from temporal analysis software packages such as the AGI software suite.
“Without this, it is challenging to build a representative model of mission truth within the GIS environment,” Smith said. Currently, major efforts are devoted to creating and recreating translators from temporal analysis tools to static, time-stepped data models within GIS. Some of these translators are stovepiped, and often user specific, which makes extensibility difficult. Moreover, inefficiencies and inaccuracies can occur with the translation of this data. After the development and acceptance of a temporal GIS data model within DoD and IC, real interoperability between systems, temporal analysis tools and GIS would have the opportunity to flourish.
“This interoperability could drastically reduce cost for the fielding of systems, increase analysis accuracy, and also increase speed of response within operational programs,” Smith said.
COMPUTER VISION
At Sarnoff’s Vision Technologies Division, meanwhile, the focus is on computer vision work in application areas such as aerial video surveillance, persistent surveillance, video data mining, video compression, robotics and medical imaging.
“Since its inception, the Vision Technologies group has passionately supported the maxim to be the best in what we do and make real-time computer vision work in the real world” said division head Peter Burt. “Applying this approach to today’s new demands for rapid and incisive geo-intelligence gives us the opportunity help our government customers get the insight they need, when they need it.”
As a technological discipline, computer vision is closely connected to geospatial intelligence. Computer vision systems provide critical enabling technology for exploitation, analysis and fusion of imagery with geospatial information. Techniques such as motion analysis, object detection and tracking, geo-registration, and scene analysis transform pixels from video, imagery, LiDAR and other sensors into geo-referenced building blocks for both real-time and incisive analysis.
Some technology gaps exist in current vision technologies that need to be improved, however, and Sarnoff approaches these gaps with a twofold strategy. The first is to collaborate with organizations such as the Office of Naval Research, Intelligence Advanced Research Projects Activity and Defense Advanced Research Projects Agency on state-of-the-art R&D. The second is to work directly within the military and intelligence communities on projects for maturing technologies in operational conditions.
One example of a technology gap is the current lack of effective search and retrieval tools that allow defense and intelligence analysts to quickly and easily interact with vast repositories of content generated from collections of video and imagery from UAVs and newly emerging wide area persistent surveillance (WAPS) sensors. While some limited approaches exist to integrate this motion imagery and WAPS data into the analytical environment, this approach requires an analyst to watch videos or wide area sensor output in a separate visualization, and manually correlate the content with other information available in a GIS environment, such as road networks, satellite imagery, GMTI and other vector or raster data.
In other cases, the rise of military operations in urban terrain has created the need for commanders, troops and analysts to have an unprecedented level of understanding of an area’s surroundings. However, sensing and modeling of territory is restricted by a number of factors in a hostile environment. An effective production pipeline must overcome these constraints and exploit and leverage available sensor data to rapidly generate accurate reference imagery and 3D models to meet situational awareness and operational needs of the warfighter. ♦