Most who have tried to reconstruct the logic that put NPIC and DMA together into the National Imagery and Mapping Agency have concluded that it was the potential, profitable convergence of imagery and geospatial processes and products. And, while it is but a few years since the inception of NIMA, it is disturbing, nonetheless, that convergence has not occurred more rapidly and more completely. There remains the cultural divide between the Imagery Analysts (IAs) and the geospatial analysts (geographers and cartographers, by another name.) Is it merely human nature to resist such change, or perhaps that the presumed competition between the two groups or functions would inevitably produce winners and losers? Or, is there something more fundamental, some logic that would keep separate the two functions? Have we just failed thus far to find the unifying theme(s)?
Belief in the convergence of imagery and mapping is not limited to this side of the Atlantic. Less than a year ago it was announced in British Parliament that the Defense Geographic and Imagery Intelligence Agency (DGIA) would be formed by merging JARIC and Military Survey--respectively, the NPIC and DMA of the UK. Each, of course, has its own history and culture: JARIC dates from the Second World War, while Military Survey recently celebrated its 250th anniversary. The logic of the merger was that
[benefits] will come as digital technology allows the work of the agencies to be increasingly integrated in future, including the production, storage and handling of similar sorts of data....It is not just increasingly common sources of data and developing digital processes that are pulling the two agencies together. There is also an increasing requirement for the agencies' outputs to contribute to a common intelligence picture required by their defense 'customers' ...8
Imagery and geospatial activities, now housed in one organization, NIMA, vice two--NPIC and DMA--continue to elude one another to a large extent. Putatively, the vision behind the amalgamation of the two organizations was the emerging construct of geospatial (digital) data that could intellectually encompass imagery and imagery analysis. This is vexing to some, while reinforcing the biases of others. Still, it is time to question the fundamentals of the assimilation argument.
A digital dataset of geospatial consequence has certain characteristics. Each "record" contains coordinates that relate it to a point, line, surface or volume about the geosphere. For most items, there is strong data typing, wherein the respective data types (or features) relate to interesting human activities and permit interesting operators to work on the items.
The dataset may include rivers and marshes, mountains and valleys, political jurisdictions, and the road to grandmother's house. The dataset can be displayed as a "map" with which we can facilitate any number of human activities. Each "record" in the dataset should also be "time-tagged" as well as geospatially referenced.
So, is an image such a dataset? Or, is it such a datum? A picture of grandmother can be geospatially referenced so that it can be viewed by clicking on grandmother's house's location on the map. How about a reconnaissance image, perhaps one from which the map was "made"--i.e., one from which the digital dataset was extracted. It, too, can be geospatially referenced and accessed via the "map," but is it more than that?
From a GIS perspective, this discussion is reminiscent of arguments about the natural superiority of raster-over digital datasets, or the reverse. To the simplistic map user, the map is "the thing" and the digital dataset is a necessary evil, about which the less heard the better. To the GIS advocate, the digital dataset is "truth" and the map is just a view of the dataset, rendered, usually, by "rastering." However, the image from which the digital dataset features may have been extracted (i.e., from which the map was made) cannot be "created" (or even "recreated") by a rendering (rasterized or otherwise) of the digital GIS dataset.9
In a totally uninteresting sense, of course, the image--as it was erected on the focal plane of the reconnaissance satellite--was pixilated and digitized by the CCD array and captured as a two-dimensional array of numbers, which incidentally are of most interest to a rastering display device. Sufficient meta-data are captured and associated with the image to describe the "camera model," the time of acquisition, the ephemeris data of the collection vehicle, and the pointing angle--that, together with information about the earth's rotation--can translate into geocoordinates of the image (and its pixels.) As a database element, an image is rather unremarkable.
However, an image is something that eons of tinkering with the human hardware and software have allowed us to collect and interpret (task, process and exploit) "with the naked eye." Consequently, an image has a primary place in our consciousness. We can relate to an image in precortical ways that we cannot relate to a map. On the other hand, over those same eons, we have acquired the capability to extract features from an image and render it so as to be able to communicate (disseminate) it to others. We have also acquired the capacity to compile geospatial datasets not only from images but from our own wanderings and from words about the wanderings of others--simply, we have learned to sketch maps.
Finding, with the help of today's technology, easier and more useful ways of moving between images, GIS datasets, renderings, and words is the key to removing today's constraints on today's TPED. Seeking convergence between cartography and imagery analysis--and merging more closely together their respective work--is particularly promising.
The products are converging, most demonstrably in "image maps" where vector data sets--road and telecommunications networks, say--are overlaid on orthorectified imagery. The advantage of such products, inter alia, is that a dated vector data set can be overlaid on an up-to-date image, allowing the end-user to "update" his perceptions. Another, compelling example of the power of fusing vector data with imagery is to "drape" the image (or pieces of several) over a terrain model to create the now classic "fly-throughs."
The systems, too, are converging. IEC, the replacement terminal for the IDEX soft-copy imagery analysis system, will have the vector capabilities better known to the modern cartographer as well as the imagery analysis functionality more familiar to the IAs.
There is reason to believe that imagery analysts can move to a higher plane if they have some of the arrows in the cartographer's quiver. And, of course, for NIMA, the more in tune with intelligence analysis the cartographer becomes, the more valuable to the enterprise he or she becomes.
The "electronic geographer"--i.e., today's cartographer, creator of GIS datasets--exploits a satellite reconnaissance image by finding, measuring, and recording natural and cultural features of interest. This extraction of "feature sets" is highly stylized and is made measurably easier if the image is a soft-copy image and if the computer has a relatively simple toolkit that references points and clicks to the image's coordinate system--i.e., georeferences the selected features--and provides a set of menu picks that embody the vocabulary of cartography--e.g., unimproved roads, bridges, etc.
The cartographer is all about making accessible a set of geographic information, which can be used subsequently--generally by others as yet unspecified--to accomplish a task. The cartographer is about making a "map", by which an aviator might navigate, or a real estate developer might site a shopping center, or an armchair traveler might experience exotic places.
By contrast with the geographer, the image analyst is about "storytelling"-like the legendary native scouts who could read subtle signs in the dust to recount the passage of game or interpret the activities of those who had camped there previously. In fact, however, the image analyst also "extracts features" such as the size and shape of new military construction, the extent and character of security fencing, and the direction of tank tracks through a trackless waste. Frequently, the extraction of these features is made easier for the imagery analyst by software tools that look suspiciously like those of the cartographer--and yielding deliciously similar digital data sets.
Alas, our image analyst does not generally regard the digital data set so derived as a product; it is frequently reduced to a textual description in an intelligence report. In this translation to intelligence prose, considerable information--all the bits and bytes that might support rendering a "real" picture vice a word picture--is lost to posterity. Worse than posterity, it is unavailable when that subject military facility is next imaged and must again be exploited, perhaps by the selfsame imagery analyst, who rereads her previous report and recreates in her mind's eye the picture.
In fact, we could capture much of the exploitation as digital datasets that would support:
The technically inclined reader will note that such a derived digital dataset supports the ultimate in smart bandwidth compression. It permits faster dynamic overlays of historical images, and can more easily travel the "last tactical mile." Automatically compatible with ELINT-derived datasets, it advances us toward the holy grail of "multi-INT" TPED.
To reiterate, a principal reason for the creation of NIMA was the recognition of the benefits of imagery and geospatial integration. The Commission has heard anecdotes of such integration (e.g., specialized, tailored products for areas in the Balkans were developed), but was unable to find evidence of a strategic plan to make such cooperation routine. A recent study sponsored by the ADCI/Collection indicated that GIS tools that link diverse information to physical locations via layers could improve analysts' understanding of their intelligence problems. Such tools can also improve multi-INT analysis, if the data are presented in the proper format. In addition, use of such tools and the collaboration of analysts and collection managers can improve collection planning and efficiency.
The imagery and geospatial community is in the process of replacing its primary image-exploitation workstation, IDEX.10 The goal was to finally move away from the light-table exploitation of film and toward soft-copy exploitation by computer. The technical challenge has always been the "need for speed." While just how big our satellite images are is classified, suffice it to say that they are Big! And they have gotten bigger just as computers have gotten faster. Simply rendering, panning, zooming, and rotating such images has remained just slightly beyond the reach of affordable desktop computers for two decades.11 Ultimately successful, IDEX was a troubled development of custom hardware and software with display power still beyond commodity desktops. It has come to incorporate a number of powerful raster-image manipulation algorithms. It does not, however, support the more commonplace vector manipulations favored by Geospatial Information Systems (GIS). So, unfortunately, it does not promote the desired convergence of disciplines.
The latest-generation IDEX "replacement" is IEC. It does support the GIS operations. Good! It does not, however, quite match the custom-designed raster-image capabilities of IDEX. Bad! Unless it is modified so it does, the fingers of the hardcore imagery analysts will have to be pried from their IDEX stations. Without widespread and enthusiastic acceptance of IEC or equivalent, the promised convergence of imagery intelligence with mapping, charting, and geodesy will remain an unrealized dream.
[The story you are about to read is true. All the details have been changed by "security."]
Washington, DC--Imagery Analysts (IAs) face the daunting task of searching a large, denied area in order to locate particular pieces of deployable military hardware. The alternative of taking high-resolution satellite imagery of the entire country and searching it, square meter by square meter, is prohibitive. Sufficient imaging capacity to do the job cannot be freed up, nor would it be feasible to image the entire country in a sufficiently short space of time to be confident that the hardware had not redeployed, hop scotch fashion, from as-yet-unimaged locations to previously imaged locations, in the interim. In any event, sufficient IA-hours are not available to conduct so brute-force a search.
St Louis, MO--Geospatial Analysts review the geography, topography, and cultural features-- road, rail, and power networks; hills and dales, forests and clearings--correlated with previous sitings (sightings) of such equipment. A factor analysis later, the Geospatial Analysts prepare a "map" (vector dataset) that provides the template for where to search--where to image and where to exploit.
Washington, DC--The IAs get the picture!
But, do they really get the picture? Is this a story about IAs who "subcontract" for collateral information? Or, is this a story about the ascendance of the Geospatial Analysts who, faced with a vexing intelligence problem--"map" the locations of subject hardware--and proceed to produce said map, showing probable future- and confirmed present-sites, with workaday assistance of trained eyeballs (to be replaced, when cost-effective, by computerized pattern recognition)? Or, is this a triumph of "collaboration?" Or, does it presage the next generation of intelligence professionals, schooled in both imagery and geospatial analysis disciplines?
More generally, NIMA is examining the feasibility of collocating regional specialists to encourage better integration of imagery and geospatial information. The Commissioners were made aware of a planned "experiment" to integrate Latin America imagery and geospatial analysts, i.e., collocate those analysts who are Latin American specialists. The Commission lauds this "experiment" but urges NIMA to include the experiment as part of the larger development of a geographic information database. Furthermore, NIMA should set explicit goals and performance metrics to determine whether collocation and integration works, how well it works, and how it may be extrapolated to other parts of NIMA.
Imagine being able to unfold a paper map and look at it "through the lens" of a computer network appliance. Suddenly the paper map would spring to life, show terrain in 3-D, show moving mobile SAMs actually moving, and see their effective threat envelope as upside down sugar loaves. And, as you moved the paper map from side to side, or rotated it, the "erected" data images would move in synchrony, allowing you to view the terrain from any perspective. Just such technology is emerging from the laboratory. Augmented or mixed reality (AR) research aims to develop technologies that allow one to mix or overlap computer generated 2-D or 3-D virtual objects on the real world. Unlike virtual reality that replaces the physical world, AR enhances the physical reality by integrating virtual objects into the physical world, which become in a sense an equal part of our natural environment.12
This fusion of computer-generated visualizations of vector data sets and paper maps is particularly intriguing as it may allow us, literally, to overlay new technology on legacy products. And, of course, it can be "multi-INT," fusing additional data derived from HUMINT and SIGINT. From the user's point of view, an especially appealing characteristic of such a "magic map" is its graceful degradation in the face of computer malfunction. We have augmented the map with computer-generated displays, but, if all else fails, the old standby map is as effective as it ever was. Moreover, the ability to overlay vector data onto maps in this way allows the soldier to simply mark up his map with traditional symbology without having to shift his gaze or attention away from the paper. Imagine sending an update to be marked on a map without having to use coordinates--sending, as it were, directly to the eye of the soldier who needs to annotate his map, or to the navigator or mariner who needs to update his chart.
8 (UK) Select Committee on Defense Fifth Report--THE DEFENCE GEOGRAPHIC AND IMAGERY INTELLIGENCE AGENCY.
9 In a technical sense, we have lost some information when we "transformed" the image into the vector data set (but, hopefully no interesting information). Of course, working with the vector dataset we also add other information.
10 The roots of IDEX go back at least a quarter-century to a research effort, IDEMS, conducted by CIA's since-disbanded Office of Research and Development (ORD) on behalf of CIA's since-absorbed National Photographic Interpretation Center (NPIC). IDEX can also trace its roots to the Air Force COMPASS COPE effort at Rome Air Development Center (RADC).
11 A lot of tricks have been tried. In the BR-90, Bunker Ramo (several times removed from TRW) married film projection with CRT technology and vector graphics. Rotating the whole CRT was also in favor briefly.
| Executive Summary and Key Judgments
| Introduction | NIMA
from the Beginning