Title: "A New Paradigm for GIS Base Mapping Data Collection: The Space Imaging Approach"

Abstract

Traditional methods for GIS base mapping data acquisition are about to change. In 1997, 1-meter resolution satellite imagery of the earth's surface will become available to commercial customers on a regularly scheduled or customer-specific basis. Beneficial aspects of satellite-derived imagery include rich, multispectral information content, high-accuracy, timeliness, frequent revisits, digital feature extraction, automated change detection, digital correction and rectification, and terrain modeling.

New methods for imagery capture will allow for customer tasking of satellites to obtain specific areas of interest within a matter of hours. In addition, when satellites are not specifically tasked, they may capture imagery that will be archived and made available to customers shortly after receipt of order. Such archives will also allow customers to order temporal sequences of respective data.

Many GIS applications require high-accuracy imagery that meets national map standards of 1:24000 without ground control. With ground control, image accuracy should be as high as 1:2400. These applications may also require color imagery with resolution as high as 1 meter.

Introduction
Why Digital Orthoimages?
Overcoming the Digital Ortho Myth
The Space Imaging Approach
Experience and Proven Technology
The Space Segment: The Satellite and Its Sensor
The Ground Segment: High-Speed Image Processing and Delivery
Conclusions

The need for information drives the development of mapping technology. Cave paintings dating back thousands of years suggest that early humans drew crude maps in attempts to gain a perspective of the land around them and their place in it. These people soon learned that higher ground provided them the overview they needed to gain additional insight into their surroundings.

Not surprisingly, mapping technology in modern times has evolved closely with other methods of earth observation, such as remote sensing. The unique perspectives of balloons, airplanes, and satellites have drastically increased the amount and quality of information that can be represented on a map.

Today, remote sensing is leading an information revolution in mapping that is moving these technologies into new applications in the workplace and the home. At the heart of this revolution is the digital orthoimage - a photographic image that has been rectified to meet the precision and accuracy standards of a map.

The need for information, of course, helped ignite this mapping revolution. Until a decade ago, aerial photographs and digital satellite images served as the primary sources of information from which appropriate vectors were extracted to create thematic maps. Most of the information, however, was left unused in the raster photo or satellite image and was usually discarded.

In fact, too much information was being lost. "Save the information" may well have become the battle cry of the orthoimage revolution.

About ten years ago, computer processing and memory capabilities became much more affordable, for the first time making orthorectification a viable process for mapping projects of virtually any size. In orthorectification, the distortions induced by the imaging platform, film and three-dimensional shape of the earth are digitally removed from the image or aerial photograph.

The final result is an image or photograph that has the precise geometry of a map. All of the information contained in the original raster image or photo is retained in the orthoimage. Compared to a vector map, the appeal of the orthoimage is the quantity of information it holds. Anything that can be seen in the image or photo is also seen in the orthoimage. And because the information has not been filtered through a cartographic interpretation, it remains unbiased.

Late next year, the quantity and quality of information available from orthoimagery will once again improve dramatically with the launch of high-resolution remote sensing satellites capable of acquiring earth imagery with one-meter spatial resolution. These satellites will answer the call for information that is "faster, better and cheaper."

Space Imaging, the developer of one of these high-resolution earth imaging systems, will leverage the 35-year satellite development experience of Lockheed Martin Corp. and the unrivaled image processing capabilities of Raytheon's E-Systems Inc. to provide GIS and mapping customers with a new generation of orthoimagery and other geographic information-related products.

Space Imaging has designed its satellite and ground system, as well as its entire operating scheme, around a new approach to viewing and using earth information. This concept will enable customers around the world to obtain accurate, high-resolution, off-the-shelf images without paying the high costs traditionally associated with custom imagery collection.

To properly understand and appreciate the advantages of creating orthoimages from satellite imagery, we must first examine the general benefits provided by orthorectification. As can be expected, these benefits relate directly to the need for information.

In many ways, today's project requirements extend beyond a simple need for information. To have practical value, information must be accurate, unbiased, interpretable, current, consistent, and in most situations, affordable. These factors are the characteristics of the orthoimage and set it apart from a traditional vector base map.

Accuracy - In the GIS environment, the base map plays the crucial role of providing geographic perspective. The overall purpose of GIS is to locate, identify and manage spatially distributed features in a comprehensive manner. Without an accurate base map, the relationships among these features cannot be determined, and the value of the information is diminished as a result.

An orthoimage should be thought of as an intelligent base map . It not only provides a correct geographic perspective in which to view information - it is also the source of that information. In the orthoimage, therefore, the relationships among land features, buildings and transportation networks, for example, are presented in their natural state without having been skewed by data conversions and interpretation.

Orthoimages made from high-resolution photos or images can achieve accuracies of better than two meters. Only very expensive field survey techniques can provide such accuracy. In many GIS, the raster-based orthoimages now serve as the backdrop with which older vector coverages are updated and corrected.

Bias - A major drawback in vector mapping is that human interpretation must be relied upon during the scanning and digitization of features. As soon as a technician decides to include one feature and omit another, the resultant map has been biased. With a photograph or image, on the other hand, no such bias is introduced.

Interpretability - Interpretation of some variety will always play a role in mapping, but it is best left to the end user, not the map maker. An important benefit of orthoimagery is that the information is accessible for nearly anyone to interpret. The orthoimage is a picture that an untrained observer understands intuitively. For instance, anyone can find his or her house on an air photo, see how close it is to a rain-swollen river, and assess the available options.

Accessibility - Today, organizations that rely on aerial imagery derived from existing sources are at the mercy of those sources' methods of delivery and format. Their delivery vehicles and processes required to obtain the desired imagery can be long and cumbersome, making up-to-date imagery less accessible. By contrast, orthophotos are available quickly and easily from off-the-shelf image "inventories." They also come in variety of digital formats, making delivery fast and efficient.

Currency - To say that information loses value with age is an understatement. Out-of-date map information can have significant negative impacts on virtually any planning project conducted in a GIS. Natural processes and human intervention are changing the surface of the earth faster than maps can keep up.

Off-the-shelf base maps are often financially attractive, but the bottom line is that Digital Line Graphs, TIGER files and even commercial products such as ETAK street maps are based largely upon U.S. Geological Survey maps, which can be several years old. Orthoimages, conversely, can be created from aerial photos or satellite images acquired just weeks before their use. Acquisition flexibility even allows the user to have the images collected during a specified time of year.

Satellites offer a tremendous advantage over aircraft in this area. Satellites can collect and deliver orders within days or weeks, while it may take four to six months to implement an aerial survey.

Consistency - Orthoimages offer outstanding consistency over large areas because the photographs and images used to create them are acquired under constant conditions, and an accurate digital terrain model (DTM) is used to systematically remove geometric errors in the image. Commercial base maps, however, are often created from several different sources of map data with widely varying levels of detail and scale.

Cost - Conventional planimetric mapping based on photogrammetry can cost $1,000/square mile, and an orthoimage created from aerial photography with one-meter resolution and two-meter accuracy can cost up to $300/square mile. Space Imaging plans to provide a satellite-derived product with superior resolution and accuracy for $100-$150/square mile.

Although in some cases the orthoimage may exceed the cost of commercial base map products, the cost-to-value must be considered: the additional information, accuracy, flexibility and quality of an orthoimage are intangible benefits that more than offset the financial cost difference, if any.

Satellite imagery has been slow to gain acceptance as the data source from which orthoimages are created. This moderate acceptance results from several myths regarding the applicability of satellite imagery for this use. As a result, many users mistakenly believe that orthophotos are too expensive, will not fit into the vector-based GIS environment, are difficult to use, or require specialized hardware and software systems.

In developing its new approach to satellite remote sensing, Space Imaging has considered each of these myths. The company's satellite, sensor and ground systems have been designed so that orthoimages and other GIS products can be created inexpensively in formats that can be used immediately in virtually any environment.

Space Imaging customers will find that satellite images are indeed less expensive than aerial photographs for the creation of orthoimage products. Space Imaging will capitalize on economies of scale to offer orthoimages at a significantly lower cost compared to customized mapping projects. This results partly from the fact that Space Imaging's products are available off-the-shelf from a digital archive. As such, multiple customers can purchase the same imagery, thereby effectively sharing the cost of collection.

Aside from dollar values, GIS users should also examine satellite imagery from a cost-benefit perspective. The overhead costs of obtaining a satellite image do not vary with project size. This means that even a small area can be imaged repeatedly with a satellite for timely updates that would be cost-prohibitive with an aircraft. Therefore, the value of satellite imagery in the GIS far exceeds that of comparable aerial photography.

Compatibility of raster-based satellite imagery with vector GIS environments is no longer an issue. During the past few years, all major GIS software vendors have integrated raster data handling capabilities into their systems.

The processing of satellite imagery is more complex than that of aerial photography, but this should be transparent to the user if the final product is packaged properly. For instance, Space Imaging will deliver products in formats that support variable tiling into manageable file sizes that can be handled by a desktop PC with eight megabytes of memory or more. Space Imaging products will arrive at the customer's door in a GIS-compatible format and registered to the coordinate system specified by the customer.

Specialized software no longer will be a prerequisite for the use of satellite-derived image products because the major GIS and image processing developers are preparing their software packages to efficiently handle the high-resolution systems. Nor will it be necessary for users to hire imaging or remote sensing specialists in order to take advantage of satellite imagery.

In addition to overcoming these myths, Space Imaging has also addressed the general needs of the orthoimage user - accuracy, lack of bias, interpretability, accessibility, currency, consistency and high value. For projects requiring one-meter resolution, Space Imaging orthoimages will offer all of these advantages over aerial photos.

Delivering map information that is faster, better, and less expensive requires superior technology and a new approach to business.

Central to Space Imaging's new business concept is the understanding that customers around the world, regardless of their location, need immediate access to high-resolution earth imagery at a competitive cost. The following are key features of the Space Imaging approach to satellite remote sensing that will make this concept a reality:

Global Archive - Space Imaging is not waiting for the satellite launch to fill its archive. The archive is being built right now with extremely accurate, high-resolution orthoimages based on aerial photography. Once the satellite is launched, newly acquired information will complement the existing products for the development and maintenance of a globally distributed information archive.

Quick Access - When fully implemented, a global archive will mean that most earth images will have already been collected and processed by the time a customer places an order. These off-the-shelf images can be delivered to the customer within hours or days.

Custom Collection - Many applications require new acquisitions due to time sensitivity, and these orders will be handled quickly.

Competitive Price - The vast image archive will allow Space Imaging to sell the same imagery to multiple users, thus capitalizing on economies of scale. Customers avoid bearing the full cost of image collection, and the cost per image is driven down. Users can expect to pay $100-$150/square mile for Space Imaging orthoimage products.

Partnerships - GIS users do not depend on one-meter imagery alone. To accommodate applications requiring sub-meter resolution, Space Imaging is developing partnerships with leading aerial mapping companies to acquire airborne imagery for inclusion in the Space Imaging product line.

Value-Added Products - Space Imaging will offer a complete line of enhanced and merged information products based on satellite imagery and additional sources of geographic data.

Affiliates - International ground receiving stations will be owned and operated by multinational companies and national governments. These stations will operate as businesses seeking to provide clients with competitive products and services worldwide.

Superior technology plays a key role in the new business paradigm. Space Imaging is capitalizing on the unparalleled experience of its strategic partners, Lockheed Martin and Raytheon's E-Systems, in developing advanced satellite and ground processing segments that will fulfill all of the mentioned user needs and dismiss any lingering myths about satellite-derived digital orthophotos.

The following are important technical capabilities of the Space Imaging system:

High Resolution - The satellite will acquire images with a 0.82 meter ground sample distance (GSD) at nadir. The entire satellite will be able to pivot in orbit to collect cross-track imagery to a distance of 725 km on either side of the ground track. Due to the satellite's 680 km altitude, imagery will maintain at least a one-meter GSD for 350 km to either side of nadir, or a 700 km swath of at least one-meter imagery.

Stereo Imagery - The satellite will collect in-track and cross-track stereo images in the same pass. Stereo pairs created from same-pass images ensure extremely high quality because each image is acquired in exactly the same ground conditions. DTMs necessary for orthorectification can be generated from these stereo pairs using photogrammetric procedures and high quality ground control points (GCPs).

Accuracy - Imagery will have a horizontal accuracy of 12.2 meters without ground control and 1.5 meters with ground control, meeting National Map Accuracy Standards for 1:24,000 and 1:2400 scale maps, respectively.

Information Content - Imagery will have exceptional spatial detail. Each pixel will have an 11-bit dynamic range capable of displaying 2,048 gray shades rather than the 256 available today. In addition, the sensor will collect four bands of multispectral imagery at four-meter GSD. The combination of multispectral and panchromatic imagery will allow users to extract significant detail related to physical and chemical characteristics of earth features.

GIS-Compatibility - Orthoimages and image-derived product will be delivered in standard formats that will be fully compatible with most GIS. When ordering, the customer need only specify a desired format and media. Although the images are acquired and processed as 11-bit data, many products will be normalized as eight-bit products and delivered in tiled formats for easy use and access.

Ordering Flexibility - Space Imaging's globally distributed imagery archive will enable customers around the world to acquire off-the-shelf imagery and order custom collections from ground station operators in regional hubs. Orders can be placed and received at any one of several regional ground stations for products that are acquired anywhere on the globe. The command queue has been designed to accommodate acquisition orders in as little as two hours before the satellite passes over the target area.

Quick Delivery - In a best case scenario, a customer's desired image will be available in the archive for shipment within hours of being received from the satellite. For orthoimage requests, the average turnaround time will vary from seven to 14 days after acquisition, depending on the availability of ground control.

Consistency - Satellites cover very large areas on a single pass. As a result, a base map can be made from one or more images acquired under absolutely identical conditions, ensuring that the entire base map has consistent qualities. More importantly, the satellite sensor will be constantly calibrated so that images acquired years apart can be processed to accurately detect changes.

Achieving the goals and providing the user benefits outlined above require Space Imaging to take full advantage of its business partners' unmatched expertise and heritage of success in the acquisition and processing of earth information. This unique legacy gives Space Imaging a distinct advantage in its mission to become a world class supplier of satellite remote sensing imagery and information.

In designing the described one-meter imaging satellite, Space Imaging is drawing upon the 35 years of satellite building experience of Lockheed Martin - the company that pioneered high-resolution satellite imaging.

In the 1960s, the Central Intelligence Agency tapped the Lockheed Missiles & Space Co. to build a series of super-secret Corona satellites. After more than 12 years and 362 missions, these satellites kept tabs on the Soviet nuclear threat for the U.S. intelligence community.

Providing photographic imagery with two-meter resolution in the 1960s, Corona was remarkably advanced even by today's standards. Since then, Lockheed has fine tuned the technology that was already decades ahead of its time. This experience will allow Space Imaging to provide one-meter imagery with a level of detail and quality unrivaled by any other commercial system.

Acquiring imagery with one-meter spatial resolution is only part of the technical challenge in building a high-resolution imaging system. For a satellite image to have practical value, the user must be able to easily extract information from it. An image with a high level of interpretability is sharp, accurate and has high contrast.

Platform stability is one of the most important factors that ensures acquisition of a sharp image. The dynamics of every moving part in the satellite must be calculated and controlled with precision gyros so that unwanted movement does not blur the image. Aerospace engineers calculate a Modulation Transfer Function (MTF) for the entire imaging system to quantify image sharpness. Space Imaging has combined a large aperture with a very stable platform that will result in a very high MTF and extremely sharp images.

The size and quality of the sensor aperture are critical factors that influence image sharpness. A larger aperture allows more light to reach the detectors which contributes to a sharper image, especially in low light conditions. The Space Imaging satellite will include a 0.7 meter aperture, significantly larger than any other planned system.

Imagery must have very high contrast for the user to extract information from surfaces that are either very bright or dark. One way of increasing image contrast is to use detectors with a high dynamic range. This enables the detector to divide light into a greater number of individual gray levels. Space Imaging will have an 11-bit dynamic range, offering 2,048 shades of gray.

Space Imaging is utilizing its high-contrast panchromatic imagery to enhance the information content of its color imagery. The company's multispectral products, which are collected at four-meter resolution and also have a very high dynamic range, can be "fused" with the one-meter imagery to create a color product with an effective resolution of one meter.

Another method of increasing image contrast is to vary the exposure of the detectors. Space Imaging will utilize a process called Time Delay and Integration to adjust the dynamic range of the detectors from the ground in accordance with localized ground lighting conditions.

Altitude is also a factor that influences image interpretability. Aerial photography users are familiar with the "leaning out" effect of buildings away from the center of a photograph. This phenomenon decreases as altitude increases. Space Imaging will fly at 680 km, about 200 km higher than any other planned one-meter systems.

Another sensor feature that has an impact on overall image interpretability is radiometric accuracy. This refers to the accuracy with which each of the thousands of sensor detectors measures incoming light and records it as a digital number. Radiometric accuracy can be measured in several ways, but it generally gauges how accurately the image represents actual light conditions of the scene.

Superior radiometric accuracy results from detectors that remain extremely stable over time. Space Imaging has determined that its detector measurements will vary a maximum of 10 percent from actual light reflection conditions. From pixel to pixel, the variation will be only 10 percent, while the linear variation from true light values will be five percent. These percentages are equal to or better than Landsat 30-meter resolution data, which means Space Imaging will have the same radiometric accuracy at a much higher resolution.

Metric accuracy refers to the degree of certainty with which objects can be located in an image relative to their actual geographic position. The described satellite will provide images that meet National Map Accuracy Standards for 1:24,000 scale without ground control and 1:2400 scale maps with ground control.

At 1:24,000 scale, 90 percent of features are located horizontally to within 12.2 meters of their actual position. For 1:2400 scale, the same percentage is located to within two meters of their correct location.

Accuracy begins with the internal geometry of the imaging system. Traditionally, remote sensing platforms have sacrificed geometric accuracy for radiometric accuracy. Space Imaging is developing an imaging system that is so robust it trades neither type of accuracy. The sensor is a photogrammetric imaging system that achieves such high levels of geometry that minimal GCPs will be needed.

GCPs can dramatically improve the accuracy of any satellite image, but they also can substantially increase the cost of the image. Collecting ground control measurements can be expensive endeavors in poorly mapped areas. A minimal number of GCPs should be sufficient to correct an image.

Accuracy without ground control becomes an important issue in those parts of the world where satellite-identifiable GCPs cannot be obtained. In the absence of GCPs, ground processing needs two important pieces of information to achieve the desired 1:24,000 scale accuracy. The satellite's position must be known to within three meters, and the line of sight in which the optical system is pointing during image acquisition must be known to within two arc seconds. Space Imaging will employ differential GPS and multiple star trackers to achieve these results.

Raytheon's E-Systems has more than 20 years of experience collecting, storing, processing, displaying and exploiting digital imagery. E-Systems has designed many segments of the Space Imaging ground processing system. This high-throughput facility will turn around new acquisition orders within days. The 10-year image archive will have a capacity of 250 petabytes (million megabytes), equivalent to 3.3 million scenes.

Space Imaging's business plan has made timely access to data a priority. To ensure that customers can have quick access to new images, E-Systems has designed a distributed ground station network. There will be transmit/receive ground stations strategically located in the Arctic, North America and around the world to provide real-time global coverage. A user can place an acquisition order less than two hours before the satellite passes over the target area.

Each of the ground stations will be identically equipped with high-throughput processing facilities capable of creating image products within two hours of acquisition. One key to quick turnaround is the downlink capability. The satellite will downlink image data at a rate of 320 megabits per second, ensuring that all image data collected on a given pass is downlinked during that pass. That rate is roughly equivalent to eleven video signals being sent simultaneously.

High-resolution satellites will lead mapping into the next phase of the orthoimage revolution. For mapping and GIS projects requiring one-meter resolution, orthoimages derived from satellite imagery are the most appealing alternative in terms of cost, accuracy, availability and currency.

Customers do not have to wait until these high-resolution satellites are launched in late 1997 to enjoy the benefits of this type of high quality imagery. These products are available right now. Space Imaging has already begun building its vast archive of high-resolution, accurate earth imagery.