Remote Sensing

Characteristics and Applications of High-Resolution Space Imagery1199rem5

An artist's impression of the IKONOS satellite in orbit (Space Imaging)

Prof. Gordon Petrie

1. Background

In this paper, high-resolution space imagery is defined as being optical imagery acquired using a space sensor with a ground resolution lying in the range 1 to 5m. The acquisition and potential use of this type of imagery is a subject that has gradually come to the fore within the civilian mapping and remote sensing communities over the last ten years or so. The trigger for developments in this particular area were some quite unexpected decisions on the part of the Russian government in the wake of the break-up of the Soviet Union in the late 1980s.

1.1 Release of Russian High-Res Photos

Thus, in 1987, not too long after the entrance into service of the SPOT satellite producing imagery with a 10m ground pixel size, came the Russian decision to allow space photography taken using its KFA-1000 camera to be sold on a world-wide basis. This photography had been taken originally for intelligence gathering purposes and had a true ground resolution in the range 5 to 10m. This was followed by the even more unexpected decision by the Russian government in 1992 to allow the sale of still higher resolution space photography. This had been acquired with the KVR-1000 and KFA-3000 cameras producing photographs having a ground resolution of 2 to 3m.

1.2 American Responses

Prior to these developments, under Presidential Directive 37 of the Carter administration issued in 1978, the US government had restricted the ground resolution of American space imagery to 10m - as in the case of the Large Format Camera (LFC) launched in 1983. The result of the first Russian decision was the easing of this restriction by the Reagan administration. After the second Russian decision, came a further response by the Clinton administration. Under Presidential Directive 23 issued in March 1994, the development of commercial satellites producing imagery to the 1m ground pixel level was allowed. On this basis, licences were issued to several American-based companies that had previously been suppliers of hardware and software to military space reconnaissance programmes.

1.3 Corona Space Photography

A further response from the US government was the decision taken in February 1995 to declassify and release into the public domain 860,000 high-resolution space photographs acquired between 1960 and 1972 through the Corona satellite reconnaissance programme. These photographs are now archived at the EROS Data Center located at Sioux Falls in South Dakota. Undoubtedly they should come into considerable use in the future, since, for many parts of the world, they offer a unique baseline for the monitoring of environmental change. With ground resolutions of 2 to 3m, they also need to be considered in the context of the present discussion.buckpal

A false colour image of Buckingham Palace in London produced from a combination of a Russian KVR image and a SPOT-XS image (Nigel Press Associates)

1.4 Recent Events

Till now - and rather unexpectedly given the considerable prior experience of the companies involved - the civilian mapping and remote sensing communities are still awaiting the first images from the commercial high-resolution satellites. In December 1997, the EarthWatch company launched its EarlyBird satellite successfully, but it then failed four days later. Shortly afterwards, in January 1998, the launch of the EROS-A satellite developed by an Israeli-American group (IAI/Core) failed. Then, in April 1999, the IKONOS-1 satellite from Space Imaging failed to go into orbit. Now, with the successful IKONOS launch, the first commercial high-resolution space imagery should become available to users. Thus it is opportune to discuss the characteristics of the different sensors and the resulting imagery and to discuss its potential markets and applications.

2. Space Photographic Cameras

Film based cameras have been used extensively by US and Russian intelligence and mapping agencies. The main characteristics of these cameras are as follows:-

(a) They are usually operated on short duration missions lasting two to four weeks, although the older American Big Bird (KH-9) satellites had a much larger film supply and a much longer operational life.

(b) The exposed film needs to be recovered for processing and delivery to the users. In the case of most US missions, this recovery operation took place using special capsules or canisters that were ejected from the satellite. Alternatively the whole satellite can be recovered - as with the current Russian missions - in which case, the camera and its auxiliary equipment can be refurbished and reused.

Table I - Characteristics of Space Cameras and Photography

Film Camera Type Format (cm) Focal Length (m) Angular Coverage Flying Height (km) Ground Coverage (km) Photo Scale Ground Resol. (m) Orbital Inclin- -ation B:Ht. Ratio
KVR-1000 18x18 1.00 8.5x8.5š 220 40x40 1:220,000 2 67š -
KFA-3000 30x30 3.00 6x6š 270 27x27 1:90,000 2-3 83š 0.04
KFA-1000 30x30 1.00 17x17š 270 80x80 1:270,000 5-10 83š 0.12
TK-350 30x45 0.35 46x65š 220 190x280 1:630,000 7-10 67š 0.52
MK-4 18x18 0.30 33x33š 280 160x160 1:930,000 10 83š 0.24


Table II - Characteristics of Space Scanner Imagery - Pan Sensors Only

Scanner System Sensor Array Type Orbital Height (km) Swath Width (km) Ground Coverage (km) Ground Pixel (m) Pointing Along Track Cross Track Orbital Inclin- -ation B:Ht. Ratio
SPOT Linear 822 60 60x60 10 No 27š 98.7š Up to 1.0
IRS-1C/D Linear 817 70 70x70 6 No 26š 98.7š Up to 1.0
MOMS-02 Linear 296 78 78x78 13.5 21.4š No 28.5š 0.8
MOMS-2P Linear 380/405 97/105 100x100 18 21.4š No 51.6š 0.8
JERS-OPS Linear 570 75 75x75 18x24 0(/15.3š No 98š 0.3
EarlyBird Areal 475 6 6x6 3 30š 30š 97.3š Variable
QuickBird Linear 600 22 22x22 1 30š 30š 66( Variable
IKONOS 1 Linear 680 11 11x11 1 45š 45š 98.1š Variable
OrbView3 Linear 460 8 8x8 1 45š 45š 97.3š Variable
EROS-A Linear 480 12.5 12.5x12.5 2 No ? 53š -
EROS-B Linear 600 16 16x16 1 Yes 45š 98š Variable

(c) The satellites engaged in these short-duration missions are placed in a Low Earth Orbit (LEO) with orbital heights in the range 200 to 270km. The cameras used in military reconnaissance flights have been operated from still lower altitudes. Thus space cameras are usually orbited as close to the Earth as possible to ensure the acquisition of high-resolution images.

(d) Furthermore the cameras used on these missions are equipped with long focal length lenses (f=1 to 3m) with a narrow angular coverage. This also helps to ensure that the resulting photographic images have a high ground resolution.

(e) If the photographic images are required in digital form, then special large format scanners may be needed for the conversion operation. In this respect, the Russian KFA-1000 and KFA-3000 cameras employ a 30 x 30cm format, while the American LFC and Russian TK-350 cameras have large rectangular formats of 23 x 46cm and 30 x 45cm respectively. These contrast with the standard 23 x 23cm format used in aerial mapping cameras and in most photogrammetric scanners.

The detailed characteristics of the current Russian space cameras and the photography taken by them are set out in Table I. From this, it can be seen that the photographs taken by the KVR-1000 and KFA-3000 cameras are those that fall into the category of high-resolution imagery having ground resolution values of 2 to 3m. Whereas the photographs taken by the KFA-1000 and TK-350 cameras fall into the class below having a medium ground resolution of 5 to 10m.

3. Space Scanner Imagers

Starting in the late 1970s, US military reconnaissance programmes were the spur for the original development of linear and areal arrays of CCDs to provide imagery from space in digital form. From the mid-1980s onwards, civilian examples of scanners using these arrays and providing images with a medium ground resolution have been deployed by a number of countries - e.g. France (SPOT); India (IRS-1C/D); Germany (MOMS), etc. The forthcoming commercial high-resolution satellites using scanner imaging technology are mainly American - e.g. those from EarthWatch, Space Imaging and Orbimaging. However the EROS satellites and imagers are being constructed by a consortium of Israeli and American companies - IAI and El-Op from Israel (who built the Israeli Ofeq reconnaissance satellite and its imager respectively) and Core Software Technology from the USA. This consortium operates under the umbrella of a company entitled West Indian Space (WIS) that is registered in the Cayman Islands.

The main characteristics of these new space scanner systems are as follows:-

(a) CCD linear arrays will mainly be used, operating in a pushbroom mode and providing digital image data.

(b) These scanner systems are designed to be operated on long duration missions - typically with a planned life of 3 to 5 years. The satellites will also be placed in very stable orbits to provide repetitive coverage as required for monitoring purposes.

(c) These long duration satellites need to be operated at much higher orbital altitudes than those used with space cameras - typically between 480 and 600km.

(d) In turn, this higher altitude leads to the use of lenses having a very long focal length - f=10m in the case of IKONOS - in order to produce the high-resolution imagery.

(e) A sophisticated and expensive ground station is needed to receive the digital image data transmitted from the satellite. If no storage or only a limited storage is provided on board the satellite, then an extensive network of such stations is needed if worldwide coverage is planned.

The detailed characteristics of spaceborne scanners and the imagery acquired by them are set out in Table II. From this, it can be seen that none of the existing satellite scanners - SPOT, IRS-1C/D, MOMS and JERS-OPS - that are in current operation produce high-resolution imagery. However all of the commercial high-resolution satellites that are planned for launch over the next few years will employ scanners that are capable of producing images with 1 to 2m ground pixel sizes.

4. Space Image Geometries

It is important, especially if mapping or monitoring operations are planned, to recognise the very different geometries that are associated with the different types of high-resolution space imagery. These have to be taken into account by users, with particular regard to the software packages that need to be employed to process and make use of these different types of

4.1 Camera Geometries

With space camera images, three rather different geometries can be identified, as follows:-

(a) With space cameras such as the current Russian TK-350 and MK-4 models (and the ESA Metric Camera and NASA LFC from the 1980s), planar type frame images are produced. Each image is acquired from a single exposure station in space and has a fixed orientation (in terms of tilts) that applies to the whole of the frame image. The frames can overlap in the conventional way to provide stereo-coverage (Fig. 1).

(b) Certain types of space camera such as the Russian KFA-1000 with its very narrow angular coverage are operated in pairs in a so-called split vertical (i.e. low oblique) configuration to provide a wider angular coverage of the terrain from a single flight (Fig. 2). These tilts have to be removed or compensated for if the images are to be used for mapping or monitoring purposes.

(c) Panoramic cameras are also used to provide a suitable combination of high-resolution and wider angular coverage of the terrain from space. Examples are the current Russian KVR-1000 camera and the older but similar American KA-80 cameras built by Itek. Also the KH-4 cameras that were used to produce the archival Corona photography were of this type. Exposure of the film is made by a moving slit passing in front of the photographic film. This gives rise to a cylindrical imaging surface (Fig. 3) on which each line is exposed from a different position in space with the possibility of varying tilt values occurring between them. As a result of the panoramic geometry, large differences in scale are experienced over the whole of the highly rectangular image format - 5.5 x 76cm in the case of the Corona KH-4; 11.5 x 115cm in the case of the KA-80; and 18 x 72cm in the case of the Russian KVR-1000 camera - with the long side of the format oriented in the cross-track direction. These large-scale changes have resulted in only the central 18 x 18cm part of the KVR-1000 photographic image being supplied to users - most often in digital form after undergoing a prior scanning

4.2 Scanner Geometries

All of the current and the forthcoming high-resolution space scanners employ linear arrays of CCDs as the imaging sensors. All of these are or will be operating in a pushbroom mode. Again three distinctive geometries can be identified - in this case, mainly on the basis of the geometry used to provide overlapping stereo-coverage of the terrain. This is designed to allow the interpretation of the terrain in 3D and the extraction of terrain elevation data either by manual/visual measurement or using automatic image matching techniques.

(a) In the case of the existing SPOT and IRS-1C/D satellites, overlapping images can be obtained from two separate flights using adjacent orbital paths (Fig. 4). The required overlap is produced through the use of a rotatable mirror placed in front of the linear array that alters the pointing in the cross-track direction during each orbital pass. Difficulties can then occur if the images are acquired with a substantial time gap between them. This can give rise to considerable differences in the appearance of the same piece of terrain - especially in respect of its vegetation and hydrology - arising from seasonal (dry/wet or summer/winter) climatic differences. Indeed these differences may well preclude the use of stereo-interpretation or stereo-correlation techniques.

(b) By contrast, in the case of the existing MOMS-02 and JERS-OPS satellites, two linear arrays are employed to image the ground simultaneously, each tilted in a fixed orientation with the one pointing forward and the other pointing backward along the line of flight (Fig. 5). This produces an overlap in the along-track direction to give stereo-coverage of the ground in a single flight. This eliminates the possibility of changes in the appearance of ground objects and areas being caused by seasonal changes.

(c) In the case of the new commercial high-resolution satellites, they will employ either gimballed mirrors or whole body movements of the satellite to point their linear array sensors in any required direction and at any viewing angle (or tilt) up to 45 (from the vertical. This arrangement will allow the acquisition of imagery in any direction through flexible pointing of the linear array sensors as well as producing overlapping cross-track and along-track imagery.

In summary, the new high-resolution scanner imagery will have quite different geometric characteristics to those currently familiar to users.

I - The images will be acquired with flexible pointing in any direction coupled with the presence of large tilt angles and the possibility of overlapping stereo-coverage with large base:height ratios.

II - The higher ground resolutions and larger image scales will really show up the tilt and relief displacements that are present in the images. These need to be eliminated if the images are to be ortho-rectified to fit existing maps or GIS data or to allow change detection for map revision or environmental monitoring purposes.

5. Need for New Software

It will be apparent to many users that most existing image processing software is not designed to cope with the geometric configurations that will be used to acquire the new high-resolution space imagery. In the case of photogrammetric software, this is mostly limited to the handling of frame photographs and, in a few cases, the imagery taken by spaceborne scanners such as SPOT and IRS-1C/D using a cross-track configuration. Only one or two packages can deal with imagery taken in the along-track configuration. In the case of remote sensing software, many packages only have a very simple geometric modelling of the imagery. Frequently space images are treated simply as 2D frames irrespective of their actual geometry and any relief and tilt displacements that may be present in the image. In which case, rubber sheeting techniques (based on the use of polynomials) are employed to fit the image to the map reference system. Needless to say, these will not work in a satisfactory manner with the forthcoming high-resolution space imagery.

6. Characteristics of the New High-Resolution Space Imagery

For most potential users, almost certainly their first questions will be to ask what will the newer types of high-resolution imagery look like and which ground features can be detected and identified on the images. In fact, it is possible to obtain a fairly good idea about the quality of the new images from an inspection of existing images from various sources - for example

(a) the high-resolution space photographs that have already been taken by the KVR-1000 and KFA-3000 cameras. Many examples are available for inspection on various Web sites, e.g. those of Eurimage ( and SPIN-2 (

(b) Aerial Photographs taken in the scale range 1:40,000 to 1:60,000 which have a true ground resolution of 1m. These form the basis of numerous simulations that are available on the Web sites of the commercial satellite imaging companies and on sample CD-ROMs, e.g. the Carterra IKONOS CD from Space Imaging.

7. Comparisons with Aerial Photography

Comparisons with conventional aerial photography taken with mapping cameras can be made on the basis of both ground resolution and areal coverage.

7.1 Ground Resolution

In the case of ground resolution, with older types of aerial cameras such as the Wild RC10 and the Zeiss RMK-TOP, the resolution of the processed negative film is around 40 line pairs per mm (40lp/mm). At 1:40,000 scale, the finest resolution of 1lp/mm on the film is equivalent to 1m in terms of ground resolution. With the newer types of aerial camera, such as the Leica RC30 and Zeiss RMK-TOP equipped with forward motion compensation (fmc) and gyro-controlled mounts and utilizing fine-grained film emulsions, the image resolution is 60lp/mm. At 1:60,000 scale, the finest resolution on the film is again equivalent to a ground resolution of 1m.qbsatmod

An artist's impression of the forthcoming QuickBird satellite (EarthWatch)

In making these comparisons, it is important to note that the ground pixel size of 1m of the new commercial high-resolution scanner images is equivalent to 2m in terms of their actual ground resolution. In this context, the Kell factor gives the relationship that 1 line pair (lp) is roughly equivalent to 2 pixels.

7.2 Areal Coverage

Regarding ground coverage, it is again interesting to note the similarities of the forthcoming commercial high-resolution images with aerial photographs in the 1:40,000 to 1:60,000 scale range.

(a) At 1:40,000 scale, a standard 23 x 23cm format aerial photo covers 9.2 x 9.2km.

(b) At 1:60,000 scale, a standard 23 x 23cm format aerial photo covers 13.8 x 13.8km.

By comparison, a single IKONOS image will cover an area of 11 x 11km, while an individual OrbView3 image will cover an area of 8 x 8km.

8. Products & Markets for High-Resolution Space Imagery

While a discussion of the products that can be acquired from the high-resolution space imagery can be carried out with a reasonable approximation to the probable reality, it is much more difficult to do so regarding potential markets and applications. This is especially so in the absence of a definitive price list that would allow meaningful comparisons to be made with aerial photography of a comparable ground resolution and with the products derived from them.

8.1 Products

The various products that are expected to be produced from the new commercial high-resolution imagery include the following:-

(a) Ortho-rectified images with relief and tilt displacements removed.

(b) Pan-sharpened images comprising the high-resolution panchromatic data with its 1m ground pixel size merged with the corresponding multi-spectral data having a 4m ground pixel size.

(c) Digital Elevation Models (DEMs) generated from the stereo-pairs and the contour plots and perspective diagrams and visualisations that can be derived from these DEMs.

(d) Topographic maps - the generation of these products from the imagery will need manual interpretation and feature extraction by a topographer.

(e) Thematic maps - the generation of these products will require expertise in the particular theme being mapped (vegetation, geology, soils, etc.) together with a detailed knowledge of the local conditions that exist in the specific area for which the mapping is being undertaken.

Items (a), (c) and (d) will require a good knowledge and understanding of photogrammetry as applied to the various space imaging geometries discussed above.

8.2 Value Added Products and Services

There has been much debate and controversy over this particular matter. For example, Space Imaging and its regional partners and distributors envisage that the various products (ortho-images, DEMs, fused pan/multi-spectral images, thematic maps, etc.) listed above will be supplied in a "GIS-ready" format. They would prefer not to supply raw data or images on which only basic processing has been carried out. Rather similar statements have been forthcoming from EarthWatch and Orb Imaging. Furthermore the former company has stated that it will retain ownership and possession of all of the data collected by its satellites. So presumably some type of licensing of the data to users is envisaged rather than its outright sale.

The reactions from potential users to these various pronouncements have been quite strong. In the first place, some doubts have been expressed as to whether the commercial satellite operators and their regional partners have the resources and the management skills to generate, in timely fashion, all the very varied forms of value-added data that a world-wide clientele of users may require. Furthermore it has been pointed out that, at present, Landsat, SPOT and IRS-1C/D image data is supplied in raw form or with basic corrections only. Users then undertake further processing using their own specialist staff who have extensive and detailed local knowledge and expertise in the field of interest. Many potential users feel that there is no way that this level of local knowledge and expertise can be duplicated either by the commercial satellite operating companies themselves or by their regional partners, distributors and re-sellers.

8.3 New Markets

Much speculation and discussion has taken place regarding the various markets that might be created for these products - especially those that cannot be served by conventional aerial photography.

A major but somewhat controversial potential market should result from the needs of foreign governments for high-resolution images for defence and security purposes - especially for images covering neighbouring countries with whom they may not have friendly relations. Many complex issues are expected to arise in this particular area. Almost certainly, some countries will not be supplied with high-resolution space images. On the other hand, it is almost certain too that the imagery will prove to be useful for the verification of international treaties and sanctions and for international "peacekeeping" operations.

Another big new market is expected to develop within the news media (TV networks, newspapers, etc.). Thus images of areas experiencing natural disasters such as floods, tornados, forest fires, earthquakes, volcanic eruptions, major landslides and avalanches, etc. or man-made disasters such as oil spills, shipwrecks, explosions, pollution, etc. are expected to achieve a ready sale. Furthermore, the timely acquisition of high-resolution imagery should also assist the assessment of the scale and impact of such disasters and the planning of relief and evacuation operations by the appropriate agencies.

9. Applications

Only the main application fields that have been identified by the commercial high-resolution satellite operators will be outlined here. Many others have been mentioned in their promotional literature and on the information pages provided on their Web sites. Undoubtedly still others will come to light when the actual data is forthcoming.

(a) Cartography - especially small-scale topographic mapping and map revision. Ortho-rectified images can be produced at scales up to 1:25,000 or perhaps even larger - in much the same way as is done with the corresponding aerial photographs taken at 1:40,000 to 1:60,000 scale.

(b) Agriculture is seen as a very important application area, especially for precision farming and crop monitoring. Frequent updates of small areas are envisaged in order to provide information on crop growth, crop damage and the early detection of disease. Over large areas, commodity forecasting and the early warning of food shortages could be other applications.

(c) Forestry - the detection and monitoring of areas of windblown trees, the detection of disease and the estimation of timber volumes have all been cited as potential applications in this particular field.

(d) Environmental Monitoring is seen as another important application area with the use of repetitive cover for coastal and estuarine studies; the monitoring of land use changes; surveillance of open-cast mining operations; etc. Governmental regulatory bodies may well find the imagery useful for impact studies of proposed new large-scale developments and to monitor compliance with environmental regulations and legislation.

(e) Rural and Urban Planning & Management could well be an application of especial importance in developing countries where cities and their populations are growing very rapidly and there is no record of the situation regarding current land occupancy on existing maps. In developed countries, the monitoring of peri-urban areas and derelict land and the changes in building and land use appear to be possible applications.

(f) Exploratory surveys for minerals, oil and gas, etc. are obvious possibilities, given the extensive use of remotely sensed image data in this field at the present time.

10. Conclusion

The future for high-resolution space imagery of a quality that is superior to that currently available is intriguing to say the least. In particular, it will be most interesting to see how it fares in competition with aerial photography of a similar quality and coverage.