Human Rights Abuses Monitored with Satellite Imagery: Myth or Reality?

Written by

Bjørn Willum (BA in Political Science)

Supervised by

Bhupendra Jasani (Dr in Nuclear Physics and Nuclear Medicine), Center for Defence Studies, King's College London

Extended Essay (10.000 words)

Submitted for the Degree of Master of Arts in War Studies

The Department of War Studies, King's College London

1 September 1999

Abstract

This dissertation aims at analyzing whether high-resolution satellite imagery could be useful to human rights organizations or other international organizations that are monitoring massive violations of human rights.

The analysis concluded that it would be possible to detect and recognize a number of objects relevant to the monitoring of massive human rights abuses. The use of satellite imagery can be summarized as having operational and political benefits. Operationally, the use of satellite imagery allows the user to cover any area at any time quite fast. Politically, satellite imagery can enhance NGO or media capabilities for verifying government-supplied information.

The disadvantages of satellite imagery are first and foremost that near-future available satellite imagery can only provide indications of activities and enhance the credibility of other reports. Further, it is very costly.

However, it remains clear from the current dissertation that high-resolution satellite imagery can indeed be useful for the monitoring of human rights violations.

Acknowledgements. The author acknowledges the supervision and support of Dr Bhupendra Jasani, without whom the present dissertation would not have been possible. He also expresses his gratitude for the invaluable advice and comments on earlier drafts provided by Marianne Ajana, Dr John Baker, and Dr Einar Bjorgø. I am indebted to Dr Einar Bjorgø who lend me illustration material from this recent PhD.

Also, I am thankful for advice on researching by Dr Christopher Simpson and Tim Brown. Obviously, I remain responsible for errors or omissions.

Introduction

There are several organizations that are interested in information on or documentation for massive abuses of human rights occurring in conflicts all over the globe. However, sites for misdeeds are often inaccessible since the perpetrators rarely want their activities exposed to the public. This dissertation aims at analyzing whether high-resolution satellite imagery could be useful to human rights organizations or other international organizations that are monitoring massive violations of human rights.

The potential value of satellite imagery can be split in the value it has before a crisis for early warning, during a crisis for crisis monitoring and after a crisis in order to hold perpetrators responsible - both politically and legally - for human rights abuses.

Before outbreak of a crisis, satellite imagery can obviously be used for monitoring in order to seek crisis prevention. This could be interesting to a number of non-governmental organizations (NGOs) or possibly governments or inter-governmental organizations (IGOs). In turn, better information gathering on small scale human rights abuses might be used to put public pressure on the authorities responsible for the atrocities and may also be used to destabilize an illegitimate regime or party before they commit grand scale massacres. As such the monitoring by satellite imagery and other sources of human rights abuses may have a preventive effect since potential perpetrators probably would think twice if they realize they are being watched. [1] Internationally, information from satellite imagery as well as other sources could be used to inform the public and, if necessary, be used by human rights organizations and the news media as a tool to create a public opinion in favour of outside intervention. A further advantage of having human rights groups with access to satellite imagery is that, A public-interest group, unencumbered by internal policy debates, would likely move ore quickly than a government in making [...] pictures [of massacres] available. [2]

During crisis NGOs, IGOs, outside governments as well as the population of the country in question benefit from humanitarian crisis monitoring in much the same way as before the crisis: the public in the affected country can be informed, the political platform of the perpetrators can be eroded, public opinion in favour of an intervention can be gathered and used as a basis for sanctions against the perpetrators. Further, when a crisis is in full swing, the news media is often interested in information and documentation, and satellite imagery is by nature a highly visible documentation, which easily can be reprinted and used for illustration. Since it is normally difficult to access areas of military conflict or ethnic cleansing, the value of satellite imagery during crisis is enhanced as compared to before a crisis.

After a crisis, archive imagery could be used as a tool for international war crimes tribunal investigators in their search for places to look for evidence of war crimes such as ethnic cleansing or mass executions. Identifying crimes and perpetrators is not only important for legal purposes, but setting the account straight is furthermore an important tool for preventing the creation of myths and the seeds of future conflict. [3]

The potential value of satellite imagery to NGOs and others is thus considerable, but is it or will it also be technically, financially and politically possible to make such use of satellite imagery as outlined above? This essay will look at the possibilities for monitoring human rights abuses from space using civilian high-resolution satellite imagery that is available on the commercial market now or in the near future; for the purpose of this dissertation I define high-resolution imagery as panchromatic imagery with a one-meter resolution and multispectral, hyperspectral and radar imagery with a 10-meter resolution or better. [4] It will consider the contribution civilian satellite imagery and its interpretation can provide for the general monitoring of human rights abuses. While the ideal option for this study would obviously be to purchase imagery to use for illustrations and thus be able to do a fully-fledged case study, this has not been possible for financial reasons. The argument will therefore only be illustrated with freely available images. Further, where images have not been available, I will rely on interviews and published information.

The structure is as follows: First, I will outline the technical possibilities of the new or up-coming commercial satellites as well as the political restrictions applied to them. Then I will analyze which objects - relevant to the monitoring of major human rights violations - can be seen. Finally, I will discuss whether the financial costs and disadvantages of using commercial satellite imagery for the purpose of human rights monitoring is outbalanced by the benefits; as part of this I will give some examples of crises in which human rights groups and other organizations might have used commercial satellite imagery as an information and advocacy tool.

The developing world of commercial satellite imagery

During the Cold War satellite technology was jealously guarded and used almost exclusively by military establishments of advanced industrial nations. Although satellites were also developed and used for civilian purposes, such satellites always had to get along with the help of yesterday's technology only.

In 1992, however, the Russian government was the first to market high-resolution satellite imagery from their military satellites, albeit only images which had been degraded to resemble a resolution of approximately two meters. [5] In March 1994 the Clinton administration then adopted a Presidential Decision Directive encouraging private companies in the United States to sell high-resolution satellite imagery. After the US government had permitted private companies to launch satellites with one-meter resolution capabilities, the Russian government has recently permitted sale of one-meter resolution imagery by autumn 1999. [6] Other countries which have allowed their military or civilian industries to market satellite imagery include India, France, Israel, and Japan. [7]

However, as of present there are severe limitations to the value of satellite imagery and, to appreciate this fully, it is necessary to outline briefly the capabilities of the commercial satellites as well as the capabilities of satellites planned to be deployed in the next few years. I will therefore go through the types of imagery available, government restrictions on satellite companies, dubbed "shutter control", the acquisition time and method, and resolution of images. [8]

Types of images

Most basically there are two types of optical satellite imagery, namely panchromatic and multispectral. Panchromatic imagery is defined by for instance the Spot Image Corporation as imagery "acquired by an optical sensor passively measuring sun energy reflected from the earth in one wide portion of the electromagnetic spectrum." Such a portion of the spectrum is also referred to as a band and with most current panchromatic sensors this single band spans the visible to near-infrared part of the spectrum and is represented on the final image as black-and-white imagery. [9] Multispectral imagery is also acquired by a sensor that passively measures reflectance from the earth, but in several bands simultaneously, usually three to seven. Hyperspectral imagery is, technically speaking, the same as multispectral imagery, but instead of measuring energy in only a few bands it measures reflectance in numerous though narrow portions of the electromagnetic spectrum, enabling it to detect very subtle characteristics and differences among surface features especially in vegetation, soil and rocks. [10] Man-made objects such as vehicles can thus often be detected in the countryside, despite their being camouflaged as the surrounding vegetation, or ground surface reflects energy in a different band of the spectrum than does metal or even camouflage cover material such as plastic, rubber or other synthetic materials. While a number of the satellites that are planned to be launched in the near future carry both panchromatic as well as multispectral sensors, no hyperspectral sensors are onboard any of the current commercial satellites. A US satellite carrying a sensor capable of acquiring hyperspectral images, Lewis, was launched in April 1999, but malfunctioned. However, the US-based company Orbital Images have plans for launching a satellite, OrbView 4, with hyperspectral imaging capabilities in year 2000. [11]

Three images of the same area taken with a panchromatic, multispectral and hyperspectral imaging sensor respectively. Unspecified images. Image copyright (c) TRW. In the hyperspectral imagery, an object is clearly detectable, which is undetectable in the panchromatic or multispectral imagery. Source: Federation of American Scientists, at http://www.fas.org/irp/imint/hyper.htm

As mentioned earlier, panchromatic and multispectral (and hyperspectral) imaging instruments are referred to as passive since they do not transmit their own source of energy. This means that they are only able to detect objects that either emit or reflect energy from the sun within one of the bands that are measured by the imaging instruments. When it is dark it will therefore be impossible to measure anything but illuminated places with such instruments since virtually no energy is then reflected from the Earth; "the one exception being passive sensors measuring thermal infrared radiation emitted from generating sources such as power plants." [12]

This is different with active imaging instruments, such as the so-called Synthetic Aperture Radar (SAR), which transmit a radar signal in the microwave portion of the spectrum and measure the strength and other characteristics of the return signal after it has been reflected by the surface of the Earth. Because SAR thus utilizes its own source of energy and operates in longer wavelengths than passive instruments, it can operate regardless of weather and time of day. This obviously makes it well suited to image objects during night-time or geographical areas frequently covered by clouds or fog.

Shutter control

A major concern for at least the new American satellite companies is government interference. The phenomenon has been dubbed shutter control and most governments who have national companies with high- or medium-resolution imagery for sale have made use of their power to temporarily or permanently forbid companies based on their territory to sell images over particular geographical areas. The Russian government, for instance, has more or less permanently refused to sell high-resolution images of North Korea, China, Serbia, Bosnia and Russia. [13] Also worth noticing is that The American Israel Public Affairs Committee successfully lobbied in the White House for an exemption that forbids US-based satellite companies to sell images of any place in Israel with a resolution better than what is routinely available from commercial sources. [14] This exemption is part of a general power given to the administration, which can then forbid the selling of images. Media organizations have complained that "the regulations permit the government to limit the collection and dissemination of satellite photography on the basis of little more than an unsubstantiated claim from the Secretary of Commerce, in consultation with the Secretaries of State and/or Defense that collection and/or dissemination of the information may compromise (whatever that means) as foreign policy interest or international obligation (again, undefined)." [15]

However, time may thus prove the exemptions obsolete if, in future, it will be possible to acquire high-resolution images of for instance Israel from providers based in other countries that do not have such friendly ties with Israel (or any other exempted countries) as the United States has. Images of Kosovo could then be acquired from a US company and so forth.

Real-time and archive images

First generation satellites used to be equipped with on-board storing capacity for large amounts of film, which was retrieved with the satellite when it was taken down at regular intervals. However, this changed in the later part of the 1970s when the American spy satellites introduced digital recording of images; the images were then conveyed to the earth when the satellite passed over these ground stations. By setting up a number of such ground stations on earth, the constant orbiting of the satellite ensured the US intelligence community near real-time capabilities when it launched the Keyhole-11 satellite in 1976, variants of which are still in use today. [16] After this invention, the use of film on-board was discontinued with US spy satellites in the early 1980s. [17] Today, most military satellites as well as coming civilian satellites utilize this method of data-retrieval allowing for near real-time image acquisition. The main exceptions are the Russian satellites, which possibly still use film stored on-board, and customers are thus first able to view images when a film capsule is retrieved from outer space. [18]

But although satellites may have these so-called electro-optical sensors on board, which can transmit the images with only a slight delay, this does still not mean that a prospective buyer can be ensured an image over the area of interest at the time of choice. The commercial satellites will use roughly two hours to orbit the Earth once and transmit their imagery electronically to the Earth when they pass over ground stations situated on various locations on Earth. This should happen at least once every two hours, since some of the ground stations usually are situated near the poles, where the satellites will pass by once per orbit. Thus, theoretically speaking, it should be possible with very rapid image acquisition, however, in practice image acquisition has - mainly for organizational reasons shown to take at least 24 hours, counting from the moment the image is taken until it is in the customer's hands. [19]

Further, due to the constant orbiting of satellites in a slightly tilted north-south direction, also known as the orbital inclination, the same place or route is only revisited once in several days, because the Earth spins in a west-east direction. [20] As Bjorgø notes the repeat coverage frequency of the existing and near future [...] satellite sensors vary between 3 and 5 days. [21] Most coming satellites, however, are constructed to enable tilting of the imaging instrument, and the satellite can then acquire images outside the ground track immediately beneath the satellite. [22] This repeat coverage frequency mentioned above is in fact only achieved by tilting the satellites slightly. However, if the satellite is tilted substantially more and images thus acquired far outside of the ground track, this will affect the resolution of these images. It would therefore only in theory be possible to have even daily updates - if one would not want to compromise the resolution of images. And even then, one would have to purchase images from a different provider every day. [23] However, the chances for achieving daily update capability will probably improve remarkably in the future when companies are expected to launch several copies of the same satellites.

Satellite	Launch	Sensor type	Resolution (m)	Swath width (km)	Revisit image coverage cycle (days)
Cosmos, KVR-1000	In orbit	Pan	2	160	-
Ikonos-2	1999	Pan, MS	1, 4	11	4
IRS-P6	1999	Pan	2.5	-	-
OrbView-3	2000	Pan, MS	1, 4	8	3
QuickBird	1999	Pan, MS	1, 4	22	5
OrbView-4	2000	Pan, MS, Hyper	1, 4, 8	8 Pan, MS	3
Landsat 7	In orbit	Pan, MS	15,30	-	-

The capabilities of today's military satellites is officially classified, but it is "generally accepted" that the KH-11B has a resolution below 10 centimeters. [26] For the commercial market the first satellite to be launched was the American Landsat in 1972 with a resolution of approximately 100 meters, and since then the capabilities of commercial satellites have steadily improved. Since the Clinton administration's permission to license the selling of high-resolution satellite imagery in 1994, several companies have planned so far unaccomplished launches of satellites in 1998 and 1999. In April 1999 the company Space Imaging Systems failed to bring a satellite with a resolution of one meter into orbit. The company has scheduled a second try to bring a satellite into orbit on 24 September 1999 and - if it is successful - it will then become the first satellite to provide images with a resolution of one meter in near real-time to private customers. [27] They will thus compete directly with the Russian satellites. While the henceforth mentioned imagery from commercial satellites (current as well as near future) have all been digital imagery, the Russians companies in fact also sell photographic imagery with a resolution of 1.5 meters, which is the equivalent of digital imagery with a resolution of about 0.80 meters. However, as previously mentioned they are known to refuse to sell images of many areas of military or human rights interest and for their handling time of orders, among others due to bureaucracy. [28] . The best available multispectral imagery in the near future is likely to be with a resolution of 4 meters and as mentioned there are no commercial satellites yet capable of acquiring hyperspectral imagery, but there are plans for one to be launched with an 8 meter resolution capability. [29] The best available radar imagery is provided by the Canadian-owned Radarsat, which can acquire images with a resolution of 10 meters.

Imagery for illustration

Allocating relevant images for illustration in the current thesis turned out to be the most difficult aspect of the research. The task was therefore to use the second-best information available, i.e. imagery with a coarser resolution or in a volume less than desired. The lack of images was naturally hampered by the fact that most of the satellites whose capabilities I wish to analyze have not been (successfully) launched yet. A brief explanation of the images chosen for illustration is necessary.

NATO recently published a number of very relevant images substantiating reports of human rights abuses in connection with the Kosovo campaign. [30] However, as is most often the case when NATO members release public imagery, they -wary of protecting claimed sources and methods - refuse to provide other details than date and place of an image taken. [31] While the images provided by NATO during the Kosovo crisis could have provided ample illustration for the current thesis, the lack of resolution details and source information unfortunately renders these images less suitable since I aim to explore the utility of civilian satellite imagery, which in the near future will have a maximum resolution of about one meter.

Research further revealed that only a few studies relevant to this dissertation seems to have been carried out, at least outside national intelligence agencies.

Another organization that has had experience using satellite imagery for humanitarian purposes is the Satellite Center of the Western European Union (WEU). The Satellite Center monitors the military situation in Europe in general, but it has also monitored the refugee crisis in Eastern Zaire 1996-97 and supplied intelligence to peacekeeping troops in Albania. [32] The analysis of the images would have been useful for the current dissertation, but unfortunately the images acquired and their analysis are only shared among the governments of WEU member states and are not publicly available.

I was thus left to use unclassified or declassified images from specified sources and resolution to use for discussing and demonstrating the utility of civilian satellite imagery. I have made use of the following three categories of imagery: commercially available imagery, leaked military satellite images and aerial photos degraded to one meter resolution.

Imagery interpretation of objects relevant for human rights monitoring

I will now analyze the possibilities for viewing certain objects on the ground that are relevant to the monitoring of human rights abuses and where possible illustrate this with freely available images and estimates in the literature where such are available.

Verification and detection of objects in general

When analyzing imagery (whether satellite or aerial) certain distinctions are necessary to keep in mind, since an object might be detectable with coarse resolution imagery while it takes higher resolution imagery to describe the object in detail. I will make use of the vocabulary provided by the Imagery Resolution Assessments and Reporting Standards Committee:

Detect is the capability to find or discover the presence or existence of an installation, object, activity, or item of interest, based on its general shape (configuration) and on other contextual information in the scene. [...]

The distinguish between level is the capability to determine that two detected objects are of different types or classes based on one or more distinguishing features.

Identify is the capability to name an object by type or class, based primarily on its configuration and detailed components. Identification depends on observation of detail in the image and not through information from non-imagery sources. [33]

The terms recognizing, establishing or the ability to give general identification of objects is also used by some analysts to describe the level necessary to distinguish between objects of different types or classes as denoted above. [34] Some analysts go even further and assign a level necessary to describe and/or a level required for doing analysis of targets. The level of imagery necessary to respectively detect, distinguish, identify or describe objects is based on the information it is possible to gather from the image itself - without the use of any collateral information. With additional information it might thus be possible to lower the requirements. Alternatively, haze, fog or clouds can impact on an image raising the image resolution requirements. [35] It will ease the interpretability of images if ground stations are operated in the target area so that information can be gathered on detected objects, which are specific for that area or region. [36]

Introduction

The developing world of commercial satellite imagery

Real-time and archive images

Resolution of images

Imagery for illustration

Imagery interpretation of objects relevant for human rights monitoring

Verification and detection of objects in general