Remote Sensing

Remote sensing is broadly defined as the act of obtaining images or data from a distance, typically using a manned spacecraft, a satellite, or a high-altitude spy aircraft. The term was invented in the 1950s to distinguish early satellite images from aerial photographs traditionally obtained from fixed wing aircraft. As such, remotely sensed images can be considered to be one kind of geospatial imagery. Although the application of unclassified remote sensing images to civil and criminal investigations has been limited, they have proven to be useful for documenting international atrocities in areas that are otherwise inaccessible to outside observers.

Sufficiently detailed satellite imagery has been used to document international crimes such as possible genocide in the Darfur region of Sudan and the existence of concealed mass graves in Iraq. In Iraq, potential gravesites were identified with the help of satellite image and aerial photograph interpretation and then investigated in more detail using ground-penetrating radar and other methods. A total of 270 mass graves were reported, of which 53 had been confirmed by early 2004, with some 400,000 bodies discovered. Features such as mass graves are generally not directly visible. Instead, analysis reveals features such as otherwise inexplicable areas of freshly moved earth or signs of heavy construction equipment used to excavate the graves. Comparison of publicly available Landsat satellite images obtained in 2003 and 2004 was also used to document the burning of 44 % of the villages in the Darfur region of Sudan during a period of civil strife, which some observers believe amounted to genocide. Burning was inferred in areas where the albedo, or amount of radiation reflected by the ground surface, had changed significantly during the times at which the two images were obtained. This was accomplished by using a computer algorithm to calculate albedo from the satellite data, then subtracting one albedo map from the other to calculate the change. This kind of mathematical operation on entire maps or digital images, as opposed to single numbers, is known as map algebra.

Modern remote sensing satellites provide panchromatic grayscale images (popularly known as black and white) and multispectral images in which channels representing discrete bands of the electromagnetic spectrum are combined. The most common multispectral images consist of some combination of red, green, blue, and near infrared bands. Hyperspectral sensors can produce images composed of dozens or hundreds of bands. Using information about the spectral reflectance characteristics of different kinds of soils, rocks, and plants, image analysts can fine tune the ratios of bands in multispectral and hyperspectral images to identify specific targets.

Image resolution has historically limited the use of satellite images, particularly those that are unclassified and easily available, in criminal and civil forensic work. The Landsat 1 satellite launched by the United States in the early 1970s, which provided the first publicly available satellite images, had a maximum resolution of 80 m. Therefore, objects smaller in size than several hundreds of meters could not be analyzed because objects must be many times larger than the maximum resolution in order to be clearly shown. Landsat 7, launched in 1999, had maximum resolution of 15 m for its panchromatic band, 30 m for its multispectral bands, and 60 m for its thermal infrared band. Although imagery with maximum resolution of 10 m or more can be useful for regional investigations, it is generally not useful for detailed forensic investigations of activities that have occurred through time on individual parcels of land. A new generation of commercial satellites such as the Quickbird satellite launched in 2001, however, has 0.61 m panchromatic resolution and 2.44 m multispectral resolution. The commercial IKONOS satellite, which was launched in 1999, has a maximum resolution of 1 m for color imagery. Although no images have been released as of early 2005, many intelligence experts believe that the most recent KeyHole surveillance satellites operated by the United States have a resolution of about 2 cm (0.02 m).

The resolution of panchromatic images is higher than that of multispectral or hyperspectral images because panchromatic information requirements are lower. In a panchromatic digital sensor, each light-sensitive photosite responds to all colors of light. In a multispectral sensor, however, the same number of photosites must be divided among each of the spectral bands. A multispectral sensor with infrared, red, green, and blue bands but the same number of photosites as a panchromatic sensor would have a resolution only 1/4 as high as the panchromatic sensor. This explains, for example, the ratio of 4 between the panchromatic 0.61 m resolution and multispectral 2.44 m resolution of the Quickbird satellite. In some cases, multispectral images can be combined with brightness information from more detailed panchromatic images. The apparent effect is a sharper image, although the resolution of the multispectral layer is not actually changed.