Blue tab
Green tab
Brown Tab

Earth Observation



Earth Observation in the frame of EO-MINERS - Remote Sensing Technologies

Electromagnetic remote sensing

Instrument addressing electromagnetic radiations are sensitive to different wavelength ranges of radiations: ranging from gamma- and X-rays, up to microwaves. The measured radiations might originate from natural sources, usually the sun, or from man produced sources. Devices based on natural sources are called passive devices, while the latter are defined as being active devices.
Most remote sensing sensor systems are composed of a few common parts: an emission source, an accumulation assembly and detector. The detector  is exposed to some form of inputs and registers the respective signal; this signal is then related to a particular location on the Earth's and projected on a detector that produce either an image view or point reading of the interaction with the matter in question. Most remote sensing systems - active and passive, spaceborne and airborne - generally have seven elements as shown in figure 1:

Figure 1: Diagram of elements of a remote sensing system. Source: Canada Centre for Remote Sensing, Fundamentals of Remote Sensing
Figure 1: Diagram of elements of a remote sensing system.
Source: Canada Centre for Remote Sensing, Fundamentals of Remote Sensing.


Energy Source or Illumination (A) - A basic requirement for electromagnetic remote sensing is an energy source to illuminate or provides electromagnetic energy to the target of interest. For passive instruments, this is usually the sun; for so-called active instruments, another sensor element emits pulses of energy.

Radiation and the Atmosphere (B) - As the energy travels from its source to the target, it will come in contact with and interact with the atmosphere it passes through. This interaction also takes place a second time as the energy travels from the target to the sensor. Inevitably there is a certain degree of atmospheric scattering of radiation.

Interaction with the Target (C) - Once the energy has made its way to the target through the atmosphere (which can be the target), it interacts with the target depending on the properties of both the target and the radiation.

Recording of Energy by the Sensor (D) - After the energy has been scattered by, or emitted from the target, a sensor (remote - not in contact with the target) collects and records the electromagnetic radiation.

Transmission, Reception, and Processing (E) - The energy recorded by the sensor has to be transmitted, often in electronic form, to a receiving and processing station where the data are processed into usable information (hardcopy or digital).

Interpretation and Analysis (F) - The processed information is interpreted, visually or digitally/electronically, to extract useful elements.

Application (G) - The final element of the remote sensing process is achieved by applying the information that has been extracted from the data about the target in order to better understand it on a spatial domain (a thematic map). This reveals some new information that assists solving a particular problem and directs the user to explore more on the phenomena in question.
Photonic RS means are characterized by different spectral and spatial configurations, spectral, spatial or temporal coverage.

Unlike for the other EO techniques discussed previously, there is only a limited number of suppliers that offer services or products covers all techniques and resolutions requirements. This holds in particular for satellite-based services and products. Their scope and availability will be discussed in the following. The focus will be hereby on those instruments and the services and product based upon them that would be of use for supporting the EO-MINERS indicators.

Spectral configurations - The term spectral configuration refers to the spectral coverage across the electromagnetic radiation spectrum, the sensor in question is sensitive, the number of bands across the selected region, their sampling intervals along with the Full Width Half Max (FWHM) that present the width of each band. The majority of remote sensing instruments utilize one or more of the following spectral ranges:

  1. The reflective range (0.4 µm - 2.5 µm, visible near and short infrared VNIR-SWIR; in this range the illumination source is obtained from the sun that penetrates through the Earth's atmosphere and reflects back to the sensor. Passive sensors sensitive to these wavelength are usually only effective during day-time.
  2. The thermal range (3 µm - 5 µm mid-wave infrared, MWIR, 7 µm - 13 µm long-wave infrared LWIR); in this range the illumination source is the Earth at high and low ambient temperatures. The average temperature at around 300k would be sense across the LWIR region (7 µm - 13 µm) whereas high temperature events (fires or volcanoes) would be sensed across the MWIR (3 µm - 5 µm) with interferences of the sun SWIR radiation around 3mm. Thermal sensors are usually passive and can be efficient at day and night.
  3. The microwave range (1 cm - 1 m); Most sensors sensitive to this range of wavelength are active, although passive technologies are emerging.
Figure 2: Improvement in spectral resolution, identification of spectral features
Figure 2: Improvement in spectral resolution, identification of spectral features

Figure 2 provides a schematic view of RS coverage of low (top) and high (bottom) spectral resolution. The high spectral resolution provides more information about the sensed area and thus, it is more accurate vehicle for RS of the earth. Generally, the high spectral resolution sensors are termed hyperspectral sensors and they are able at detection and quantification of minor changes in the enquired target (e.g. different minerals in soils, vegetation types etc.). The current limitations of hyperspectral sensors still prevent achieving products that a laboratory spectrometer can provide, but it is anticipated that with the progress technology this gap will be bridged. The limitations are: heavy load of data for transmitting to the ground and to process, relatively low signal-to-noise ratio and low spatial resolution as well as expensive to operate sensors from any platform. These limitations are the reason why we do not see in orbit much hyperspectral sensors whereas airborne sensors are more attractive and manufactured by several vendors quite intensively (, e.g. Specim company sold about 90 hypersepctral sensors during the past decade).

Spatial resolution - Different sensors will have different spatial resolutions (Figure 3). This variation, for optical passive sensors, is a function of the spectral resolution. To receive good information, a reasonable amount of photons have to be accumulate on the sensor’s detector. If high spectral resolution is required then the same amount of photons has to be split between all bands according to their FWMH. To achieve enough photons to all bands, and accordingly a favourable SNR, the pixel size must to be enlarged (Figure 4). In high spatial resolution, where the pixel size is small, the amount of photons are kept favourable in term of SNR by enlarged the FWHM of the band(s) in question. Also the spectral region plays a major role in the spatial resolution: According to the Sun's emission curve, a sensor detecting light in the visible wavelengths range (0.4 µm - 0.7 µm) receives a relatively large amount of radiation reflected from Earth. This fact enables the sensor to obtain high spatial resolution (smaller pixel size) in the visible spectral range. Longer wavelengths such as thermal infrared (TIR) are less energetic and thus forcing sensor developers to increase the pixel size in order to retrieve sufficient radiation in the same amount of time. For the same reason, when spectral resolution increases (i.e. more spectral bands for the same range), pixel size must increase as well, reducing the spatial resolution. As a result, the majority of high spatial resolution spaceborne sensors (pixel resolution 0.5 m - 10 m) are located in the VIS region consisting of 1-4 spectral bands, while high spectral resolution spaceborne sensors (more than seven bands) have low spatial resolution (15 m - 120 m). For RADAR sensors, spatial resolution is usually dependant of the pulses, chosen wavelengths and antenna size. Please also note that some RADAR sensors have a volumetric spatial resolution.

Figure 3: Comparison of ASTER (15 m ground resolution) and IKONOS (1 m ground resolution) Imagery, Identification of sources of contamination seen by IKONOS
Figure 3: Comparison of ASTER (15 m ground resolution) and IKONOS (1 m ground resolution) Imagery,
Identification of sources of contamination seen by IKONOS


Figure 4: The corelation between spatial resolution and spectral bandwidth (a component of spectral resolution).
Figure 4: The core la ti on between spatial resolution and spectral bandwidth (a component of spectral resolution).


Figure 5: Suggested hierarchical flow chart of the environmental indicators groups and the EO instruments which can be used to detect and monitor those indicators

Click image to enlarge
Figure 5: Suggested hierarchical flow chart of the environmental indicators groups and the EO instruments which can be used to detect and monitor those indicators

The spatial resolution of optical passive sensors is linked to the size the instrument as a heavy telescope has to be mounted on-board. Multispectral and hyperspectral sensors can be built with higher spatial resolution, if deployed on a platform other than a satellite. If a hyperspectral imager is deployed airborne, spatial resolution can be relatively high (2 m - 5 m), although it will still be lower than a low spectral resolution airborne sensor (5 cm - 50 cm). For a ground based instruments even higher resolution can be achieved. For microwave sensors, spatial resolution is linked to other parameters, including the wavelength and technology used. Similar resolutions as optical sensors can be achieved.
Calibration and ground-truth - Remote sensing as most other spectroscopic techniques do not give absolute signals, but signals need to be calibrated against some sort of known standard. Usually calibration of sensors is done on the ground prior launch to orbit or routinely for airborne sensors. Some orbital sensors would have also onboard calibration capabilities where an illumination source and white standard on integrating sphere is placed and used routinely. Some orbital sensors also are calibrated by using the moon surface which is constant with very thin atmosphere or using the ice-cap of the poles and deserts.

Instrument selection - There is a wide variety of EO instruments for the monitoring of environmental indicators. One should choose an instrument according to the objectives and limitations of its project with respect to the capability of the available technology. Figure ‎5 and 6 is an example for a simplified hierarchical chart made to aid choosing the appropriate instrument for the detection and monitoring of various environmental indicators. EO instruments are marked in a green ellipse; environmental indicators are marked in an orange rectangle.


Mapping environmental indicators - RS based mapping methods include the utilization of diverse spectral, spatial and statistical methods, taking into account the spectral or geometrical properties of the desired target.

Figure 6: An example of the spectral information
Figure 6: An example of the spectral information that can be extracted from each pixel: here, a pixel showing clay soil polluted by hydrocarbons. In this case the pixel was measured at the ground, but the information obtained from the measurement can now be used for the mapping of hydrocarbon pollution, iron oxides and clay content of the entire bare soil area in the hyperspectral image.

For example, in hyperspectral optical remote sensing, by applying a wide range of mathematical methods on a spectrum, it is possible to detect not only the type of the material, but also more detailed properties such as its chemical and physical composition. Hence, a target after mapping can be defined as follows: "Soil"→"Clay soil"→"Contains the minerals Montmorillonite & Kaolinite"→"Obtains an Electrical Conductivity value of 3.4 ds/m"→ "Polluted by 0.1% of unleaded fuel" etc. This way, each pixel in the image receives a value that represents the mapping subject (in the case of the example above it can be soil type, minerals, salinity or fuel content). The result is a set of maps, each one representing one soil property on a fitted scale. Each pixel in those maps is equivalent to a lab analysis of a sample from the surface. This allows the end-user to obtain continuous data on large areas, without the need to go and sample each and every point.