SECTION 2
To understand how remote sensing operates it is necessary to understand some of the physical principles which make it possible. This requires knowledge of electromagnetic radiation (EMR) and its interactions with various components of the environment. The elements involved in obtaining a remotely sensed signal are: the energy source, the atmosphere, the target and the sensor. This section deals with the first three of these elements. Sensors and the processing and interpretation of the signal will be explored in later sections.
Electromagnetic radiation is a form of energy which can only be observed by its interaction with matter. EMR is made up of both electric and magnetic components and is affected by the electric and magnetic properties of the matter with which it comes in contact. Two hypotheses are generally used to describe the behaviour of EMR: the wave model and the particle model. Both models are valid and both are important for remote sensing; for application specialists, however, the wave model is generally favoured.
Figure 2.1 illustrates a series of electromagnetic waves travelling through space. The electric and magnetic components are in phase and are always perpendicular to each other as well as perpendicular to the direction of travel. For this reason, it is simpler to think of the wave as a single entity without distinguishing between the two components. The orientation of the wave (i.e. the plane along which it travels) is referred to as the polarization. Naturally produced EMR contains waves which are randomly polarized; polarizing filters may be used to select those waves which have a particular orientation. Manmade sources of EMR such as radar systems sometimes produce waves with a single polarization, usually vertical or horizontal with respect to the surface. Certain types of materials can be distinguished by their tendency to “depolarize” these waves.
The physical distance from one wave crest to the next is termed the wavelength and is usually designated by the Greek letter lambda (λ). The number of wavelengths passing a given point inspace in a specified period of time is called the frequency and is designated by the letters f orν . Since each wavelength represents one complete cycle of the wave, frequency is generally expressed as a number of cycles per second or Hertz (Hz). One Hertz equals 1 cycle per second.
Regardless of its wavelength, all EMR travels at the same velocity (c) which in a vacuum is approximately 300 million metres per second. The relationship between velocity, wavelength and frequency is given by:
C=λf
It is thus possible to determine either the frequency or the wavelength of an electromagnetic wave, providing that the other value is known. The change in velocity of EMR as it passes from one medium to another is defined as refraction. For most purposes, however, the velocity (c) may be regarded as a constant.
Figure 2.1 An electromagnetic wave and its components.
With the exception of some radar systems, wavelength rather than frequency is more commonly used to describe EMR. Figure 2.2 illustrates how different portions of the electromagnetic spectrum are designated in terms of wavelength bands. The bands of interest to remote sensing are as follows:
(i) the photographic ultraviolet, with wavelengths between 0.3 and 0.4 micrometres (μm) (300–400 nanometres); shorter wavelength ultraviolet radiation is absorbed by ozone in the upper atmosphere;
ii) the visible, with wavelengths between 0.4 and 0.7 μm; this region contains all the colours of light which can be perceived by the human eye;
iii) the near infrared, with wavelengths between 0.7 and 3 μm; although not visible, this radiation can be detected by films with infrared sensitive emulsions in the range of 0.7 to 1.3 μm;
iv) the middle infrared, with wavelengths between 3 μm and 8 μm; as with the shorter wavelength bands, energy in this region is primarily from reflected solar radiation and does not contain information on the thermal properties of materials;
v) the far (thermal) infrared, with wavelengths from 8 to 1000 μm; this region includes terrestrial radiation related to thermal emission;
vi) the microwave, with wavelengths between 1 millimetre and 100 centimetres; this region can be used to measure terrestrial emission but is also important for active sensors such as radar systems.
It should be understood that these divisions are arbitrary and that the electromagnetic spectrum, by definition, is the continuum of energy ranging from kilometres to nanometres in wavelength. The waves travel at 3 × 108 metres per second and are capable of propagation through a vacuum such as outer space. The broad categories listed above may be further subdivided. Within the visible band, for example, wavelengths between 0.4 and 0.5 μm correspond approximately to blue light, those between 0.5 and 0.6 μm to green light, and between 0.6 and 0.7 μm to red light. Much finer discrimination is also possible depending upon the spectral resoltinon of a particular sensor.
Figure 2.2 Electromagnetic spectrum showing bands employed in remote sensing.
While the wave model is the more appropriate for describing how EMR travels through space, the particle model is better for explaining how it is detected and measured. In accordance with the particle model, EMR is emitted in discrete units called quanta or photons. When a photon emitted from an object reaches a sensor it causes a physical reaction which can be amplified and measured. This may be the exposure of a silver halide grain in a film emulsion or a voltage signal in an electronic system. Planck's quantum theory states that the energy of a photon can be described by:
E= hf
where E = the amount of energy in Joules,
h = Planck's constant of 6.626 X 10-34 Joules-seconds, and
f= the frequency of the radiation.
Thus it is clear that the higher the frequency of EMR the more energy there is available, which facilitates its detection. In addition, the relationship between velocity, wavelength and frequency (c = λ f) indicates that the shorter the wavelength, the greater the amount of available energy. Since there is less energy in the longer wavelength bands, sensors operating in these regions generally have a coarser spatial resolution because they must collect photons from a wider area in order to receive a measurable signal.
With the exception of active sensor systems, which provide their own illumination, remote sensing relies on radiation from natural sources. Whether dealing with reflected solar radiation or with radiation emitted directly from the earth's surface and atmosphere, it is important to understand certain fundamentals. All matter which is at a temperature greater than absolute zero (273°C or 0° Kelvin) constantly emits electromagnetic radiation. The amount of EMR which is emitted by an object and its spectral distribution (i.e. the range of wavelengths) can be graphically described by a spectral emittance curve as shown in Figure 2.3. This curve is for an idealized, hypothetical object called a blackbody which is capable of absorbing and re-emitting all of the energy which it receives, regardless of the wavelength.
There are two relationships which describe the amount and spectral characteristics of EMR emitted by a blackbody as a function of temperature. The Stefan Boltzmann Law states that the radiant emittance increases as the fourth power of the temperature increase, while Wein's Displacement Law states that as the temperature increases the emission reaches a peak at progressively shorter wavelengths. The shift of wavelengths with temperature can be seen when a piece of metal is heated: when cool, it emits no visible light but as it becomes hotter it begins to glow a dark red, then orange, then yellow and finally white at high temperatures.
Figure 2.3 Plot of radiance from a blackbody aganist wavelength, with temperature as a varibale. (After T.M. Lillesand and R.W. Kiefer, 1979)
Although blackbodies are hypothetical, actual materials generally exhibit a similar pattern of emission, at least within certain wavelength ranges. These materials, however, emit less EMR than would be emitted by a blackbody at the same temperature. The ratio between the actual and theoretical emission is referred to as the emissivity of the material. Because of differences in emissivity, it is sometimes possible to distinguish between two materials which have the same surface temperature, due to the difference in the amount of radiation which they emit.
Radiation from the sun can be approximated by the spectral emission curve for a blackbody at 6000° Kelvin, while terrestrial radiation (considered as a whole) resembles that of a 300° K blackbody. Solar radiation peaks at around 0.5 micrometres (500 nm) in the visible portion of the spectrum while the earth's energy peak is approximately 9.7 micrometres in the thermal infrared. Thus, if one is interested in studying phenomena which can be observed only at short wavelengths, data collection is generally restricted to daylight hours (the detection of bioluminescence being an exception) or to using an active sensor.
The energy which a sensor receives from a target of interest must pass through the atmosphere. The gaseous components and particulate matter within the atmosphere can affect the intensity and spectral distribution of the energy and may impair or prevent observations of surface features. The magnitude of atmospheric effects depends upon such factors as the path length, the wavelengths being observed and the day-to-day variation in atmospheric conditions. In the case of reflected energy, the atmosphere intervenes between the source of illumination and the target, as well as between the target and the sensor. For emitted wavelengths the situation is simpler since the target is the source of illumination.
The most easily discernible effect of the atmosphere at visible wavelengths is that of scattering. Scattering is simply the reflection of energy by particles in the atmosphere; while the individual particles are small, their net effect can be quite significant if they are present in appreciable quantities. Rayleigh scattering is caused by atmospheric molecules and minute particles which are considerably smaller than the wavelengths of radiation which they affect. This type of scattering primarily occurs in the ultraviolet and blue portions of the spectrum and, in fact, is responsible for the blue appearance of the daytime sky. It is also one of the major causes of haze in imagery. Mie scattering is caused by spherical particles such as dust or water vapour which are approximately the same size as the wavelengths which they affect. Although Mie scattering occurs throughout the photographic spectrum (ultraviolet to near infrared), it tends to affect longer wavelengths than does Rayleigh scattering. When EMR encounters particles which are several times larger than the affected wavelength, nonselective scattering occurs. The term nonselective means that all of the reflected solar wavelenghts are affected more or less equally. Water droplets in clouds or fog banks are nonselective scatterers; they appear to be white because all of the visible wavelengths are reflected. Due to their long wavelengths, microwave sensors are not affected by atmospheric scattering and are therefore able to “see” through clouds.
Because of scattering, the energy which is received by the sensor includes reflections from the atmosphere as well as the reflection from the target. The atmospheric reflection component is referred to as the path radiance and complex algorithms are needed to correct this effect. In the case of nonselective scatterers, such as clouds, no radiation from the target reaches the sensor, at least in the visible and infrared bands. In the case of Rayleigh and Mie scatterers, it is possible to measure the influence of path radiance by looking at very dark objects, which reflect relatively little energy, in several bands acquired at the same time. Visually, the effect of scattering is a more diffuse image with lower overall contrast.
Unlike scattering which redirects EMR and causes a loss of detail, atmospheric absorption actuallyreduces the amount of energy in certain wavelength bands. Although the atmosphere is essentially transparent in the visible portion of the spectrum, there are several regions in which it is partially or totally opaque. Various component gases in the atmosphere absorb or ‘take in’ energy at these wavelengths which are called absorption bands. Figure 2.4 illustrates the percentage of EMR which can pass through the atmosphere as a function of wavelength. Sensor systems are only designed to operate in atmospheric windows, nonabsorption regions of the spectrum where transmission is high.
When gas molecules absorb EMR, their energy level is raised; this energy is subsequently re-emitted as heat which is thermal infrared radiation. Atmospheric emissions can degrade the signal which reaches a sensor from a target in much the same way that scattering affects reflected energy. Again, it is possible to correct for this effect by comparing measurements made simultaneously in different bands.
Figure 2.4 Transmission of energy through the atmosphere as a function of wavelength. Wavelength regions of high transmittance are atmospheric window. Gases responsible for absorption are noted. (After F.F. Sabins, Jr., 1978)
When EMR interacts with matter it may be reflected, absorbed or transmitted. In the previous section these interactions were discussed with reference to the atmosphere. Remote sensing, however, is primarily concerned with how EMR is modified by the terrestrial and marine environments. In some respects, terrestrial interactions are easier to describe since, in most cases, these take place at the earth surface where transmission is not a significant factor. In the marine environment, on the other hand, certain wavelengths are transmitted; the energy which reaches the sensor may come from the water surface, from substances in the water column or from bottom materials.
The reflection of energy from a surface is generally described as being specular or diffuse (refer to Figure 2.5). Specular reflection occurs when the energy which is reflected by the surface continues to travel in one direction and the angle of reflection is equal to the angle of incidence. This is the type of reflection which is seen in mirrors or from a smooth sea surface, at visible wavelengths. In diffuse reflection the reflected energy, in contrast, is broken up or scattered in all directions. In reality, most surfaces are neither perfectly specular nor perfectly diffuse reflectors but fall somewhere between these two extremes.
For remote sensing purposes, it is the spectral content of diffuse reflection which generally provides the most information, at least with regards to the composition of surface materials. Specular reflection, however, can be useful in characterizing the surface roughness and geometry of different areas. In fact, it is the characteristic of surface roughness relative to the wavelength of the incident energy which, together with the incidence angle, determines whether a surface is a specular or diffuse reflector. The bright glint of sunlight reflected from calm (i.e. smooth) water is an example of specular reflectance; as the water surface becomes rougher, the reflectance becomes more diffuse unless the sun is low on the horizon. By measuring the relative roughness of the water, it may in turn be possible to infer the wind speed. Moreover, surfaces which are diffuse reflectors at one wavelength may be specular reflectors at a longer wavelength. Simultaneous measurements made at two or more wavelengths can thus be used to differentiate, for example, between sand and gravel beaches or between smooth ice and rough ice.
Figure 2.5 Specular and diffuse radar reflection. (After T.E. Avery and G.L. Berlin, 1986)
In most applications, however, specular reflectance is a hindrance rather than an asset. It is the pattern of spectral reflection rather than surface roughness which usually permits a determination of the chemical or biological processes which are active at the surface. When EMR strikes an object, some of the wavelengths are reflected while others are absorbed or transmitted. In the visible spectrum this selective reflectance of certain wavelengths is perceived as colour. The amount of energy which is reflected by an object in different wavelengths (relative to the energy which it receives) is termed the spectral reflectance and is an intrinsic property of each material. The reflectance characteristics of different materials can be illustrated graphically on a spectral reflectance curve as shown in Figure 2.6. In this figure, typical reflectance values are given for ocean water and crude oil. As can be seen, at certain wavelengths some materials can be readily distinguished while at others they may appear quite similar. Ideally it would be desirable to find a single band which would allow all of the features of interest to be separated. In practice this is not always possible. It is therefore necessary to have simultaneous measurements from several bands. When it is necessary to distinguish two or more features with similar spectral reflectance curves, such as different plant species, it may be possible to acquire data from a very narrow band where the reflectance differences are maximized. Alternatively, it may be possible to use a sensor system which is sensitive to small changes in reflectance values, i.e. has a fine radiometric resolution. These techniques may also be used to detect variations or changes within a given type of material.
Figure 2.6 Spectral reflectance of ocean water and a thin layer of crude oil. (After F.F. Sabins, Jr., 1978)
For materials which do not transmit EMR the incident energy which is not reflected is absorbed. As with the atmosphere, the absorbed energy is subsequently re-emitted usually as heat. As previously described, the amount of energy which is emitted is a function of the temperature and emissivity of the material (refer to Figure 2.7). Since the earth emits most strongly in the thermal infrared region and because the emissivity of water is essentially constant within this range, infrared measurements can be used to determine surface water temperatures with a reasonable degree of accuracy, if the measurements are calibrated. While such measurements can also be made for terrestrial materials, the variations in emissivity from one material to another make absolute rather than relative temperature determination much more difficult. It should be noted that energy is only emitted from the surface and that subsurface conditions may be quite different. For example, a film of oil on the surface of water will appear to be cooler than oil-free water at the same temperature since oil has a lower emissivity. This characteristic can be utilized to detect unauthorized discharges or to direct cleanup operations.
Figure 2.7 Effect of emissivity differences on radiant temperature. (After F.F. Sabins, Jr., 1978)
There is another type of emission, called fluorescence, in which absorbed radiation is re-emitted at a longer wavelength without first being converted into thermal energy. Many minerals fluoresce in the visible spectrum when exposed to ultraviolet radiation. The wavelengths emitted by fluorescent materials are generally in a number of very narrow well-defined bands which are characteristic of specific materials. For marine applications fluorescence may be used to identify chlorophyll, algae and various types of pollutants. Special fluorescent dyes also may be used to trace water masses. Fluorescence occurs naturally due to solar radiation or it canbe induced by active sensors equipped with lasers. In order to detect fluorescence, however, it is necessary to have a sensor with sufficiently fine spectral resolution.
The transmission of EMR through water is important when information is needed on conditions or phenomena below the surface. Transmission, however, is essentially limited to the visible spectrum and is greatest at the blue and green wavelengths (refer to Figure 2.8a). As might be expected, there is greater transmittance through clear water than through turbid water (refer to Figure 2.8b). It is perhaps less obvious that the peak transmittance shifts to slightly longer wavelengths as the water becomes more turbid. Transmittance can be altered by certain dissolved organic materials, both natural and man-made.
Figure 2.8a Light absorption by 10 m of pure water as a function of wavelenght. (After P.K. Weyl, 1970)
Figure 2.8b Variation of light transmission as a function of depth for various sea waters. (After P.K. Weyl, 1970)
The energy which is detected by a sensor may be reflected by the water surface, by particles suspended in the water column or by bottom materials. As the percentage of light which is transmitted decreases, the ability to “see” into the water is also decreased because the energy is attenuated both approaching and leaving the reflector. Suspended particles produce a volume scattering effect which is analogous to that caused by atmospheric aerosols; if the concentration is relatively low, their reflectance is super-imposed on top of the reflectance from bottom materials. At higher concentrations, suspended particles can effectively block out transmission to and from lower depths. In the absence of bottom reflections, the reflectance from water can beused to measure the concentration of suspended materials. The presence of chlorophyll is of particular interest for fisheries management as it is an index of primary productivity. As the level of chlorophyll increases, there is a decrease in reflectance between 0.4 and 0.5 μm and, at higher concentrations, there is an increased reflectance in the 0.5 to 0.6 μm range. Suspended sediments discharged by rivers are particularly strong reflectors. Mapping these sediment plumes is a means of studying water mixing and circulation patterns. Measurements in the 0.6 and 0.7μm band show a high degree of correlation with suspended sediment concentrations. Because they reflect so strongly over a wide range of wavelengths, however, high sediment concentrations can interfere with or prevent chlorophyll measurements.
Different investigators have different targets of interest; what may constitute “noise” in one study may be the “signal” in another. The energy which eventually reaches a sensor is reflected or emitted by a variety of environmental components. By knowing the spectral response patterns of these components it is possible to determine the optimum spectral regions for observation. The marine environment presents particular opportunities and challenges because of its spectral characteristics and its dynamic nature. In the next few chapters, we will examine the instrumentation which is available for its study.
