Tải bản đầy đủ - 0 (trang)
III. Physical Basis for Remote Sensing

III. Physical Basis for Remote Sensing

Tải bản đầy đủ - 0trang



0 .I


0.7 1.0Wavolongth




tmic ron s)

FIG.1. The electromagnetic spectrum. The lower part emphasizes the regions of

primary importance in remote multispectral sensing. (From Hoffer, 1967.)



Sensors that measure the energy naturally radiated by objects are passive

remote sensors. Active remote sensors, such as radar, transmit energy to

the object and measure the portion which is reflected back. Solar radiation

is the ultimate source of energy for passive systems. A portion of the incident solar radiation (at wavelengths from 0.4 to 3 pm) is immediately



reflected away from objects at the earth’s surface (Fig. 2). This energy

is a function of the reflectance of an object’s surface and is related to its

physical properties.

Another portion of incident solar radiation is absorbed and later emitted

as thermal energy at wavelengths of 3-1 5 pm. Emitted energy is a function

of the temperature and nature of the surface of a body. The nature of

radiating surfaces is described by their emissivities and is again related

to the physical properties of particular materials. The characteristics of

plant and crop reflectance and emittance will be discussed in a subsequent


Sun angle is an important consideration because it influences both the

quantity and quality of incident radiation. Illumination levels are affected

by latitude, season of the year, and hours before or after solar noon (Heller, 1970). Sun angle also affects the distribution of energy by wavelength.

At low angles, the blue portion of sunlight is almost completely scattered

owing to the extreme depth of atmosphere that the light must penetrate.

For these reasons, remote sensing missions dependent upon reflected radiation are generally best flown within 2 hours before and after solar noon,


While the sun is a relatively constant source of energy above the atmosphere, the amount of energy reaching the earth’s surface depends upon

FIG.2. Reflected (R) and emitted (E) energy exchange in a natural environment.

Remote sensing largely utilizes measurements of electromagnetic energy which is

reflected or emitted by objects receiving, and then returning, energy from the sun.

(From Landgrebe, 1973.)



the atmospheric conditions. Through scattering, reflection, and absorption,

the atmosphere alters the amount of solar energy striking the earth. Further

alteration occurs as energy reflected or emitted by a feature on the earth’s

surface travels back through the atmosphere before the sensor records it.

The chief cause of energy reduction in the visible portion of the spectrum

is scattering by aerosols, haze, smoke, and dust. The principal cause in

the infrared is absorption by water vapor, carbon dioxide, and ozone. In

summary, only the wavelengths that have a high atmospheric transmission

can practically be used for remote sensing.




In addition to the illumination and atmospheric effects described above,

vegetation and other terrestrial features affect the return of energy to a

remote sensor. An understanding of the interaction of energy between

plants, soil, water, etc., is crucial to the successful acquisition and interpretation of remote sensing measurements.

The spectral quality and intensity of crop reflectance and emittance depends on such factors as leaf morphology and pigmentation, canopy geometry, crop maturity, soil background, management and cultural practices,

and weather. In this section, the physical and physiological characteristics

of plants that are significant for multispectral remote sensing of agricultural

crops will be summarized. Some of the more comprehensive reviews and

discussions on this topic have been published by Gates et al. (1965), Knipling ( 1967, 1970), Gausman et al. (1972), and Sinclair et al. (1973).

I . Fundamentals of Leaf Reflectance

The reflectance of plant leaves is relatively low in the visible portion

of the spectrum (0.4 to 0.7 pm), with a slight peak at 0.53 pm (Fig. 3).

In the near-infrared, reflectance is quite high but gradually decreases to

a very low level (2.5 pm), at which emittance begins to dominate. Portions

of the incident energy are also absorbed or transmitted. The transmission

spectrum closely resembles the reflectance spectrum, but generally at a

lower level.

The low reflectance and transmittance of visible radiation is attributed

to the high absorption by leaf pigments, primarily the chlorophylls (Gates,

1965 ) . However, these pigments are highly transparent to infrared radiation, and the internal cellular structure of the leaf appears to determine

the high reflectance at these wavelengths. The primary evidence is the similarity between reflectance and transmittance (Knipling, 1967). The low

reflectance and transmittance at about 1.45, 1.95, and 2.6 pm are due to

strong water absorption (Allen and Richardson, 1968).












2 0

2 2

2 4



FIG. 3. Characteristic reflectance of green leaves. The three primary regions of

reflectance are: (1 ) the visible wavelength region, in which plant pigments dominate

the spectral response; (2) the region from 0.72 to 1.3 pm, where there is very little

absorption and most of the energy is either transmitted or reflected; and (3) the

water absorption region extending from about 1.3 to 3.0 rm. (From Lab. for Agr.

Remote Sensing, 1970.)

The reflectivity of leaves is determined by their internal structure; however, it is not known exactly where the reflecting surfaces are or to what

extent various surfaces or subcellular structures contribute to the total reflectance (Knipling, 1967). The Willstatter and Stoll (1913) theory explains leaf reflectance and transmittance as a result of critical or total reflectance of light at the cell wall-air interfaces of spongy mesophyll tissue;

the theory is based on observations of the spectral properties of leaves

in the visible wavelengths. However, recent research by Sinclair et al.

(1973) has shown that the Willstatter and Stoll theory does not account

for the infrared reflectance of leaves; they have offered an alternative hypothesis that reflectance arises from the diffuse characteristics of cell walls.

Their theory satisfactorily explains observed reflectance phenomena in both

the visible and infrared wavelengths.

Among the other important physiological factors commonly affecting

leaf reflectance are: maturation, senescence, and water content. In general,

as leaves mature, their visible reflectance decreases and infrared reflectance

increases. Gausman et al. (1972) attributed this effect to the greater numbers of intercellular air spaces in the mesophyll of mature leaves, compared

to those of more compact young leaves. Senescence produces the opposite



effect of maturation; i.e., visible reflectance decreases (Knipling, 1967).

Water content has a major effect on leaf reflectance (Fig. 4). Both visible and infrared reflectance are increased as leaf water content decreases

(Hoffer and Johannsen, 1969; Sinclair et al., 1973; Gausman et al., 1972).

However, changes in reflectance are not substantial until the leaves reach

about 75% relative turgidity; thus the change in reflectance is not a sensitive indicator of the initial stages of water stress (Knipling, 1967; Thomas

et al., 1966).

Nutrient stress is another factor that significantly affects the reflectance,

transmittance, and absorptance in both the visible and near-infrared wavelengths. Al-Abbas et al. (1974) compared the spectral characteristics of

normal maize leaves and leaves deficient in nitrogen, phosphorus, potassium, sulfur, magnesium, and calcium. The nutrient deficiencies caused reductions in chlorophyll concentration and absorptance at 0.53 and 0.64

pm. Positive correlations were found between moisture content and absorptance at 1.45 and 1.93 pm. And leaves from P- and Ca-deficient plants

absorbed less energy than those from normal plants in the near-infrared

wavelengths, while leaves from the S-, Mg-, K-, and N-deficient plants absorbed more than the normal.

2 . Reflectance of Crop Canopies

The reflectance characteristics of single leaves are basic to understanding

the reflectivity of crop canopies in the field but cannot be applied directly

without modification (Knipling, 1970). There are both quantitative and

qualitative differences in the spectra of single leaves and canopies. And

the reflectivity of canopies is considerably more complex than that from

single leaves because of the many more interacting variables in canopies.









FIG.4. Effects of differences in moisture content on leaf reflectance spectra. Percent moisture:

, 0-40; -. .-. ., 40-54; . . , , 54-60; -,

66-98. (From

Lab. for Agr. Remote Sensing, 1970.)



There are several variables encountered in remote sensing of crops (Hoffer,

1967), i.e., canopies grown under field conditions, that are not factors when

the spectra of single leaves are measured. These variables are listed below.

1 . Variations in amount of leaf area and ground cover due to differences

in planting date, soil type, soil moisture, uneven germination, and/or disease conditions which cause stunted, small plants. When ground cover is

not complete, soil type and soil moisture conditions per se may cause

marked differences in response.

2. Variations in maturity due to differences in variety, planting date,

soil type, and soil moisture.

3. Differences in cultural practices, such as fertilizing or harvesting.

4. Changes in reflectance and emission characteristics of the plants

caused by disease and/or moisture stress.

5 . Geometric configuration of the crop due to differences in row width,

row direction, or lodging of plants.

6. Environmental variables, such as atmospheric conditions, wind, angle

of reflection in relation to angle of solar incidence, and soil moisture conditions as affected by amount of previous rainfall and length of time and

weather conditions since the last rainfall.

The effects of some of the above crop variables affecting the multispectral response of crops have since been investigated by several researchers.

Two of the important factors affecting reflectance-leaf area and percent

ground cover-were studied by Bauer and Cipra (unpublished data). Using a field spectroradiometer similar to the one described by Learner el

al. (1973), they measured the reflectance of the canopies of five planting

rates of corn grown on dark and light-colored soils throughout three growing seasons. They found the strongest relationships between leaf area index

(LAI) and reflectance in the near-infrared region. Reflectance increased

linearly between LAIs of 0.5 and about 3; further increases in LA1 had

relatively little influence on reflectance. There was not as strong a relationship between ground cover and reflectance as for LAI. Bauer and Cipra

also found a significant interaction between soil type and canopy reflectance, particularly for lower levels of leaf area and ground cover. They

also noted that infrared reflectance decreased with maturity after the plants

had reached their maximum vegetative growth.

In other field experiments, Bauer and Cipra (unpublished data) measured the reflectance of corn canopies affected by H . maydis (southern

corn leaf blight) and nitrogen deficiency. The nonsystemic stress of blight

and the systemic stress of nitrogen deficiency both affected the spectral

response. Compared to healthy corn, blighted corn displayed increased reflectance in the chlorophyll absorption wavelengths and decreased reflectance in the green and reflective infrared wavelengths. The changes in re-


28 1

flectance were attributed to changes in canopy geometry as well as reflectance of individual leaves. Nitrogen deficiency caused increased reflectance

in the visible wavelengths and reduced the infrared reflectance compared

to the reflectance of canopies with adequate nitrogen fertilization. The

changes in reflectance were attributed to lower levels of chlorophyll in the

leaves and less leaf area and ground cover.

In one of the early investigations of the utility of aerial infrared-sensitive

film, Colwell ( 1956) demonstrated that the changes in reflectance recorded

on such film can be used for early detection of loss of vigor due to blackstem rust of cereal grain crops. Since then many investigators have used

Ektachrome Infrared Aero film for crop and vegetation studies. Among

them are Carneggie (1968), Heller (1968), Manzer and Cooper (1967),

and Norman and Fritz ( 1965 ) .

Healthy plant foliage characteristically appears as a bright red or magenta color with different species often distinguished by varying shades,

whereas unhealthy, damaged, or dying vegetation tends to deviate from

the red color. Reviewing the interaction between the characteristics of leaf

reflectance and the properties of infrared film, Knipling (1969) concluded

that there has been a tendency to attribute all deviations from the red color

of plants on infrared film to a lack of or decline in infrared reflectance.

He pointed out, however, that the image color of the film is sensitive to

reflected radiation of the green and red portions of the visible spectrum

as well as the near infrared (Fritz, 1967). In addition, a number of researchers have reported that most of the color differences of plants observed on false color, infrared-sensitive photographs also can be found visually and with conventional color photography (Ciesla et al., 1967; Heller,

1968; Knipling, 1 967). Disease, damage, and physiological stresses in

plants also change the geometry and density of foliage as well as the infrared reflectance of the individual leaves. These changes are manifested in

the visible as well as the infrared portion of the spectrum.

While there may typically be numerous possible variables present in a

remotely sensed agricultural crop scene, their effects on reflectance can

be quantitatively described to a considerable degree. Allen and Richardson

( 1968) and Allen, et al. ( 1970) applied the theory of Kubelka and Munk

( 193 1 ) for attenuation of light in a diffusing medium to a crop of constant

depth and random leaf orientation canopy and showed that spectral reflectance and transmittance of a plant canopy are functions of total leaf area,

an absorption coefficient, a scattering coefficient, and background reflectivity. The coefficients are related to the geometry of the canopy and optical

properties of individual leaves. Suits (1972) has expanded the model of

Allen et al. (1970) to include multiple layers having different biological

components. Suits calculated the directional reflectance rather than assum-



ing that canopies are Lambertian reflectors. Colwell (1974) examined various cause-effect relationships influencing canopy reflectance and concluded

that the following factors are important to understanding canopy reflectance: leaf hemispherical reflectance and transmittance, leaf area and orientation, characteristics of other components of the canopy (stalks or petioles), soil reflectance, solar zenith angle, look angle, and azimuth angle.


The following brief discussion of sensor systems will be limited to sensors that operate in the visible, reflective infrared, emissive or thermal infrared, and microwave portions of the electromagnetic spectrum. These

are the spectral regions that appear to have the most potential for agricultural surveys. The primary sensors for these spectral regions are photography, multispectral scanners, thermal infrared scanners, radar, and passive

microwave radiometers. The platform on which the sensors are mounted

is an important consideration, The two basic platforms are aircraft and


1 . Sensors

a . Aerial Photography. The oldest and most developed remote sensing

device is aerial photography. The U.S.Department of Agriculture has used

aerial photography operationally since the 1930s to record land use and

serve as a soil-mapping base. The Manual of Photographic Interpretation

( 1960) covers the many applications of aerial photography.

Among the advantages of aerial photography are: superior spatial resolution, the relative simplicity of aerial photography and film processing,

the relatively low cost of equipment, and the considerable amount of information it provides human interpreters. Its disadvantages include: film return to earth for processing is more difficult than by telemetering electronic

signals, and, since the medium is film, the range of sensitivity (0.4 to 1.0

pm) is confined by film-emulsion technology to the visible and near-infrared regions (Heller, 1970).

b. Multispectral, Optical-Mechanical Scanners. These scanners are capable of collecting data in the visible and thermal portions of the spectrum

(0.3-14.0 pm). They are usually mounted on aerospace platforms, either

aircraft or satellite. Figure 5 is a diagram of a typical multispectral, opticalmechanical scanner. The energy reflected and emitted from a small area

of the earth’s surface is “seen” by the scanning mirror, then reflected

through a system of optics that disperses the energy spectrally. In this example, the energy in the visible wavelengths is spread by a quartz prism;

dichroic gratings serve as dispersive devices for the infrared energy. The




Scan Raster Line


Ground Resolution Patch

FIG.5. An airborne multispectral scanner. The scanner senses the reflected and

emitted energy of a scene in a line-by-line fashion. The optics of the system separate

it into wavelength bands, and the response in each wavelength band is then

stored on magnetic tape. (From Landgrebe, 1973.)

detectors, carefully selected for their sensitivity in the various portions of

the spectrum, measure the energy in specific wavelength bands. The size

of the resolution element, the instantaneous field or view of the scanner,

is a function of the scanner configuration and the altitude of the platform.

As the platform passes over an area, the mirror scans the ground surface

in successive strips or scan lines. The rotating motion of this mirror allows

the energy along a scan line to be measured. The simultaneous forward

movement of the platform, which is perpendicular to the scan line, brings

successive strips of terrain into view. Thus, a continuous area of the earth’s

surface can be sensed by using several wavelengths bands, which can encompass the entire optical portion of the electromagnetic spectrum.

The output signals from the detectors are amplified and then simultaneously recorded on magnetic tape or transmitted directly to the ground.

An important feature of this sensing system is that sampling the output

of all bands produces single data sets containing all the spectral information

available for a given resolution element. This is a convenient way to pack

the data for machine processing.

While photographic data collection systems tend to have better spatial

accuracy, optical-mechanical scanner data have better spectral resolution.



This is because in the latter the parameters of the detectors can be set

for much narrower wavelength bands. Data reformatting, calibration, and

registration need to be performed before the data is ready for analysis.

However, machine processing methods can be applied to multispectral

scanner data most easily since the data either are recorded in digital form

or can be converted from analog to digital form.

c. Thermal Infrared Scanners. In the thermal infrared, the reemitted

portion of the absorbed solar energy is recorded. This thermal sensing is

possible due to the fact that an object’s total radiation is temperature dependent, according to the Stefan-Boltzman law for a black body. Because

thermal sensors do not depend upon reflected energy; they may be operated

during the night as well as the day. However, sensors can obtain imagery

only in the atmospheric “windows” of 3.5-5.5 pm and 8-14 pm due to

atmospheric absorption by water, carbon dioxide, oxygen, and ozone.

Optical-mechanical scanners, such as those described above, are the

most frequently used sensors for obtaining thermal measurements. Thermal

infrared response is a constantly changing function of the diurnal cycle;

hence, relative differences between objects are of little inherent value except for facilitating boundary detection between communities.

d. Radar. Radar is an active remote sensing system which operates in

the microwave portion of the electromagnetic spectrum. Its imaging capability depends upon the return of energy from the target, which is supplied

by the target itself. Some radar systems are side-looking; that is, they scan

a path to one side of the aircraft’s flight path. A pulse of energy is transmitted from the radar antenna, and the relative intensity of the reflections

from objects being sensed produces an image.

The advantages of radar include day-or-night and nearly all-weather capability, and the ability to penetrate vegetation to show soil patterns (Holter

et al., 1970). Radar also possesses some undesirable features, including

rather coarse spatial resolution and the fact that presently side-looking

radars are uncalibrated. And, radar’s single-frequency characteristic images

are extremely limiting for vegetation identification (Morain, 1974). Nevertheless, radar can be used to good advantage in cloud-covered areas, where

it would be nearly impossible to obtain data with sensors dependent upon

reflected solar energy.

e. Passive Microwave Radiometers. These radiometers differ from radars in that they sense the natural radiation emitted by objects rather than

artificial illumination. Passive microwave systems generally operate in the

shorter-wavelength portions of the microwave spectrum. Microwave systems image primarily emitted radiation, as do thermal infrared devices,

but microwave imagery is more linearly a function of temperature than

is infrared imagery.

Drawbacks of passive microwave are their low spatial resolution. The



limited information available concerning the use of passive microwave radiometers suggests that it would have little value for crop identification,

but it probably has the most potential for measuring soil moisture (Ulaby

et al., 1974).

2 . Aerospace Platforms

For remote sensing applications, sensors mounted in either an aircraft

or a satellite collect the data. A sensor such as a spectroradiometer may

also be mounted on the bucket of an aerial-lift truck to gather research

data from altitudes of 5-20 meters. Until 1972, most remote sensing data

were acquired from aircraft at altitudes ranging from 500 to 20,000 meters.

The astronauts obtained infrared and multiband photography of earth during the Gemini and Apollo missions, which was used for research on the

use of satellite-acquired imagery. However, remote sensing data for earth

resources was not routinely acquired until the launch of the first Earth

Resources Technology Satellite (ERTS-1 ), conducted by the National Aeronautics and Space Administration (NASA) in July 1972. The ERTS-1

system mission demonstrated the feasibility of utilizing multispectral remote

sensing from space in practical earth resources management applications.

In January 1975 ERTS-1 was renamed LANDSAT-1 and a second sattelite, LANDSAT-2, was launched.

The orbit of LANDSAT is sun synchronous and crosses the equator

in a north-south direction at about 0942 local time (Boeckel, 1974).

Successive orbits are separated by about 2870 km at the equator. The orbit

moves approximately 159 km to the west each day, resulting in a repetition

of the same ground track every 18 days.

The primary sensor is a multispectral scanner that measures energy in

four spectral bands, 0.5 to 0.6, 0.6 to 0.7, 0.7 to 0.8, and 0.8 to 1.1 pm.

The data are either telemetered directly to receiving stations or recorded

for playback when the satellite is over a receiving station. The instantaneous field of view or the size of each resolution element is approximately

80 meters. In the first 18 months of operation more than 100,000 scenes

( 1 85 km or 100 nautical miles square) were obtained. The data are available as either 70-mm or 9-inch imagery of each band, color composites

of three bands simulating false color infrared photography, or in digital

computer-compatible tape form. An example of LANDSAT-1 imagery is

shown in Fig. 6.





Until recently, manual interpretation was the primary method of extracting information from remotely sensed imagery. However, the volume of

Tài liệu bạn tìm kiếm đã sẵn sàng tải về

III. Physical Basis for Remote Sensing

Tải bản đầy đủ ngay(0 tr)