Monthly Archives: May 2014

General Properties of the Ionosphere

Basic Structure. The ionosphere is an ionized region in the upper atmosphere that, by generally ac­cepted convention, lies between an altitude range of 60 km to 1000 km. Nevertheless, the region above 1000 km but below 2000 km, called the protonosphere, is also ionized and may be important when considering the to­tality of ionization effects on radio systems. As a matter of convenience, some specialists have combined the ionosphere and protonosphere into a single region of ionization. For example, the integrated electron density from a ground station to a geosynchronous satellite (referenced to the vertical) is referred to as the total electron
content of the ionosphere (TEC), even though both ionospheric and protonospheric electrons contribute to the integral. For the purpose of this article, we shall use the more restricted definition for the ionosphere, generally placing the upper limit at approximately 1000 km. While there are equal numbers of free electrons and pos­itive ions within the ionosphere, it is the electron number density that characterizes the array of interesting phenomena associated with the region. The ionosphere is imbedded in the earth’s magnetic field, and this situation influences the distribution of the ionized constituents. A clear indication of this may be seen in the worldwide distribution of electron density in the upper ionosphere, which tends to be described by geomagnetic rather than geographic coordinates. Moreover, being a magnetoionic medium, the ionosphere has a profound effect upon radiowaves that interact with the medium.

To sun

10ОО I—

10a 104 106 N(h)

Fig. 1. Depiction of the ionospheric layers and the diurnal variation. [By permission of J. M. Goodman and Kluwer Academic Publishers, Norwell, MA (11).]

The ionospheric electron density distribution is logically evaluated first in terms of its height profile, followed by its geographical and temporal variabilities. Though there is abundant evidence suggesting a rather complex electron density profile comprised of several peaks and valleys, the basis for understanding fundamental properties of the ionosphere comes from a simple picture of an ionized medium dominated by a single region, or layer, having a distinct maximum in electron density. This is not without justification, since the highest and thickest component region, the so-called F layer, typically exhibits the greatest electron density. Moreover, in many radiowave applications, it is the F layer that exhibits the dominant interaction. Figure 1

Fig. 2. Various atmospheric and ionospheric layers, the depth of penetration of solar radiation, and the thermospheric temperature profile. [From a National Research Council report (62).]

depicts the various regions or layers of the ionosphere in terms of the electron number density. It has been observed that the height profile varies diurnally, seasonally, and as a function of solar activity.

Formation of the Ionosphere. The sun exerts a number of influences on the upper atmosphere, but the interactions of most importance for our discussion are photodissociation and photoionization. Figure 2 depicts the neutral atmosphere, its various regions, and the depth of penetration of the various components of solar flux.

In the lower atmosphere, species such as N2 and O2 dominate the constituent population, though other species such as water vapor, carbon dioxide, nitric oxide, and trace element gases are influential in specific contexts. In the upper atmosphere, however, molecular forms are dissociated by incoming solar flux into separate atomic components. Formally the lowest portion of the ionosphere is the so-called D layer at an altitude of ~ 60 km ± 20 km, but the free-electron and ion population rises dramatically at an altitude of ~ 100 km, which is the median altitude of the E layer. Two things occur at this altitude. First, oxygen becomes dissociated as a result of solar UV radiation. Secondly, the mixing of the atmosphere, so efficient below 100 km, ceases rather dramatically, and the region where this occurs is called the turbopause.

The process of dissociation is so efficient that we treat the distribution of neutral species in a vast segment of the upper atmosphere (i. e., above 200 km) as that of a monatomic gas. In the lower atmosphere (i. e., below

Fig. 3. (a) Profiles of ion concentrations, as a function of height, for daytime conditions. (b) Electron density distributions for day/night and solar maximum/minimum conditions. [From Jursa (8).]

roughly 200 km), the gas is largely polyatomic, although the transition between the two regimes is rather gradual between 100 km and 200 km. This has implications for the lifetime of ion-electron pairs created through photoionization. Also, in the altitude regime above about 200 km and well above the turbopause, collisions become a rarity, so that mixing of the various species becomes unimportant in comparison with diffusive forces. As a consequence, diffusive separation occurs, with constituents of the neutral gas seeking their own height distributions dictated by their atomic masses, the gas temperature, and the acceleration of gravity. Figure 3(a) shows height profiles of ionic species in the upper atmosphere, and Fig. 3(b) shows typical distributions of midlatitude electron density for daytime and nighttime under solar maximum and minimum conditions.

It may be seen that ionized monatomic oxygen is the majority ion between roughly 180 km and 800 km, and is wholly dominant between about 200 km and 500 km. Atomic hydrogen ions become important above 500 km, and the region from about 800 km to 2000 km is called the protonosphere. It should also be noted that above 500 km (i. e., the base of the exosphere), the neutral atmosphere is virtually collisionless and particles tend to move about freely. On the other hand, electrons and ions in the exosphere are still influenced by the earth’s magnetic field and electrodynamic forces.

The electron density distributions in the ionosphere and protonosphere are variable. Because of this, the boundary between the ionosphere and the protonosphere is not sharply defined, being dependent upon a number of factors including time of day, season, and solar activity. The protonosphere is often referred to as the plasmasphere, especially by magnetospheric scientists and those engaged in transionospheric TEC measurements.

Ionospheric Layering. Table 1 provides information about the various ionospheric layers, the altitude ranges of each, the principal ionic constituents, and the means of formation. A comment is appropriate here on the nature of ionospheric layering, with some emphasis on the historical distinctions made between the words layer and region as they pertain to the ionosphere. Often the terms are used interchangeably, and while neither is generally preferred, region is the more accurate description. This is because it does not convey the incorrect impression that sharp discontinuities in electron density exist at well-defined upper and lower boundaries. This is especially the case for the F region, and to a lesser extent for the D and E regions. From a historical perspective, the concept of layering derives from the appearance of the ionospheric regions on vertical – incidence ionospheric soundings, called ionograms (see the subsection “Sounder Measurement Method” below). Furthermore, the alphabetic designation of the ionospheric regions was also based upon the early sounding studies. On the other hand, there are certain situations for which the restrictive term layer is acceptable. For example, the normal E region may occasionally be characterized by an electron density profile displaying a degree of boundary sharpness. Aside from this, the most significant localized concentration of free electrons in the ionosphere, called sporadic E (or Es), exists as an isolated layer within the boundaries of the normal E region (see the section “Sporadic E” below). It is termed sporadic because it exhibits a generally unpredictable temporal and geographical distribution, and because of its limited geographical extent, it is sometimes referred to as a sporadic E patch.

As indicated above, the ionosphere is often described in terms of its component regions or layers. These were the so-called D, E, and F regions. These designations are largely based upon data obtained from crude sounder (i. e., ionogram) measurements undertaken in the 1920s and 1930s. These early measurements often exhibited evidence for an additional layer between regions E and F in the daytime ionosphere. This led to the notion that the F region is actually composed of two distinct regions (F1 and F2) having different properties.

The lowest region of the ionosphere, the D region, is important in the characterization of absorption losses for short-wave systems, and also as a reflecting layer for long-wave communication and navigation systems. There is also evidence for a bifurcation in the D region, with the upper portion (above 60 km) being produced by solar flux, and with the lower portion (below 60 km) being produced by galactic cosmic rays.

Ground-based vertical-incidence sounder measurements have provided the bulk of our current informa­tion about ionospheric structure (see the subsection “Sounder Measurement Method” below). Through applica­tion of ionogram inversion technology to allow for the radio-wave interaction effects, individual sounder stations provide information about the vertical distribution of ionization to the altitude of the F2 maximum (i. e., 300 km to 400 km). In addition, the worldwide distribution of these systems has allowed a good geographical picture to be developed using sophisticated mapping algorithms. These measurements are somewhat limited in the characterization of certain features such as the so-called E-F valley, and they cannot evaluate ionization above the F2 maximum. There is also a paucity of data over oceanic regions. Satellite measurements (viz., topside sounders and in situ probes) have been invaluable in the characterization of the F-region ionization density

Table 1: Properties of the Ionospheric Layers

Region Height Range

JW Range



Basis of






D 70-90



La X rays

E 90-130

~10n (day)

~0.3 (night)


Lft X rays;


~1010 (night)

(smooth diurnal variation)

~3.0 (day)

Chapman layer

Es 90-130

-ID12 (highly variable)

Metallic ions

Wind Shear & meteoric debris;



Auroral electrojet and precipitation

Fi 130-210

~2 X 10’1 (day)

~ 3-6 (day)


Helium II line;

Kmi™ 180

~ 0 (night)

(merges with F2

UV radiation;

layer at night)

(smooth diurnal

Chapman layer


F2 200-1000

~1012 (day)

~ 5-15(day)



hmax’v 300

~2 x 1011 (night)

~ 3-6 (night)

diffusion from



the F1 layer;

diurnal variation)



over oceanic regions. Rocket probes and incoherent backscatter radar measurements, which provide a clearer representation of the true electron density profile, typically reveal a relatively featureless profile exhibiting a single F-region maximum with several underlying ledges or profile derivative discontinuities. Nevertheless, a valley of ionization may often be observed between the E and F regions. Ionization above the F2 maximum may be deduced from satellite probes and Thomson-scattering radars, but a large amount of information has been derived from total electron content measurements using Faraday rotation or group path measurements of signals from geostationary satellites or Global Positioning System (GPS) satellites. Hunsucker (10) describes various ionospheric measurement techniques.

Simple layering occurs as the result of two factors. First, the atmospheric neutral density decreases expo­nentially with altitude, while the solar ionizing flux density increases with height above sea level. This leads to the formation of single region for which the ionization rate is maximized, and ultimately results in a layer having the so-called Chapman shape. This shape is based upon a simple theory advanced by Sidney Chapman in 1931 (Ref. 17; see Fig. 4). We observe nonetheless a degree of structure in the ionosphere, which suggests more than one layer. One cause for multilayer formation is the existence of a multicomponent atmosphere, each component of which possesses a separate height distribution at ionospheric altitudes. But there are other factors. Solar radiation is not monochromatic as suggested in the simple Chapman theory, and it has an en­ergy density that is not evenly distributed in the wavelength domain. Furthermore, its penetration depth and ionization capability depend upon wavelength and atmospheric constitution. All of this results is a photoion­ization rate, and an associated electron density profile, that are structured functions of altitude. It has been shown that the Chapman model is valid for the D, E, and F1 regions but is not generally valid for the F2 region.

Chapman Layer Theory. One of the basic tenets of Chapman theory is that solar radiation will penetrate to an altitude for which the total number of atoms or molecules, P (populating a column of unit cross sectional area directed toward the sun) is equal to the reciprocal of the absorption (or interaction) cross

Fig. 4. An idealized representation of ionization production in the atmosphere as the solar radiation encounters a neutral gas with exponentially increasing density. [By permission of J. M. Goodman and Kluwer Academic Publishers, Norwell, MA (11).]

section a that is P = 1/a. The peak in ionization will be produced in the neighborhood of that altitude, and the concept is valid for oblique solar illumination as well as for the case in which the sun is directly overhead. It is convenient to look at the production rate in terms of its deviation from the peak (overhead) value at height h0. For this it is useful to define a reduced height z, corresponding to the normalized departure of an arbitrary value of ionospheric height h from h0.

where h0 is the peak height for vertically incident radiation from the sun, and H is the neutral scale height given by the following expression:


H — kT/mg

where k is Boltzmann’s constant, T is the absolute gas temperature, m is the atomic or molecular mass, and g is the acceleration of gravity. Within the thermosphere (with hfi 100 km), the gas temperature is monotonically increasing, reaching an asymptotic level near the base of the exosphere. The temperature rises from « 180 K at the mesopause (and incidentally near the turbopause) to levels approaching a diurnal range of 600 K to 1100 K at solar minimum and 800 K to 1400 K at solar maximum. The heat sources include solar radiation, the dissipation of atmospheric gravity waves, and particle precipitation. The asymptotic levels of T are due to limits on the thermal conductivity of the gas. The scale height H is a convenient parameter, since it may be used as a measure of layer thickness for an equivalent fixed-density slab. More importantly, it has a physical meaning. If the atmosphere is in diffusive equilibrium governed by the force of gravity and the gas pressure
gradient, and N is the atomic or molecular gas density (as appropriate), we have

H = (*

where N0 is the atomic or molecular density at some reference height.

In a diffusively separated atmospheric environment, each constituent has its own unique scale height governed by its own molecular (atomic) mass. In an ionized gas in which the electrons and ions are coupled by electrostatic forces, the effective value of the mean molecular mass is «1 the mass of the positive ion. This is because the mass of the electron is essentially zero in comparison with the ion mass.

Figure 5 depicts the production-rate curves associated with an ideal Chapman-like production profile and a range of solar zenith angles x. It is seen that there are a number of curves, parametrized in terms of x, for which production rate maxima qmax, may be observed. The largest qmax occurs for x = 0 (overhead case corresponding to q = q0), and we see that other values for qmax, corresponding to oblique geometries wherein x = 0, will decrease in magnitude and occur at increasing heights as x becomes larger (i. e., the sun moves toward the horizon). Chapman theory yields the following rate-of-production formula:

At altitudes well above the peak in q, the rate of electron production drops off in an exponential fashion imitating the exponential decrease in gas pressure with height. In order to relate Chapman production curves to actual electron density distributions, we must examine loss processes and certain dynamic factors.

Cardiac Output Monitoring

Cardiac output is the volume of blood ejected by the heart per unit time. Because the capacity of circulatory transport is proportional to cardiac output, and the level of metabolism is limited by the capacity of oxygen transport, cardiac output has to be maintained above a level corresponding to the oxy­gen demand. Even if blood pressure is normal, cardiac output drops when peripheral vascular resistance is increased. Thus, to establish the state of circulatory function correctly, moni­toring both blood pressure and cardiac output is desirable. However, there is no well approved noninvasive method of cardiac output monitoring. In practice, the thermodilution method has been used most commonly in critically ill patients both during surgery and in intensive care units, although it is an invasive procedure.

In the thermodilution method, a Swan-Ganz thermodilu­tion catheter is introduced into the pulmonary artery through the superior vena cava and right heart as shown in Fig. 4. Approximately 10 mL of cold saline at near 0°C is injected instantaneously into the right atrium, and the temperature change is recorded by a thermistor placed in the pulmonary artery. Cardiac output is then obtained by the amount of cold saline divided by the area of the blood temperature record that lies under the baseline (9). It has been confirmed by many studies that the thermodilution method of cardiac out­put measurement is reliable and accurate enough for most clinical purposes. Although the monitoring is not continuous, measurement is repeatable, and the catheter can be placed for several days while the patient is in an intensive care unit.

The measurement of thoracic impedance has been studied as a method of assessing cardiac output. According to the original Kubicek’s method (10), four band electrodes are used so that two are attached around the neck and two around the upper abdomen. An ac current in the 20 to 100 kHz range at a current level within the range from 10 ^A to a few milli-

Figure 4. Thermodilution method for cardiac output measurement.

amps is supplied through the outer electrode pair, and the induced voltage is measured from the inner electrode pair. The stroke volume is computed from the slope of the thoracic impedance change in the ejection phase assuming a homoge­neous cylindrical model for large arteries. The cardiac output is expressed by the stroke volume times the heart rate. Al­though inconsistencies remain among the reports that have evaluated the impedance method in contrast to other methods of cardiac output measurement, its noninvasiveness is a great advantage and thus further improvements are desirable.


The ionosphere poses an interesting challenge for many radio systems that make use of signal transmission through all or some portion of that medium. Being a magnetoionic medium imbedded in a background neutral atmosphere, it exhibits very interesting refractive properties, including anisotropy, dispersion, and dissipation. The laminar ionosphere introduces an array of effects, which are related to the ionospheric component of radio refractivity. These include ray path bending, phase path increase, group path delay, absorption, Doppler shift, pulse dispersion, Faraday rotation, and magnetoionic path splitting. Inhomogeneities in the ionosphere give rise to temporal and spatial variations in the effects just cited. An understanding of the ionospheric personality provides information about a wide range of solar-terrestrial interactions, and it has significant space-weather implications. Space weather is a new discipline that includes a wide range of exoatmospheric phenomena of major importance to space systems and their operational effectiveness.

The main features of the ionosphere are well known, although details are subjects of continuing research. There are many excellent sources of information about the ionosphere, from both a theoretical and an exper­imental perspective. The books by Davies (1,2,3), Ratcliffe (4), and Giraud and Petit (5) should be consulted. Theoretical and plasma-physics aspects of the ionosphere have been discussed in a book by Kelley (6). A read­able account of the basic physics of the ionosphere has been given by Rishbeth (7). Other useful references, which place the ionosphere within a larger context of the geospace weather system, include the Air Force Handbook of Geophysics and the Space Environment (8), and an Introduction to the Space Environment by Tascione (9). Various techniques for probing the ionosphere have been described in a monograph by Hunsucker (10). From a practical perspective, Goodman (11), Johnson et al. (12), and McNamara (13) have published ex­positions on the ionosphere in connection with radio system applications. There are also proceedings of topical conferences and workshops. The Ionospheric Effects Symposia (14) have chronicled ionospheric research ac­tivities and applications since 1975; and the Commission of the European Communities has published reports dealing with ionospheric prediction and modeling (15,16).

The purpose of this article is to provide a general understanding of the ionosphere. The emphasis is on those ionospheric processes and phenomena that are encountered by users of radio propagation systems. More complete descriptions of the underlying physical processes may be found in various references cited in the text. A final section on the current status of ionospheric research is provided as an aid to specialists and graduate students.

Spatial Matched Filter Classifiers for SAR (2)

A typical target recognition using SAR is done in multiple stages and is illustrated by the block diagram in Fig. 7 (13). In the first stage, a CFAR detector prescreens by locating potential targets on the basis of radar amplitude. Since a single target may produce multiple detections, the CFAR detections are clustered (grouped together). Then a region of interest (ROI) around the centroid of each cluster is passed to the next stage of the algorithm for further processing.

The second stage takes each ROI as its input and analyzes it. The goal of this discrimination stage is to reject natural-clutter false alarms while accepting real targets. This stage consists of three steps: (1) determining the position and orientation of the detected object, (2) computing simple texture features, and (3) combining the features into a discrimination statistic that measures how “targetlike” the detection object is.

The third stage is classification, where a 2-D pattern-matching algorithm is used to (1) reject clutter false alarms caused by man-made clutter discretes (buildings, bridges, etc.) and (2) classify the remaining detected objects. Those detected objects that pass the second stage are matched against stored reference templates of targets. If none of the matches exceeds a minimum required score, the detected object is classified as clutter; otherwise, the detected object is assigned to the class with the highest match score.

Matched filters are investigated in 2 as pattern-matching classifiers in the target recognition sys­tem shown in Fig. 7. They are synthetic discriminant function (SDF), the minimum average correlation energy (MACE) filter, the quadratic distance correlation classifier (QDCC), and the shift-invariant 2-D


Spatial Matched Filter Classifiers for SAR (2)

Fig. 6. Edges refined using IFSAR result.

pattern-matching classifier. The basic structure of the SDF and MACE filter is characterized in the frequency domain by

where H denotes the DFT of the spatial matched filter. The matrix X is composed of a set of target training vectors obtained by taking the DFT of the target training images. The vector U represents a set of constraints imposed on the values of the correlation peaks obtained when the training vectors are run through the spatial matched filter. The matrix A represents a positive definite weighting matrix. A is an identity matrix for SDF

Spatial Matched Filter Classifiers for SAR (2)

Fig. 7. Block diagram of a typical baseline target recognition system. [Adapted from Novak et al. (13)].

and is the inverse of the following matrix D.

where N is the number of training images and p is the dimension of the training vectors.

In the QDCC, the DFT of the spatial matched filter is expressed by

where m1 and m2 are means of the DFTs of the training images for classes 1 and 2, respectively. S is a diagonal matrix defined by

where M1 and M2 are matrices with elements of m1 and m2 placed on the main diagonal, and Xi and Yi are ith training vectors from classes 1 and 2, respectively.

In the shift-invariant 2-D pattern-matching classifier, the correlation scores are calculated by

where T is the DFT of the dB-normalized test image and Ri is the ith reference template.

Novak et al. (2) did extensive experiment with the high-resolution (1 ft x 1 ft) fully polarimetric SAR data. In the four-class classification experiment using four types of spatial matched filter classifiers, it is reported that all targets are correctly classified (2).

Threshold Network Synthesis

The significance of the preceding results is that assuming TGs can be built with a cost and delay comparable to that of logic gates, many basic functions can be computed much faster and/or much cheaper using TGs than using logic gates. This is one of the motivations for investigating devices able to implement TGs. However, the usefulness of threshold logic as a design alternative, in general, is determined not only by the availability, cost, and capabilities of the basic building blocks but also by the existence of synthesis procedures. The problem to be solved at this level can be stated as given a combinational logic function, described in the functional do­main (by means of truth tables, logic expressions, etc.), derive a network of the available building blocks realizing f that is optimal according to some design criteria.

Many logic synthesis algorithms exist for targeting conven­tional logic gates but few have been developed for TGs, al­though the problem was addressed as early as the beginning of the 1970s by Muroga. The procedure described by this au­thor (1) transforms the problem of deriving the smallest (low­est gate count) feed-forward network realizing a given func­tion in a sequence of mixed integer linear programming (MILP) problems. The problem of determining whether a given function f can be realized by a feed-forward threshold network with M gates, and if it can, determining the weights and the threshold for each of the M elements can be formu

Concerning two-level (depth-2) threshold networks, an al­gorithm called LSAT (23), inspired in techniques used in clas­sical two-level minimization of logic circuits, has been devel­oped. The core of the algorithm performs as follows. Suppose we have a two-level threshold network satisfying the follow­ing conditions: (1) weights of first-level TGs are restricted to the range [— z, +z] and (2) weights of the second-level gate are all equal to 1 and the threshold of this gate is S. Another threshold network that also satisfies previous conditions (1) and (2) with a minimal number of gates is obtained. This op­eration is repeated increasing S by 1 until a value of S is reached for which no solution is found. As a two-level AND-OR network is a threshold network of the type handled by the procedure with z = 1, S = 1, the algorithm is started with this network. Such a two-level circuit is easy to obtain and in fact is a standard input for other synthesis tools. LSAT has a run-time polynomial in the input size given by n X z, where n stands for the number of variables and z defines the allowed range for the weights. This means central processing unit (CPU) time increases if large weights are required.

The practical use of synthesis procedures for TGs is not restricted to the design of integrated circuits but to areas such as artificial neural networks or matching learning. Dif­ferent problems encountered in these fields are naturally for­mulated as threshold network synthesis problems