Layers of ionization that appear in the ionospheric E region sporadically in time have been named sporadic-E layers, or Es. These sporadic ionization layers play a crucial role in small-scale irregularity generation (as they guarantee a plasma density gradient necessary for gradient-drift processes to develop) and in providing reliability of radio communication and navigation. Despite the fact that Es layers have been studied since the late sixties, a method to visualize the Es horizontal structure was suggested for the first time only recently [Kagan et al., 2000]. In the present paper we propose a new type of experiment for the reconstruction of the patchy-type Es structure and dynamics using a combination of the above-mentioned method with the method of artificial periodic irregularities (API) [Belikovich et al., 1999]. This combined method allows us for the first time to carry out Es tomography, to trace the three-dimensional (3-D) dynamics of sporadic ionization clouds and their associated 3-D neutral motions, and to study the effects of high-frequency radio waves on the sporadic-E layers. The procedure includes recording of the Es horizontal structure and dynamics with an all-sky camera via radio-wave-induced optical emissions in combination with measurements of the Es vertical structure and dynamics by the API method. Both methods complement each other and can be used simultaneously. The experiment is to be equipped with a powerful HF radio wave transmitter, all-sky cameras, an ionosonde, and an HF digital receiver.
 Layers of ionization that appear in the ionospheric E region sporadically from time to time have been called sporadic-E (Es) layers. Es layers are common phenomena all over the globe. First observed with ionosondes, now they are routinely recorded remotely by ionosondes and incoherent scatter radars (ISR) and are occasionally measured in situ with rocket-borne probes. These Es layers have attracted special scientific interest because they may significantly affect radio wave propagation and act as a source for small-scale irregularity generation caused by gradient-drift processes [see, e.g., Fejer and Kelley, 1980; Ecklund et al., 1981; Tanaka and Venkateswaran, 1982; Sudan, 1983; Riggin et al., 1986; Kagan and Kelley, 1998, and references therein].
 Sporadic-E phenomena have been studied since the late sixties and the basic mechanism for their generation is well developed and understood; that is, a vertical shear in the neutral wind gathers plasma in an ionization layer, which may exist for a long time due to long-lived metallic ions (for details, see the reviews by Whitehead  and Mathews ). However, there are still outstanding issues concerning the sporadic-E phenomena that have to be understood. One of those is the spatial structure of sporadic ionization clouds. Both ionosondes and ISRs, in most cases, provide information only about the Es altitude-time evolution.
 An exception is the ISR study by Miller and Smith [1975, 1978]. They performed experiments using the Arecibo incoherent scatter radar that showed that an Es layer seen in the ionograms simultaneously with F-layer branches (and called up to that time “transparent”) in fact had horizontal structure. In their experiment the Arecibo incoherent scatter radar showed the variation in plasma density as the radar beam was held at a fixed zenith angle and was scanned in azimuth with an angular speed of about 23 deg/min. They measured a one-dimensional horizontal structure of sporadic-E layers and thus confirmed the Whitehead theory of Es. Ever since their work, the transparent type of Es has been called “patchy,” that is, composed of ionization clouds.
 Circumstantial evidence that the Es layers may have complicated structure can be deduced from ionograms (Figures 1a and 1b) and ISR data (Figure 1c). So, for example, when no reflected signal from the F region is observed in the ionograms, it is because the Es completely hides the F region up to the Es critical frequency f0Es (Figure 1a). Thus the layer must fill the ionosonde beam. This type of Es was named “blanketing.” When both the F and E region branches are seen in the ionogram at frequencies ≤f0Es (Figure 1b), it means that some part of the transmitted radio waves is able to reach the F maximum and to come back. This type of sporadic-E layer has been called patchy since it should either have holes in it or be composed of separate ionization clouds to allow radio wave propagation up to the F region. The sporadic-E layer seen in the altitude-time-plasma frequency (density) plot measured by the Arecibo ISR on February 20, 1999 (Figure 1c), also argues in favor of the fact that the Es may have a rather complicated spatial distribution. Here the radar swings in a circular arc at 15° zenith angle. The remarkable variability is due to a combination of spatial gradients, plasma motion, and beam swinging.
 Visualization of a two-dimensional Es horizontal structure became possible for the first time only recently due to the optical method proposed by Kagan et al. . This method is based on HF-radio-wave-induced 557.7-nm airglow, which was for the first time observed in the ionospheric E region in connection with a sporadic-E event [Djuth et al., 1999; Kagan et al., 2000].
 Herein we further develop this method and propose a project and a scheme for an experiment for the reconstruction of the patchy-type Es structure and dynamics using a combination of the above-mentioned method with the method of artificial periodic irregularities (API) [Belikovich et al., 1999]. This combined method will for the first time make it possible to do Es tomography and simultaneously to trace the dynamics of the plasma and the neutral atmosphere. We start with a presentation of the optical method in section 2, followed by a description of API diagnostics in section 3. In section 4 we discuss the benefits of both methods and of their combination, and present a possible scheme for an experiment. We give our conclusions in section 5.
2. Method to Observe the Horizontal Structure of Sporadic-E Layers
 Artificial airglow occurs when, due to the interaction of a powerful electromagnetic wave with ionospheric plasma, electrons (either directly or having been accelerated by heater-induced Langmuir waves) acquire enough energy for collisional excitation of neutral species [Mantas and Carlson, 1996; Newman et al., 1988; Haslett and Megill, 1974; etc.]. Clouds of artificial airglow can be used to study the dynamics and characteristics of the neutral atmosphere and the ionosphere. Recently, Bernhardt et al.  have demonstrated how the observations of 630.0.nm airglow clouds induced by HF radio waves may be used for determining plasma drift velocities, neutral winds, diffusion coefficients, and collisional quenching times in the ionospheric F region. In this paper we propose a procedure to derive plasma and neutral parameters in the E region using images of 557.7 nm artificial airglow simultaneously with the method of artificial periodic irregularities, both induced by a ground-based HF radio wave source.
 Until the first observations of sporadic-E associated artificial airglow enhancements in the 557.7-nm line of atomic oxygen and in the N2 first positive molecular bands (Arecibo, January 1998) [Djuth et al., 1999; Kagan et al., 2000], the optical emissions were believed to come from the ionospheric F region only. The measurements showed strong airglow enhancements at 630.0 nm, which sometimes were accompanied by much weaker enhancements in 557.7 nm emission [Haslett and Megill, 1974]. The excitation energy of atomic oxygen for the 1D state is 1.97 eV (the transition to the ground state is accompanied by emitting a photon with a wavelength of 630.0 nm), while for the 1S state it is 4.19 eV (the transition to the lower-energy state is accompanied by emitting a photon with a wavelength of 557.7 nm). Thus to excite the atomic oxygen to the 1S state requires much more energetic electrons than to excite O(1D). The O(1D) lifetime is about 108 s in the natural ionosphere [Shunk and Nagy, 2000] and about 36 s in the heater-modified F region, since it is reduced by collisions [Bernhardt et al., 1988, 1989]. Such a relatively long lifetime of O(1D) leads to collisional quenching of the 1D state in the E region much faster than the atom is able to emit the 630.0-nm photon. The altitude range for the 630.0-nm airglow lies within the range of 250 ± 50 km in the ionospheric F region. This fact is used to determine the region of the 557.7-nm artificial airglow (which may come from both E and F regions) by the absence/presence of 630.0 nm airglow enhancements.
 Under natural ionospheric conditions the green-line airglow occurs at altitudes of 250 ± 50 km and 95 ± 2 km. The radiative lifetime of O(1S) is 0.71 s [Shunk and Nagy, 2000]. There is effectively no collisional quenching of O(1S) above 100 km. The collisional quenching of O(1S) is predominantly caused by O2 with a collisional quenching rate of 0.15 s−1 at 100 km and of about 0.008 s−1 near 120 km altitude [Torr, 1985]. This means that the green-line artificial airglow decays radiatively almost immediately following the pump wave turn-off. As we show further, this gives us the advantage of being able to trace transport processes in the E region, but in contrast to 630.0 nm artificial airglow clouds, it cannot be used to study diffusion. The E region diffusion, as we discuss below, may be measured via API relaxation times.
 The mechanism of the HF-radio-wave-induced airglow has been widely discussed over the years [see Mantas, 1994; Mantas and Carlson, 1996, and references therein]. Recent theories tend to explain airglow enhancements observed in ionosphere modification experiments by means of excitation by energetic electrons that have been accelerated in the interaction with HF-radio-wave-induced plasma waves, rather than by energetic electrons from the tail of the heated thermal gas. Such a preference is understandable, since the excitation of 557.7 nm emission (the excitation energy is 4.19 eV) and 640–675 nm emissions from the N2 first positive molecular bands (excitation energy of the order of 9 eV) would require electron temperatures much more than 10,000 K. Such temperatures are considered impossible to attain with heating facilities. The Arecibo HF transmitter's effective radiated power was 80 MW during the Arecibo campaign in January 1998 when the N2 first positive molecular bands were observed. Newman et al.  have shown that low plasma density (solar minimum conditions) is more important in obtaining larger temperature enhancements than is the transmitter power attained. Nowadays it is generally accepted that ionospheric electrons are accelerated via nonlinear interaction with Langmuir waves [Grach et al., 1984; Gurevich and Milikh, 1997; Newman et al., 1998]. Then the accelerated electrons collisionally excite atomic oxygen to the 1D and 1S states in a manner similar to the natural aurora caused by particle precipitation.
 For the first time we observed E region artificial airglow during an ionosphere modification campaign performed at Arecibo Observatory, Puerto Rico in January 19–29, 1998 (for more details, see Djuth et al.  and Kagan et al. ). The campaign was aimed at measuring large electron temperature enhancements and the formation of plasma cavities in the ionospheric F region under solar minimum conditions (low plasma density). The high power radio wave source was operated at a frequency of 3.175 MHz, with the highest HF level attained corresponding to an effective radiated power (ERP) of about 80 MW. Two photometers with 5° fields-of-view were pointed at 33° azimuth and 4° zenith angles to target the center of the heater beam at 250 km altitude. The photometers were used to measure 630.0 nm (the filter band pass was 0.45 nm) and 557.7 nm (filter band pass 0.45 nm) emissions. An all-sky CCD camera belonging to Cornell University was running either the 557.7 nm filter only or 557.7 nm and 630.0 nm filters in turn. The CCD imager of the Naval Research Laboratory was operated at 630.0 nm [Djuth et al., 1999; Kagan et al., 2000].
 The strongest 557.7 nm airglow was measured during the nights of January 22–23 and January 26, 1998 (AST). For both events the 630.0 nm emission was either completely absent or incomparably weaker. For both nights the induced 557.7 nm emission was observed in close correspondence with sporadic-E layers. On the night of January 26 the 557.7-nm airglow cloud showed a rather complicated structure and an emission intensity distribution (see Figure 2 of Kagan et al. ). Since the ionograms for the time of interest showed the patchy type of Es, Kagan et al.  assumed that the sporadic-E layer was composed of separate ionization clouds that modulated the HF radio transmitter beam. We reproduce their simplified diagram and the horizontal distribution of the 557.7 nm emission intensity in Figure 2. According to their scenario, the higher electromagnetic energy density under the Es plasma clouds (due to their proximity to the transmitter) results in more energetic accelerated electrons, which, in turn, collisionally excite the atomic oxygen 1S state.
 For visualization of the Es structure and dynamics there are two important facts. The first is that the lifetime of the atomic oxygen in the 1S is very short, 0.71 s. Hence the O(1S) excitation is almost immediately followed by emission of 557.7 nm photons.
 The second is that the mean free path for 10 eV super-energetic electrons near an altitude of 120 km is about 120 m, what is less than the altitude resolution of the instrumentation. Therefore, the 557.7 nm airglow represents the Es ionization clouds in real time and space. Thus if the altitude of the sporadic layer is known (due to either ionograms or ISR data or the API technique described below), then we do not even need to use triangulation to find the altitude of the 557.7 nm emission. However, in this case it is important to be sure that the emission comes from the E region. This may be tested with simultaneous 630.0 nm imaging: the absence of (or weaker) red-line airglow will confirm that the green-line emission comes from the E region.
 It is possible, in principle, that the patchy-type Es (as seen in the ionograms) may modulate the HF transmitter beam, but the major part of the electromagnetic wave energy penetrates to the F region. Then either only red or both red and green-line emissions should be observed. In this case green-line emissions may come from both the E and F regions. For the F region emissions, the 630.0 nm airglow will be stronger than the 557.7 nm one, and the patches in the green images will correspond to the holes in the sporadic-E layer projected to the HF radio wave reflection level in the F region. In such situations, the green-line artificial airglow clouds may be observed shifted from the electron acceleration region by a mean free path of the energetic electrons (which is about 2–3 km along the geomagnetic field for the conditions of the 1998 Arecibo campaign). Also, the green-line airglow clouds may be displaced from the clouds of artificial red-line airglow, since the O(1D) lifetime in the heater-modified F region is much longer. Then the images of the 557.7 nm clouds will correspond to holes in the sporadic-E layer projected on the altitude, which is the difference between the pump wave reflection level and the F region accelerated electrons' mean free path. They also will mark the regions of atomic oxygen collisional excitation by the accelerated electrons. Displacement between the centers of respective red- and green-line airglow clouds may provide additional information on the thermosphere parameters complementing the one derived from the heater-induced 630.0 nm airglow developed by Bernhardt et al. .
 Note that to excite artificial airglow in the ionospheric E region the frequency of the HF radio wave transmitter should be less than the Es critical frequency, and the effective radiated power must be enough to accelerate electrons up to about 5 eV.
 In Figure 3 we present an example of the sporadic-E horizontal structure as it was seen with the 557.7 nm airglow camera over Arecibo on January 26 (AST), 1998. The image is flat-fielded and projected for the altitude 122 km since the simultaneous ISR data showed the sporadic-E layer to be near 122 km in altitude. The image exposure time was 20 s. The exposure time plus the time needed for computer data recording will define the time resolution for the ionospheric parameters (Es and neutral wind horizontal velocities) derived from a sequence of all-sky images. The horizontal space resolution is defined by the angular resolution of the all-sky camera and the emission altitude.
3. Method of Artificial Periodic Irregularities
 Artificial periodic irregularities (API) are generated in antinodes of the standing electromagnetic wave formed due to interference of the HF radio waves transmitted vertically and reflected from the ionosphere. The API are horizontally aligned with a vertical scale of one-half of the wavelength λ of the transmitted wave. In different ionospheric regions the API are caused by different processes: a ponderomotive forcing out in the F region, electron pressure redistribution caused by electron thermal diffusion in the E region, and due to a change in the rate of the chemical reactions (the electron attachment/detachment coefficient in particular) in the D region (see Belikovich et al.  and references therein for details).
 The method of studying the ionosphere and neutral atmosphere with the use of these artificial ionospheric irregularities has been developed in the Radiophysical Research Institute (Nizhny Novgorod, Russia) since 1978 [Belikovich et al., 1999; Bakhmet'eva et al., 1998, and references therein]. The API diagnostic procedure is comprised of three steps: API generation, plasma diagnostics, and diagnostics of the neutral atmosphere.
 The method of using artificial periodic irregularities as tracers to study the neutral atmosphere and ionosphere is based on HF ionosphere modification and generation of API with a known wavelength (λpump/2) followed by ionospheric probing with radio waves of the same wavelength λprobe = λpump using the resonant backscatter condition. Information about the neutral atmosphere is obtained by measurements of the API relaxation time, which is defined after the pump turn-off and occurs by known processes, namely by ambipolar diffusion in the E region, by transport and diffusion in the F region, and by electron attachment/detachment in the D region.
 There are two ways to satisfy the condition of Bragg resonant scatter:
 The first is when the ionosphere pumping and probing waves are of the same frequency and polarization:
 The second is to use waves of different polarization and frequency for ionosphere pumping and probing,
where n is a refractive index.
 Condition (2) is satisfied over all the altitude range of the API generation up to the electromagnetic (pump/probe) wave reflection level. So, it is very similar to the regular backscatter echo from irregularities with a scale of one-half of the radar wavelength observed with coherent scatter radars. We present examples of backscatter from API, measured at the Sura heating facility (56.15°N, 44.3°E), when fprobe = fpump and both the pump and probe waves are of X-mode polarization. In Figure 4a we show an altitude-time-intensity plot for a typical example of API probing under quiet ionosphere conditions (no natural ionosphere turbulence). Here the vertical axis is altitude and the horizontal axis is number of scans lasting 3 minutes in total. Under perturbed ionosphere conditions, backscatter may show a very complicated structure with several ionization clouds and layers (Figure 4b). Continuous observations of backscatter from API after each cycle of pump turn-off-turn-on from 19:00 LT, August 16, 2000, until 5:30 LT, August 17 are presented in Figure 4c. Disappearance of D region plasma in the nighttime after 20:40 LT until 3:50 LT the next day is clearly seen. Also, starting from about the same time one can see the sporadic-E layer near 105 km and near 98 km the next day, lasting until about 4:20 LT.
 The fact that n in equation (3) is a function of plasma density N allows measurements of N by a stepwise change of the probe frequency. Obviously this procedure works when the frequencies of the pump and probe waves are less than the frequency of the F maximum and are distinguishable, so (fpump − fprobe) > Δf, where Δf is the spectral width of the sounding pulse. The API method allows us to reconstruct the plasma density profiles starting from altitudes where the electrons are strongly magnetized and O- and X-waves do not propagate together. To cover the frequency range of 1 MHz when fpumpX is fixed and fprobeO is changed with a step of 20 kHz (50 pulses) takes about one second. In Figure 5 we present an example of an electron density profile reconstructed by 15-min averaging of successive records of backscatter intensity for 16:30 LT on October 4, 1991, at the “Sura” heating facility. The thin sporadic-E layer with a maximum density of about 9.5 · 104 cm−3 near 98 km altitude and a much wider plasma density peak (Ne ≅ 7.3 · 104 cm−3) are clearly seen.
 For building the electron density profile for altitudes up to the F maximum, the use of the API method is an alternative to that of incoherent scatter radar (ISR) and beyond comparison for sites without ISR such as HAARP (High Frequency Active Auroral Research Project) and Sura.
 A routine scheme of the API diagnostics is shown in Figure 6. API generation takes from 0.1 up to 3 s, depending on effective radiated power (ERP), and necessitates ERP ≥ 20 MW [Benediktov et al., 1997]. The probe pulse has an effective radiated power of about 100 kW, a duration of about 25 μs, and a repetition rate of about 50 Hz.
 API diagnostics of the neutral atmosphere are based on measurements of the API relaxation time after a pump turn-off (see Figure 4a). In the E region this relaxation time is defined by ambipolar diffusion
where M is the ion mass, νin is the ion-neutral collisional frequency, κT is Boltzmann's constant, Te and Ti are electron and ion temperatures, respectively, and k is the irregularity wave number. At the altitudes of interest, the electron, ion, and neutral temperatures may be assumed to be the same. We may find the neutral density Nn and the temperature Tn by complementing (4) with the condition of atmosphere equilibrium
 Note that k in equation (4) is altitude dependent because the refractive index is a function of plasma density, which changes with altitude. Thus for the correct determination of Nn and Tn, it is necessary first to measure Ne(z) either with the API technique or with any other instruments, for example, ISR.
 Finally, Doppler measurements within the API method allow us to study vertical velocities of neutral particles [Bakhmet'eva et al., 1998], while turbulent neutral velocities are derived from the deviation of the altitude dependence of the relaxation time. To illustrate the capabilities of the API technique in neutral atmosphere diagnostics, we present in Figure 7 the time dependences for the neutral mass density and the neutral temperature near 106 km altitude, and the neutral vertical velocity near 117 km for the gravity wave event on February 27, 1997.
 Timing of the API measurements assumes we start with observations of the plasma density profiles (which takes ≤10 s) and then measure the neutral parameters (∼30–50 s).
 Note that to make the N measurements, the HF radio wave transmitter should be able to change frequency step-by-step. This is possible with the HF transmitters at the Sura, HAARP, and Tromsø facilities. The HIPAS and Arecibo heaters (before the last was destroyed by a hurricane) were able to transmit only at several fixed frequencies. In this case the problem may be mitigated if other instrumentation exists to measure electron density, for example, ISR, whose data for N may be used as input for calculating the neutral atmosphere parameters from the API relaxation time.
4. Es Optical/API Experiment
 The time for the experiment should be chosen to correspond with the local time of the maximum of dense sporadic-E occurrence during sufficiently long periods of dark time (neither sun nor moon). Hence it is reasonable to plan the experiment for the time of the new moon. If modification of the natural aurora by HF radio waves isn't of special interest [Blagoveshchenskaya et al., 2001; Sergienko et al., 1997], it is desirable to avoid natural aurora events when the experiment is carried out at high-latitude sites (Tromsø, HIPAS, HAARP). Thus the preferable time for an optical/heating experiment at high-latitude facilities is January–February. We have discarded December based on its proximity to the equinox and the observation of a higher occurrence of aurora during this time. At the Sura facility the experiment may be scheduled in late May or August, when sporadic-E layers are regularly observed and there are sufficiently long periods of dark time. However, August in Russia may be cloudy and rainy, preventing optical observations.
 The equipment necessary for the type of experiment proposed includes a powerful HF radio wave transmitter, an ionosonde, a digital HF receiver for receiving the backscatter signals, and all-sky camera(s). It is better to observe with the all-sky camera with no filter simultaneously with taking 557.7 nm images to distinguish between the sporadic-E ionization clouds and clouds in the lower atmosphere. As we have discussed above, simultaneous red-line images will help to distinguish between E and F region green-line emissions.
 The necessary requirements for the execution of the experiment are as follows: (1) The HF radio wave transmitter should be able to (a) radiate O- and X-polarization waves, (b) change the transmitter frequency stepwise, and (c) radiate HF radio waves of sufficiently high effective radiated power at low frequencies. (2) To induce sporadic-E associated artificial optical emissions, the pump frequency should be less than the Es critical frequency.
 The requirements 1c and 2 mean also that, for successful use of the method proposed herein, the Es critical frequency should exceed the lowest available radiated frequency of the HF radio wave transmitter. The latter is 3.175 MHz at Arecibo, 4.3 MHz at the Sura facility, 2.8 MHz at the HAARP facility and 3.85 MHz at the EISCAT heating facility in Tromsø.
 The other very important condition is that the effective radiative power (ERP) of the HF radio wave transmitter be high enough to excite the parametric decay instability. The threshold of the electric field near the reflection level ωpump = ω0 for excitation of the parametric decay instability is [Gurevich, 1978].
The heater intensity is given by W0 = E2/16πNkTTe, so the threshold heater intensity for PDI generation is
This gives a threshold PDI heater intensity of 4.5 · 10−4 and 4 · 10−3 for 120 and 105 km, respectively. We did these estimates based on the ionosphere parameters given by Gurevich  and assuming the electron and ion temperatures are equal. In performing the experiment the electron collisional frequency and plasma density should be found for the site location and the time of the experiment using the MSIS-E-90 model (available at the National Space Satellite Data Center web site http://nssdc.gsfc.nasa.gov/space/model/). One should also keep in mind that continuous heating will increase the electron temperature and thus reduce the threshold heater intensity.
 According to the estimates of Newman et al.  the threshold heater intensity for PDI generation for an Es gradient scale of 500 m, ERP = 80 MW and altitude of 120 km is of about 10−3 or less. This result is in very good agreement with our estimates above. The actual heater intensity for Es gradient scale 500 m, ERP = 80 MW and altitude 120 km is 1.5 · 10−2 [Newman et al., 1998]. At 105 km altitude the actual heater intensity is 1.7 · 10−3 for Es gradient scale 500 m, ERP = 80 MW. So, to induce PDI (and thus optical emissions) in the Es near 105 km altitude, the effective radiative power should be of the order of about 40 MW or higher. Note that the minimum ERP required also should be estimated in accordance with the time and location of the specific experimental campaign.
 The combined optical/API experiment assumes initial ionosphere diagnostics for Es occurrence and determination of the Es-type with the ionosonde. From then on, the ionosonde operates every 10 minutes until the Es disappears. After detecting the Es, the ideal procedure is to create the artificial periodic structure by HF O-mode radio waves and then to sound API with pulses of X-mode radio waves. In doing so the continuous O-mode heating would also induce artificial emissions. All-sky cameras (running 557.7 nm, 630.0 nm filters and no filter) provide simultaneous data on artificial airglow and the angular emission intensity distribution. After excluding the lower atmosphere clouds, we separate the sporadic-E associated emission by comparing the red and green-line images and simultaneous ionograms. Then we reconstruct the Es structure by flattening the images correspondingly to the Es altitude shown by the electron density profiles derived by the API technique.
 The information on the horizontal Es dynamics may be traced with a set of successive 557.7 nm images. During the Arecibo experiment described above we were able to use a 20 s exposure time for taking the 557.7 nm images. Since plasma motions in the E region are mainly due to neutral motions, we may as well trace horizontal neutral winds near the Es altitude.
 Altitude-time dependences for vertical neutral velocities (and thus for the plasma velocities also), neutral mass density and temperature are calculated in accordance with equations (4)–(7) using the API technique.
 Applying the inverse scheme, heating by X-mode HF waves and subsequent pulsing by O-mode radio waves, and comparison with the evidence from the first scheme, may help to understand the mechanism for generation/acceleration of super-energetic electrons.
 The combined optical/heating/API experiment proposed makes it possible to visualize the E region sporadic ionization clouds and their vertical and horizontal motions, to measure neutral density, temperature, and neutral vertical and horizontal velocities, and to study the effects of Es modification by HF radio waves.
 This information is of potential importance for the Space Weather Program. The sporadic-E structure affects ionospheric propagation of radio waves and thus the reliability of navigation and communication. Es modification by powerful HF electromagnetic waves may show the possibility of affecting radio wave propagation in a desirable way. The sporadic clouds of ionization and neutral motions influence each other and both are active components in small-scale irregularity generation. The Es clouds provide the gradients and the neutral winds provide the driving force necessary for the gradient drift and thermal processes to develop [Kagan and Kelley, 1998, 2000]. Finally, neutral atmosphere winds, temperature, and mass density are necessary components for testing upper atmosphere modeling.
 The work has been supported by Russian Foundation of Basic Research grants 01-05-65025, 00-05-64695, and 99-02-16479. Work at Cornell University was supported by the office of Naval Research under grant N00014-00-1-0658.