Modeling 3-D artificial ionospheric ducts

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH: SPACE PHYSICS, VOL. 118, , doi: /2013ja018823, 2013 Modeling 3-D artificial ionospheric ducts K. A. Zawdie, 1 J. D. Huba, 2 and T.-W. Wu 2 Received 4 March 2013; revised 26 August 2013; accepted 22 October 2013; published 26 November [1] The injection of powerful HF waves into the ionosphere leads to strong electron heating followed by a pressure perturbation, which can drive electrons along the field line to the conjugate side, creating what is known as an artificial duct. The electron temperature along the duct is above the ambient temperature; the electron density is depleted on the heating side and increased on the conjugate side. Duct formation due to HF heating has previously been studied using a modified version of SAMI2 (Sami2 is Another Model of the Ionosphere). We use a similarly modified version of SAMI3 to examine the effects of zonal E B drifts on interhemispheric ducts created by heating over Arecibo. We found that the longitudinal E B drifts, particularly those caused by the zonal neutral winds, significantly suppress the effects of HF heating on the conjugate side, reducing the temperature and density increases by about 90% and 75%, respectively. Citation: Zawdie, K. A., J. D. Huba, and T.-W. Wu (2013), Modeling 3-D artificial ionospheric ducts, J. Geophys. Res. Space Physics, 118, , doi: /2013ja Introduction [2] Powerful HF heaters at High Frequency Active Auroral Research Program (HAARP) and other facilities can create strong electron heating by injecting HF waves into the ionosphere. The electron heating is followed by a depletion of electrons in the heated region, which can create a pressure perturbation that propagates along the entire magnetic field line such that an artificial duct is formed. Such ducts have been shown to act as waveguides for VLF/ELF waves. These types of experiments are crucial to help us understand the complex physical processes that are associated with high-power wave interactions with plasmas. [3] A number of campaigns have been conducted over the years where density structures formed from HF heating have been observed and analyzed [Duncan et al., 1988; Bernhardt et al., 1988; Milikh et al., 2010a]. The majority of these investigations have focused on high latitudes because the operational radiowave heaters (HAARP, European Incoherent Scatter (EISCAT), SURA) are located at magnetic latitudes above 40 ı. A number of scientific investigations were carried out at Arecibo, which is located at 28 ı magnetic latitude, in the 1980s and 1990s [Duncan et al., 1988; Bernhardt et al., 1988; Newman et al., 1988; González et al., 2005] before the radiowave heater was destroyed by a 1998 hurricane. An updated heating facility is currently being installed at Arecibo, which should be available for 1 Space Science Division, Naval Research Laboratory, Washington, DC, USA. 2 Plasma Physics Division, Naval Research Laboratory, Washington, DC, USA. Corresponding author: K. Zawdie, Code 7633, Space Science Division, Naval Research Laboratory, Washington, DC, 20375, USA. (kzawdie@ssd5.nrl.navy.mil) American Geophysical Union. All Rights Reserved /13/ /2013JA experiments in the near future. The conjugate latitude for Arecibo is just off the coast of Argentina and the apex height of the magnetic field line connecting them is about 2200 km. The heater is expected to operate using 5.1 and MHz bands. Once the heater is operational, there will be many opportunities to produce ionospheric heating effects that may be observable in the conjugate ionosphere. A reliable model of duct formation is necessary to analyze the data and guide future expeditions. [4] Several modeling studies have investigated the processes that could lead to field-aligned duct formation using strong HF ionospheric heating. Some of these inquiries occurred during the time that the Arecibo heater was operational, but they were restricted to 1-D models and limited to altitudes below 500 km [Newman et al., 1988]. More recently, there have been a series of studies using modified versions of SAMI2, which is ideal for such investigations since it includes the ion inertia parallel to B in the ion momentum equation. This is critical to modeling artificial ducts since it allows for the study of sound wave propagation in the plasma. Also, SAMI2 models the plasma along the entire field line so conjugate effects can be investigated. [5] The first study using a modified version of SAMI2 was by Perrine et al. [2006]. They added a localized, Gaussianshaped heating source to an early version of SAMI2 to demonstrate that ducts could form in the ionosphere from strong thermal perturbations. The authors demonstrated that HF heating generates a thermal wave in the plasma, which propagates through the topside ionosphere and down the other side of the magnetic field line. The wave drove ion outflows and displaced the ambient plasma leading to the formation of a density duct that stretched along the magnetic field line to the conjugate point. The study focused on the HF heating response of plasma confined to a single magnetic field line, so it was essentially 1-D. The authors also made a number of comparisons to determine the effect of the basic heating parameters on the creation of artificial ducts.

2 [6] A second study, by Milikh et al. [2010b], also used a localized, Gaussian-shaped heating source, but it used a more robust version of SAMI2 as the basis for the heating model. The purpose of the study was to validate the new heating model against two experiments: the first experiment was done at the Sura heating facility and used the DEME- TER (Detection of Electro-Magnetic Emissions Transmitted from Earthquake Regions) satellite as a diagnostic tool; the second was an EISCAT heating experiment using the EIS- CAT incoherent scatter radar (ISR) as a diagnostic tool. The authors showed that the new heating model reproduced the experimental observations. [7] A recent study by Wu et al. [2012] examined the impact of HF heating at Arecibo on the conjugate ionosphere. In particular, they studied the development and formation of enhanced electron temperature along the heated field line and the associated changes in electron density. They found that the conjugate effects occurred during periods of relatively low electron densities, such as during the night, or during periods of low F10.7. The effects on the conjugate ionosphere were restricted to the topside F layer and did not penetrate below the F peak on the conjugate side. The primary inhibitor of conjugate effects was heating losses associated with inelastic electron-neutral (N 2 ) collisions. [8] Despite the successes of the artificial heating models, questions still remain. In particular, SAMI2 does not model phenomena in the zonal direction, so the effect of the longitudinal E B drift is not included. Longitudinal E B drifts drive electrons to other field lines and could reduce the effectiveness of artificial duct formation in the ionosphere. The neutral winds are the main driver of the E B drifts in the ionosphere; we use the SAMI3/ESF (Equatorial Spread F) [Huba et al., 2008] to investigate the effects of these winds. In SAMI3/ESF, the E B drifts are determined by the perpendicular electric field E = rˆ which is computed self consistently. This differs from SAMI2, which uses the Fejer/Scherliess empirical model [Scherliess and Fejer, 1999] for the E B drifts and thus does not provide insight into the specific physical mechanisms that affect the creation of artificial ducts. The objective of this paper is to examine the effects of E B drifts on the generation of artificial ducts in the ionosphere. [9] This paper is organized as follows: Section 2 describes the details of the 3-D ionospheric model used in the study. Section 3 describes the simulation conditions and results, and section 4 describes the conclusions and suggestions for future work. 2. The SAMI3/ESF Model [10] SAMI3/ESF is a three-dimensional, physics-based model of the ionosphere derived from the two-dimensional model SAMI2 [Huba et al., 2000]. SAMI2 and SAMI3 have been benchmarked against various ionosphere data sources [Swisdak et al., 2006; Kralletal., 2009; Shim et al., 2011, 2012]. SAMI3/ESF models the plasma and chemical evolution of seven ion species (H +,He +,O +,O2 +,N +,N + 2, and NO + ). The code includes 21 chemical reactions, radiative recombination, and simulates plasma along the entire dipole electric field line. SAMI3/ESF uses a nonorthogonal, nonuniform fixed grid, which is periodic in longitude and simulates a 4 ı longitudinal wedge of the ionosphere ZAWDIE ET AL.: MODELING 3D DUCTS with a fine longitudinal grid. The code solves the electron temperature equation as well as the complete ion temperature equation for three ion species (H +,He +,ando + ). Quasi neutrality is assumed, so the electron density is determined by summing the densities of each ion species. The neutral composition and temperature are specified using the empirical NRLMSISE00 model. The neutral wind is specified using the horizontal wind model (HWM07) or it can be set to a constant for simplicity. The magnetic field is a nontilted dipole so geographic and magnetic latitude are the same. [11] SAMI3/ESF includes E B drift perpendicular to B (both vertical and longitudinal), and ion inertia parallel to B, which is included in the ion momentum equation for motion along the geomagnetic field. The E B drifts are determined by the perpendicular electric field E = rˆ which is computed self consistently. The electric potential equation used in SAMI3/ESF in dipole @ˆ @F Here p and are dipolar coordinates. The represent the field line integrated Hall and Pedersen conductivities. The F terms are integrated functions of the gravitational force and neutral wind velocities along the field line. The full equation and terms are described in detail in Huba et al. [2008]. By selecting the Hall, neutral winds, and gravitational terms individually, we can determine the effect of each term on HF heating. For this study, we neglect the background global neutral wind dynamo; and only consider the local wind dynamo associated with the density modification caused by the heating. [12] We add a source term to the electron temperature equation in SAMI3/ESF to simulate HF heating, similarly to Milikh et al. [2010b] and Wu et al. [2012]. The original electron temperature equation in the SAMI3/ESF includes heating terms from electron-neutral collisions (Q en ), electron-ion collisions (Q ei ), and photo-electron heating (Q phe ). We add a source term (Q RF ) to represent the high-power radio wave heating, as 2 3 n e k b2 e = Q en + Q ei + Q phe + Q RF (2) In equation (2), the second term on the left hand side is a diffusion term, e is the parallel electron thermal conductivity, k is the Boltzmann constant, and b s is the component of the magnetic field in the field line direction. These terms are defined in detail in Huba et al. [2000]. [13] Similar to earlier studies [Wu et al., 2012; Milikh et al., 2010b; Perrine et al., 2006], we use a Gaussianshaped source term (also known as the hot brick model): dte Q RF = dt 0 exp[ (z z 0 ) 2 / z 2 ] exp[ ( 0 ) 2 / 2 ]exp[ ( 0 ) 2 / 2 ] (3) In equation (3), (dt e /dt) 0 is the total heating rate per electron in K/s. The other parameters are the electron number density n e, Boltzmann s constant k, the heated spot s center altitude z 0, the vertical extent of the heated region z, the heated 7451

3 centered average of F10.7), Ap = 4, and day of year = 130. The simulation includes geographic longitudes from 2 ı, so universal time and local time are the same. The plasma is modeled from 35 ı magnetic latitude, the peak altitude at the magnetic equator is 2400 km. The plasma parameters at 00:00 UT of the second day are used to initialize the 3-D model. The modeled F peak maximum electron density above Arecibo is about cm 3, which is above the critical frequency for heating at Arecibo s lower heater frequency of 5.1 MHz. Figure 1. The relative electron temperature (T h e /Ta e ) along the heated field line after 1 h of heating for simulations with different neutral wind values. The different cases are as follows: the baseline case (0 m/s meridional and neutral winds); a meridional wind of 60 m/s; a meridional wind of 60 m/s; andaneutralwindof60m/s. spot s center latitude 0, the latitudinal extent of the heated region, the heated spot s center longitude 0,andthe longitudinal extent of the heated region. Note that the heating source term is calculated at each gridpoint; thus, z is the altitude of the grid point, is the latitude, and is the longitude. We estimate the total heating rate per electron to be approximately 1000 K/s for the Arecibo transmitter running at its lower intended frequency of 5.1 MHz. This agrees with previous studies using the Sura heater [Milikh et al., 2010b] and is justified since we expect Arecibo to have an effective radiative power very similar to the Sura heater. [14] In SAMI3/ESF, the magnetic field is not tilted; thus, the magnetic latitude and geographic latitude are the same. For the heating latitude, 0, we choose the magnetic latitude of the Arecibo Observatory, 28 ı, to ensure that we have a realistic field line. The latitudinal extent of the heated region is set to =0.25 ı. The particular longitude is irrelevant since the magnetic field is not tilted, so we define a grid which covers 2 ı to 2 ı in longitude, and we set 0 to 0 ı longitude. The longitudinal extent is set to =0.25 ı. [15] The physical mechanism we study is anomalous absorption as described by Gurevich et al. [1996] which dominates in an altitude range having the vertical extent z between the wave reflection point and the upper hybrid height. The wave reflection point is the altitude, z 0, where the wave frequency is equal to the plasma frequency! pe.the upper hybrid height is defined by the altitude of the upper hybrid resonance frequency! =[! 2 pe (z 0 z)+ 2 e ]1/2 where e is the electron cyclotron frequency. Similarly to Wu et al. [2012], we select the altitude where the electron density is in the range cm 3 <n e < cm 3 (about 325 km here) as our heating altitude. This corresponds to the intended heater frequency of 5.1 MHz. Based on previous simulations, we set z to 10 km. Output from the SAMI2 model is used as the initial state of the SAMI3/ESF. We ran SAMI2 for 48 h using the following geophysical conditions: F10.7 = 100, F10.7A = 100 (F10.7A is the 81 day Results [16] The main drivers of longitudinal E B drifts are the neutral winds. The meridional wind primarily drives the electrons up and down the field lines, but a small component of it can create zonal E B drifts. The zonal wind, on the other hand, can create strong zonal E B drifts. To analyze the effect of E B drifts on the artificial ducts, we ran the SAMI3/ESF code using constant values for the neutral winds. The code is run from 0 to 4 LT; the HF heating occurs between 1 and 2 LT. We use the same geophysical conditions as the SAMI2 run, so F10.7=100, F10.7A = 100, Ap = 4, and day of year = 130. These scenario parameters were chosen based on the findings in Wu et al. [2012] that the heating of the conjugate region tends to be more effective during periods of relatively low electron densities, such as for low F10.7 values and at night, since less energy is lost to inelastic electron-neutral (N 2 ) collisions. Using these scenario parameters, we ran SAMI3-ESF with four different neutral wind configurations. In the first configuration, referred to in the rest of this paper as the baseline case, the meridional and zonal neutral winds were set to 0 m/s everywhere. The second and third configurations have nonzero values for the meridional wind, specifically 60 m/s, which is a moderately high value at Arecibo. The fourth and final configuration has a nonzero zonal wind of 60 m/s. [17] Figures 1 and 2 show the relative electron temperature (T h e /Ta e ) and the relative electron density (nh e /na e ) along Figure 2. The relative electron density (n h e /na e ) along the heated field line after 1 h of heating for simulations with different neutral wind values.

4 Figure 3. Contour plots of electron temperature (K) as a function of latitude and altitude for different times during HF heating in (left) the baseline case and (right) the case with 60 m/s zonal winds. The HF heater is on between 1 LT and 2 LT. the heated field line after 1 h of heating for each of the simulations. Here the superscripts h and a refer to heated and ambient conditions, respectively. In the baseline case, both the maximum temperature increase and the maximum density decrease are found at the heating latitude. The density along the field line is enhanced directly above and below the heated spot and also at the conjugate point above the F2 layer. The SAMI2/3 models assume that an ambipolar electric field along the geomagnetic field is supported by the electron pressure gradient and this electric field affects the ion motion. In this case, the intense electron heating in the bottomside F layer sets up a strong ambipolar electric field that accelerates ions away from the heated region: both up the field line to higher altitudes and down the field line to lower altitudes. [18] The meridional winds primarily drive electrons up (for equatorward winds) and down (for poleward) the field lines. In the cases where the meridional wind is set to 60 m/s, the temperature perturbation due to the HF heating is slightly suppressed. In these cases, the temperature increases by about 210% at the heating latitude, compared to the 240% increase seen in the baseline case. Since the maximum temperature perturbation is decreased, the temperature perturbations along the entire field line are similarly decreased. There are also differences in the electron densities; when the meridional wind is set to 60 m/s, the E B drifts drive electrons up the field line on the conjugate side, since the winds are blowing toward the equator in the Southern Hemisphere. This causes the area of increased density to expand to cover more of the field line on the conjugate side. For the case where the meridional wind is set to 60 m/s, the E B drifts drive electrons up the field line on the heating side, since in this hemisphere, the winds are equatorward. In this case, the relative density distribution on the heated side is affected, and the relative density distribution at the conjugate side is nearly identical to the baseline case. These changes are 7453

5 Figure 4. Contour plots of (logarithmic) electron density as a function of latitude and altitude at different times for (left) the baseline case and (right) the case with 60 m/s zonal winds. The HF heater is on between 1LTand2LT. relatively small and primarily due to the effect of meridional winds driving particles up and down the field line. Thus, relatively strong meridional winds do not create strong enough zonal E B drifts to significantly change the effects of HF heating. [19] The final case has the zonal wind set to a constant value of 60 m/s. For this case, there were large differences in the response to HF heating, particularly on the conjugate side. The zonal E B drifts are strong enough to suppress the effects of HF heating at the conjugate point; after 1 h of heating the temperature has only increased by about 1% (compared to 10% in the baseline case) and the densities by about 12% (compared to 40%). So, a relatively strong zonal wind creates zonal E B drifts that suppress the effects of HF heating at the conjugate side. To better understand why the HF heating effects are suppressed, we look more closely at the differences between the baseline case and the case with a 60 m/s zonal wind. [20] Figures 3 and 4 show the electron temperature and density at 0 ı longitude as a function of latitude and altitude for the baseline case (left) and the case with a 60 m/s zonal wind (right). After 3 min of heating, the electrons at the heated location in both cases show a large increase in temperature, about 240% above the ambient temperature of 1000 K. Temperature increases are localized at the heated spot and do not spread elsewhere along the field line this early in the heating. In the baseline case, the electron density at the heating location is slightly perturbed after 3 min of heating, but the nonzero zonal wind electron densities are unchanged. [21] As the HF heating continues, the temperature at the heating location remains constant, and the temperature of electrons further along the field line begins to increase. The temperature increase causes a pressure gradient, which drives electrons along the field line to the conjugate side, decreasing the electron density at the heated spot; this is 7454

6 Figure 5. The relative electron density (n h e /na e ) at the heating latitude and altitude as a function of longitude for different times during the heating. The zonal wind is set to a constant of 60 m/s. known as the snowplow effect. In the baseline case, the electron perturbation is driven to the apex of the field line after 15 min of heating and driven to the conjugate point after another 15 min, or 30 min total. After 30 min, the electron temperature at the conjugate point has increased and the electron density has decreased. On the heating side of the field line, the electron densities have been depleted. The nonzero zonal wind case is similar but the timescales are longer. The effects of heating take 20 min to reach the apex and another 20 min to reach the conjugate point. [22] In the baseline case, after an hour of continuous HF heating, the densities at the heated location have decreased by nearly 50% and the temperature remains 240% higher than ambient. The effects of HF heating have spread along the field line to the conjugate point, where the temperature has increased by 10% and the densities have increased by about 40%. Thus, the heated plasma has expanded along the flux tube, compressing the density ahead of it and plowing the plasma to the conjugate side, creating an artificial duct in the ionosphere. In the 60 m/s zonal wind case, after an hour of heating the temperature is constant at about 3400 K and the densities have only decreased by about 15% at the heating location. The electron temperature at the apex has increased by only about 1%, and the electron density has increased by about 10%, which is much smaller than in the baseline case. [23] Figure 5 shows the time progression of the relative densities at the heating latitude and altitude for the case with a 60 m/s zonal wind. Shortly after heating begins, the density hole is nearly symmetric about 0 ı longitude. As time progresses, however, the density hole becomes deeper and shifts in the positive longitude direction. For the negative longitudes, electrons are driven down the field line to the conjugate side by the HF heating and to positive longitudes from the zonal E B drifts, but are also replenished by E B drifts that bring electrons from more negative longitudes. For the positive longitudes, there is a depletion of electrons due to heating and the zonal E B drifts, but the density is not replenished as quickly since the electrons at more negative longitudes have been depleted. [24] This can also be seen in Figure 6, which shows contour plots of electron temperature and electron density at the Figure 6. Contour plots of electron temperature (K) and electron density (cm 3 ) as a function of longitude and altitude at the heating location at (top) onset of heating and (bottom) shortly after heating has ceased. Note that the scales are different between the top and bottom panels. 7455

7 heated location as a function of longitude and altitude at the onset (top panel) and cessation (bottom panel) of heating. At onset (0102 UT), the electron temperature increase and electron density decrease are nearly symmetric around 0 ı longitude. Shortly after the heater has been turned off (0202 UT), the density hole is centered on 0.5 ı longitude and the temperature enhancement has drifted to 0.25 ı longitude. Figure 7 shows the electron temperature along the heated field line as a function of longitude after 1 h of heating. The increased temperature does not propagate to the conjugate region because it is convected in longitude by the zonal drift which inhibits conduction along the field. [25] In the baseline case, once the heater is turned off, the electron temperature decreases quickly at the heating and conjugate locations; within 15 min, the temperature has returned to ambient levels. The temperatures at higher altitudes, however, take longer to return to ambient because the primary mechanism for cooling is electron-neutral interactions and there are fewer neutral particles at high altitudes. After an hour of cooling, the temperatures at high altitudes near the apex of the field line are still elevated about 600 K above the ambient temperature. These results are consistent with the findings of Wu et al. [2012] using SAMI2. As the temperature at the heating location returns to ambient values, the electrons, which were held on the field line to either side of the heating location by the pressure gradient, slowly refill the density hole. The densities do not completely return to their ambient values, however, since some have been lost to the conjugate side. The density enhancement on the conjugate side decreases slightly after heating ends, but without E B drifts or pressure gradients to move the electrons elsewhere, the density returns to ambient values very slowly. [26] In the 60 m/s zonal wind case, once the heater is turned off, the temperature decreases quickly along the field line, even at high altitude. Within 15 min, the temperature has returned to ambient levels. Unlike the baseline case where the only primary cooling mechanism at high altitudes was electron-neutral interactions, here electrons are driven to other field lines, thus cooling the temperature at the heated field line. The electron densities return to ambient values within 45 min after the HF heating ends, which is also much faster than the baseline case. Figure 7. Contour plot of the electron temperature (K) along the heated field line as a function of longitude. Electrons drift from the heated field line at longitude 0 ı to higher longitudes due to the 60 m/s zonal wind. ZAWDIE ET AL.: MODELING 3D DUCTS 7456 Figure 8. The relative electron temperature (T h e /Ta e ) along the heated field line after 1 h of continuous heating for different zonal wind speeds. Strong zonal winds result in large zonal E B drifts which dampen the heating effects on the conjugate side. [27] Figure 8 shows the relative electron temperature (T h e /Ta e ) along the heated field line after 1 h of continuous heating for different zonal wind speeds. In general, the strong zonal winds result in large zonal E B drifts, which we expect to dampen the heating effects on the conjugate side. As shown in Figure 8, the higher the zonal wind speed, the more suppressed the effects of HF heating will be. 4. Discussion [28] The purpose of this study was to determine the effect of zonal E B drifts on the formation of artificial ducts. We have found that zonal E B drifts do not prevent the creation of artificial ducts, but zonal E B drifts can partially suppress them. [29] We have ran simulations using multiple values for the neutral winds and found that the meridional neutral winds do not significantly change the effects of HF heating in the ionosphere. The E B drifts from the zonal neutral wind term, however, have a considerable effect on the HF heating. The heated electrons drift away from the heated field line, making the heating far less effective. The addition of a 60 m/s zonal wind reduces the temperature increase at the conjugate point by about 90% and reduces the increase in electron density by about 75%. The E B drifts from the zonal wind slows progress of the temperature and density perturbations along the field line; without zonal winds, the perturbations reach the conjugate point in 30 min, with zonal winds, the perturbations take 40 min to reach the conjugate point. [30] We also discovered that in the case with no E B drifts, the density perturbation at the heating and conjugate points took several hours to dissipate. In addition, the temperature at high altitudes remained elevated above the ambient temperature for several hours after heating. In the case of a zonal wind of 60 m/s, the density perturbations dissipate much more quickly, and the temperatures at high altitudes return to ambient levels within an hour.

8 [31] It should be noted that these studies were performed with neutral winds that are constant over space and time; however, the results do not change significantly with more realistic winds. To confirm this, we ran a set of simulations using HWM07 for the meridional and zonal neutral winds. Our findings were very close to the results using 60 m/s zonal winds, which we expected since horizontal neutral winds above 300 km are generally assumed to be constant. [32] Since zonal winds depress the effects of heating significantly at the conjugate point, we would not necessarily expect to see Arecibo heating experiments result in electron density or temperature perturbations at the conjugate point in Argentina. Although the direct density and temperature perturbations may not appear at the conjugate point, there may be other effects of heating that can be measured there, such as the change in the propagation of VLF waves [Starks, 2002]. On the other hand, there are times when the zonal wind is zero (or close to it) [Krall et al., 2009] and under these conditions there may be observable effects in the topside conjuate ionosphere. [33] There are still many improvements to be made in the model, some of which may enhance the effect of HF heating at the conjugate location. In particular, it has been shown that HF waves can refract upon density cavities in the ionosphere. As the density depletion convects away from the heating location due to neutral winds, the HF waves refract off the outer edges of the depletion, altering the propagation of the ray path[bernhardt et al., 1988]. To address this issue, one could use a ray-tracing code at each time step to determine the appropriate heating location. Another possible improvement to the heating model would be to account for the self-action effect as in the recent study by Milikh et al. [2012]. In their new model, the absorption efficiency of the pump wave and the heating altitude change at each time step to approximate the interference of the self-action effect. Finally, another improvement would be to include the effects of the background global neutral wind dynamo, which was excluded from this study. [34] Acknowledgments. This research has been supported by NRL Base Funds. The authors are grateful to K. Papadopoulos for helpful discussions. This work is from a dissertation to be submitted to the Graduate School, University of Maryland, by Katherine Zawdie in partial fulfillment of the requirements for the Ph.D. degree in Physics. References Bernhardt, P. A., L. M. Duncan, and C. A. Tepley (1988), Artificial airglow excited by high-power radio waves, Science, 242, Duncan, L. M., J. P. Sheerin, and R. A. Behnke (1988), Observations of ionospheric cavities generated by high-power radio waves, Phys. Rev. Lett., 61, González, S. A., M. J. Nicolls, M. P. Sulzer, and N. Aponte (2005), An energy balance study of the lower topside ionosphere using the Arecibo incoherent scatter radar and heating facilities, J. Geophys. Res., 110, A11303, doi: /2005ja Gurevich, A. V., A. V. Lukyanov, and K. P. Zybin (1996), Anomalous absorption of powerful radio waves on the striations developed during ionospheric modification, Phys. Lett. A, 211, , doi: / (95) Huba, J. D., G. Joyce, and J. A. Fedder (2000), Sami2 is Another Model of the Ionosphere (SAMI2): A new low-latitude ionosphere model, J. Geophys. Res., 105, 23,035 23,053. Huba, J. D., G. Joyce, and J. Krall (2008), Three-dimensional equatorial spread F modeling, Geophys. Res. Lett., 35, L10102, doi: /2008gl Krall, J., J. D. Huba, and C. R. Martinis (2009), Three-dimensional modeling of equatorial spread F airglow enhancements, Geophys. Res. Lett., 36, L10103, doi: /2009gl Milikh, G. M., E. Mishin, I. Galkin, A. Vartanyan, C. Roth, and B. W. Reinisch (2010a), Ion outflows and artificial ducts in the topside ionosphere at HAARP, Geophys. Res. Lett., 27, L18102, doi: / 2010GL Milikh, G. M., A. G. Demekhov, K. Papadopoulos, A. Vartanyan, J. D. Huba, and G. Joyce (2010b), Model for artificial ionospheric duct formation due to HF heating, Geophys. Res. Lett., 37, L07803, doi: /2010gl Milikh, G. M., A. Demekhov, A. Vartanyan, E. V. Mishin, and J. Huba (2012), A new model for formation of artificial ducts due to ionospheric HF-heating, Geophys. Res. Lett., 39, L10102, doi: / 2012GL Newman, A. L., H. C. Carlson Jr., G. P. Mantas, and F. T. Djuth (1988), Thermal response of the F-region ionosphere for conditions of large HF-induced electron-temperature enhancements, Geophys. Res. Lett., 15, Perrine, R. P., G. M. Milikh, K. Papadopoulos, J. D. Huba, G. Joyce, M. Swisdak, and Y. Dimant (2006), An interhemispheric model of artificial ionospheric ducts, Radio Sci., 41, RS4002, doi: /2005rs Scherliess, L., and B. G. Fejer (1999), Radar and satellite global equatorial F region vertical drift model, J. Geophys. Res., 104, Shim, J. S., et al. (2011), CEDAR Electrodynamics Thermosphere Ionosphere (ETI) Challenge for systematic assessment of ionosphere/thermosphere models: NmF2, hmf2, and vertical drift using ground-based observations, Space Weather, 9, S12003, doi: / 2011SW Shim, J. S., et al. (2012), CEDAR Electrodynamics Thermosphere Ionosphere (ETI) Challenge for systematic assessment of ionosphere/thermosphere models: Electron density, neutral density, NmF2, and hmf2 using space based observations, Space Weather, 10, S10004, doi: /2012sw Starks, M. J. (2002), Effects of HF heater-produced ionospheric depletions on the ducting of VLF transmissions: A ray tracing study, J. Geophys. Res., 107(A11), 1336, doi: /2001ja Swisdak, M., J. D. Huba, G. Joyce, and C.-S. Huang (2006), Simulation study of a positive ionospheric storm phase observed at Millstone Hill, Geophys. Res. Lett., 33, L02104, doi: /2005gl Wu, T.-W., J. D. Huba, G. Joyce, and P. A. Bernhardt (2012), Modeling arecibo conjugate heating effects with SAMI2, Geophys. Res. Lett., 39, L07103, doi: /2012gl