SEARCH

SEARCH BY CITATION

Keywords:

  • dip equator;
  • ionosphere;
  • sunrise

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method of Analysis
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[1] This paper shows that, contrary to previous explanations, the apparent undulating motion of the equatorial F region peak at sunrise is produced by photochemistry rather than dynamics. Our study is based on an investigation of the behavior of the early morning ionosphere observed by a Digital Ionosonde at Trivandrum, India. The phenomenon is rooted in the production of new plasma at the upper F region altitudes soon after sunrise. As the peak photoproduction rate moves down in altitude and increases in magnitude the newly formed plasma follows a similar trend. Once the density becomes large enough to be detected by an ionosonde, a jump is observed in the F region peak altitude. The jump is followed by a quick downward motion of the increasingly strong F peak. Chemistry causes the downward motion of the F peak to end near 250 km. Electrodynamics is not responsible for the sunrise undulation, but plays an indirect role in the detection of the sunrise effect by simultaneously lowering during the night the peak height and decreasing the density. When detectable, the remnant plasma introduces a lower peak height that facilitates the observation of the initial increase in peak height, while the lower background density allows the relatively small initial density increase from photoionization to be observed.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method of Analysis
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[2] The equatorial region is rich in ionospheric phenomena observed exclusively at low latitudes. A primary example is the steep climb in the peak height of the F-region during local sunset hours at the dip equator [Balsley, 1969]. Studies based on incoherent scatter radar observations have shown the PRE phenomenon to be clearly related to a reversal in the sign of the zonal electric field [Woodman, 1970; Fejer, 1981; Scherliess and Fejer, 1999] and numerous models have been developed to understand low latitude electric field structures [Heelis et al., 1974; Richmond et al., 1976; Scherliess and Fejer, 1999] documenting, in the process, the origin of the PRE. Competing explanations for the exact physical origin of the enhancement in the electric field prior to its reversal have been provided by Rishbeth [1971], Farley et al. [1986] and Haerendel and Eccles [1992].

[3] A less documented but similar undulation in the altitude of the F region peak has also been reported shortly after sunrise. Given the similarities of the F peak motion with the evening PRE undulation and the fact that the zonal electric field changes sign around sunrise, it has been assumed that the sunrise undulation is also the result of an electrodynamic phenomenon [Woodman, 1970; Aggson et al., 1995; Nayar et al., 2009; Mathew et al., 2010]. However, a closer look suggests that things are not so simple: if the rapid F region uplift was due to an eastward electric field, there should have been a quick and strong reversal from the night time value. The subsequent downward motion would also indicate another reversal in the zonal field. The explanation would no longer imply just a pre-reversal enhancement, but rather, a double oscillation from a negative to strongly positive back to strongly negative zonal electric field. Since the phenomenon is a regular occurrence at least at equinox, this would imply that some unexplained and complicated electrodynamics operates on a frequent basis.

[4] Here we have used observations from an advanced Digisonde system located at the dip-equator to perform a more detailed study of the F region undulation at sunrise. We have focused on data obtained in March 2010 at a time when the sunrise undulation in the F region was particularly clear and repeatable. We show that chemistry explains not only the sudden increase in the F region peak altitude after sunrise, but also a previously unnoticed F peak density increase during the undulation. The main role played by the electrodynamics is to remove, prior to sunrise, F region plasma produced on the previous day.

2. Data and Method of Analysis

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method of Analysis
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[5] F region electron density information was retrieved from a DPS-4D digital ionosonde (Digisonde) installed at Trivandrum (8.47°N, 76.92°E, 0.17°S dip-latitude). The Digisonde sweeps frequencies in the range 1–30 MHz and provides reliable estimates for the electron density below the F region peak through the ‘true height inversion algorithm’ embedded in the SAO Explorer software package [Reinisch et al., 2009].

[6] We compared the Digisonde findings with calculations from a time-dependent one-dimensional numerical ionospheric model that solved the ions and electron continuity equations. The model was limited to a computation ofO+, NO+, N2+ and O2+densities based on the photo-ionization ofO, O2, and N2, the conversion of O+ to molecular ions via charge exchange reactions, and the subsequent dissociative recombination reactions of molecular ions [St.-Maurice and Torr, 1978; Schunk and Nagy, 2000]. The neutral densities and temperature were taken from the MSIS model (http://omniweb.gsfc.nasa.gov/vitmo/msis_vitmo.html) while the electron temperature came from the IRI model (http://ccmc.gsfc.nasa.gov/modelweb/models/iri_vitmo.php). The EUV flux was obtained from the EUVAC model [Richards et al., 1994] and the relevant cross-sections and resulting photo ionization and photo absorption rates were taken from the references given inSchunk and Nagy [2000].

[7] For the situation at hand, the model was used to establish the basic chemistry for the sunrise effect and was not meant to describe all chemical processes in detail. Thus, the effect of metastable oxygen ions on the chemistry was not included. Between 250 and 350 km in particular, neglecting the effects introduced by metastable O+(2P) could introduce as much as a 50% overestimation in the net ion density, owing to the quenching of O+(2P) by N2 to create N2+ and molecular ions that quickly recombine [Richards, 2011]. Photoelectrons are an important source of ionization, particularly at large solar zenith angles where the longer wavelength EUV fluxes are severely attenuated and ionization is mainly due to photons with wavelengths below 40 nm. For 95 < SZA < 103 approximately 50% of the total photoionization rate may come from photoelectrons. For SZA < 95, the photoelectron contribution reduces to 35% of the total (P. Richards, private communication, 2012). We simply considered here the ratios given by Richards and Torr [1988], which are only strictly valid when the solar zenith angle (SZA) is 90° or less. We also neglected photoelectron transport. Near the equator and around equinox, photoelectron transport should be unimportant for the altitudes considered here. For the large zenith angles needed for sunrise effects we used the algorithm provided by Smith and Smith [1972]. The model also incorporates the effects of vertical electrodynamic drifts resulting from zonal electric fields [Choudhary et al., 2011]. It does not, however, include diffusion: with the altitudes of interest well below 400 km, where the magnetic field lines are still very horizontal, the time scale for diffusion to operate was assumed to be long compared to the time scales of interest.

3. Results and Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method of Analysis
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[8] Figure 1presents the temporal evolution of the F-region peak height (hmF2) and virtual height (hF) derived from the Trivandrum Digisonde observations during March 2010. We note that the hmF2determination from the Digisonde has been shown to match Incoherent Scatter Radar-derived values with a median difference of 5 km within a 13 km standard uncertainty [Scali et al., 1997]. The plots represent the monthly averaged value in 5 minutes intervals, along with the standard deviations for each 5 min bin, both for the sunset (1800 to 2200 IST) and sunrise (0500 to 0700 IST) periods (IST = UT + 05:30). For the sunset case, the average F-region base height (hF) is close to 240 km around 1800 IST. From that time, it rises swiftly to exceed 280 km by 2000 IST, a 40 km uplift in approximately two hours. Such a rise in the evening is quite normal at the dip-equator over the Indian zone, particularly near equinox [Rastogi, 1980]. The peak height of the F-region (hmF2) during the same period displays a similar trend, ascending from about 330 km at 1800 IST to about 380 km by 2000 IST for a 50 km uplift in two hours. Thereafter, both the hmF2 and hFdecrease monotonically only to regain their normal altitudes by 2200 IST. As discussed in the introduction, the evening F region undulation is well-known, particularly in March, and describes the PRE.

image

Figure 1. Temporal variations in (top) monthly averaged hF and (bottom) hmF2 and of their standard deviations, observed (left) between 0500 and 0700 IST and (middle) between 1800 and 2200 IST during March of 2010. (right) hmF2 histogram derived from the Digisonde measurements for the month of March 2010 during the time interval between 04:00 LT and 10:00 LT. Near vertical pink lines: terminator position as a function of time for the beginning and the end of the month. Thick black line: average hmF2 for the month, as already shown by the red line in the bottom left panel.

Download figure to PowerPoint

[9] A similar trend of an initial elevation followed by a depression in hmF2 and hF was also found around sunrise (Figure 1, left). The hFmoves up on average from about 218 km to about 242 km and then slides down until 0700 IST. This represents a 25 km upswing in the base height of the F-region after 0550 IST, which compares to the sunset rate of increase. More precisely, whilehFascends by 40 km in approximately two hours during the post-sunset period, the gain during pre-sunrise is of the order of 25 km, but in only half the time. The same can be said of the F peak itself.

[10] Figure 1(right) clarifies what lies behind the sunrise F region oscillation, through the introduction of a 2-D histogram of thehmF2 generated for the month of March 2010 during the time interval 04:00 to 10:00 IST. To generate the histogram, the hmF2values for the 31 days of March 2010 were sorted in 2 km altitude bins and 15 min time intervals. The colors and their associated contours describe the number of counts recorded in particular time-altitude bins. The changes are so fast at sunrise that small day-to-day variations in the jump onset time show up as a standard deviation in ofFigure 1(left): through its 2 km altitude binning, the 2-D histogram shows that a rapid rate of descent is behind the observed standard deviations. Note that the spread in values seen in the histogram around a given 15 min bin should not be confused with the number of events. For instance individual bins could register only 6 events even though the fast descending trend itself was observed on a daily basis. It should be clear fromFigure 1that there is a strong trend around 0630 IST for a descending F peak that becomes detectable around 300 km altitude. Within a half hour the peak comes down to 250 km only to move back up from 0730 IST onward. The 2-D histogram shows that the apparent jump in the F-region altitude is a result of two separate factors: an erratic remnant from the night before, which is typically found below 260 km altitude, and a fast descending component that starts at an altitude far in excess of the ‘remnants’ mean altitude.

[11] In Figure 2, our model calculations are compared with observations for 3 distinct cases, namely: (a) a case (March 7) for which there were almost continuous observations during the morning transition; (b) a case (March 23) for which there were observations before and after local sunrise, but no reflections around local sunrise; and (c) a case (March 25) for which there were no echoes recorded from the ionosphere before sunrise. We note that the absence of data did not mean that there was no plasma, but just that the foF2 was too low to scale the ionograms, indicating that the plasma density was less than 1.25 × 1010 m−3. For these calculations, we initialized the model with zero electron densities so as to only track down pure sunrise effects.

image

Figure 2. (left) A comparison between the model-derived and the Digisonde-inferredhmF2 for (top) March 07, (middle) March 23 and (bottom) March 25, 2010. The blank portion in the observations represents the time during which no Digisonde echoes were received while the slanted green lines indicate the ionospheric shadow height. (right) Comparison of the model and observed foF2 for the three days. The blanketed green region is to emphasize the lowest density levels that could possibly be recorded with the Digisonde.

Download figure to PowerPoint

[12] The left panels of Figure 2 compare the hmF2observed between 0500 and 0730 IST with the model-derived values for the three days that were selected. The solar zenith angle associated with the posted local time, is given on the top x-axis. The blue dashed lines show the model values while points connected by a red solid line represent Digisonde-derivedhmF2values. The solid green curve represents the ‘shadow height’ of the sun (by ‘shadow height’ we mean here the altitude below which the ionosphere was completely shielded from solar radiation). As is well-known, the calculations show that, as soon as the sun rises, photo-ionization starts to generate electrons, starting with higher altitudes. The resulting plasma densities, however, are too low at first to be recorded by the Digisonde. The panels on the right show that with an absolute lower limit of 1.25 × 1010 m−3, it should not have been possible to see the newly formed plasma above 330 km altitude at that time of year.

[13] As time progresses, the solar zenith angle decreases and solar radiation penetrates further down to the lower part of the ionosphere, where the photoionization rate is larger, thereby shifting the ionization peak to a lower altitude. An apparent downward motion of the F peak follows, even though there is no actual plasma movement involved. The descending motion continues until the F peak reaches approximately 250 km, by 0700 IST. Interestingly, when the Digisonde starts to record the reflections from the newly produced plasma shortly after 0600 IST (on March 23, and 25), the model-derivedhmF2 closely matches the observed hmF2.

[14] In the right panels of Figure 2, we show the modeled F peak plasma density (NmF2) and compare it with the Digisonde-observedNmF2. To facilitate the comparison, we have converted NmF2 to the foF2, which is the Digisonde-derived parameter, using the relation (NmF2 = (foF2)2/80.3) where the density is in m−3. We have hatched the area where the foF2 is less than 1.25 MHz for easy guidance. The modeled density was less than 1.25 MHz until 0600 IST and the Digisonde correspondingly produced no data during this time interval on March 23, and 25, 2010. Interestingly, during the time when hmF2 presented an abrupt increase, the observed electron density had an almost similar magnitude on all the three days despite of the fact that background conditions on those days were quite disparate. This shows that the old densities were not involved in the plasma rebuilding, as expected from a purely chemical effect involving the rebuilding of plasma at sunrise.

[15] Figure 2also clearly illustrates that the model-derived and the observedhmF2 values descend in unison, while the density keeps increasing steadily. The descent, however, becomes increasingly slow and finally comes to a stop after the loss rates have increased sufficiently. The point is that, by providing an adequate description of the observations, the model confirms that the sunrise effect can be explained entirely by chemistry, meaning that there is no need to invoke complicated electrodynamics. Not only does the chemical model produce densities that are only somewhat underestimated (as expected from the exclusion of photoelectrons as a production term and the fact that the model was initialized with zero density) but it reproduces very well the observed F peak motion after sunset and explains the gain in the F peak density in a straightforward manner. It should be noted that, if we include the dates for which there was no density recorded until sunrise, the fast jump in the F peak was seen on 29 of 31 days during March 2010. The two missing dates simply produced no density data at all.

[16] As further evidence against an electrodynamic mechanism, we can consider in more detail the March 7 observations, when reflections were recorded by Digisonde almost continuously during period of our interest. Till 0608 IST, the NmF2 decreased monotonically and so did the hmF2. At around 0611 IST, when NmF2 started to increase, there was a sudden jump in hmF2which appeared to move the F-region peak from 245 km to 310 km. This amounted to an increase of about 65 km in 3 minutes. In order to introduce such a huge jump inhmF2 through electrodynamics, the required vertical drift would have had to be of the order of 360 m/s, which is much larger than expected from zonal electric fields. This result is similar to the decrease in hmF2after sunrise quite often seen at mid-latitudes, where it is generally associated with the rapid production of ionization in the lower F-region [Rishbeth and Garriott, 1969]. In the latter case, however, the process gets abated by the northward meridional thermospheric winds during morning which play an important role [Zhang et al., 1999]. In any case, in the absence of photoionization, electric fields could not produce an increase in electron density at the dip equator.

[17] One remaining issue is the residual ions from the prior night and how they impact the Digisonde measurements. To study that issue, we present in Figure 3the results of model runs obtained with and without a remnant density at 0415 IST, and we compare them with the electron density profiles estimated from the Digisonde measurements. The Digisonde-derived electron density profiles are shown by the blue dashed lines inFigure 3. In one case (black lines) we initialized the model with negligible electron densities, as discussed above. In another case (red lines) we used an initial plasma density profile that matched Digisonde-derived parameters obtained late at night, namely thehmF2 and NmF2obtained at 0415 IST on March 23, 2010. A 20-km wide Gaussian distribution was assumed for the initial electron density in this second case. A completely drift-free case was assumed for the negligible initial electron density case, while a negative vertical drift of 10 m/s had to be introduced in the latter case to push the remnant plasma down to the lower altitudes and make its density decrease so as to match both the position and value of the F peak inferred from the Digisonde data until their disappearance below the detection threshold after 0530 IST.

image

Figure 3. Comparison of model derived electron density profiles with densities derived from Digisonde measurements. Black lines: cases for which the model was initiated with a negligible electron density. Red lines: cases for which the model was initiated with a electron density profile matching the Digisonde-derived hmF2 and NmF2 at 0415 IST.

Download figure to PowerPoint

[18] Figure 3 illustrates the potential role that remnant densities could have in masking the sunrise effect if the remnant plasma is not pushed down. However, a 10 m/s downward drift on March 23 (and similar drifts on other days) eliminated the remnant plasma from the previous day and in turn caused a one order of magnitude depletion in the NmF2 magnitude between 0422 IST and 0530 IST. Our computations also confirm that after 0530 IST, the NmF2would have become too small to be detectable in the presence of a continuing downward plasma drift. The situation reversed itself after sunrise when the fresh production of ions due to photo-ionization made the plasma density increase at higher altitudes. There was no “rebirth” of the low altitude plasma. Rather, a secondary peak in the electron density appeared at higher altitudes, along with the invisible (to the Digisonde) remnant that was already present lower down. As time progressed, a competition built up between the two peaks, but by 0600 IST, the upper peak took over. Finally, when the upper density peak became large enough for the Digisonde to detect it, it was indeed observed. After 0600 IST, when the Digisonde started to observe reflections, the model-derivedhmF2 and NmF2 were a close match to the Digisonde observations.

4. Conclusion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method of Analysis
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[19] The behavior of the ionosphere shortly after sunrise has been characterized using high time resolution observations from a Digital Ionosonde (Digisonde) at Trivandrum, India. For March 2010, there was sudden F peak upward motion on every single day that data were obtained. This motion was quickly followed by downward motion that was almost equally fast. Our study makes it clear that the hmF2 jump from 250 km to 300 km, 30 min after the upper F region sunrise, and its subsequent downward excursion after 0630 IST is entirely explicable by chemical effects associated with the production of new plasma after sunrise. Our chemical calculations not only predict the right altitudes at the right time for the F peak, but also, very near the right densities. The up and down oscillations that follow sunrise are therefore not due to an oscillating vertical drift. Having stated this, we have also found that the sunrise effect, as described here, should only be visible to an ionosonde if the plasma from the previous day has been pushed down fast enough and low enough to be destroyed by dissociative recombination so as to allow newly created plasma from higher altitudes to be visible. This requires a strong enough westward electric field to be present before sunrise, which was the case on basically every day during March 2010. Given that the zonal electric field changes sign after sunrise, then, in a sense, the sunrise effect could be said to be associated with a zonal electric field oscillation, but not in the way that had been proposed before.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method of Analysis
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References

[20] We thank R. S. Simi for carefully scaling the ionograms. Insightful discussions with D. C. Thompson and V. Eccles and comments/suggestions by the 2 referees are likewise gratefully acknowledged. This work was completed in part when K.M.A. visited the Univ. of Saskatchewan on a fellowship. K.M.A. is also supported by an ISRO Research Fellowship. The work was supported in part by the Canadian NSERC.

[21] The Editor thanks Phil Richards and an anonymous reviewer for their assistance in evaluating this paper.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method of Analysis
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References