Empirical models of storm-time equatorial zonal electric fields

Transcription

1 Utah State University From the SelectedWorks of Bela G. Fejer November, 1997 Empirical models of storm-time equatorial zonal electric fields Bela G. Fejer, Utah State University L. Scherliess Available at:

2 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 102, NO. All, PAGES 24,047-24,056, NOVEMBER 1, 1997 Empirical models of storm time equatorial zonal electric fields Bela G. Fejer and Ludger Scherliess Center for Atmospheric and Space Sciences, Utah State University, Logan Abstract. Ionospheric plasma drifts often show highly complex and variable signatures during geomagnetically active periods due to the effects of different disturbance processes. We describe initially a methodology for the study of storm time dependent ionospheric electric fields. We present empirical models of equatorial disturbance zonal electric fields obtained using extensive F region vertical plasma drift measurements from the Jicamarca Observatory and auroral electrojet indices. These models determine the plasma drift perturbations due to the combined effects of short-lived prompt penetration and longer lasting disturbance dynamo electric fields. We show that the prompt penetration drifts obtained from a high time resolution empirical model are in excellent agreement with results from the Rice Convection Model for comparable changes in the polar cap potential drop. We also present several case studies comparing observations with results obtained by adding model disturbance drifts and season and solar cycle dependent average quiet time drift patterns. When the disturbance drifts are largely due to changes in magnetospheric convection and to disturbance dynamo effects, the measured and modeled drift velocities are generally in good agreement. However, our results indicate that the equatorial disturbancelectric field pattern can be strongly affected by variations in the shielding efficiency, and in the high-latitude potential and energy deposition patterns which are not accounted for in the model. These case studies and earlier results also suggest the possible importance of additional sources of plasmaspheric disturbance electric fields.!. Introduction The ionospheric electric field plays a dominant role on the dynamics, distribution of ionization, and generation of plasma waves in the low-latitude and equatorial thermosphere. In the last 2 decades, significant progress has been achieved in the study of equatorial electrodynamic plasma drifts and electric fields. Radar measurements have determined the seasonal and Copyfight 1997 by the American Geophysical Union. Paper number 97JA /97/97 J A The response of equatorial electric fields and currents to geomagnetic disturbances has been examined in detail in a large number of case-by-case and statistical studies, but with conflicting results. On one hand, there is extensive evidence for the occurrence of large electric field and current perturbations associated with changes in the interplanetary magnetic field and in the high-latitude current system [e.g., Gonzales et al., 1979; Fejer, 1986, 1991; Fejer et al., 1990; Sastri et al., 1992; Forbes et al., 1995; Kikuchi et al., 1996], consistent with results from theoretical and numerical models [e.g., Senior and Blanc, 1984; Spiro et al., 1988]. On the other hand, average diurnal patterns of equatorial F region vertical drifts (driven by the zonal electric fields) during solar cycle dependent average quiet time patterns, theoretical studies identified the electric field generation mechanisms and, more recently, largely improved theoretical electrodynamic plasma drift models have been developed [e.g., Richmond, 1995; Fejer, 1997]. The progress in the understanding of equatorial electric fields and plasma drifts during magnetically quiet and disturbed conditions are essentially geomagnetically active periods has been slower even though identical, as illustrated in Figure 1. Furthermore, the equatorial magnetometer, radar, and ionosonde measurements have long electric fields do not always seem to respond to high-latitude suggested the storm time electrodynamicoupling between the disturbances, whereas, on other occasions, the observed high- and low-latitude ionosphere. The two basic mechanisms disturbances are apparently inconsistent with simultaneous for the generation of mid-latitude, low-latitude, and equatorial higher-latitude electric field and current perturbations. In the storm time electric field disturbances have been identified as last few years, it became clear that the basic reason for these the solar wind-magnetosphere dynamo [e.g., Senior and Blanc, conflicting results is the frequent simultaneous occurrence of 1984; Spiro et al., 1988; Zakharov et al., 1989], and the competing disturbance processes. ionospheric disturbance dynamo [Blanc and Richmond, 1980]. Recent experimental studies examined the local and storm However, the highly variable signatures of the currents and time variations of the disturbance electric fields and have plasma drifts during disturbed conditions have made the presented empirical models for their representation. The identification of prompt penetration and disturbance dynamo penetration of magnetospheric zonal electric fields in the electric fields difficult and, until recently, still the subject of nightside plasmasphere following IMF B z turnings was debate. 24,047 investigated by Deminov and Deminova [1992]. These electric fields, which under stationary conditions have timescales of 1-2 hours, were modeled with a cosine local time dependence. Fejer and Scherliess [1995] have shown that equatorial zonal electric field disturbances driven by magnetospheric and ionospheric disturbance dynamos can be

3 24,048 FEJER AND SCHERLIESS: STORM TIME EQUATORIAL ELECTRIC FIELDS Jicamarca 40 (I)adj=130 t,, i, I ' (2 21 ß o AE > Mar-Apr,j -, I,,,, I,,,, I,,,, Local Time Figure 1. Average F region vertical plasma drifts (positive upward) for two ranges of AE indices. These average drifts correspond to a solar flux index = 130 units. separated provided that their storm time evolution is taken into account. The plasma drift perturbations from the combined effects of these mechanisms which have different storm and local time dependence can explain most of the frequently observed apparent inconsistencies between highand low-latitude disturbances. Mikhailov et al. [1996] studied equatorial vertical drifts inferred by solving a simplified electron continuity equation for given F region peak electron densities and heights. The response of the derived drifts to high-latitude disturbances was generally consistent with results from radar observations. In this work, we describe initially a methodology for empirical studies of the storm time dependence of ionospheric electric fields, and present time-dependent equatorial zonal electric field models obtained from extensive incoherent scatter radar observations of equatorial F region vertical plasma drifts over Jicamarca. These analytical models, which have auroral electrojet (AE) indices as input parameters, describe the complex storm time dependent plasma drifts resulting from direct penetration and disturbance dynamo electric field effects. We use the models first to examine the nt resolution of km and with an integration time of about 5 min. The values used here represent averages between about 300 and 400 km where the signal-to-noise ratios are highest and the vertical drifts do not change much with altitude. The uncertainty on these measurements is about 1-2 m/s during the day and somewhat larger at night when the signal-to-noise ratios are much smaller, particularly near solar minimum. In the F region over Jicamarca, an upward drift velocity of 40 m/s corresponds to an eastward electric field of about 1 mv/m. We have chosen the auroral electrojet AE index as the highlatitude disturbance parameter. This index is based on magnetic field observations from 11 high-latitude stations from 1968 to 1973 and from 12 stations between 1974 and AE indices based on a larger number of high-latitude current measurements are not available for the entire 20-year period covered by our radar measurements. This parameter, which is available with a time resolution of minutes, has been empirically related to both the polar cap potential drop and the hemispheric power input by several authors [e.g., Ahn et al., 1983, 1992; Richmond et al., 1990; Lu et al., 1995]. We will use these relationships between AE and the polar cap potential and hemispheric power input to compare our empirical results to those from global convection and disturbance dynamo models. Our database consists of 4926 hours of vertical drift measurements and AE indices obtained from March 1968 through June We have not used data from 1976 through 1977, and after June 1988 since the corresponding auroral indices were not available. These data yielded 15,162 quarter hourly averaged drifts. In the following section, we describe the procedure for the derivation of our disturbance drift models. 3. Methodology and Model Properties 3.1. Data Preparation The equatorial plasma drifts vary with the phase of the solar cycle, season, and magnetic activity and also exhibit significant short time (periods from minutes to several days) variability even during magnetically quiet periods [e.g., Fejer, 1991 ]. Therefore these drifts can be expressed as V(t,d,,AE) = Vq(t,d, ) + Vq(t,d, )+ Vd(t,d,,AE ) (1) where t is local time, d is the day of the year, and is the solar signatures of equatorial electric field perturbations associated decimetric index. The first two terms on the right-hand side with changes in the high-latitude convection. In a companion paper, $cherliess and Fejer [this issue] study in detail the characteristics of equatorial storm time dynamo zonal electric fields driven by enhanced energy deposition into the highlatitude ionosphere. Finally, we consider some case studies to denote the quiet time average and fluctuation values, respectively, and the third represents disturbance drifts driven by high-latitude processes. Since we are interested in the storm time (i.e., a time period after a given change in the highlatitude current system) dependence of the equatorial electric highlight the effects of prompt penetration and disturbance fields, we will examine the plasma drifts in response to a time dynamo electric fields and the importance of other processes series of AE indices. This index does not fully specify the and discuss the limitations of the models and possible future upgrades. state of the high-latitude ionospheric currents, and therefore our model will average the effects of a number of high-latitude processes such as substorm and convection enhancements, and 2. Data of energy deposition in different local time sectors. The seasonal and solar cycle variations of the Jicamarca The equatorial vertical plasma drifts used in this study were quiet time average drifts are well understood [Fejer, 1991, obtained from incoherent scatter radar measurements at the 1997]. For the present study, we have developed a solar cycle Jicamarca Radio Observatory (12.0 ø S, 76.9 ø W; magnetic dip 2øN), in Peru. The experimental procedure was described by Woodman [1970]. In general, these measurements are made over an altitudinal range of about 250 to 600 km with a height dependent 15-day sliding bimonthly vertical drift model for extended quiet time periods (defined as corresponding to periods when the hourly AE indices had been continuously smaller than 300 nt for at least 28 hours) as a function of the

4 FEJER AND SCHERLIESS: STORM TIME EQUATORIAL ELECTRIC FIELDS 24,049 decimetric solar flux (L. Scherliess, unpublished results). The average AE index for these extended quiet time drifts was about 130 nt. The low, moderate, and high solar flux average seasonal drift patterns for extended quiet conditions are quite model described by Scherliess and Fejer [this issue]. This model can be readily adapted for use with other hourly averaged disturbance input parameters. The higher time resolution model, derived from 15-min-averaged data, is used for similar to those presented by Fejer et al. [ 1991]. The equatorial comparisons of model results with Jicamarca data. drifts show considerable day-to-day and shorter term perturbations produced by variability in the E and F region dynamos associated, for example, with changes in tidal, gravity, and planetary wave forcing, and in the field-integrated conductivities. It is possible that high-latitude convection driven equatorial electric fields (especially disturbance dynamo electric fields)are also season and/or solar cycle dependent, but these effects will not be investigated here. In our analysis, we initially subtract from each quarter Fejer and Scherliess [1995] showed that the zonal electric field disturbances can be well reproduced in terms of prompt penetration and disturbance dynamo effects. In this study, the direct penetration vertical drift (zonal electric field) at a given time t was modeled in terms of changes in the AE indices at times t-30 min and t- 90 min. They also showed that the time constant for the decay of direct penetration electric fields following a step function change in AE (i.e., in the polar cap potential drop) is about 1 hour, in agreement with results from hourly or hourly averaged drift measurement the corresponding theoretical models [e.g., Senior and Blanc, 1984; Spiro et al., season and solar flux dependent quiet time average drift. This 1988]. In our companion paper, Scherliess and Fejer [this procedure is illustrated in Figure 2 for the period of September 23-24, 1986, where the top, middle, and bottom panels show the AE indices, the measured and average quiet time drifts, and the perturbation drifts, respectively. The perturbation drifts in Figure 2 are due to both magnetic activity as well as to quiet time variability; that is, they correspond to the last two terms in the right-hand side of (1). Then for given high-latitude current conditions (specified by a time series of AE indices), the disturbance drifts Vd(t,d,,AE) can be determined, provided that the database is large enough for the quiet time perturbations to cancel out. In the next step, the disturbance issue] show that the effects of disturbance dynamo electric fields on the equatorial vertical drifts can be described in terms of (positive) average values of AE d = AE -130 nt over the preceding periods 1-6, 7-12, and hours. Therefore the total disturbance vertical plasma drifts were expressed as 9 Vd (t, AE ) = {ai, l AAE(t - 30 min) + ai, 2 AAE(t - 90 min) i=1 +ai, 3AEd(1-6 hours)+ai, 4OAEd(7-12 hours) (2) +ai,5 [AEd(22-28 hours)- 200 nt]}ni, 4 (t) drifts are associated with a time series of AE indices covering Here the first two terms under the summation account for the periods of the preceding 28 hours. The final step consists in modeling these perturbation drifts in terms of the prompt penetration and disturbance dynamo components. prompt penetration drifts at time t and are related to changes i n the AE indices with average time delays of 30 and 90 min, where 3.2. Model Representation In this section we describe our empirical models. The first one, determined from hourly averaged AE indices and drift measurements, combines the prompt penetration component by Fejer and Scherliess [1995] with the disturbance dynamo U.T. 1' '... '... '... '... ' 'l - ^ September 23-24, /\, -I 501 ø o ø, I, i, I i,, I,, I,,, I, 08 1'2 1'6 20 2'4 0'4 LT Figure 2. AE indices, measured and average vertical drifts, and perturbation drifts obtained by subtracting the average from the observed velocities shown in the center panel. AAE(t-30min) = AE(t)-AE(t-1 hour) AAE(t-90min)= AE(t-1 hour)-ae(t-2 hours) The last three terms are used to account for ionospheric disturbance dynamo effects. Scherliess and Fejer [this issue] showed that the efficiency of disturbance dynamo processes with time delays of 7-12 and hours are controlled by shorter-term average activity indices AE(1-6 hours) and AE(1-12 hours), respectively. These short-long term coupling can be described by I 0 AE(1-6hours) (3) AE(1-6hours) < 200nT (z = nT < AE(1-6hours) < 300nT (4) AE(1-6hours > 300nT and [3 = exp [-AE(1-12 hours)/90 nt] for AE(1-12 hours) > 70 nt, and [3 = 0.46 for AE <70 nt. The last term in the righthand side of (2) accounts for long-term disturbance dynamo drifts for AE d (22-28 hours) > 200 nt. For smaller values of AE d, this term is set to zero. The local time dependence of the vertical drifts is described by nine normalized cubic-b splines of order 4, Ni.4(t ). Eight nodes were placed in equally separated local time intervals at 0,3, hours, and one additional node was placed at 4.5 local time to account for rapid changes near dawn. The B-spline functions used in this study are shown in Figure 3. These functions are nonvanishing over limited intervals and add up to one at all local time [e.g., DeBoor, 1978]. The 45 coefficients ai, j of the model were constrained to make the fit periodic in '24 hours and were determined by minimizing the mean square error defined by

5 24,050 FEJER AND SCHERLIESS: STORM TIM EQUATORIAL ELECTRIC FIELDS [ [ [ [ I ' ' [ [ [ [ [ ' ' I ' ' [ [ I ' [ [ N9,4 N8,4 N1 4 N9 4 N3'4 N4'4 N5'4 N6'4 N7' NL N2, N3, N4, Ns, N6, N7, Local Time Ns, N9A Figure 3. Normalized cubic-b splines used for the local time representation of the prompt penetration and disturbance dynamo vertical drifts. were again determined by minimizing the mean squarerror of the perturbation drifts but now using 15,162 quarter-hourly m=4926 averagedrift values. The resulting coefficients for the (5) m! dm ( t,ae)- Vd ( t,ae)12, disturbance dynamo effects are essentially identical to those obtained using the hourly averaged values. The coefficients of the direct penetration terms in (7) are given in Table 2. where V dm denotes the hourly measured perturbation drifts. These coefficients, given in Table 1, can be used to estimate For time delays shorter than 75 min, we developed an hourly disturbance drifts based on AE indices over periods of interpolation scheme based on (6) to account for changes in the AE indices at 15-min intervals, whereas for longer time 28 hours. This model provides a significantly more complete description of the storm time equatorial vertical perturbation delays, we assumed an exponential decay associated with the drifts than given by Fejer and Scherliess [1995]. These AAE(t-75 min) response. A least squares fit analysis to the disturbance drifts superposed on the corresponding average direct penetration responses at storm times of 30 and 75 min at all local times showed that the average exponential decay time quietime patterns, will provide our low time resolution model constant is about 70 min, consistent with theoretical results for equatorial vertical drifts. Our higher time resolution model uses quarter-hourly [e.g., Spiro et al., 1988; Fejer et al., 1990], although this time constant varies somewhat with local and storm time. This averaged AE indices for the representation of direct penetration velocity perturbations. Here the direct penetration effects at a exponential dependence provides a reasonably realistic longer given time t are related to changes in the AE indices at average term decay for the prompt penetration drifts but, as will be discussed later, it neglects the small disturbance drifts due to time delays of 7.5, 30, and 75 min obtained from the leakage of high-latitudelectric fields to the equator under AAE(t- 7.5 min) = AE(t)- AE(t- 15min) stationary conditions. AAE(t - 30min) = AE(t - 15 min) - AE(t - 45 min) (6) AAE(t- 75min) = AE(t- 45min)- AE(t- 105rain) 3.3. Model Properties and Limitations For the longer timescale disturbance dynamo drifts, we use the same representation given above. Then, the coefficients for our high time resolution model defined by 9 Vd(t, AE ) =,{ai,oaae(t min) + ai, laae(t - 30 min) i=1 +ai, 2AAE(t- 75 min) + ai,3aed(1-6 hours) (7) +ai,4txaed(7-12 hours) + ai, sis[aed(22-28 hours) -200 nt]}ni,4 (t) Table 1. Coefficients for the Vertical Disturbance Drift Model (1 Hour Time Resolution) for (2) B Spline ai, 1 ai, 2 ai. 3 ai. 4 ai.5, m/s nt N1, N2, N3, N4, Ns, N6, N7, Ns, N9, Table 2. Prompt Penetration Model Coefficients (15-min Time Resolution) for (7) B Spline ai, o ai, 1 ai, 2, m/s nt We have seen that the basic feature of our empirical models is the inclusion of storm time effects which allows for the separation of direct penetration and disturbance dynamo perturbations. The use of the AE index as the model input parameter allows us to compare our empirical results with those from global convection and disturbance dynamo theoretical models driven by the polar cap potential drop and hemispheric energy input into the high-latitude ionosphere, respectively. Ahn et al. [1983, 1992] used a magnetometer inversion technique to show that the relationship of the polar cap potential and the AE index measured at 12 stations over a two day equinoctial period is approximately linear, although there is a large degree of scatter. The linear regression obtained was (kv) = AEi2(nT) with a correlation coefficient of 0.83 for AE up to 900 nt. This relationship is not realistic for very small potentials. Similaresults were obtained using the magnetometer inversion technique complemented b y incoherent scatter radar measurements [Richmond et al., 1990; Lu et al., 1995], and by comparing polar cap potentials derived from electric field measurements from the DE 2 satellite with simultaneous AE indices measured with ground-based magnetometers [Weimer et al., 1990a, b]. These studies also showed that there is no significant difference in the errors between linear and nonlinear fits. Several experimental

6 FEJER AND SCHERL[ESS: STORM TIME EQUATORIAL ELECTRIC FIELDS 24,051 studies have shown that the Joule heating production rate and the particle energy injection rate can be expressed as linear functions of the A U, AL, and AE indices [e.g., Ahn et al., the Blanc-Richmond disturbance dynamo model is presented by Scherliess and Fejer [this issue]. Finally, we illustrate the use of the model for case studies. 1983; Richmond et al., 1990; Lu et al., 1995] and also as a function of the Kp index. Ahnet al. [1983] derived a relationship between the global energy injection rate and the AE index given by U(MW) = 290 AEi2(nT), and a ratio of Joule heating and particle energy injection rate of about 4, 4. Prompt Penetration Zonal Electric Fields Figure 4 shows the prompt penetration vertical drift patterns obtained from our high time resolution (15 min) independent of magnetic activity. Somewhat larger Joule model at three storm times following a step function increase heating rates were obtained by Baumjohann and Kamide in the AE index by 400 nt over our quiet time value. The [1984], Richmond et al. [1990], and Lu et al. [1995], as a pattern at t o + 30 min, which corresponds to the initial time result of the use of different ionospheri conductivities. direct penetration response presented by Fejer and Scherliess The AE index has well-known limitations which should be [ 1995], can also be obtained from our model using hourly AE kept in mind when estimating the polar cap potential and indices. The results for storm time 7.5 min shows that the hemispheric power input and therefore the prompt penetration nighttime downward perturbation drifts are largest in the and disturbance dynamo electric fields. These parameters postmidnight sector with a maximum value of about 15 m/s cannot be accurately estimated from the AE index during small (i.e., a westward electric field of about 0.4 mv/m), whereas the potentials when AE estimates are not accurate due to poleward daytime upward drifts have typical values of about 5 m/s. Fejer motions of the auroral currents and during strongly disturbed and Scherliess [1995] showed that the perturbation drifts conditions due to both equatorward motions of the auroral oval associated with identical increases and decreases in the AE and to precipitation induced conductivity enhancements. For a given polar cap potential, the AE indices are higher by a factor of about 1.5 in the summer hemisphere than in the winter hemisphere due to the higher ionospheric conductivity [e.g., BerthelJer, 1976; Weimer et al., 1990a]. As a. result, there is a larger increase of the polar cap potential with AE during the northern hemisphere winter than in the summer. In addition, index have the same amplitude but opposite signs. Following a sudden increase in the AE indices (or in the polar cap potential drop), the disturbance electric fields and drifts decrease with an average decay time constant of about 70 min. The prompt penetration of high-latitudelectric fields into the middle- and low-latitude ionosphere has been studied using a number of elaborate theoretical and numerical global the Joule heating rate and the ionospheric electrodynamics are convection models [e.g., Nopper and Carovilano, 1978; Spiro also affected by neutral wind effects [e.g., Lu et al., 1995]. et al., 1981, 1988; Senior and Blanc, 1984; Tsunomura and This results in a range of possible AE values for a given polar Araki, 1984; Zakharov et al., 1989; Denisenko and Zamay, cap potential and high-latitude energy input. Finally, the 1992]. These models determined the global ionospheric equatorial electric perturbations associated with these highlatitude effects could also depend on season and solar flux. These effects are averaged out in our model but in principle they could be taken into account if we had a significantly electric fields and currents resulting from given electrostatic potential distributions or field aligned currents at highlatitudes by solving the continuity equation of ionospheric currents on a two-dimensional (thin shell) ionosphere with larger database. specified ionospheric conductivity distributions. The In the following section, we compare our prompt equatorial zonal electric field (vertical drift) perturbation penetration results with those from global convection models patterns predicted by the different models for comparable by using the same idealized input parameters considered in the sudden changes in the polar cap potential (or in the hightheoretical models. The comparison with the predictions of latitude field-aligned currents) are quite similar [e.g., Fejer, 1991]; the minor discrepancies are largely due to the use of i i i i i! i i i i i! i i i i i i i i i i i l different conductivity models. The local time variation of prompt penetration electric fields depends mainly on the potential equatorward of the shielding layer and on the t +7.5 min -I distribution of the ionospheric conductivities. These perturbations decrease as a function of storm time following the readjustment of the shielding charges in the inner -10 magnetosphere the new polar cap potential drop. The shielding time constant depends on the high-latitude 10 ionospheric conductivity and on the temperatures in the plasma sheet [e.g., Jaggi and Wolf, 1973; Spiro et al., 1988]. Here we will compare our empirical results with the predictions -10 from the Rice Convection Model (RCM), which includes the coupled electrodynamics of the inner magnetosphere and 10 ionosphere [e.g., Wolf et al., 1986; Spiro et al., 1988]. o , The logical setup and parameters of the Rice Convection Model used for mid-latitude and low4atitude direct penetration,of - I i [ I [.I I l i I I i I I, I,,,, I,,, I electric field studies were described by Spiro et al. [1988]. In this model, the electrostatic potential at the poleward Local Time boundary of the calculation (equatorward edge of the electric Figure 4. Empirical prompt penetration vertical drift field reversal region) is given by patterns at three storm times following a step function increase in the AE index by 400 nt. V(UT, ) = Vpc (UT) g( ) (8)

7 24,052 FEJER AND SCHERLIESS: STORM TIME EQUATORIAL ELECTRIC FIELDS 700 [ t7'5 t10 ',,,, 60,,, ' ' ' ' i ' ' ' ' ' ' i t' ' 300 J[- < Storm-Time (Minutes) I,,, I, i,, I,,,, i, i,, i, i, 30 Empirical Model } RCM..- ' A(I) =33kV Initial Time Response E v 0 t+10min...,,.--, 'E.o '-' x ". / - > 10 t+60min o - o.o -lo Local Time Figure 5. Comparison of prompt penetration zonal electric fields obtained from our empirical model (solid curves) and from the Rice Convection Model following an increase in the polar cap potential drop by 33 kv. where Vpc is the polar cap potential drop and g( ) is a function which gives the local time angular dependence of the electrostatic potential at the poleward boundary as specified by Heelis et al. [ 1982]. The local time and latitudinal variations of the zonal and meridional prompt penetration electric fields predicted by this model were discussed by Fejer et al. [1990]. Figure 5 compares the vertical perturbation drifts obtained from our empirical model at t min and t + 60 min for the increase in the AE index by 400 nt shown in the top panel with the response predicted by the Rice Convection Model for the corresponding increase in the polar cap potential drop by 33 kv. In this case, we have used the relationship between the polar cap potential drop and the AE index derived by Ahn et al. [1983], have scaled down the RCM pattern shown in Figure 4 of Fejer et al. [1990] to correspond to 33 kv, and have neglected the small perturbation drifts due to the leakage of the high-latitude electrostatic potential into the equatorial ionosphere under stationary conditions. The results for t + e are probably not realistic since is it is not clear if significant polar cap potential changes can occur in timescales shorter than about 10 min, as assumed in the simulation. Figure 5 shows that the local and storm time dependence, and the amplitudes of the disturbance drifts predicted by the RCM are in good agreement with the results from our empirical model except in the evening sector. As mentioned earlier, these results from the RCM are also consistent with those from other numerical models [Senior and Blanc, 1984; Tsunornura and Araki, 1984; Zakharov et al., 1989; Denisenko and Zarnay, 1992]. Theoretical and simulation studies indicate that the shielding time constant increases with the plasma sheet temperature from about min to 1 hour. We cannot determine this time constant accurately, but here also the simulation and empirical results are consistent. Our representation of prompt penetration disturbances is based on the assumption that identical sudden increases and decreases in AE lead to velocity perturbations with the same amplitude and opposite sign. Fejer and Scherliess [1995] provided experimental evidence for this symmetric response, which is also a property of global convection models (e.g., the RCM) when steady state electrostatic potential effects are negligible. Theoretical models also indicate that the time constants for the build up and decay of electric field perturbations associated with sudden increases and decreases (undershielding and overshielding conditions, respectively) are comparable. These results have been thought to be inconsistent with equatorial observations which typically show larger and longer lasting velocity perturbations following sudden decreases in convection (or large IMF B z noahward turnings) than after identical sudden convection increases [e.g., Kelley et al., 1979; Fejer, 1986]. Fejer et al. [1990] suggested that the larger and longer lasting equatorial perturbations under overshielding conditions could be explained by the combined effects of fossil winds and magnetic reconfiguration, The study by Fejer and Scherliess [1995], however, showed that this asymmetrical equatorial response can be largely accounted for by disturbance dynamo drifts. Although fossil wind and magnetic reconfiguration effects do not seem to play a dominant role on prompt penetration equatorial electric fields in general, our results do not rule out their possible importance under very large and fast magnetic quieting conditions. As mentioned earlier, our empirical model based on the AE index does not take into account some potentially important processes and averages out the effects of a number of ionospheric and magnetospheric parameters which should play important roles on the amplitude and phase of the electric field perturbations and on the shielding time constant. For example, model results presented by Zakharov et al. [1989] suggest that IMF B v might cause large changes in the prompt penetration equatorlal zonal electric field pattern. Preliminary examination of our data suggests that for large IMF By values both the prompt penetration and the disturbance dynamo patterns might differ significantly from our average results. A comprehensive study of these additional processes will require a significantly larger database. 5. Case Studies The equatorial plasma drifts often exhibit a complex interplay of prompt penetration, short- and longer-term disturbance dynamo electric fields. Figure 6 shows an example of our data and model results. The upper panel presents the AE indices for this period, the center panel gives the prompt penetration and the disturbance dynamo drift components derived from our high time resolution model, and the lower panel shows the original data, the corresponding average quiet time pattern, and the result from the superposition of the quiet time pattern and the perturbation components. During this period, the disturbance drifts had small amplitudes until about 0230 LT although the AE indices showed moderate geomagnetic activity. After about 0300 LT, the prompt penetration drifts were initially downward and later upward when combined with the disturbance dynamo drifts they produced large upward velocity perturbations between about 0330 and 0530 LT. The large late night upward drifts lead to the generation of equatorial spread F over an increasing range of heights sampled by the radar which initially contaminated and later precluded further incoherent scatter radar measurements until sunrise.

8 -,, FEJER AND SCHERLIESS: STORM TIME EQUATORIAL ELECTRIC FIELDS 24, LU 500 E o 20 o v r UT ' i,,,, i,,, i,,,, i,,,, i, - -, i,,,, i,,,, i,,, ' i,,, ß I, Disturbance Dynamo....-/%/'X... Vo ' / --'" Prompt Penetrati - - Jicarnarca Data I I I ' ' ' I,,,, I, [,, I,,,, I, LT from changes in the high-latitude convection. They also support earlier suggestions that prompt penetration electric fields can also result from other high-latitude or magnetospheric processes not currently included in the model. Figure 7 also shows large prompt penetration and disturbance dynamo drift perturbations in the postmidnight sectors. The perturbation drifts between about 0030 and 0600 LT on May 14 are partially reproduced by the model and explained as due to disturbance dynamo electric fields. For May 15, the model results indicate comparable prompt penetration and disturbance dynamo drifts of opposite signs between about 0300 and 0600 LT. In this case, the model provided the correct timing of the prompt penetration drifts but it underestimated the amplitudes. Notice that a slight change in the timing of large and rapid AE changes and/or a shift of the disturbance dynamo pattern in the postmidnight sector can sometimes affect significantly th e amplitudes of the predicted drift perturbations. We should also keep in mind that the Jicamarca measurements usually have a higher time resolution than the model. F!gure 8 presents drift measurements and model results for another period of extended geomagnetic activity. This example illustrates the generally fair to good agreement between the model resul ts and th.e data. We now examine further the propc.rties of the electric field perturbations and the limitations of the model resulting from the use of our AE indices as the input parameter. Figure 9 Figure 6. (top) Auroral electrojet indices. (middle) Calculated prompt penetration (solid line) and disturbance dynamo drifts presents the interplanetary magnet!c field (IMF)da. ta, the (dashed line). (bottom) Measured (thin solid line), average auroral AU, AL, and AE indices, and equatorial vertical drifts quiet time (dashed line), and model vertical drifts obtained by adding the average quiet time and the perturbatio n drifts (thick solid line). during a period of long lasting magoetic activity when large daytime drift perturbations were observed. On March 7, the data show a large upward drift followed by two downward - Figure 7 shows the high-latitude and equatorial data and the U.T. model results during an extended period of magnetic activity in 500- May In this case, large upward and downward velocity - May 14-15,. / J. - perturbations were observed in the postmidnight sector (about UT) on May 15. The upward velocity perturbations between 0300 and 0500 UT, which were not reproduced by the model, were most likely due to prompt penetration eastward 0 ', : I I I i I I I electric fields. There are two possible explanations for the Jicarnarca absence of large upward velocity perturbations in the model drifts during this period. First, the AE indices shown in the top panel in Figure 7 and used in our model do not indicate the 20--,, - occurrence of a large increase followed by a decrease in the intensities of the high-latitude ionospheric currents during this period, which according to our model, would produce the observed drift variation. Second, even if such a change in AE had occurred (and was missed as a result of the small number of stations used to obtain these AE indices), it would not explain the large measured perturbation since our empirical model (an o ,.," i. i also the theoretical convection models) predicts relatively weak zonal prompt penetration electric fields in the LT sector (see Figure 5). Another example of a large amplitude, short timescale upward velocity perturbation near the midnight sector which cannot be explained by our model is shown in Figure 9 of Scherliess and Fejer [this issue]. Gonzales et al. [1979] showed that this large upward velocity I,,,, I,,,, I, i I perturbation at about 0400 LT on August 9 was indeed LT associated with a prompt penetration event. These results Figure 7. Auroral electrojet indices; measured, average quiet highlight the fact that since our empirical model uses the AE time, and model drifts; and prompt penetration (solid line) and index as the high-latitude disturbance parameter, it can explain disturbance dynamo (dashed line) drifts for the disturbed period prompt penetration electric fields resulting predominantly of May 14-15, 1969.

9 24,054 FEJER AND SCHERLIESS: STORM TIME EQUATORIAL ELECTRIC FIELDS IJ. T. O LT Figure 8. Same as Figure 7, but for the disturbed period of November 12-14, perturbations. On the following day, magnetometer data from Huancayo, Peru, showed the occurrence of a large eastward current perturbation in the equatorial electrojet at about 0900 LT followed by a large westward perturbation about 1000 LT [Fejer, 1986]. These signatures are indicative of large upward and downward F region vertical plasma drifts, respectively. The Jicamarca radar observed a large upward perturbation at 1300 LT on March 8, consistent with the magnetometer data. Most of the upward (downward) perturbations observed during this period were associated with southward (northward) changes in B z, as is often the case [e.g., Fejer, 1986]. The model results for March 7 indicate downwar drift perturbations only at about LT. The absence of a noticeable downward velocity perturbation at about LT is a result of the small amplitude of prompt penetration electric fields driven by convection changes in the afternoon sector (see Figure 5). These and other observations of large vertical drift perturbations in the afternoon sector [e.g., Fejer, 1986] suggest that the shielding between high- and low-latitude electric fields can sometimes be considerably weakened and/or The examples above illustrate the general agreement between the model results and observations and have also highlighted some of the difficulties introduced by the use of the AE index which does not account for some important disturbance processes. As mentioned earlier, under unusual magnetospheric conditions, such as during periods of magneticlouds when IMF B v is large, the model results often show large disagreement wil h the measuredrifts. However, even when the disturbance processes are correctly modeled, disagreement between measured and modeled drifts can still occur as a result of our use of average quiet time drifts which do not account for day-to-day variability of the low-latitude ionospheric dynamo. This variability is particularly severe near solar minimum and December solstice. Therefore our model results have largest uncertainties under these conditions. Figure 10 shows measured drifts and model results during a moderately disturbed solar minimum period which was preceded by two days of very low geomagnetic activity. Although the model results underestimated the perturbation drifts, the large disagreement between the daytime measured and modeled drifts is most likely due to the considerably weaker electric fields driven by the low-latitude wind system during this period compared to the average values. In fact, for low solar flux conditions, the large day-to-day variability of the undisturbed wind dynamo poses the main challenge for the development of realistic low-latitude ionospheric predictive models. The examples above indicate that the disturbance drift patterns can depart significantly from those presented earlier as a result, for example, of the rotation of the potential pattern equatorward of the shielding layer and also of large ' 10- o ca -10 ' 10-5c 0 ca" -10 : 1000, < 0 < that the potential pattern equatorward of the shielding layer can undergo large changes. Recent RCM studies also suggest the occasional occurrence of very weak shielding (R. Wolf, private communication, 1997). 20 The model predicts large upward and downward perturbations consistent with the data for March 8, but it underestimates the C3 0 amplitudes between about 1100 and 1600 LT. This can probably be attributed to the large underestimate of the polar -2o cap potential drop as a result of the large equatorward motion of the auroral zone. Notice, for example, the negative values ' lo, of the A U indices at about LT. This shows that the o accurate description of the state of the high-latitude currents is ":' -10' essential for accurate predictions of the plasmaspheric disturbance electric fields U.T o6 12 LT Figure 9. IMF B v and Bz; AE, AU, AL; measured (thin solid line), average quiet time (dashed line), and modeled drifts (thick solid line); and calculated prompt penetration (solid line) and disturbance dynamo drifts (dashed line) for March 7-8, 1970.

10 FEIER AND SCHERLIESS: STORM TIME EQUATORIAL ELECTRIC FIELDS 24, U.T. 000 Oct 20-21, ]!,!!,,, i,,!,!,,!,,,,!,!,,, i, 20-,' o ,...,' 0 ' : - -10,-, O LT Figure 10. Same as Figure 7, but for the period of October 20-21, those of global convection models. There are indications that under certain conditions (probably high plasma sheet temperatures) the shielding between the high- and low-latitude ionosphere is considerably weakened. Electric field leakage also occurs under stationary conditions, but at the equator these electric fields appear to be noticeably weaker than the disturbance dynamo electric fields which have opposite polarity and occur at about the same time. The empirical disturbance dynamo patterns, presented in detail in our companion paper, are in good agreement with those from the Blanc-Richmond model. The model presented in this paper can reproduce fairly well perturbation drifts due to convection changes and disturbance dynamo effects during and after typical high-latitude disturbance conditions. It does not properly account for electric fields associated with processes which lead to the rotation in the potential pattern equatorward of the shielding layer and to weak shielding between the high- and low-latitude electrical circuits. Our data suggest that additional electric field sources are most important in the noon-midnight sector during periods of frequent large and short-lived high-latitude disturbances. The inclusion of these effects and of the longitudinal dependence of energy deposition into the highlatitude ionosphere would provide a major improvement on the representation of the disturbance drifts. longitudinal changes in the high-latitude energy deposition sectors. The accuracy of the model results depends strongly on the accuracy of the input parameter and also of the quiet time drifts used to reconstructhe full drift pattern. We use a solar flux dependent 15-day average quiet time pattern to account for the ionospheric dynamo drifts. However, the low-latitude ionospheric dynamo drifts exhibit considerable variability with periods of up to several days, especially close to solar minimum [e.g., Richmond, 1995]. 6. Summary and Conclusions We have described a methodology for the study of the storm time dependence of electric field (plasma drift) disturbances. In this approach, we determine initially the perturbation drifts by subtracting seasonal and solar cycle effects. These perturbation drifts are then modeled as a function of the AE index, which can be easily related to both the polar cap potential drop and to the hemispheric energy input into the high-latitude ionosphere. Our model assumes a linear dependence between the disturbancelectric fields and the AE indices which might not be correct for very large AE values. We could have used the same procedure to examine the dependence of the disturbance drifts on season and solar cycle, but this would have required a significantly larger database. Currently, it is not possible to account for the day-to-day variability of the ionospheric electric fields since it depends on a large number of parameters and also has seasonal and solar cycle dependence. This variability is particularly severe near solar minimum. For an increase in the polar cap potential drop, the prompt penetration electric fields are eastward during the day and westward at night with largest amplitudes in the postmidnight sector. The time constant of these electric field perturbations is about 1 hour. These results are in very good agreement with Acknowledgments. We thank R. Wolf, R. Spiro, P. Riley, D. Hysell, A. Richmond, and M. Mendillo for useful discussions. B. G. Fejer thanks his wife Fran for her support and encouragement. This work was supported by the Aeronomy Program, Division of Atmospheric Sciences of the National Science Foundation through grant ATM The Jicamarca Radio Observatory is operated by the lnstituto Geofisico del Peru, with support from the National Science Foundation. The Editor thanks D. L. Hysell and G. W. Pr61ss for their assistance in evaluating this paper. References Ahn, B.-H., S.-I. Akasofu, and Y. Kamide, The Joule heating production rate and the particle energy injection rate as a function of the geomagnetic indices AE and AL, J. Geophys. Res., 88, , Ahn, B.-H., Y. Kamide, H. W. Kroehl, and D. J. Gorney, Cross polar potential difference, auroral electrojet indices, and solar wind parameters, J. Geophys. Res., 97, , Baumjohann, W., and Y. Kamide, Hemispherical Joule heating and the AE indices, J. Geophys. Res., 89, , Berthelier, A., Influence of the polarity of the interplanetary magnetic field on the annual and diurnal variations of magnetic activity, J. Geophys. Res., 81, , Blanc, M., and A.D. Richmond, The ionospheric disturbance dynamo, J. Geophys. Res., 85, , DeBoor, C. A., A practical guide to splines, Appl. Math. Sci., 27, Deminov, M. G., and G. F. Deminova, Low-latitude nighttime ionospheric reaction to B z of IMF turnings on IKI-19 satellite data, Geomagn. Aeron., 28, , Denisenko, V. V., and S.S. Zamay, Electric field in the equatorial ionosphere, Planet. Space Sci., 40, , Fejer, B. G., Equatorial ionospheric electric fields associated with magnetospheric disturbances, in Solar Wind Magnetosphere Coupling, edited by Y. Kamide and J. A. Slavin, pp , Terra Sci., Tokyo, Japan, Fejer, B. G., Low latitude electrodynamic plasma drifts: A review, J. Atmos. Terr. Phys., 53, , Fejer, B. G., The electrodynamics of the low-latitude ionosphere: Recent results and future challenges, J. Atmos. Terr. Phys., 59, , Fejer, B. G., and L. Scherliess, Time Dependent response of equatorial ionospheric electric fields to magnetospheric disturbances, Geophys. Res. Lett., 22, , 1995.

11 24,056 FEIER AND SCHERLIESS: STORM TIME EQUATORIAL ELECTRIC FIELDS Fejer, B. G., R. W. Spiro, R. A. Wolf, and J. C. Foster, Latitudinal activity inferred from combined incoherent scatteradar and ground variation of perturbation electric fields during magnetically disturbed magnetometer observations, J. Geophys. Res., 95, , periods: 1986 SUNDIAL observations and model results, Ann. Sastri, J. H., K. B. Ramesh, and H. N. Ranganath, Transient composite Geophys., 8, , electric field disturbances near dip equator associated with auroral Fejer, B. G., E. R. de Paula, S. A. Gonzalez, and R. F. Woodman, substorms, Geophys. Res. Lett., 19, , Average vertical and zonal plasma drifts over Jicamarca, J. Scherliess, L., and B. G. Fejer, Storm time dependence of equatorial Geophys. Res., 96, , dynamo zonal electric fields, J. Geophys. Res., this issue. Forbes, J. M., R. G. Robie, and F. A. Marcos, Equatorial penetration of Senior, C., and M. Blanc, On the control of magnetosphericonvection magnetic disturbanceffects in the thermosphere and ionosphere, J. by the spatial distribution of ionosphericonductivities, J. Geophys. Atmos. Terr. Phys., 57, , Res., 89, , Gonzales, C. A., M. C. Kelley, B. G. Fejer, J. F. Vickrey, and R. F. Spiro, R. W., M. Harel, R. A. Wolf, and P. H. Reiff, Quantitative Woodman, Equatorial electric fields during magnetically disturbed simulation of a magnetospheric substorm, 3, Plasmaspheric electric conditions, 2, Implications of simultaneous auroral and equatorial fields and evolution of the plasmapause, J. Geophys. Res., 86, measurements, J. Geophys. Res., 84, , , Heelis, R. A., J. K. Lowell, and R. W. Spiro, A model of high-latitude Spiro, R. W., R. A. Wolf, and B. G. Fejer, Penetration of high-latitudeionospheri convection pattern, J. Geophys. Res., 87, , electric-field effects to low latitudes during SUNDIAL 1984, Ann Geophys., 6, 39-50, Jaggi, R. K., and R. A. Wolf, Self-consistent calculation of the motion of Tsunomura, S., and T. Araki, Numerical analysis of equatorial a sheet of ions in the magnetosphere, J. Geophys. Res., 78, enhancement of geomagnetic sudden commencement, Planet. Space 2866, Sci., 32, , Kelley, M. C., B. G. Fejer, and C. A. Gonzales, An explanation for Weimer, D. R., N. C. Maynard, W. J. Burke, and C. Liebrecht, Polar cap anomalous equatorial ionospheric electric fields associated with a potentials and the auroral electrojet indices, Planet. Space Sci., 38, northward turning of the interplanetary magnetic field, Geophys , 1990a. Res. Lett., 6, , Weimer, D. R., L. A. Reinleitner, J. R. Kan, L. Zhu, and S.-I. Akasofu, Kikuchi, T., H. Liihr, T. Kitamura, O. Saka, and K. Schlegel, Direct Saturation of the auroral electrojet current and the polar cap penetration of the polar electric field to the equator during a DP2 potential, J. Geophys. Res., 95, , 1990b. event as detected by the auroral and equatorial magnetometer chains Wolf, R. A., R. W. Spiro, and G. A. Mantjoukis, Theoretical comments and the EISCAT radar, J. Geophys. Res., 101, , on the nature of the plasmapause, Adv. Space Res., 6, (3), , Lu, G., A.D. Richmond, B. A. Emery, and R. G. Robie, Magnetosphere ionosphere-thermosphere coupling: Effect of neutral winds on Woodman, R. F., Vertical drift velocities and east-west electric fields at energy transfer and field-aligned current, J. Geophys. Res., 100, the magnetic equator, J. Geophys. Res., 75, , , Zakharov, V. E., M. A. Nikitin, and O. A. Smirnov, The response of Mikhailov, A. V., M. F/Jrster, and T. Y. Leschinskaya, Disturbed vertical low-latitude electric field to the action of magnetic source, ExB plasma drifts in the equatorial F2 region at solar minimum Geomagn. Aeron., 29, , deduced from observed NmF2 and hmf2 variations, Ann. Geophys., 14, , Nopper, R. W., and R. L. Carovillano, Polar equatorial coupling during B.G. Fejer and L. Scherliess, Center for Atmospheric and Space magnetically active periods, Geophys. Res. Lett., 5, , Sciences, Utah State University, Logan, UT ( Richmond, A.D., Ionospheric electrodynamics, in Handbook of bfej er@cc. usu.edu; scherlie@logan.cass. usu.edu) Atmospheric Electrodynamics, vol. 2, edited by H. Volland, pp , CRC Press, Boca Raton, Fla., (Received May 30, 1997; revised July 22, 1997; Richmond, A.D., et al., Global measures of ionospheric electrodynamic accepted July 24, 1997.)