2 Comment on ‘‘Unified PCN and PCS indices: Method of calculation,

3 physical sense, dependence on the IMF azimuthal and northward

4 components’’ by O. Troshichev, A. Janzhura, and P. Stauning

6 Renata Lukianova1

Received 30 June 2006; revised 17 April 2007; accepted 26 April 2007; published XX Month 2007.8

9 Citation: Lukianova, R. (2007), Comment on ‘‘Unified PCN and PCS indices: Method of calculation, physical sense, dependence on

10 the IMF azimuthal and northward components’’ by O. Troshichev, A. Janzhura, and P. Stauning, J. Geophys. Res., 112, XXXXXX,

11 doi:10.1029/2006JA011950.

13 1. Introduction

14 [1] Troshichev et al. [2006, hereinafter referred to as T06]15 presented newly calculated unified PC indices from both16 hemispheres that are ‘‘well consistent one with another in17 their value and behavior and linearly correlate with the solar18 wind merging electric field (MEF)’’ (p.1 of T06). Signifi-19 cant differences in the magnitude between the northern20 (PCN) and southern (PCS) indices produced by the Danish21 Meteorological Institute (DMI) and the Arctic and Antarctic22 Research Institute (AARI), respectively, was revealed after23 transition to the 1-min values [Lukianova et al., 2002].24 Since the technique traditionally applied for calculation of25 the PCN and PCS is different to some degree, elaboration of26 the consensus between two institutions, the producers of the27 indices, is quite appreciated. However, careful examination28 of the changes introduced by T06 to the procedure and the29 results of these changes reveals some contradictions. T0630 alleged that the newly applied smoothing of the coefficients31 and the choice of quiet daily variation (QD) as the level of32 reference for magnetic disturbances have eliminated the33 systematic exceeding of the PCS over the PCN. The34 similarity of two indices was achieved in such a manner35 that the PCS became smaller while the PCN became larger.36 [2] The AARI technique always took into account devia-37 tions of the magnetic field from the QD curve. In the DMI38 technique an appropriate daily quiet level (QWL) was39 deduced from interpolation between the magnetic field’s40 absolute values determined at midnight hours of quiet41 winter days in 2 consecutive years [Vennerstrøm et al.,42 1991]. The analysis performed in preparation of the paper43 [Lukianova et al., 2002] revealed that the exceeding of44 PCS over PCN quickly increased with the growth of45 magnetic activity and mostly during the northern summer46 months. In order to check if the discrepancy in the indices47 arises from the different procedure, the AARI technique48 was applied to the magnetic data from Thule. The calcu-49 lations by AARI method showed that the northern50 PCN_AARI became larger and closer to PCS than pub-

51lished PCN from DMI. The increase of PCN_AARI accel-52erated with the growth of magnetic activity. The similarity53of PCN_AARI to PCS was achieved solely due to an54increase of the former. An example of the 1-min PC time55series for 1998 is presented in Figure 1. In this figure, the56former published PCN and PCS are presented along with57the northern PCN_AARI.58[3] T06 did not mention the specific character of the59discrepancy between PCN and PCS. T06 stated that the60former PCN was an ‘‘underestimate,’’ whereas the PCS was61an ‘‘overestimate.’’ These authors argued that the main62reason for the discrepancy was ‘‘too high’’ and ‘‘extremely63weak’’ of a smoothing applying to the normalization coef-64ficients (p. 4 of T06). The first goal of present comment is to65demonstrate that the choice of the level of reference,66namely replacing the QWL line by the QD curve, is crucial67for increasing the magnitude of PCN in response to the solar68wind (SW) and IMF forcing. Other factors (better capture of69the UT variation of fitting parameters as well as the70procedure for derivation of the QD curve) are of secondary71order with respect to the leveling of PCN and PCS ampli-72tude especially during disturbed periods. As is shown in this73comment, for the unified index, T06 followed the AARI74technique and used the QD in both hemispheres. Conse-75quently, a question of why does the amplitude of the unified76PCS turn out smaller than that of former PCS arises. Also, a77larger increase should be expected from the unified PCN. It78particularly concerns the extreme values of the index. The79appearance of many new negative values is troubling.80[4] At the present time, there is no doubt that the polar81cap electrodynamics during the disturbed periods is com-82plicated and nonlinear. The transpolar part of the DP283ionospheric current associated with the two-cell convection84system which determines the behavior of the PC index is85controlled by different factors, not only by the interplane-86tary merging electric field (MEF or Em). The second goal of87the Comment is to caution against a simplification of the PC88response to the MEF. T06 stated that the new ‘‘indices89ensure the best linear correlation’’ with MEF under con-90ditions of southward IMF (p. 10 of T06). They ignored the91fact that during magnetic storms, convection bays, SW

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, XXXXXX, doi:10.1029/2006JA011950, 2007ClickHere

for

FullArticle

1Arctic and Antarctic Research Institute, St. Petersburg, Russia.

Copyright 2007 by the American Geophysical Union.0148-0227/07/2006JA011950$09.00

XXXXXX 1 of 8

92pressure events and substorms the linear correlation is often93violated.

942. Data and Calculation Techniques

95[5] In this section the question of how the QD choice or96the smoothing of fitted parameters affects the magnitudes of97PCN and PCS is addressed. The influence of each modifi-98cation proposed by T06 on the final value of the PC index is99briefly considered.

1002.1. Level of Reference for Magnetic Variations

101[6] Figure 2 gives an example of the QD curve and the102QWL line for the horizontal geomagnetic components from103Thule and Vostok in 1998–2000. Both components show104the same regularities before 1998 and after 2000. The105presented 3-year time series exhibits the well-known sea-106sonal and diurnal variations of the QD. The right panels107give the QDs used for routine day-to-day calculation of PCS108at AARI. Straight lines in the left panels represent the QWL109which were used for routine calculation of PCN at DMI.110Sine-like curves in the same panel show the QDs used for111the PCN_AARI from our earlier paper [Lukianova et al.,1122002]. All QDs were averaged from the 5 quietest days of113each month taking into account three (D, H, Z) components114of geomagnetic field measured at Thule and at Vostok.115Arbitrary fluctuations were removed to obtain the smoothed116wave-like variation. Following the AARI technique, T06117rejected the QWL and adopted the QD for calculation of the118unified PC. T06 introduced a construction of the QD curve119for H and D components by 1-min quiet intervals snapped120up from the previous 30 days. The advantage of the

Figure 1. Example of the PC indices for 1998: published(former) PCS and PCN, and the northern PCN_AARIcalculated by former AARI method.

Figure 2. Level of reference for disturbances in horizontal geomagnetic components for (left) Thule and(right) Vostok. Thin and thick lines represent the behavior of QD curve and QWL line, respectively.

XXXXXX LUKIANOVA: COMMENTARY

2 of 8

XXXXXX

121 proposed method is sliding of the days, but an artificial122 composition of the curve from separated pieces indepen-123 dently for two geomagnetic components suffers by lack of a124 physically interpretable meaning. In the context of the125 present comment, we note that both methods provide126 similar QD curves.127 [7] It is important to understand that the factor that128 crucially affects the final value of PC is the choice of QD129 curve versus QWL line. If we choose the QD approach, the130 PC is more sensitive to the variations of the IMF and SW.131 Choice of the QWL reduces the final value of PC. Figure 3132 gives a sketch illustrating this statement. By definition, the133 PC is based on the linear regression analysis between the134 MEF (called here a ‘‘signal’’) and geomagnetic horizontal135 disturbance dF (called here a ‘‘response’’) as shown in the136 main formula for PC (equation (3) from T06):

PC ¼ Em ¼ z � dF� bð Þ=a ð1Þ

137 where z = 1 m/mV is a scale coefficient and a and b are139 fitted parameters. We first collect statistics for several years140 of the linear regression between the signal and response at141 each UT moment of the year. These statistics are mainly142 represented by the coefficients of slope (a) and intercept143 (b), where a is of the major influence on PC. For simplicity144 in the following discussion we will neglect b and ignore the145 angle f contained in dF. Having a set of coefficients, we are146 able to reconstruct the signal at any time based solely on the147 actually measured response. This sequence of steps is148 shown in Figure 3. The horizontal and sine-like thick solid149 lines represent QWL and QD, respectively. Note that this is150 not the actual D or H component but a representation of the

151geomagnetic disturbance dF. First, we calculate the152coefficients. In the sketch, the thin solid line indicates a153statistical response to a statistical signal for a given interval154of time t1t2. The signal is not shown because its actual value155is not important for the present discussion. Assuming the156signal equals unity at t = t0, the distances (ac) and (bc)157correspond to the coefficients (or to the slope a if b is158neglected) in the case of QWL and QD, respectively. It is159clear that (ac) exceeds (bc). That is because the former160includes both the response to solar illumination and the161response to MEF while the latter includes the pure response162to MEF. The effect is clearly seen in the amplitude of a163given below in the top of Figure 4.164[8] As a second step, we apply the coefficient to the165actually measured response in order to obtain the PC. If for166example, the magnetic activity increases a factor of two, the167response reaches point (d). In the sketch the dashed line168indicates the corresponding behavior of the horizontal169geomagnetic field dF. It is easy to see that the value of170PC is different in the QD case from the QWL case. Taking171ab = bc = cd for simplicity and following the equation (1),172we obtain PC � (ad)/(ac) = 1.5 and PC � (bd)/(bc) = 2,173respectively. Further, if the magnetic activity increases, for174example, by a factor of 6 (point e), than the ratio (ae)/(ac) =1753.5 and the ratio (be)/(bc) = 6. The difference in the first and176second ratios increases progressively with the growth of177activity. This simple illustration gives the explanation of178why the PCS exceeded PCN and why the discrepancy179increased with the growth of activity and mostly in northern180summer months when (ab) is larger. That can also explain181why the PCN from the unified technique of the T06182becomes larger than former PCN. However, it is harder to183explain the reported decrease of unified PCS.

1842.2. Smoothing of the Coefficients

185[9] T06 said ‘‘mainly the unified smoothing procedure186(for the coefficients) has led to similarity of the newly187unified PCN and PCS indices’’ (p. 4). It is difficult to agree188with this statement. Figure 4 shows a time series of189coefficients obtained by Vennerstrøm et al. [1994] (avail-190able at http://web.dmi.dk/projects/wdcc1/pcn/coef24g3.txt)191and used for routine calculation of the PCN at DMI (Figure 4,192left), the coefficients obtained for calculation of the northern193PCN_AARI by the AARI technique [Lukianova et al.,1942002] (Figure 4, middle), and the coefficients used for195routine calculation of the PCS at AARI (Figure 4, right).196The coefficients are given in series for each month of the197year. The diurnal variation is seen within each month. In the198AARI technique, the original data sets of coefficients were199obtained by correlating the 5-min averaged values of dF200with 5-min values of MEF over 1977–1980 from IMP-8/9201and combined monthly to increase statistics. The DMI202technique used the 15-min values of the same parameters.203The original coefficients were then smoothed through the204entire year. In Figure 4, gray and black lines represent the205original and smoothed datasets, correspondingly. We were206unable to locate the original 15-min data sets from DMI. At207AARI, FFT filtering was used to eliminate high harmonics208from the time series. A similarity in rhythm of yearlong209curves of the slope a (i.e., the main part of the dF response)210has been selected as a criteria for the degree of smoothing211applying to the southern coefficients. That was merely

Figure 3. Sketch illustrating how the choice of QD (thicksolid curve) or QWL (thick horizontal line) affects the finalvalue of PC using the statistical variation of the horizontalgeomagnetic component (thin solid line c), and the variationwith the growth of magnetic activity (dashed lines d and e),where letters a, b, c, d, e indicate the responses at t = t0.

XXXXXX LUKIANOVA: COMMENTARY

3 of 8

XXXXXX

212 because the sets of coefficients at AARI were calculated213 later than at DMI. Additional difficulty in the appropriate214 choice of the smoothing degree applied to the entire215 yearlong time series arose from the fact that southern216 coefficients showed more frequent variations than northern217 ones. That is mostly caused by the difference in location of218 the stations. A unified degree of smoothing applied to both219 sets of coefficients therefore may not answer properly to its220 real behavior. The fitting parameters a, b, and f once221 obtained were used to calculate the index for each current222 year.223 [10] At p. 4 T06 wrote ‘‘when deriving the former 1-min224 PC indices, too high of a smoothing was applied in case of225 the PCN index, and extremely weak of a smoothing was226 applied in case of the PCS index. As a result, the amplitudes227 of the former PCN index turned out to be underestimated,228 whereas the former PCS index was overestimated.’’ What229 does this mean mathematically in terms of the former230 calculation methods? Figure 4 of this comment demon-231 strates that there were no ‘‘arbitrary fluctuations of the232 coefficients’’ (p. 10 of T06). Because smoothing is a233 nonlinear procedure applied to the original curve, some234 points of the resulting curve are located above the original235 curve whereas other points are below. Better capture of236 seasonal and UT variation is undoubtedly the important237 issue for more precise calculation of the PC. How can the

238proposed change in the coefficients affect the general239magnitude of PC compared to the former one?240[11] We use the formula (1) to answer the question. We241simply substitute the coefficients from the former and newly242calculated sets then roughly estimate the result. Figure 5243visualizes (1) for winter and summer at both polar caps.244From the two-dimensional plots given in Figure 3 of T06245and from the time series given in Figure 4 of the comment,246one can see that both a and b reach their extremes at about2470400 and 1600 UT (except the Antarctic summer when the248UT variation is more complicated). The peaks are likely the249combined effect of the local noon at each station and the UT250variation of ionospheric conductivity due to the largest251interhemispheric asymmetry in solar illumination at 0440252and 1640 UT. The extremes would be affected first by the253applied smoothing. Thus it seems reasonable to compare the254dependence of PC on dF obtained from the former and new255calculations at 0400 and 1600 UT. The values of the256corresponding a and b are given in Table 1. Figure 5a257gives the result for the northern polar cap. Note that only the258northern PCN_AARI, for which the QD curve were used as259the level of reference, can be explicitly compared with the260unified PCN because of approximately the same value of261dF. Four lines in each plot depict the equation (1) using a262and b from Table 1. Thin and thick lines indicate the former263and unified method, respectively. Black and gray colors

Figure 4. Yearlong time series of the slope a, intercept b, and angle f for (left) Thule using the formerDMI method and (middle) Thule and (right) Vostok using the former AARI method. Gray and black linesrepresent the original 5-min and smoothed values, respectively.

XXXXXX LUKIANOVA: COMMENTARY

4 of 8

XXXXXX

264 indicate UT = 0400 and UT = 1600, respectively. Figure 5a265 one can see that all lines of the unified PCN are above these266 of PCN_AARI. It implies that the newly calculated coef-267 ficients should produce PCN even a bit larger than the268 northern PCN_AARI which are, in turn, very close to the269 former ‘‘overestimate’’ PCS.270 [12] Figure 5b gives the result for Vostok. In January271 (left) at 0400 UT the unified PCS is larger than the former272 PCS. At 1600 UT under quiet conditions (PC < 2) the line273 of unified PCS is below the line of the former. With the274 growth of activity (PC > 2) the lines gradually become close275 to each other and eventually makes the new PCS larger than276 the former. In July (right) the lines of the unified PCS are277 below those of the former PCS. At last the new fitting

278procedure does seem to result in smaller PCS values in the279southern winter. However, two details are troubling. First,280the unified PCS becomes close to the former with the281growth of dF in July. At 0400 UT the thin and thick lines282are almost parallel resulting in the fact that the ratio of283former PCS to unified PCS decreases from 2.1 down to 1.3284for dF = 20 and 100 nT, respectively. The most crucial285moment is 1600 UT when the unified PCS is progressively286smaller than the former PCS. However, again in July, the287ratio between them slowly decreases with the growth of288activity from 1.4 down to 1.3 for dF = 20 and 100 nT,289respectively. On p. 5, T06 wrote ‘‘Figure 4 shows as an290example the run of the PCN and PCS indices during 1998–2912001. One can see the remarkable agreement in behavior of292the positive PC indices in the northern and southern hemi-293spheres, with the index reaching as large value as 20 at both294the Thule and Vostok stations.’’ However, simple arithmetic295shows that if the former PCS exceeds 25 (for example, the296former PCS reached the upper level of 35 on 25-09-1998 or29715-07-2000), the expected value of the new PCS exceeds 20.298[13] The second doubt arises from the comparison of the299lines representing different seasons in Figure 5. If we want300PC about the same in each hemisphere, the coefficients a, b301must provide approximately the same value of PC in302response to a given level of SW forcing irrespective of303the season. Indeed, in Figure 5 in both polar caps the lines304are more vertical in local winter, when the geomagnetic305effect (dF) to MEF is minimal, and more horizontal in local306summer, when the effect of MEF is maximal. We can easily307calculate the ratio between the winter and summer PC for308fixed dF, for example dF = 100 nT. (The same can be done309for fixed PC, but Figure 5 is visually easier when using a310fixed dF.) The results are given in Table 2. One can see that311for the former indices the ratios are 1.9 and 1.8 for312PCN_AARI and PCS, respectively. It implies that the313amplitude of geomagnetic variations decrease by a factor314of 1.8–1.9 going from summer to winter in both hemi-315spheres to provide generally no seasonal difference in PC in316response to a given MEF. For the unified indices we obtain317the smaller values: 1.5 (PCN) and 1.3 (PCS). Why is the318seasonal difference, especially in the southern dF, so small?319The geomagnetic records show larger differences. Figure 6320gives an example. Figure 6a shows the 1-hour time series of321D and H components from Vostok for January and July3221998. Figure 6b shows the quiet daily variations for the323corresponding month (shown above in Figure 2 with lower324resolution). We now estimate the seasonal change of the325amplitude of D and H taking into account only the response

Figure 5. (a) Dependence of PCN on dF obtained from theformer and newly calculated coefficients for January andJuly; (b) the same but for PCS. Lines in each plot depict theequation (1) with corresponding a and b from Table 1. Thinand thick lines indicate the former and unified method,respectively. Black and gray colors indicate UT = 0400 andUT = 1600, respectively.

t1.2 Table 1. Values of a and b Used in Figure 5

IndexPCN_AARIFormer

PCNUnified

PCSFormer

PCSUnifiedt1.3

Coefficient a b a b a b a bt1.5January, UT = 0400 20 0 20 �10 50 0 45 �5t1.7January, UT = 1600 30 �2 30 �5 45 �35 35 �10t1.9July, UT = 0400 40 0 40 �20 30 �15 32 2t1.11July, UT = 1600 70 �30 90 �120 22 �8 28 �6t1.13

t2.1Table 2. Values of the Winter, w, and Summer, s, PC for dF =

100 nT and the Ratio Between Them

IndexPCN_AARIFormer

PCNUnified

PCSFormer

PCSUnified t2.3

UT = 0400, January 5.0 (w) 5.5 (w) 3.7 (s) 3.1 (s) t2.5UT = 0400, July 2.5 (s) 3.2 (s) 2.0 (w) 2.4 (w) t2.7UT = 1600, January 3.3 (w) 3.4 (w) 4.9 (s) 3.8 (s) t2.9UT = 1600, July 1.9 (s) 2.4 (s) 2.9 (w) 3.1 (w) t2.12UT = 0400 2.00a 1.72a 1.85a 1.29a t2.14UT = 1600 1.73a 1.42a 1.69a 1.22a t2.16Averaged over UT 1.9a 1.6a 1.8a 1.3a t2.18

aRatio between the winter (w) and summer (s) PC for dF = 100 nT. t2.20

XXXXXX LUKIANOVA: COMMENTARY

5 of 8

XXXXXX

326 to MEF. We choose the approximate zero level and then327 calculate the mean value of positive (PA) and negative (NA)328 deviations. The sum of PA and NA gives us the mean329 amplitude of the original geomagnetic variation (from330 Figure 6a) and the mean amplitude of the QD curve (from331 Figure 6b). In order to remove the effect of solar illumina-332 tion we subtract the latter from the former. The values of333 PA and NA are given in Figure 6. We obtain the following334 estimates for the amplitude of D and H components: for335 January dD = (133 + 112) � (30 + 20) = 195 nT and dH =336 (54 + 62) � (21 + 32) = 63 nT, for July dD = (57 + 60) �337 (6 + 3) = 108 nT and dH = (23 + 18) � (3 + 3) = 35 nT. One338 can see that for both dD and dH the ratio between the339 southern summer and winter variations is 1.8. This value is

340in good agreement with what we have from the former PCS.341However, the new PCS implies that the summer variations342of dD and dH exceed the winter ones only by a factor of 1.3,343which is smaller than the actually measured geomagnetic344field.

3452.3. Correspondence Between the MEF and PC

346[14] Comparing the ground magnetic data with space data347from the ACE satellite for 4 years free of gaps, T06 used the3485-min time resolution from the AARI technique. Before, the349periods with Dst < �50 nT were excluded from IMP data.350The use of ACE instead the IMP, although it brings more351statistics, gives rise to confusion with the arrival time of the352SW structures and corresponding ground response. Unlike

Figure 6. (a) One-hour time series of D and H components and (b) the run of the QD curves fromVostok for January and July 1998. The mean values of positive and negative deviations from the zerolevel are denoted by ‘‘PA’’ and ‘‘NA’’, respectively. Black and gray lines in Figure 6b represent the D andH components, respectively.

XXXXXX LUKIANOVA: COMMENTARY

6 of 8

XXXXXX

353 IMP, ACE is located much farther from the magnetopause354 (at L1 point) and the arrival time can vary from 20 min to355 100 min depending on the SW speed and the precise356 position of the satellite. T06 did not use any specific tracing357 technique. Without appropriate tracing, the utilization of358 ACE data can give rise to uncertainties in statistics, espe-359 cially of the upper values.360 [15] Another concern is many large negative values361 appeared in the newly calculated PC. Even visual exami-362 nation of Figure 4 of T06 reveals more negative PC values363 than those from the former time series (compare with Figure364 1 of the comment). An appearance of many PC < 0 can365 mean that the regression relationship between MEF and dF366 was found inaccurately. Equation (1) shows that the slope a367 is equally effective for both large and small values of dF,368 whereas any change in the intercept b would affect mostly369 the lowest values of PC. Just b is responsible for zero or370 positive PC under northward IMF conditions where dF is371 small or even negative. Indeed, when the geomagnetic data372 are compared with space data to lay the statistical back-373 ground for the calculation of PC, the situation of dF < 0374 (antisunward ionospheric transpolar current) while MEF �375 0 (IMF BZ > 0) occurs sometimes. This forms a statistical376 threshold ensuring PC � 0. At a given UT/day this377 threshold determines the magnitude of b. The PC remains378 positive until the actual measured dF exceeds b during the379 events of very strong northward IMF. Since b is more380 negative for the unified PCN compared to b for the former381 PCN_AARI (e.g., b is �120 and �30, respectively, for July382 at 1600 UT in Table 1), one might expect less negative new383 PCN values. However, instead, from Figure 4 of T06, more384 negative PC values occur in the local summer at both385 stations.

387 3. Response of PCN/PCS to Variations of MEF388 and Other SW Parameters

389 [16] T06 alleged that differences in values of the previous390 1-min PCN and PCS indices gave rise to discrepancies in391 results of various analyses and to ‘‘quite dissimilar’’ or even392 ‘‘erroneous’’ physical conclusions of previous works utiliz-393 ing the PC index. T06 suggested the replacement of the394 existing sets by newly calculated unified PCN and PCS. In395 particular, T06 said that they planned ‘‘to revise the PCN/396 PCS relationship with the solar wind dynamic pressure and397 auroral substorms’’ using newly calculated ‘‘unified PCN398 and PCS’’ which ‘‘under conditions of southward IMF399 linearly correlate with the MEF.’’ In general, the linear400 correlation exists. Since the algorithm for the derivation of401 the PC index is based on the regression analysis of an402 assumed linear relationship between MEF and dF observed403 near the geomagnetic pole [Troshichev et al., 1988;404 Vennerstrøm et al., 1991], with large statistics under quiet405 and moderate conditions the index does linearly correlate406 with MEF. T06 showed correlation coefficients of PCS and407 PCN with MEF in Figure 10 for all conditions to be about408 0.63–0.66, while Troshichev and Andrezen [1985] found409 better correlation coefficients of r > 0.8.410 [17] Since PC originates in ground magnetic variations, it411 responds not to the MEF itself but to the overhead electric412 currents which are not solely controlled by the MEF. It has413 recently been demonstrated that the linear correlation is

414often violated during disturbed periods. For example, there415is a loss of correlation between the PC and MEF during416magnetic storms [Troshichev and Lukianova, 2002], sub-417storms [Huang, 2005], and SW pressure pulses [Lukianova,4182003; Lee et al., 2004]. The SW pressure pulses are much419more geoeffective under conditions of intense southward420IMF [Boudaridis et al., 2005] with poleward and equator-421ward expansion the polar cap as well as the intensification422of the nightside aurora [Liou et al., 1998]. Under extreme423conditions the PC reflects a strengthening of convection and424a nonlinear magnetospheric response to SW forcing. Refer-425ring to Nagatsuma [2002], T06 wrote ‘‘the use of the former426underestimated PCN index provides effect of the PC index427saturation.’’ Troshichev et al. [2000] reported the effect of428saturation of the ionospheric electric field based on the PCS429statistics. These various results reflect a nonlinear behavior430of the magnetosphere/ionosphere system under very dis-431turbed conditions and further investigation is needed, so the432linearization of the PCN and PCS may not the best way to433resolve the problem.

4344. Conclusion

435[18] The unified technique proposed by T06 seems to be436similar to the former AARI technique in the choice of the437quiet day curve which governs the magnitude of the PC438index. With the same technique it is difficult to see how the439different smoothing in the slope, intercept and angle can440level up and down the former ‘‘underestimate’’ PCN and441‘‘overestimate’’ PCS, respectively. Examination of the coef-442ficients presented by T06 shows that the former northern443PCN_AARI, the new unified PCN, and the unified PCS are444relatively close to the former PCS in January. In July, the445larger a in the unified PCS results in smaller values, while446the more negative b in the unified PCN results in larger447values as shown in Figure 5. However, these coefficient448changes lead to smaller summer to winter seasonal changes449in the assumed geomagnetic variations with ratios of 1.5450(unified PCN) and 1.3 (unified PCS). The observed geo-451magnetic records require a summer to winter ratio of 1.8 at452Vostok, not 1.3, whereas the former PCS coefficients have a453seasonal variation of about 1.8.454[19] How does the different smoothing lead to a better455linear correlation between PC and MEF, including the456disturbed periods when linear dependence between PC457and MEF is often violated? Further explanation would be458helpful for the solution of the long-standing problem of the459discrepancy between the PCN and PCS. The agreement460between two institutions, DMI and AARI, about the usage461of the same basic technique for both indices is very462important, but clearer understanding of the procedure and463more accurate analysis is needed.

464[20] Acknowledgments. Zuyin Pu thanks Barbara Emery and another465reviewer for their assistance in evaluating this paper.

466References467Boudaridis, A., E. Zesta, L. R. Lyons, P. C. Anderson, and D. Lummerzheim468(2005), Enhanced solar wind geoeffectiveness after a sudden increase in469dynamic pressure, J. Geophys. Res., 110, A05214, doi:10.1029/4702004JA010704.471Huang, C.-S. (2005), Variations of the polar cap index in response to solar472wind changes and magnetospheric substorm, J. Geophys. Res., 110,473A01203, doi:10.1029/2004JA010616.

XXXXXX LUKIANOVA: COMMENTARY

7 of 8

XXXXXX

474 Lee, D.-Y., L. R. Lyons, and K. Yumoto (2004), Sawtooth oscillations475 directly driven by solar wind dynamic pressure enhancements, J. Geo-476 phys. Res., 109, A04202, doi:10.1029/2003JA010246.477 Liou, K., P. T. Newell, C.-I. Meng, M. Brittnacher, and G. Parks (1998),478 Characteristics of the solar wind controlled auroral emissions, J. Geo-479 phys. Res., 103, 17,543.480 Lukianova, R. (2003), Magnetospheric response to sudden changes in solar481 wind dynamic pressure inferred from polar cap index, J. Geophys. Res.,482 108(A1), 2428, doi:10.1029/2002JA009790.483 Lukianova, R., O. Troshichev, and G. Lu (2002), The polar cap magnetic484 activity indices in the southern (PCS) and northern (PCN) polar caps:485 consistency and discrepancy, Geophys. Res. Lett., 29(18), 1879,486 doi:10.1029/2002GL015179.487 Nagatsuma, T. (2002), Saturation of polar cap potential by intense solar488 wind electric fields, Geophys. Res. Lett., 29(10), 1422, doi:10.1029/489 2001GL014202.490 Troshichev, O. A., and V. G. Andrezen (1985), The relationship between491 interplanetary quantities and magnetic activity in the southern polar cap,492 Planet. Space Sci., 33, 415.493 Troshichev, O. A., and R. Lukianova (2002), Relation of the PC index to494 the solar wind parameters and substorm activity in time of magnetic495 storm, J. Atmos. Solar Terr. Phys., 64, 585.

496Troshichev, O. A., V. G. Andrezen, S. Vennerstrøm, and E. Friis-Christensen497(1988), Magnetic activity in the polar cap—A new index, Planet. Space498Sci., 36, 1095.499Troshichev, O. A., R. Yu. Lukianova, V. O. Papitashvili, F. J. Rich, and500O. Rasmussen (2000), Polar cap index (PC) as a proxy for ionospheric501electric field in the near-pole region, Geophys. Res. Lett., 27, 3809.502Troshichev, O., A. Janzhura, and P. Stauning (2006), Unified PCN and PCS503indices: Method of calculation, physical sense, and dependence on the504IMF azimuthal and northward components, J. Geophys. Res., 111,505A05208, doi:10.1029/2005JA011402.506Vennerstrøm, S., E. Friis-Christensen, O. A. Troshichev, and V. G. Andrezen507(1991), Comparison between the polar cap index PC and the auroral508electrojet indices AE, AL and AU, J. Geophys. Res., 96, 101.509Vennerstrøm, S., E. Friis-Christensen, O. A. Troshichev, and V. G. Andrezen510(1994), Geomagnetic Polar Cap (PC) index 1975–1993, Rep. UAG-103,511World Data Cent. A for Sol. Terr. Phys., Natl. Geophys. Data Cent.,512Boulder, Colo.

�����������������������513R. Lukianova, Arctic and Antarctic Research Institute, 38 Bering Street,515St. Petersburg, 199397, Russia. (renata@aari.nw.ru)

XXXXXX LUKIANOVA: COMMENTARY

8 of 8

XXXXXX