Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.kosmofizika.ru/pdf1/clilverd_jastp98.pdf Äàòà èçìåíåíèÿ: Mon Oct 31 10:44:00 2005 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 00:58:00 2012 Êîäèðîâêà:

\
PERGAMON
Journal of Atmospheric and Solar!Terrestrial Physics 59 "0887# 0936ï0945

Increased magnetic storm activity from 0757 to 0884
M[ A[ Clilverda\ \ T[ D[ G[ Clarkb\ E[ Clarkeb\ H[ Rishbeth
c

c

British Antarctic Survey "NERC#\ Madin`ley Road\ Cambrid`e\ CB2 9ET\ U[K[ b British Geolo`ical Survey "NERC#\ West Mains Road\ Edinbur`h\ U[K[ Dept[ of Physics and Astronomy\ University of Southampton\ Southampton SO06 0BJ\ U[K[ Received 5 January 0887 ^ revised 5 April 0887 ^ accepted 8 April 0887

a

Abstract The aa index provides the longest continuous data set which can be used in the analysis of magnetospheric and ionospheric phenomenology[ All phases of the solar cycle show increases in activity since cycle 03[ The activity increase does not appear to be associated with any instrumental\ ionospheric or magnetospheric e}ects[ Barely signi_cant e}ects "in terms of the results presented in this paper# have been identi_ed in the long!term change in magnetic latitude of the observatory sites\ the positions of high!latitude ionospheric features such as the cusp\ and ionospheric Pedersen and Hall conductivities due to changing magnetic _eld orientation and strength[ The prime cause of the change in geomagnetic activity is an increase in solar activity[ The number of storms at solar minimum has typically increased by 39) more than the other phases[ This is principally due to increased recurrent storm activity to such an extent that conditions at minimum in recent cycles could be thought of as being more representative of the declining phase[ ÷ 0887 Elsevier Science Ltd[ All rights reserved[

0[ Introduction Previous workers have pointed out an increasing trend in the aa index with time "Feynman and Crooker\ 0867#[ This paper describes an increase with time in the number of magnetic storms as de_ned by aa - 39 nT\ par! ticularly during periods of solar minimum activity[ An analysis of the occurrence rate of magnetic storms is undertaken for the whole data set "014 years#[ Possible causes of the increase in magnetic activity are discussed in terms of observational\ solar\ ionospheric\ and mag! netospheric changes[ Measurements of geomagnetic variation have been undertaken since 0739[ The measurements help to quan! tify the in~uence of solar disturbances on the Earth|s magnetic _eld[ One index of geomagnetic disturbance is the aa\ which uses data from antipodal magnetic observ! atories to describe the level of geomagnetic activity[ Values of the index are typically 1ï199 with -39 indica! tive of disturbed conditions[ The aa index has been retro! spectively calculated from 0757 "Mayaud\ 0861# and pro!

Corresponding author[ Tel[ ] 9933 112 140430 ^ fax ] 9933 112 110115 ^ e!mail ] m[clilverdùbas[ac[uk

vides one of the longest continuous data sets which can be used in the analysis of magnetospheric and ionospheric phenomenology[ In comparison\ routine ionospheric soundings have only been in operation since 0820 and satellite measurements only since 0846[ Other long term data sets include sunspot number and also the occurrence of aurora[ These long term data sets can be useful in describing geomagnetic changes since the aa index started in 0757[ The sunspot number is well known to vary with a period that is on average 00 years[ The length of this cycle varies typically by 20 year[ Currently solar cycle lengths are shortening\ with a maximum length having been observed in about 0899 "cycle 02#[ Variations much longer than 00 years have also been reported\ with well! de_ned 79 and 199 year periods observed "Hughes\ 0866#[ Long!term smoothed sunspot numbers are now observed to be increasing after passing through a minimum in about 0809ï0804 "cycle 03#[ Observations of aurorae have been noted for more than two millennia[ Mid! and low!latitude aurora trig! gered the interest of the Chinese and more recent obser! vations have taken place at higher latitudes[ The occur! rence of aurorae shows a 199!year cycle as well as the 00!year one "Lassen and Friis!Christensen\ 0882#[ The

S0253ï5715:87 ,*see front matter ÷ 0887 Elsevier Science Ltd[ All rights reserved PI I ] S02 5 3 ï 57 15 "87 #9 993 8ï 1


0937

M[A[ Clilverd et al[:Journal of Atmospheric and Solar!Terrestrial Physics 59 "0887# 0936ï0945

activity minimum of the current 199!year cycle occurred in 0894[ An important limitation of this data set can be indicated by the observation that the number of aurorae detected after 0604 was much higher than before[ This sudden change apparently coincides with a well!recorded display towards the end of the 69!year Maunder mini! mum which produced much interest in the phenomenon and hence more observations "Halley\ 0605 ^ Hughes\ 0866#[ Another change that has occurred during the lifetime of the Aa index is the slow decrease in the Earth|s dipole _eld strength "B#\ where a change of -09) in the _rst three terms of spherical harmonic magnetic _eld models "e[g[ IGRF# has been observed in the last 049 years[ This may produce decreases in magnetotail loading as given by with the Akasofu "0879# loading factor and hence corresponding changes in the substorm occurrence rate[ Additionally\ the decrease in B may be expected to pro! duce changes in the ionospheric response to magnetic disturbance by changing the Hall and Pedersen con! ductivities in~uence[

observatories used in forming the index were initially Greenwich "0757ï0815# and Melbourne "0757ï0815#[ In the northern hemisphere these were superseded _rst by Abinger "0815ï0846#\ and later by Hartland "0846ï#[ In the South\ Toolangi "0815ï0879# and Canberra "0879ï# have been used to continue the series[ At each observ! atory change a correction was made for changes in geo! magnetic latitude and local induction e}ects[

2[ Results Figure 0 shows the solar cycles in Aa by Bartels solar rotation format[ The panels increase in year from left to right and top to bottom[ For each solar cycle the mini! mum "as de_ned by smoothed sunspot numbers# occurs on the leftmost edge of the plot\ while solar maximum occurs near the middle of each panel[ In the _rst few cycles solar minimum coincides with an extended period of relatively quiet magnetic activity[ However\ in the more recent solar cycles the period of quiet magnetic activity is much less apparent\ suggesting that solar mini! mum conditions have changed signi_cantly in the last 014 years[ The following _gures show this in more detail[ Figure 0 shows recurrent solar activity with periods of about 16 days as patches that extend over a number of rotations "e[g[ day 5 of the later half of cycle 08#[ Typically this type of activity is associated with the declining phase of the solar cycle\ although there is some suggestion from this plot that recurrent activity during solar minimum has become more common during recent cycles[ This hypothesis is considered in more detail later in the paper[ Figure 1 s a plot of the number of magnetic storms with aa - 39 nT as a function of year since 0757[ Super! posed on the _gure is the variation in smoothed sunspot numbers which is used to de_ne solar minimum con! ditions and a vertical dashed line is plotted to indicate when this occurs[ The plot suggests that the rate of occur! rence of magnetic storms has increased during most solar conditions\ but particularly during the minimum[ The _rst four cycles show almost no signi_cant solar activity at solar minimum whereas the later cycles have activity levels which are similar to early solar maximum conditions[ Similarly\ smoothed sunspot numbers are also non!zero in the most recent cycles[ By breaking the solar cycle down into four well de_ned periods ] maximum "20 year#\ declining\ minimum "20 year# and ascending\ we can determine the change in storm occurrence for each phase during solar cycles 00ï 11[ To e}ectively compare the numbers of storms in each phase of the solar cycle the rate of storm occurrence was calculated for each phase and then normalised such that each phase is of equivalent length "25 months# whatever the length of the cycle[ Figure 2 is such a plot and shows that all the phases exhibit an upward trend in activity since cycle 03[ A least squares linear _t to the data since

1[ The aa index In producing an index of magnetic activity two sys! tematic variations have to be taken into account ] the local time variation\ which is related to the nighttime maximum of substorm activity ^ and the annual variation that has a maximum in each hemisphere during the sum! mer solstice[ By using two geomagnetically near!anti! podal observatories both of these variations should be approximately cancelled and as a result only random planetary activity would have any e}ect on the index[ However\ it should be noted that the semiannual vari! ation in activity is not a}ected by the use of near!anti! podal observatories[ At each observatory\ K indices are produced and scaled to be equivalent to a mid!latitude station[ The aa index is de_ned by the average\ for each 2 h period\ of the K indices from two near!antipodal stations after the transformation of K into amplitudes "nT#[ The magnetic _eld variations observed at the ground\ from which the activity index is calculated\ depend on the E!region Pedersen and Hall conductivities in the local ionosphere[ Mayaud "0860# showed that the cancellation of the local time dependence of storm activity is e.cient when four or more 2 h values are averaged and that averaging over 13 h "denoted by Aa# gives an excellent correlation with other planetary indi! ces[ For these reasons this paper used 7!point running means to generate 13 h average values of aa\ denoted by aa [ A geomagnetic storm is counted when aa - 39 nT and considered to be over when aa drops below 39 nT for two consecutive 2 h periods[ Despite its inherent simplicity the index is a meaningful measure of global geomagnetic activity[ The antipodal


)
M[A[ Clilverd et al[:Journal of Atmospheric and Solar!Terrestrial Physics 59 "0887# 0936ï0945 0938

Fig[ 0[ Daily mean values of the aa index "Aa# for 0769ï0885\ arranged by 16!day solar rotations[ Red represents the most disturbed days and blue represents the quietest days[ The horizontal panels represent a complete solar cycle from minimum to minimum[ In this plot recurring geomagnetic activity appears as horizontal patches of red and green[

CMYK Page 0938

)


0949

M[A[ Clilverd et al[:Journal of Atmospheric and Solar!Terrestrial Physics 59 "0887# 0936ï0945

Fig[ 1[ Histogram showing the annual number of magnetic storms with aa - 39 nT[ Superposed is the smoothed sunspot number[ The dashed lines indicate solar minimum and the dotted lines indicate solar maximum[ The lowest levels of storm activity occur at solar minimum[

cycle 03 indicates that the phases Min]Asc]Max]Dec have increasing gradients of 09]7]6]4 storms:cycle respectively[ The con_dence levels in the correlation coe.cients exceed 86[4) for all phases apart from the maximum\ which has greater than 89) con_dence[ This analysis suggests that conditions during solar minimum are changing more than at other phases[ As already mentioned\ Fig[ 0 suggests an increasing prevalence of recurrent magnetic activity at solar mini!

mum\ i[e[ increasing recurrent features[ Figure 3 is a plot of the percentage number of recurrent storms "i[e[ aa - 39 nT with a 1621 day repetition period# that occurred during the combined declining and minimum phase of each solar cycle[ Although the proportions vary from one cycle to another\ the minimum phase con! tributes a greater proportion of recurrent activity in more recent cycles "06ï11# than in most of the earlier ones

Fig[ 2[ The variation of the number of storms "aa - 39# with each phase of the solar cycle[ The plot shows that storm occur! rence in solar minimum has undergone a marked change since cycle 04[

Fig[ 3[ The percentage number of recurrent storms occurring during the declining and minimum phases of the solar cycle[ From cycle 00 to cycle 04 the vast majority of recurrent activity occurred during the declining phase\ but\ later cycles show an enhanced contribution from the minimum phase[


M[A[ Clilverd et al[:Journal of Atmospheric and Solar!Terrestrial Physics 59 "0887# 0936ï0945

0940

"00ï03#[ The increase in recurrent storm activity at solar minimum can also be observed in Fig[ 0[ Typically the recurrent activity lasts for more solar rotations in recent solar minima than in past ones\ i[e[ about 4 rotations in cycles 00ï04 and about 8 rotations in cycles 07ï11[ There are about the same number of active groups in each minimum phase\ i[e[ approximately 3\ but the most recent cycles have shown an increase from 1 to 4 days in the number of days that each recurrent storm lasts[ In sum! mary the observed increase in geomagnetic activity in the declining and minimum phase of the solar cycle result from a doubling both of the number of recurrent storms and their duration[ A comparison of the average sunspot number "R# and the average Aa index during the solar minimum phase is shown in Fig[ 4[ The plot indicates that both R and Aa vary in a similar fashion\ and this suggests a link between solar activity and the trend in the Aa index[ However\ by using a threshold "i[e[ aa - 39 nT# in the analysis it is possible that some of the long!term changes reported in this section could be due to instrumental e}ects\ such as changing sensitivity[ Figure 5 is a plot of yearly mean of the lowest Aa value observed in each month[ Even the most disturbed part of the solar cycle tends to have per! iods of quiet activity and thus it would be expected that the lowest Aa values recorded each month would remain constant[ This assumption is supported by the monthly minimum sunspot number superposed on the plot which shows that during every solar minimum phase there are occasions when no sunspots are observed[ Figure 5 shows an upwards trend in quietest Aa in the last 099 years but only about 2ï3 nT[ The corresponding rise at sunspot maximum is rather similar in magnitude[ This could be caused by a subtle instrumental e}ect or possibly by local ionospheric current:conductivity e}ects[ Both of these e}ects will be discussed in the next section[ Figure 6 shows the e}ect of having a threshold value other than 39 nT on the storm numbers detected during the solar minimum phase of solar cycles 01ï11[ A similar trend is seen for all four threshold values[

3[ Discussion We have shown that the occurrence of high geo! magnetic activity has changed markedly since the begin! ning of the Aa index in 0757[ The number of storms at solar minimum has increased by a factor of about 1 in 00 solar cycles compared with the declining phase\ with the most signi_cant increase being observed in recurrent storm activity[ In this next section we discuss possible mechanisms through which this change may have occurred[ 3[0[ Observational chan`es Although o}set corrections have been made for chan! ges in observatory site\ no long term corrections have been made for the slow change in geomagnetic latitude of the observatories because of the secular change of the geomagnetic _eld[ Figure 7 shows how the corrected geomagnetic latitudes of Hartland and Canberra have altered since 0769[ In the northern hemisphere an equ! atorward drift of about 3> has occurred\ while in the South there has been a drift of about 1> polewards[ The scale used at each observatory to convert the 2 h ranges into K index depends on the angular distance from the observatory to the band of maximum auroral activity "Mayaud\ 0879# which is de_ned to be at a corrected geomagnetic latitude "c[g[l[# of 58>[ Note that the angular distance between the observatory and the auroral oval is not simply the di}erence between the c[g[l[ of the two because of the distortion of the corrected geomagnetic coordinate system[ The observatories in the northern hemisphere have used the same K scale throughout the whole timespan of the aa index\ but the c[g[l[ of the observatories have decreased during this time\ so it may be expected that\ given a constant level of global geomagnetic activity\ more quiet intervals "K 9 or 0# would be scaled in the later years than the earlier years[ We would expect to see the opposite trend in the southern hemisphere[ However an examination of the annual occurrence of intervals of K 9 and 0 at the northern and southern observatories indicates that the occurrence of K 0 has remained fairly constant throughout the timespan of the aa index "cycle 11 has 84) K 0 occurrence compared with previous cycles#[ The occurrence of K 9 has decreased in both hemispheres "cycle 11 has 29) K 9 occurrence com! pared with previous cycles#\ although it appears to have decreased more in the northern hemisphere "Clark et al[\ 0886#[ Figure 5 indicates some increase in the amplitude of Aa during the quietest magnetic periods\ which would have the e}ect of decreasing the occurrence of K 9[ This is discussed in detail by Clark et al[ "0886# who concluded that the scaling of the quietest activity is a very sensitive process "because of the logarithmic scale of K indices# and even the component of the rise since 0874

Fig[ 4[ Average sunspot number "R# versus average Aa index during the solar minimum phase[ The con_dence level in the correlation coe.cient exceeds 88[8)[


0941

M[A[ Clilverd et al[:Journal of Atmospheric and Solar!Terrestrial Physics 59 "0887# 0936ï0945

Fig[ 5[ The variation in the yearly mean of the lowest Aa values observed in each month "squares# and sunspot numbers "circles#[

Fig[ 6[ The e}ect of using di}erent threshold values on the number of storms detected during solar minimum[


M[A[ Clilverd et al[:Journal of Atmospheric and Solar!Terrestrial Physics 59 "0887# 0936ï0945

0942

Fig[ 7[ The variation in corrected geomagnetic latitude of the Hartland and Canberra observatories since 0769[

associated with changing to digital instrumentation does not in~uence the scaling of higher index values in any signi_cant way[ 3[1[ Ionospheric chan`es Secular changes in the strength of the Earth|s main _eld will in~uence the electric currents in the ionosphere\ which will in turn a}ect the magnetic signature at the ground[ Rishbeth "0874\ 0886# suggests that progressive weakening of the dipole _eld is causing a gradual increase in height of the ionospheric dynamo layer[ As mentioned in section 1\ the geomagnetic activity observed at a station depends mainly on currents ~owing locally in the ionosphere[ The ionospheric currents depend on the e[m[f[s that drive them and the electrical conductivity[ The currents associated with magnetic activity probably originate from magnetospheric processes[ Roughly speaking\ the e[m[f[ generated in the magnetosphere should be proportional to "solar wind speed#â"magnetic _eld strength#\ at least on time scales much longer than those of reconnection events\ i[e[ minutes[ Although this idea may be too simple as regards the currents that produce magnetic activity\ we see no reason to expect the e[m[f[s to change greatly as a result of secular decrease of the geomagnetic _eld[ For a given e[m[f[ the ionospheric current\ which mainly ~ows in the E!layer\ depends on the Pedersen and Hall components of conductivity given by ] sP Ne vene vini ¦ 1 1 B ve ¦ne ni1¦vi1

s

H

1 Ne vi1 ve - 1 1 B ve ¦ne ni1¦v

0

1 i

1

"1#

where N is the electron:ion concentration\ -e is the charge on an electron\ and m\ v\ n "respectively# represent mass\ angular gyrofrequency Be:m\ and collision fre! quency with neutral particles[ Subscripts e and i apply to electrons and ions respectively[ For simplicity\ we assume there is a single ion species with positive charge ¦e[ It seems most unlikely that our conclusions would be alt! ered by taking account of multiple ion species[ It can be seen from "0# and "1# that the conductivities take the form s s
P

"Ne:B#Ï fP"ve:ne#¦fP"vi:ni#L "Ne:B#Ï fH"ve:ne#-fH"vi:ni#L

"2# "3#

H

0

1

"0#

where\ fP\ fH are dimensionless algebraic functions[ The heights at which Pedersen and Hall conductivities are largest are determined by the heights at which ve:ne 0 and vi:ni 0[ These heights are a}ected by changes in geomagnetic _eld intensity B[If B undergoes a long!term secular decrease then ] "a# the multiplying factor "Ne:B# increases ^ and "b# the height pro_les of sP and sH move upwards\ because the resulting change in the ratios "ve:ne# and "vi:ni# which are in proportion to B[ These changes are not very signi_cant for the 015 year period considered in this paper[ During this time\ B has decreased by approximately 09)\ thereby increasing the conductivities by 09)\ because of "a# above ^ and moving the conductivity pro_les upwards by about 09) of an


0943

M[A[ Clilverd et al[:Journal of Atmospheric and Solar!Terrestrial Physics 59 "0887# 0936ï0945

atmospheric scale height\ i[e[ by about 0 km in the E! layer\ because of "b#[ We have no reason to believe that a change of magnetic _eld intensity a}ects the ionospheric electron concentration:height pro_le N"h# "except pos! sibly in the magnetic equatorial region where the well! known F!layer fountain operates#\ so the likely long!term change in the multiplying factor "Ne:B# is that due to the decrease of B[ These simple considerations are con_rmed by detailed calculations of the height!integrated conductivities in the range 79ï199 km[ The height pro_le of N was estimated using the International Reference Ionosphere model "Rawer and Bilitza\ 0878# and the height pro_le of ni was estimated using the relationships given by Banks "0855#[ We note here the limitations of the IRI model using sunspot number as a proxy for solar UV and EUV ~uxes "Donnelly et al[\ 0872#[ The height pro_le of B was mod! elled using the International Geomagnetic Reference Field "Barton et al[\ 0886#[ Figure 8 shows the variation of the local magnetic _eld intensity at Hartland and Can! berra since 0899 modelled by the IGRF[ There has been an increase of 1[7) since 0829 at Hartland and a decrease of 0[5) at Canberra since 0849[ If we compute the inte! grated conductivities for local noon at summer solstice assuming solar minimum conditions\ we _nd that both the height!integrated Pedersen and Hall conductivities have decreased by 2[6) since 0829 in England and increased by about 1[1) in Australia[ The heights of the peaks in conductivity are estimated to have changed by amounts smaller than 0 km[ If there were no other factors a}ecting ionospheric currents\ then we would expect that magnetic variations recorded on the ground would change in amplitude by roughly the same amount[ Since the upper limit for K 9 at Hartland is 4 nT\ and the resolution of the variometers has ranged between about 0[9ï9[0 nT\ any change in the amplitude of magnetic disturbances of the order of 3) would hardly a}ect the occurrence of K 9[ Increased concentrations of greenhouse gases are expected to cause a cooling in the thermosphere "Roble and Dickinson\ 0878#[ Rishbeth "0889# predicted a low! ering of 1[4 km for the peak height of the E!layer for doubled mixing ratios of carbon dioxide and methane[ This level of height change is at the limit of detectability for normal ionosonde soundings and no trends in E!layer peak height or critical frequency could be found in noon measurements from ionosonde data at Slough and Can! berra[ We conclude that changes in ionospheric con! ductivity due to changes in the local geomagnetic _eld intensity or greenhouse gases do not contribute sig! ni_cantly to the long!term change in the Aa index[ 3[2[ Ma`netospheric chan`es In the previous subsection we mentioned the e}ect of the decreasing total _eld intensity of the Earth[ In the

magnetosphere this would tend to have the e}ect of reducing the sunward loading area as de_ned by Akasofu "0879#[ A 09) decrease on the _eld strength would typi! cally result in an Earthwards displacement of 4) in the magnetopause position and thus a consequent reduction of the sunward loading area by about 09)[ However\ it is not clear that such small changes in loading area would signi_cantly alter the coupling of solar energy into the magnetosphere since the solar wind {aspect angle| also has a signi_cant e}ect on energy coupling[ Several authors have addressed the problem of a large decay of the dipole component of the Earth|s magnetic _eld during _eld reversals "e[g[ Siscoe and Crooker\ 0865 ^ Rishbeth\ 0874#\ but no work is known to the authors that has been undertaken on the e}ect of small changes[ This would be an interesting area of study but one that is beyond the scope of this paper[ One possible result of a decreasing _eld intensity could be the relocation of the average position of high latitude ionospheric features[ If the auroral oval moves closer to the observatories then the response of Aa to magnetic activity would increase[ Newell and Meng "0883# showed how the dayside ionospheric features responded to chan! ges in solar wind pressure[ During a 3!fold increase in solar wind pressure "but with constant velocity# iono! spheric features moved equatorward by ¾4>[ Decreasing the _eld intensity would have the same sort of e}ect as increasing the solar wind pressure although in the case of a 09) change in _eld strength the ionospheric dis! placement would be proportionally less i[e[ ½0> equa! torwards[ This is less than the drift of the observatories reported in section 3[0 and\ as discussed before\ is small compared with the distance to the auroral oval[ It is unlikely to produce any signi_cant change in the observed levels of magnetic activity "Mayaud\ 0879#[ 3[3[ Solar chan`es Section 0 of this paper indicated that 79 and 199 year periods are observable in solar activity using very long data sets such as low!latitude aurorae observations and sunspot numbers[ The most recent minima in the 199 year cycle occurred in the 0809ï0804 or cycle 03ï04[ At solar minimum\ even though the 79 or 199 year period minima is not clearly detectable in total number of storms\ a signi_cant change in the number of recurrent storms has occurred since cycle 03 or 04[ Before that time\ recurrent storms were a rare occurrence but became much more common afterwards "see Fig[ 3#[ The appar! ent incursion of recurrent activity into solar minimum from the declining phase suggests a link between the two phases\ or rather\ that our somewhat arbitrary delin! eation between the two phases is no longer valid[ Figure 4 also suggests a link between solar activity and the reported increase in the Aa index during solar minimum conditions[ Thus it appears that although there


M[A[ Clilverd et al[:Journal of Atmospheric and Solar!Terrestrial Physics 59 "0887# 0936ï0945

0944

Fig[ 8[ The variation in local total _eld intensities of the Hartland and Canberra observatories since 0769[ No annual mean values for Canberra are available before 0782[

is a general increase in storm activity in all phases of the solar cycle since the beginning of the Aa index\ solar minimum has undergone the most dramatic change\ with storm numbers increasing typically 39) more per cycle than the maximum phase[ This change has been brought about principally through increased recurrent storm activity since cycle 04 to such an extent that conditions could be thought of as representative of the declining phase rather than minimum[

4[ Conclusions This paper reports an increase in the number of mag! netic storms "de_ned by aa - 39# during the minimum phase of solar cycles since 0757[ Although all phases show increases in activity with time since 0809ï0804\

solar minimum has undergone the most marked change[ The number of storms at minimum has typically increased by 39) more than the other phases since cycle 03 "about 0802#[ The underlying cause of this increase is principally the incursion of declining phase conditions into solar minimum "de_ned by minima in the variation of smoothed sunspot numbers#\ so that {solar minimum| is much shorter than it used to be[ We have no grounds to believe that any signi_cant increase in activity can be due to instrumental e}ects "section 3[0[# or changes in ionospheric conductivity "sec! tion 3[1[#[ As for changes in magnetospheric con! _guration "section 3[2[#\ the decrease in the dipole _eld reduces the cross!section on the magnetosphere\ and thus decreases the solar wind energy intercepted[ As the geo! magnetic _eld may be changing in shape as well as in strength\ the {aspect angle| at which the solar wind


0945

M[A[ Clilverd et al[:Journal of Atmospheric and Solar!Terrestrial Physics 59 "0887# 0936ï0945 Instruments\ Data Acquisition and Processing\ Germany\ 7ï 04 September[ Donnelly\ R[F[\ Heath\ D[F[\ Lean\ J[L[\ Rottman\ G[J[\ 0872[ Di}erences in temporal variations of solar UV ~ux\ 09[6 cm solar radio ~ux\ sunspot number\ and Ca!K plage data caused by solar rotation and active region evolution[ J[ Geophys[ Res[ 77\ 8772ï8777[ Halley\ E[\ 0605[ An account of the late surprising appearance of the lights seen in the air on the sixth of March last\ with an attempt to explain the principal phenomena thereof[ Phil\ Trans[ for the months of Jan[ Feb[ March 236\ 395ï318[ Hughes\ D[\ 0866[ The inconsistent Sun[ Nature 155\ 394ï395[ Lassen\ K[\ Friis!Christensen\ E[\ 0882[ Critical Assessment of Selected Solar Activity Parameters[ Danish Meterological Institute Scienti_c Report 82!09\ Copenhagen[ Mayaud\ P[N[\ 0860[ Une mesure planetaire d|activity mag! netique basee sur deux observatories antipodaux\ Annales de Geophysique 16\ 56ï69[ Mayaud\ P[N[\ 0861[ The aa indices ] A 099!year series charac! terizing the magnetic activity[ J[ Geophys[ Res[ 66\ 5769ï5763[ Mayaud\ P[N[\ 0879[ Derivation\ meaning\ and use of geo! magnetic indices[ Geophysical monograph 11\ American Geo! physical Union\ Washington\ DC\ U[S[A[ Newell\ P[T[\ Meng\ C[!I[\ 0883[ Ionospheric projections of mag! netospheric regions under low and high solar!wind pressure conditions[ J[ Geophys[ Res[ 88\ 162ï175[ Rawer\ K[\ Bilitza\ D[\ 0878[ Electron density pro_le description in the International Reference Ionosphere[ J[ Atmos[ Terr[ Phys[ 40\ 670ï689[ Rishbeth\ H[\ 0874[ The quadrupole ionosphere[ Ann Geophys[ 2\ 182ï187[ Rishbeth\ H[\ 0889[ A greenhouse e}ect in the ionosphere< Planet[ Space Sci[ 27\ 834ï837[ Rishbeth\ H[\ 0886[ Long!term changes in the ionosphere[ Adv[ Space Phys[ In press[ Roble\ R[G[\ Dickinson\ R[E[\ 0878[ How will changes in carbon dioxide and methane modify the mean structure of the meso! sphere and the thermosphere< Geophys[ Res[ Lett[ 05\ 0330ï 0333[ Siscoe\ G[L[\ Crooker\ N[U[\ 0865[ Auroral zones in a qua! drupole magnetosphere[ J[ Geomag[ Geoelectr[ 17\ 0ï8[

impinges on the magnetosphere may also change[ Detailed magnetospheric modelling would be required to investigate the consequences[ Small and barely signi_cant e}ects may arise from secular changes in magnetic lati! tudes of the observatory sites\ in relation to features such as the auroral ovals and dayside cusp[ Since the solar wind has been systematically measured for only 29 years\ we cannot assess its possible long!term changes[ Long term trends in the E!region parameters measured by ionosondes close to the location of the observatories were not observed\ although this area of study continues to be a topic for future work[ We are left with the increase of solar activity "section 3[3[# as the likely prime cause of the change in geomagnetic activity[

Acknowledgement The authors would like to thank M[ Lockwood for his helpful discussion of solar wind:magnetosphere inter! actions[

References
Akasofu\ S[!I[\ 0879[ The solar wind!magnetosphere energy coupling and magnetospheric disturbances[ Planet[ Space Sci[ 17\ 384ï498[ Banks\ P[M[\ 0855[ Collision Frequencies and Energy Transfer[ Planet[ Space Sci[ 03\ 0974ï0092[ Barton\ C[E[ 0886[ International Geomagnetic Reference Field ] the Seventh Generation[ J[ Geomagn[ Geoelectr[ 38\ 012ï037[ Clark\ T[D[G[\ Clarke\ E[\ Clilverd\ M[A[\ 0886[ The Role of Magnetic Observatories in Monitoring Long!term Change in the Solar!Terrestrial Environment[ Submitted to Proceedings of VIIth IAGA Workshop on Geomagnetic Observatory