Atmospheric greenhouse effect and ionospheric trends. Hari Om Upadhyay and K.K. Mahajan. Radio and Atmospheric Sciences Division, National Physical ...
GEOPHYSICALRESEARCHLETTERS,VOL. 25, NO.17,PAGES3375-3378,SEPTEMBERI, 1998
Atmospheric greenhouse effect and ionospheric trends Hari Om Upadhyay and K.K. Mahajan Radio and Atmospheric SciencesDivision, National Physical Laboratory, New Delhi, India
positionwould lower the heightsof the E-layer peak by about 2 km and of the F2 peak by 15-20 km. Rishbeth also pointed out that while this lowering would km decreasein the F2 peak height(hmF2) and a mi- have noticeableeffecton HF/VHF radio wavepropanor decreasein the F2 layer criticalfrequency(foF2) gation via the ionosphere,the changesin the E and F2 IRishbethand Robie1992]. In this paper we analyze layer electron density would be rather small. Rishbeth ionosondedata for some 31 stations to study the long andRoble[1992]later madea moredetailedcalculation term trends in hmF2 and foF2. Regressioncoefficients of the greenhouse effectby usingthe NCAR Thermofor hmF2 and foF2 as a function of solar activity, are sphere/Ionosphere generalcirculationmodeland basiobtaine'dfor each station and departures(anomalies) cally confirmedthe earlier results. from expectedvaluesderivedfor both theseparameters. Ionosphericdata from severalstationsare now availAn analysisof hmF2 and foF2 anomaliesindicatesneg- able, especiallysincethe IGY (1957-58)and one can ative trends for some stations and positive trends for thusstudythe longterm trendsin the ionosphere.These others. These varied between +29 to-20 km for hmF2 data contain severalionosphericparameters,including and +0.7 to -0.6 MHz for foF2 in the 34 year period E-layer critical frequency(foE), F2 layer critical fresince1957,the InternationalGeophysicalYear (IGY). quency(foF2) and F2 layer M factor- the ratio of We conclude that the present data do not provide a the maximum usable frequency divided by the critidefinitive evidenceof any global long term trend in the cal frequency(M(3000)F2). The last parametercan ionosphere. be usedto deduceF2 layer peak height(hmF2) by an Abstract. Theoretical calculationspredict that COs. doubling would produce a 50 K decreasein the thertoospherictemperature which can result in about 20
empiricalformulationgivenby $himazaki[1955]. Indeed data from two of the stations have been analyzed
to study the long term trends in the ionosphere;for
Introduction
Anthropogenicactivities have been steadily enhancing the concentrationof greenhousegaseslike COs., CH4 and NOs. in the Earth's atmosphere. For example COs. at ground level at Mauna Loa, Hawaii, has increasedfrom 314.5 to 355 ppmv between 1958 and 1990[Keelinget al, 1995]andis expectedto doublefrom
its IGY (1957-58)valuebeforethe end of 21st century with Business as usual[seee.g. Houghtonet al., 1995]. Robleand Dickinson[1989]werethe first to examine
Juliusruh/Riigen (54.6øN,13.qøE) byBremer[1992] and for Sodankyl/• (67.4øN,26.7øE)by UlichandTurnhen [1997]. Theseauthorshavefoundnearlya linear decreasein hmF2 during the last 30-40 years. This result has been used as an evidenceof long term coolingof the upper atmosphere,therebylinking this effectto the increasein the concentrationof the greenhousegases. Sincethe upper atmosphericcoolingis expectedto be a globalphenomenon, it is desirableto examinethe ionosphericresponseat as many stationsas possible.In this paper we thereforeanalyzeionosphericdata for a large number of stations with geomagneticlatitude varying
the responseof the upper atmosphereto the increased concentrationof these greenhousegases. By using a global averagemodel of the coupledmesosphere,ther- from 52øSto 69øN,and all longitudes to lookfor any mosphereand ionospheresystem, Roble and Dickinson signaturesdue to the globalcoolingof the upperatmofound that mesosphereand thermospheretemperature sphere. will cool by 10 K and 50 K respectively,if the mixing ratios of COs.and CH4 are doubledat mesospheric
heights(at 60 km). Robleand Dickinsonpointedout
Data Analysis
that when the mixing ratios are halved, these regions are heatedby similar amounts. Theseauthorspredicted that the ionospherestructure would alsobe altered with loweredE and F-regionspeak densitiesand smallertopsideplasmascaleheights,althoughno numericaldetails
The World Data Center A at Boulder, CO, U.S.A. has producedtwo CD-ROMs containingdata covering the periodfrom IGY to the year 1990. Thesedata have been compiledafter-applyingsomestringenttests and thus can be consideredto be of high quality. The to-
werepresented.Rishbeth[1990]madesimplecalcula-
tal number of stations listed in these CDs is 328. For
tions to examine numerically the effect of a 50 K decreaseon the ionosphereand he concludedthat 50 K
studyinglongterm trends,we haveselectedonly those stations which have nearly continuousdata for 30 or coolingand the associatedchangesin the neutral com- more years [seealso Bremer, 1992]. Thirty one stations qualified this criterion; 18 stations with 34 years of data, 8 stationswith 33 yearsof data and 5 stations Copyright 1998 by the AmericanGeophysicalUnion. with 32 years of data. The parametersanalysedare foF2 and hmF2. Since Paper number 98GL02503. theseCDs do not contain the parameter hmF2 for ma0094-8534/98/98GL-02503505.00 3375
3376
UPADHYAY AND MAHAJAN: LONGTERM IONOSPHERIC TRENDS
jority of the stations, we obtained it from the follow-
include thesefour stations in our analysisfor studying
ing empiricalrelationgivenby $himazaki[1955]and trendsin hmF2. Another station Sloughhad somedata improvedby Bilitza, [1979]- a proceduresuccessfullygapsin M(3000)F2 . This stationtoo wasnot included followedearlierby Bremer[1992]andby Ulichand Tu- for studying trends in hmF2. We thus derived hmF2 runen[1997]:
anomalies for 26 stations and foF2 anomalies for 31 sta-
tions and studied trends in theseparameters by a linear fit. Table I givesthe geomagneticco-ordinatesand the
1490
hrnF2[M(3000)F2 4-AM) ]- 176 (1) values of hmF2
and foF2 anomalies for the various sta-
tions. It can be noted that anomaliesare both positive
Theoriginalequation developed by $himazaki [1955], and negative. For hmF2, 14 stations show negative The correction term
had the correction term AM-0.
trend while 12 show positive trend. For foF2, 17 sta-
wasintroducedby Bilitza, [1979]to take into account tionsshownegativetrend while 14 showpositivetrend. the influenceof underlyinglayersand is givenas: Figure I showsexamplesof thesetrendsfor hmF2 for F1F4
AM--[(foF2/foEF2)]4.F3
(2)
with F1 - 0.00232R 4. 0.222, F2 = 1.2- 0.0116 exp
(0.0239R),F3 - 0.00064(R-25), F4 - 1-(R/150)exp(O2/1600),R - sunspot number,andO is thegeomagnetic latitude.
We haveemployedequation(1) and (2) to obtain hmF2 from the M(3000)F2 factor givenin the WDC data set. Equation(2) alsoneedsinformationon foe which was not available
for some of the stations.
three of the stations - one showsa positive trend, the secondshowsnearly no trend and the third a negative trend. Figure 2 showssimilar examplesfor foF2. From Figures I and 2 and from Table 1, no definitive evidence of any global long term trend in foF2 and hmF2 can be seen.
To examine if the regressioncoefficientb for any sta-
tion (seeEquation5) is significantly differentfromzero, we calculatedthe Fisher's F-parameter from the equation:
We
usedthe IRI model[Bilitza,1990]for this purpose.To verify the validity of this procedure,IRI derivedfoe and the observedfoe was compared for a few stations like
Maui, Uppsalaand Juliusruh/Riigen,and a reasonable agreementwas found between the two. To study the long-term trends,we examinedthe mid-
day data only. FollowingUlichand Turnhen[1997],we averagedthe monthly median data for 5 hours, 10:00LT
to 14:00LTfor foE,foF2andM(3000)F2. Sinceall these parameters are strongly influencedby solar activity, it is necessaryto filter out the solar activity effect in order to look for any long term changesin foE, foF2 and hmF2. To achieve this each of these parameters was correlated with monthly averaged 10.7cm solar radio flux. Assuminga linear dependence,we obtained the regressioncoefficientsA, and B from the equation:
Xth = A + BF10.7
(3)
Table
1. Linear Trends in hmF2
and foF2
Sr.
Name of
Geomagnetic hmF2
No.
the Station
Lat.
1. 2. 30 4, 5. 6. 7. 8. 9. 10. 11.
Aldta Alma ata Ashkhabad Boulder Canberra Churchill Dourbes Gorky Hobart Huancayo Irkutsk
29.6 33.4 30.4 48.9 -43.8 68.7 51.8 50.2 -51.5 -0.6 41.2
Long. km/yr 206.2 151.4 133.9 317.4 225.2 323.9 88.4
-0.15 0.40 0.86 -0.22 0.19 -
foF2
MHz/yr -0.001 0.005 0.003 -0.005 0.001 0.000 0.009
127.3
0.06
225.2
-
0.000
354.6
-0.36
175.0
-0.02
-0.006
-0.020
-0.003
0.017
12.
Johannesburg -27.1
92.2
-0.58
a procedure similar to the one adopted by Ulich and
13.
Juliusruh
54.3
99.3
-0.21
-0.003
Turuncn[1997]. Usingthesecoefficients, we then obtained the deviation AX of the observed data from the theoretical value as follows
14. 15. 16.
Kokubunji Leningrad Lycksele
25.6 56.1 62.7
206.1 117.8 111.5
-0.13 0.03 0.16
-0.002 -0.002 -0.007
AX = X•p - Xtn
17. 18.
Maul Moscow
21.0 50.8
268.9 121.1
-0.12 0.24
0.000 -0.000
-43.3
187.2
0.11
-0.013
15.4 56.4
196.3 351.8
0.06 0.21
0.015 0.001
(4)
Theseanomalies(AX) in foF2 and hmF2 werestudied as a function of time, and linear trends estimated from the equation
AX = a + b(Year - 1957)
Long Term Trends in hmF2 and foF2 As stated in the last section,only 31 stationssatisfied the conditionof having more than 30 years of
continuous data. However 4 of thesestations(viz. Canberra, Dourbes,Hobartand Townsville)had someobviouserror in the parameterM(3000)F2. We did not
19.
Mundaring
20. 21.
Okinawa Ottawa
22. 23. 24.
Port Stanley Slough Sodankyl•
-40.5 54.2 63.7
9.8 83.9 120.4
-0.33 -0.38
-0.004 0.004 -0.017
25. 26. 27.
Sverdlovsk Tomsk Townsville
48.5 46.0 -28.7
139.1 160.2 219.9
0.29 0.02 -
-0.006 -0.000 0.004
28.
Uppsala
58.4
106.5
-0.58
-0.016
29. 30.
Wakkanai Yakutsk
35.4 51.1
206.8 194.4
-0.13 -0.17
0.003 -0.016
31. Yamagawa 20.5 198.6 -0.15 0.000 indicate trends in the 4th place of decimal.
0.004
UPADHYAY F2
80
PEAK
AND MAHAJAN-
HEIGHT
LONGTERM
ANOMALY
IONOSPHERIC
TRENDS
3377
anything very definitive. There is some evidenceof a positive trend in hmF2 in the Asian stations, but most
60
of these are below the 95% confidence level and there-
fore one can not derive very definitive conclusionsfrom
-• 20
T y;......t•.•.
_
"-40 E -GO
80 60
I I I I
i
i
these.
It is known that the COa during the period under
considerationincreasedby about 15% [Keelinget al, 1995].If we assumea linearchangein exospheric temperature with increasing CO2 , the expected decrease in thermospheric temperature would be about 10 K. This change in the thermospheric temperature would lower hmF2 by about 5 km and produce a very minor
i
40
20
decreasein foF2. Our analysisof Shimazaki-hmF2sug-
o
-
:
:.
gest that this parameter has not shown any uniform changeover the globesincethe IGY. Of the 26 stations analyzed, the changein hmF2 is below the 95% con-
_
-20 -40
fidence level for 12 stations.
-60 80
i
i
i
i
i
ASH
60
_...
I'qHABAD
For stations
which have
showntrends above 95% confidencelevel, sevenhave exhibited a decreasevarying between 20 km to 6 km in hmF2 while an equal number has shown an increase varying between29 km to 5 km. These large and varying trends in hmF2 are not consistentwith the theoretical modelsof Rishbeth[1990]and Rishbethand Robie [1992]and thus do not provideany definitiveevidence for uniform coolingof the thermospheredue to increase in the greenhousegases.However,thesevarying trends
i
--
RFIT• ..... ,'
u--m -40
I• -60
remain unexplained.
191
YEAR
Figure 1. The F2 layerpeakheightanomaliesAhmF2 showntogetherwith best linear fits, from three selected stations - one showingpositive trend, secondshowing nearly no trend and the third showinga negativetrend.
F- [(1- 72)1(• - •)
The analysisof foF2 (which is directly scaledfrom ionograms)has shownthat out of the 31 station, 17 F2
CRITICAL
FREQUENCY
ANOMALY
JOHANNEsBuRG
3
LINEAR
--
FIT
....
(6)
for eachof the stations. In equation(6) 7 is the correlationcoefficientbetweenAX and (Year- 1957)and n is the number of months included in the investigations. When the F-parameter so derived exceedsgiven significancelevel in the F-distribution, it can then be concludedthat the derived regressioncoefficientis sig-
•
-3 4
3
•
2
nificantlydifferentfrom zero. Panel (a) in Figure 3 showsthe valuesof the F-parameter derived for hrnF2 anomalies
for each station.
The number
0
on the X-axis
is the stationserialnumbergivenin Table- 1 (e.g. 1 for Akita, 2 for Alma Ata .. and soon). The 95% confidencelevel in the Fisher's F-distribution is shownby
,?
-2
•
-3 i
brokenlines. Panel (b) givesthe valuesof the trends (coefficient b) for hrnF2. It canbe notedthat trendsin
i
HUANCAYON EAR
i
FIT
....
hmF2 are statistically significantwith a confidencelevel of 95% or above for 14 stations out of 26 stations. Panel
- .-2
(c) in Figure3 showsthe valuesof the F-parameterderived for foF2 anomalies for the various stations.
Values
of trendsin foF2 are givenin panel(d). It canbe noted that for foF2, trends are statisticallysignificantwith a confidence level of 95% or above for 6 stations out of the 31 stations.
Discussion
We tried to look for any latitudinal, regionalor longitudinal dependenceof these trends but did not find
1955
1965
1975
1985
YEAR
Figure 2. The F2 layer peak frequencyanomalies AfoF2 showntogether with best linear fits, from three selectedstations- one showingpositivetrend, second showingnearlyno trend and the third showinga negative trend.
UPADHYAY AND MAHAJAN' LONGTERM IONOSPHERIC TRENDS
3378
FISltF•'S PARAMETER,LINEAR TRENDS AND 95% CONFIDENCELEVELS
ing the solaractivity effect,trendsin the peak height anomaly(AhmF2) and the criticalfrequencyanomaly (AfoF2) werestudiedfor thesestations. Trendswere negativefor somestations and positivefor other stations. We concludethat ionospheric data do not provideanydefinitiveevidence of a globallongtermtrend.
60.00
_
(a)
FISHER'S ARAMETERF -
40.00
20.00
ß
ß
-
95%
ß
ß
ßß
Acknowledgements: One of us (HOU) held the Senior
ß
- - • - 1.00•[ - - -,ß- - - I - - -o• ©- -.,.L-ø-o ß (b)
ResearchAssociateshipof Council of Scientific and Indus-
trial Research,New Delhi, India duringthe progress of this work. KKM acknowledges fundingfrom Departmentof Sci-
LINEAR TREND IN hmF2 &
ß
ß
ß
0.00
ß
ß
ß
&
ence and Technology,New Delhi on the project "Global AtmosphericScience". Helpful commentsby J. Kar and refereesof this paper are gratefullyacknowledged.
ß
ß
ß
ß
ß
ß
ß
ß
ß ß
-1 .oo
I
I
I
I
References
•
F]SI-•P,.' S ?AEAMETER
15.00
Bilitza, D., N.M. Sheikh,and R. Eyrig, A globalmodel for the height of the F2-peak using M3000 values from the CCIR numericalmap, Telecom.J., 46, 549-
F
10.00
553, 1979.
Bilitza, D., (Ed.), International ReferenceIonosphere 1990, N SSDC 90-22, World Data Center A for Rockets and Satellites, Greenbelt MD, USA, 1990. Bradley, P.A., and J. R. Dudehey,A simplemodel of the
9• ø ß 0.02
-.le
ß
...
__
ß
I..o
*co
-I.
vertical distribution
ß
LINEAR TREND IN foF2
.
1973.
ß
_
&&
ß
ß
AA
AAA
ß
ß
ß
ß
J. Atmos. Terr. Phys., 54, 1501-1511, 1992. J.T. Houghton, L.G. Meira Filho, B. A. Callander, N.
ß
ß
in the
Bremer, J., Ionospherictrendsin mid-latitudesas possibleindicator of the atmosphericgreenhouse effect,
ß
o.oo
of electron concentration
ionosphere,J. Atmos. Terr. Phys., 35, 2131-2146,
ß
ß
Harris,A. Kattenbergand K. Maskell(Eds.), IPCC,
(Intergovernmental Panelon ClimateChange) Cli-
-0.02
5
10
15
ST^TION
Figure 3.
20
25
mate ChangeCambridgeUniversityPress,Cambridge,
30
NU]V•EP,,
1995.
Keeling, C.D., T.P. Whoif, M. Wahien, and J. van der
Fisher's parameter F and linear trends in
Pfiicht, InterannuM extremes in the rate of rise of atmosphericcarbon dioxide since 1980, Nature, 375,
hmF2andfoF2.The brokenlinesin panels(a) and(c)
showFisher'sparameter F for 95% confidencelevel.
666-670, 1995.
Rishbeth,H, A greenhouse effectin the ionosphere ?,
stationsindicate a decreasingtrend while 14 stations indicate an increasingtrend. However,trendsfor 25
Planet. SpaceSci., 38, 945-948,1990.
Rishbeth,H., and R. G. Roble,Coolingof the upper atmosphereby enhancedgreenhouse gases- modellingof thermosphere, Planet. SpaceSci., 40, 1011-
stations were below the 95% confidence level. Results for these 25 stations are consistent with the models of
1026,1992.
Rishbeth[1990]andRishbeth andRoble[1992],which
Roble,R. G., andR. E. Dickinson, Howwill changes in
wouldpredicta very insignificantchangein foF2 due
carbondioxideand methanemodifythe meanstructure of the mesosphere and thermosphere?, Geophys. Res. Lett., 16, 1441-1444,1989.
to 15% increasein CO2. Howeversix of the 31 stations
haveshowna significantchangein foF2 (above95% confidence level)sincethe IGY. Surprisingly onesta-
Shimazaki, T., World-wide variations in the heightof the maximumelectrondensityof the ionospheric F2
tion showedan increasingtrend in foF2 whilethe other
five showeda decreasing trendwhichis muchlarger than that predictedby Rishbeth[1990]and Rishbeth andRobie[1992]. Theselargetrends(bothincreasing and decreasing) thusremainunexplained. Conclusion
Shimazaki-hmF2 was computedfrom monthlymedian ionosonde data for the stations which have con-
tinuous data for more than 30 years. After remov-
layer,J. RadioRes.Labs.Japan,2(7), 85-97,1955. Uli.ch,T., andE. Turunen,Evidence forlong-term cooling of the upperatmosphere in ionosonde data, Geo-
phys. Res. Lett., 24, 1103-1106,1997.
Hari Om Upadhyayand K. K. MahajanNationalPhysical Laboratory, NewDelhi110012,India. (e-mail' hariom@csn pl.ren.nic.in; mahajan @csnpl.ren.nic.in)
(Received February12, 1998;revisedJuly8, 1998; acceptedJuly 16, 1998)