T h e areally averaged change of mean temperature from a reference time
O
to timey,
for a region containing a well-distributed network of stations, can be estimated as(5)
i = l s=1
where dsi is the difference of reported m e a n temperature at station s from time
i-1
to time i, and where ni is the number of stationswith
data sufficient to compute a&.
T h e standard error of estimate of the regional tem- perature difference, given
by
equation(5),
is closely approximatedby
Y
i = l i=2
where ni, 1-1 is the number of stations
in
the region supplying temperature differences between times i- 2
and i
- 1
and also between times i- 1
andi.
Heres(e) is the standard error of the
Cl,i
as estimators of the regional temperature difference, assumed to be quasi- constantwith
time (see Mitchell,1961~).
T h e curve6
in
Figs. 1-3 are based on weighted averages of the series of relative temperaturein
100-latitude bands. Each of the latter series, in turn, has been derivedby
means of equation(5)
applied to five-year (pentadal) m e a n station temperatures, accumulated backwardin
time from the1955-59
pentad(zo). The
relative tempe- rature for each pentady
of the series possesses an error variance givenby
squaring equation(6),
and hereafter denotedby v(Eg)
for brevity. Values of ( T ~-
T ~ ) andof
u(E,)
for annual and winter mean temperaturesin
each 100-latitude band are given back to the1880-84
pentadin
Tables5
and6
respectively.T h e temperature changes shown in Tables
1 (right-
hand column),
2,
and3
are the differences between two temporal averages of curves such as those illustratedin
Figs.1-3. The
standard errors of such change values, also givenin
Tables2
and3,
are derivedby
the equationSE(A) =
[Zwiav(%-Ëb)i]t/.,
(7) where wi is the fraction of the total surface area of the ith constituent loo-band, andv(Ë, - Ëb)
is the error variance of the temperature change in the ith band calculatedby
a procedure noted below, andin
which176
World-wide pattern of secular temperature change
T A B
LE5.
Mean temperature by pentads1, expressed as departures in OF. from 1955-59 pentad, each 100-latitude band from 800N.
to 600S.
Winter mean temperature 0.3 0.1
errors of estimate of temperature change in the consti- tuent bands are assumed to b e mutually independent.
T h e error variance
OP
the digerence between t w o was derived from the equationwhere the summations extend over the
N
pentads involved in each interval being compared, where .(e) is the square of 4 8 ) previously defined,and
where n is the average number of stations whose data are used to compute the (7"-
T ~ ) in these interva1s.l-
where the numerical subscripts apply to the pentad 1. If the intervnls ore not contiguously chosen,-
n is the average niimherof stntiona for the total period of record involved.
177
Clwrngcs of climate
1
Los changomcrtts d e climatnumbers, counted backwards from the
1955-59
pentad (number O).lThe
derivation of the equationsin
this appendix, togetherwith
a more rigorous discussion ofthe
validity of their application, are givenin
[3].S U P P L E M E N T A R Y D A T A
Various supplementary data are
given for
each 100-latitude bandin
Table7.
These include : (a) changes of mean temperature and their associated standard errors as estimatesfor
the data areas, corresponding to the periods of record shown in Tables2
and3;
(b)
values of .(e) andC, defined in
the text, which are178
requised for evaluation of standard errors
w(Ey)
givenby
equation(6) and
of other quantities givenby
equa- tions(2)
and(3) ;
and (c)the
percentage of each latitude band that is land covered, and the percentage of the total earth's surface containedin
each latitude band [proportional tothe
weights zuiin
equations(2),
(3), and(7)l.
1,. Combination OC equations (6) nnd (9) yields
where the n/, 1-1
-
symbols arc as defined under equation (6). W eform of (9) is more uaefui for direct calculation.T A B L E
6 (continued)Temperature change
1
stnudard error of estimate for data areas (OF.)Percentage p ~ ~ { ~ ~
cover =ea of
C land
de)
Band 30-year change 1890-1919 to 1920-49 IO-year change 1940-49 to 1950-59
earth
Changcs ofclimate
/
Les cltangeni.ents de climatACKNOWLEDGMENTS In
addition, he thanks Messrs.H. H.
L a m b andA. I.
Johnson for providing sea level pressure charts T h e writer is indebted to the Staff of the Weather on whichFig. 6
is based. Finally,he
gratefully aclcnow- Bureau National Weather Records Center, Asheville, ledges the assistance ofMi. J. S.
Barry, Mr.W. B.
Buck, North Carolina, whose assistance in locating recent Jr., andMrs. E. B.
Buchanan, in preparing the temperature data for theworld
m a d e this study possible. manuscript for publication.R É S U M É
La
répartition mondiale des variations séculaires de température(J.
Murray Mitchell, Jr.)L’auLeur examine les changements enregistrés depuis
1880
dans la température moyennedu
globe etil
évalue la mesure dans laquelle les données disponibles per- mettent de déterminer ces variations avec exactitude.La
tendance a u réchauffement observée depuis quelques dizaines d’années dans le monde entier paraît s’être renversée peu après1940. Les
températures moyennes semblent avoir évolué de manière parallèle dans les deux hémisphères, mais l’insuffisance desdonnées rela tives aux régions océaniques empêche de déterminer avec précision l’ordre de grandeur de ces changements.
L’auteur expose et compare la répartition géogra- phique des variations annuelles et hivernales enregistrées de
1900
à1960
durant diverses périodes dedix
et de vingt ans. Les variations de la circulation générale sont nettement démontrées c o m m e la cause principale de ces écarts régionaux.Les variations observées dans la température mondiale seraient d’une grande importance pour étudier les causes définitives des changements séculaires
du
climat.D I S C U S S I O N
E. KRAUS.
Perhaps the most interes% part of the evidence presented byDr.
Murray Mitchell,Dr.
Rodewald and some of the other speakers is the way in which it falls into a pattern.Not only air temperature, but also subtropical rainfall, the tendency of hurricanes to move along certain tracks or sea- surface temperatures, show a reversal of the preceding climatic trend during the last one or two decades. The true physical signiñcance of
Dr.
Murray Mitchell’s result lie? perhaps in the combined evidence, based on so many different variables.B.
DZERDZEEVSKII.I think
thatDr.
Mitchell’s paper is very interesting and that the results he obtained are extremely important.I
a m certain that progress can be made on investi- gations into climatic change only by using world-wide data and by analysing the “ world-wide connexions ” among the meteorological observations and circulation characteristics embracing at least one hemisphere.I
also believe that dividing the long-period data into shorter periods is very useful;
but it seems to m e that it is preferable to use the natural periods or epochs ” of different duration.The dividing of climatic data can be fulfilled by analysing the circulation data. Indeed it is not easy. Although such analyses considerably increase the amount of work, the results obtained prove to be more valid.
J. MURRAY
MITCHELL. Dr. Dzerdzeevskii’s remarks are greatly appreciated.With
regard to his suggestion to parti-180
tioning the climatological data according to natural periods,
I
a m very sympathetic to this view and, in a way, this is whatI
have tried to do.I
believe that my curves of world mean temperature (see Fig. 1) might be interpreted as part of a rhythm somewhere between 60 and 90 years in period. Evidence for periods near 80 or 90 years has of course been suspected in the past. It may indeed be worth while to study changes of world-wide climate in and between the various phase8 of such a presumed rhythm. However, if the pre-eminence of such rhythms is ever refuted by new information, as it may well he, w e should,I
think, he willing to abandon this approach without further ado.J. NAMIAS. In
examining Dr. Mitchell’s figures 5(f) and 6(a),I
wouldlike
to point out an important connexion between the pressure distribution and its change and the field of temper- ature change-both for the decades between the 1940s and 1950s. The areas of warming over the United States and Europe whichDr.
Mitchell mentioned and indeed associated with the strong development of the Gulf of Alaska centre action following repeated outburst of cold Alaskan air.It
is well known from my own studies that this situation produces fast westerlies, rapid zonal movements of cyclones along the Canadian border, and flooding of the United States by mild Pacific air-the polar air being “contained” in Canada.The
European warmth is associatedwith
a well-developed Icelandic low, which is in turn associated with cold outflowingWorld-wide pattern of secular temperature change
air masses from the east Canadian tundra.
This
system drives mild Atlantic air masses into Europe. Hence the cooling in northern laiitudes especially over Alaska and the Yukon, and the development of this cold air over the Gulf of Alaska in effect sets up areas of warmth which lessen the degree ofapparent hemispheric or global cooling.
J.
MURRAY MITCHELL.
Dr. Namias’ interpretations are a valuable addendum to my own brief comments on this matter, undI
thank him.C. C. WALLEN.
Like oLher speakersI
wish to express my great interest in and admiration for the contribution given by Dr. Mitchell. There are just two points whichI
would like to raise because they have bearing on my o w n home area.It
was surprising to m e to find that Dr. Mitchell refers to the Scandinavian area as one where pressure has fallen during the recent period of decreasing world temperature indicating rather an increase of temperature. This necessarily must he clue to the selection oí‘ the period of comparison : in this case the forties and the fifties. Already seven years ago, using temper- ature records €rom Sweden,I
pointed out that the so-called“ recent climatic fluctuation ” upwards trend seemed to have changed into a decline.
A
recent study of the differences in temperature and pressure between the periods 1901-30 and 1931-60 shows a definite decline of temperature in Scandinavia coupled with an increase of 9 mh. in pressure in the northern parts. Splitting up the period 1931-60 and looking separately at the forties and fifties evidently means that a slight recovery in tbe overall downwards trend that has occurred in theñfties, in northern and central Europe, seems to bave been given too much emphasis.
I
wish also to state how muchI
agree with Dr. Kraus that this downwards trend in temperature should be considered significant from a physical point of view although it m a y not be so from a statistical point of view. It certainly renders itself extremely well to studies of the relationship between changes in climate elements on one side and fluctuations in the general circulation.In this connexion
I
should wish to mention that in the recent study for Scandinavia mentioned aboveI
have also tried to find some causes for the temperature decline in winter from 1901-30 to 1931-60 in terms of changes in the general circu- lation. Evidence definitely shows that the decline is coupled with an increase of easterly to north easterIy circulation types indicating a shift westwards in the mean position of ridges and troughs. This seems to go well with the explanation for overall decline in temperature during historical periods that has been given earlier by Dr. Lamb.J. MURRAY MITCHELL.
Let m e comment on Dr. Wallén’s first point merely by observing that, in studies of the relation between circulation changes and climate changes, one fre- quently comes to somewhat different conclusions depending on which averaging periods are compared. In many parts of the world, this is probably due to the use of inhomogeneous data, butI
doubt very much that this is the case in Scandinavia.W h e n such discrepancies occix, there seems to be no recourse but to make a more refined analysis of available data, of the exemplary kind for which Dr. Wallén himself is well known.
B I B L I O G R A P H Y / B I B L I O G R A P H I E
CALLENDAR,
G. S.
1961. Quart. J. R. met. Soc., vol. 87, p. 1.- .
1961b. Solar variations, climatic changes, and related LANSBERG, H.E.
; MITCHELL, Jr.,J. M.
Quart.J.
R. met. Soc., geophysical problems, Ann. N.Y.
Acad. Sei., vol. 95, MITCHELL, Jr.,J. M.
1961a. Researchpaper, no. 43. Washington,WILLETT,
H. C. 1950. Centenary proceedings. London, Royalvol. 87, p. 435.
D.C., United States Weather Biireau.
p. 235.
Meteorological Society, p. 195.