It is patently clear that the zonally integrated trends thus far discussed are not representative of trends at all longitudes
within
the zones.In
order to reveal the pattern of changes,the
data for solid-record stations havebeen
differenced between certain10-
and 20-year periods, and these differences mapped as shown inFig. 5
(a$).Figs.
5
(u) and5 (b)
show the 20-year changes of annual-average temperature and winter-season temper- ature, respectively, from1900-19
to1920-39.
These charts are directly comparable to those publishedby
Willett
(1950, his
Figs.5
and6),
but they have been analysed independently. Comparison of the two sets of analyses reveals a number of digerences, primarilyin
areas of poor data coverage. These differences, which give an indication of the degree of subjectivity involvedin
making such analyses, are not, however, sufñcient to obscurethe
salient features of the changefield
evident in both these and Willett's figures.The
greatest changes are seen to have been the warming ofthe
Arctic, parti- cularlyin
the Atlantic sector, and,in
the winter season,TABLE
4. Hypothetical temperature change(OF.)
in no-data areas,Arc,
such that planetary average change estimated by observed change in data areas,A,
is not significantly different from zero;l and evaluation of reasonableness2Thirty-year change 1890-1919 to 1920-49 Ten-year change 1940-49 to 1950-59
Annual Winter Annual Winter
Latitude zone
World
800 N.-60"
S. <
-0.27(U) <
-0.4'6 (U) >-0.09( ?)
>-0.30(R)
600 N.-600S. <
-0.12(U) <
-0.08 (U) >--0.10(R)
>-0.34(R)
Northern Hemisphere800 N.-OO
<
-0.68(U) <
-1.22(U)
>-0.23( ?)
>-o.go(R)
600 N.-Oo
<
-0.35(U) <
-0.28(U)
>-O21(R)
>-0.92(R)
TropicsSouthern Hemisphere
300 N.-300
S. <
+O.ll( ?) <
+O.il( ?)
>-0.09(R)
>-0.19(R)
00-600S. <
+0.243(R) <
+0.39(R)
>-0.21(R)
>-0.25(R)
1. Computed by equation (4) with 1,
=
2.58, corresponding to 99 per cent signiEcance level. Values of A for comparison arc ahown in Tables 2-3.2. R ?
= =
mnrginnlly definitely reasonable reaaonahle{
hence, true plnneiary trend m a y have been zero.U
=
unreasonnblc, hence true planetary trend WEIS non-zero.166
World-wide pattern of secular temperature change
a broad zone of cooling throughout southern Eurasia.
S o m e cooling is also evident over Canada, over m u c h of South America and southward to Antarctic, and over parts of southern and western Africa. Warming extended in a more or less continuous zonal belt between about
200
and400N.,
being interruptedonly
in the Asian sector. This warming took in most of the arid zones of the Northern Hemisphere. T h e proportion of the total earth’s surface which experienced warmingduring
this period was evidently quite large, perhaps as m u c h as85
per cent for the case of annual temperatures.Figs.
5
(c) and5
(d) show the subsequent 20-year changes from1920-39
to1940-59. Although his
period coincidedwith
a very small net change of world mean temperature, as can be inferred fromFig. 1, it
is clear from these figures that m a n y parts of the world were experiencing appreciable net temperature changes at the time. T h e earlier warming o€ the ArcLic had apparently terminated, and a fairly extensive area of cooling also appeared over (or at least surrounding) the southern Indian Ocean. Cooling also developedi n
a meridional band approximately parallel to the principal mountain ranges of the Americas.While
the
Arctic as a whole appears to have cooled since1940
the degree of cooling has not been geogra- phically uniform. T h e greatest cooling appears to have been confuied to northern Siberia,the
west coast of Greenland (probably including Bafñn B a y and the eastern Canadian archipelago), andthe St.
Elias Range- Rocky Mountain region in Alaska and extreme western Canada. Another area of marked cooling included western South Africa, and this m a y have extended southward over a considerable portion of the sub- Antarctic.Fig.
5
(e) indicates that, in all, about80
per cent of the total earth’s surface has probably been involved in the net annual cooling since1940.
Four or five areas of net warming nevertheless stand out in Figs.5
(e) and5(f3.
These include theUnited
States and south- eastern Canada, eastern Europe,the
Pacific coast of Asia, the Brazilian Plateau, and western portions of the Indian Ocean.Comparing the period of net world-wide warming [Figs.
5
(a) and5
(b)]with
the period of net cooling [Figs.5
(e) and5 (J)],
w e are drawn to the following conclusions.First, although, as w e have previously noted, zonally integrated trends over large portions of the earth appear to have been parallel,
the
trends in individual geogra- phical locations have borne very little systematic rela- tionship to these. That is to say, the trendsin
most individual localities are not well correlated-either posi- tively or negatively-with the global average trends.Second, the geographical patterns of change are not predominantly a matter of oscillation of temperature from one time interval to another, as one might expect
if
a simple strengthening and weakening of the general circulation were responsible. Rather, the patterns appear to have changed in more or less independent modes.In order to clarify the extent to which the temperature change fields are related to concomitant changes of the general circulation, the writer has differenced the January and the
July
decadal mean sea level pressure charts for194049
and1950-59,
copies of which were winter temperature change from194049
to1950-59
shown in Fig.5 (f).
F r o m such a comparison, it is evident that most of the areas of strong interdecadal m e a n temperature changes can be attributed with little difficulty to interdecadal m e a n wind-field changes.Specifically, one m a y note the following.
T h e marked winter-season cooling
in
south-eastern Alaska and the Yukon evidently coincidedwith
an equally impressive change of pressure gradient in that areain
January, which favoured an increased local frequency of Arctic air masses from the Canadian interior.T h e warming
in the
north-eastern United States and south-eastern Canada coincidedwith
a weakening of the normal north-westerly gradientwind
from the Hudson B a y region, which permitted a higher incidence of w a r m maritime air masses to reach the area in the1950s.
M u c hthe
same relationship can be noted between the warming and the circulation change in the vicinity of Japan fromthe 1940s
to the1950s.
T h e change of the pressure field across North America, like that across Europe, w a s such as to favour a n inten- sified zonal flow in those areas.
In
the case of North America, this m a y have resulted in increased pseci- pitation and downslope heating consistent with the observed warming over and eastward of the Rocky Mountain Plateauin
theUnited
States. In the case of Europe, the strengthened westerlies probably de- creased the incidence there of polar outbreaks from Siberia.In the Southern Hemisphere, similarly direct rela- tionships between the winter temperature and
July
circulation changes can be noted. However, even rela- tively large pressure changes are insufficient to produce radical changes of the enesgetic zonal circulation of the Southern Hemisphere. Consequently, it is not sur- prising that the temperature changes associated with the pressure changes have been smaller
in
magnitude there and havein
some instances been smeared over broader ranges of longitude than is true ofthe
Northern Hemisphere.167
168
169
O Y
170
172
173
a
B
L74
175
Cltanges of clintute