1
2
3
submitted to Journal of Climate
4
5
6
7
Understanding the U.S. east-west differential
of heat extremes in terms of record
temperatures and the warming hole
8
9
10
11
Gerald A. Meehl1*, Julie M. Arblaster1,2, and Grant Branstator1
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
1. National Center for Atmospheric Research+, Boulder, CO, U.S.A.
2. CAWCR, Bureau of Meteorology, Melbourne, Australia
November 8, 2011
* Corresponding author: meehl@ncar.ucar.edu 303-497-1331
(+) The National Center for Atmospheric Research is sponsored by the National Science
Foundation
38
1
39
Abstract: A linear trend calculated for observed annual mean surface air temperatures over
40
the U.S. for the second half of the 20th century shows a slight cooling over the southeastern part
41
of the country, the so-called “warming hole”, while temperatures over the rest of the country rose
42
significantly. This east-west gradient of average temperature change has contributed to the
43
observed pattern of changes of daily record temperatures as given by the ratio of daily record
44
high temperatures to record low temperatures, which has a comparable east-west gradient.
45
Ensemble averages of 20th century climate simulations show a slight west-east warming gradient,
46
but no warming hole, suggesting that it is not the product of external forcing but is internally
47
generated. A warming hole appears in one ensemble member of simulated 20th century climate
48
in CCSM3, and in this model the warming hole can be produced from internal decadal timescale
49
variability originating mainly from the equatorial central Pacific associated with the Interdecadal
50
Pacific Oscillation (IPO). Analyses of a long control run of the coupled model, and specified
51
convective heating anomaly experiments in an atmosphere-only model, trace the forcing of the
52
warming hole to positive convective heating anomalies in the central equatorial Pacific Ocean
53
near the Dateline. Cold air advection into the southeastern U.S. in winter, and low level moisture
54
convergence in that region in summer, contribute most to the warming hole in those seasons.
55
Projections show a future larger increase in the ratio of daily record high temperatures to record
56
low minimum temperatures in the southeastern U.S. compared to the past 50 years.
57
58
1. Introduction
59
60
As daily record highs have increased relative to record lows across the U.S. over the last
61
several decades, there has been a relatively larger ratio of daily record high temperatures to
2
62
daily record low minimum temperatures in the western part of the U.S. compared to the east
63
during that time period (Meehl et al., 2009a). This ratio has been about a factor of two larger
64
in the west compared to the east over the past decade (using 100°W as the dividing line). A
65
small average warming can produce significant changes in extremes with more record high
66
temperatures and fewer record low temperatures, so daily temperature records are closely
67
tied to changes in average temperature (Anderson and Kostinski, 2010). Thus, to gain
68
insight into the reason for the east-west differential in daily temperature records, we must
69
understand what produces the east-west differential in average warming.
70
71
Observed trends in annual mean temperature in the second half of the 20th century show that
72
warming averaged over the western half of the country outpaced warming in the eastern half
73
by about a factor of two (Fig. 1a). Due to the relationship between changes in average
74
temperature and changes in record temperatures (e.g. Anderson and Kostinski, 2010), the
75
pattern of mean temperature change over the U.S. (greater warming in the west compared to
76
the east) is similar to the differential changes in the ratio of record highs to record lows
77
(Meehl et al., 2009a). Some areas in the southeastern U.S actually show a slight cooling
78
trend of annual mean surface air temperatures during this time period (Fig. 1a), termed the
79
“warming hole” (Pan et al., 2004; Kunkel et al., 2006). This greater average warming in the
80
west compared to the east is reflected in other observed changes of temperature extremes in
81
the second half of the 20th century. For example, there have been proportionately more
82
decreases in observed frost days (defined as daily minimum temperature below freezing) in
83
the west compared to the east (Easterling, 2002). Trends in minimum and maximum
3
84
temperatures (May-June) also show less increase in the eastern U.S. compared to the western
85
U.S., or even a decrease (Portmann et al., 2009).
86
87
To understand the reason for the differential in warming between the eastern and western
88
U.S., this paper explores the factors that could have caused the warming hole in the eastern
89
U.S. in the second half of the 20th century. We first identify an ensemble member from a
90
global coupled climate model simulation of 20th century climate that simulated a warming
91
hole in the second half of the 20th century in the southeastern U.S. (Fig. 1c). We then use
92
that member as a case study to examine aspects of the warming hole in the context of decadal
93
timescale climate variability and response to external forcings, mainly increases in
94
greenhouse gases. Previous studies have shown that both tropical Pacific and North Atlantic
95
SSTs can affect climate anomalies over the U.S. in models and observations (e.g. Robinson
96
et al., 2002; Kunkel et al., 2006; Schubert et al., 2010; Shin and Sardeshmukh, 2010). We
97
will analyze connections of the warming hole to SST patterns in both basins in terms of the
98
processes involved. The dominant role of decadal variability in the tropical Pacific will then
99
be studied for connections to processes that can produce a warming hole. Specified
100
convective heating anomaly experiments will demonstrate the link between SST and
101
precipitation (and associated convective heating) anomalies in the central equatorial Pacific
102
and the warming hole. Finally, the processes that differ with season to produce the warming
103
hole year-round will be described.
104
105
2.
4
Case study of the warming hole in a climate model simulation
106
Ensemble average multi-model simulations of 20th century climate have shown a somewhat
107
greater average warming trend in the western U.S. compared to the east, but with no warming
108
hole (Hegerl et al., 2007). In an analysis of changes in frost days, Meehl et al. (2004) showed
109
the west versus east warming differential in the second half of the 20th century (projected to
110
continue into the 21st century) was associated with a comparable west versus east differential
111
of decreases of frost days (i.e. more decreases in frost days in the west compared to the east
112
as in the observations). That study went on to note that in the simulations of the future in that
113
model, this west versus east difference was produced primarily by greater ridging over the
114
western U.S. compared to the east as the climate warmed, with consequently warmer
115
temperatures in the west compared to the east. This was due in part to the change in base
116
state of the midlatitude circulation consisting of a zonal wave-5 pattern of upper tropospheric
117
circulation that produces relatively greater ridging over the western U.S. (Meehl et al.,
118
2006a; Meehl and Teng, 2007).
119
120
Thus at least part of the greater warming in the western U.S. compared to the east over the
121
second half of the 20th century could be connected to forcing by increases of GHGs (e.g. in
122
CCSM3 by Meehl et al., 2009b). A seven member ensemble average from the CCSM3 (for a
123
description of CCSM3 see Collins et al., 2006) also shows this west-to-east warming
124
differential (Fig. 1b) but does not show the observed relative cooling trend in the
125
southeastern U.S. in the second half of the 20th century, and also shows warming across the
126
entire Pacific and Atlantic (Fig. 1b). In contrast, as noted previously, the observed warming
127
trend from observations for the second half of the 20th century shows a distinct pattern over
128
the oceans, with warming in the eastern tropical Pacific, cooling in the northwest and
5
129
southwest Pacific, warming in the tropical Atlantic and cooling north of about 30N in the
130
northern Atlantic (Fig. 1a).
131
132
This trend pattern of SSTs is similar in other observed SST data sets (not shown) and is
133
thought to have driven the cooling trend (warming hole) over the southeastern U.S. (Fig. 1a).
134
This feature is centered in somewhat different locations in the eastern part of the country
135
depending on the period and season being analyzed (e.g. Kunkel et al., 2006), and these
136
details will be discussed later in this paper. The warming hole previously has been attributed
137
to possible altered hydrologic feedback (Pan et al., 2004; Portmann et al., 2009), increased
138
cloudiness (Robinson et al., 2002), or internally generated decadal timescale variability of
139
either tropical Pacific SSTs, northern Atlantic SSTs, or some combination of both (Robinson
140
et al., 2002; Kunkel et al., 2006; Wang et al., 2009; Shin and Sardeshmukh, 2010).
141
142
The single ensemble member from the CCSM3 set of seven 20th century simulations that was
143
analyzed by Meehl et al. (2009a) shows about a 20% greater warming trend in the west
144
compared to the east for the second half of the 20th century, and a comparable differential in
145
the ratio of daily record high temperatures to record low minimum temperatures. Yet there is
146
no indication of the observed cooling trend over the southeastern U.S. in that model
147
ensemble member.
148
149
However there is another of the CCSM3 20th century ensemble members that does show a
150
warming hole (Fig. 1c) with about a factor of three greater warming in the west compared to
151
the east. By applying the standard relationship between mean temperatures and record
6
152
minimum and maximum temperatures (Anderson and Kostinski, 2010), there is a comparable
153
ratio of record high maxima to record low minima with a somewhat greater than the observed
154
differential of this ratio of about two to one. A comparison of the observed time series of
155
area-averaged surface temperatures over the southeastern U.S. (area average over 24°N-
156
40°N, 70°W-90°W) and this single ensemble member that simulates the warming hole shows
157
in both an increase of temperatures until the early 1950s, then a decline for the next couple of
158
decades, followed by an increase again (Fig. 1d). There is not an exact match in the time
159
evolution of temperatures between this ensemble member and observations in Fig. 1d since
160
the decadal variability is not quite the same. However, both feature temperatures in the late
161
20th century over that area of the country that do not quite return to the mid-century values,
162
thus producing the eastern U.S. cooling trend shown in Fig. 1a,c from 1950-99. It is also
163
worth noting that the warming hole reverts to a positive temperature trend in the future in that
164
figure as will be discussed later.
165
166
The fact that the warming hole is evident in one ensemble member of the CCSM3 20th
167
century simulations is not unique to this model. Kunkel et al (2006) surveyed the CMIP3
168
multimodel ensemble and showed that a few ensemble members from several models
169
simulated a warming hole comparable to observed. This suggests that the warming hole is
170
not completely a product of external forcing (assuming that first order aspects of
171
anthropogenic and natural forcings are represented in the multi-model ensemble), but likely
172
has a contribution from internally generated decadal timescale variability. In fact, comparing
173
the observed surface temperature trends in Fig. 1a with the trends in the single CCSM3
174
ensemble member in Fig. 1c, there are some striking similarities. Both show warming trends
7
175
in the tropical Pacific and northeast Pacific off the west coast of North America, and cooling
176
north of Hawaii and in the North Atlantic north of about 30N. Kunkel et al. (2006) show
177
that, in their definition of the warming hole, it is connected to opposite-sign SST variations in
178
the tropical eastern Pacific and North Atlantic (warm in the tropical Pacific, cool in the North
179
Atlantic off the east coast of the U.S. and north of about 40N). This result has been noted in
180
model simulations that have isolated the relative contributions to precipitation and
181
temperature anomalies over the central U.S. of Pacific and Atlantic SSTs cited previously.
182
183
The most notable decadal timescale feature of temperature variation over the second half of
184
the 20th century was the so-called mid-1970s climate shift when SSTs in the tropical Pacific
185
transitioned from negative to positive on the decadal timescale (Trenberth and Hurrell, 1994).
186
The mid-1970s shift has subsequently been attributed to a combination of forcing from
187
increasing GHGs, and internally generated variability associated with the Interdecadal
188
Pacific Oscillation (IPO, Meehl et al., 2009b). Since trends calculated over the second half
189
of the 20th century would feature the mid-1970s climate shift prominently, it is not surprising
190
that the observed SST trends in Fig. 1a show a pattern in the Pacific that resembles the
191
positive phase of the IPO.
192
193
However, Meehl et al. (2009b) implicated greenhouse gas forcing changes in the mid-1970s
194
as making some contribution to the climate shift, so this source of variability could also play
195
a role in the warming hole phenomenon. Since sulfate aerosol concentrations are greater in
196
the eastern compared to the western U.S., it is possible that there could be a contribution to
197
the warming hole from the response to those aerosols. However, in a model simulation
8
198
forced only by changes in the direct effect of sulfate aerosols, there is almost uniform cooling
199
across the country (Meehl et al., 2004). Thus this source of forcing does not appear to
200
provide the kind of east-west spatial differentiation suggested by the warming hole. There
201
may be aerosol forcings not accounted for in the models that could have contributed to the
202
warming hole (e.g. secondary organic aerosols, Weber et al., 2007). But the fact that some
203
ensemble members from some models produce the warming hole, while other ensemble
204
members using the same forcings from those same models do not, implicates internally-
205
generated decadal timescale variability as at least a significant player in producing the
206
warming hole.
207
208
To examine further the effects of SST forcing versus radiative forcing on simulating the
209
warming hole, we analyze five member AMIP-type experiments with the CAM3 atmospheric
210
component of CCSM3 (Deser & Phillips, 2009). The first set includes only the time-
211
evolving SST forcing (Fig. 2c), while the second set has only the time-evolving radiative
212
forcing with climatological SSTs (Fig. 2d). The radiative forcings are the full set from the
213
CCSM3 20th century simulations and include natural (volcanoes, solar) and anthropogenic
214
(GHGs, ozone, sulfate aerosol direct effect) as described by Meehl et al. (2006b). These are
215
compared to the SST plus radiative forcing experiment (Fig. 2b) and to observations (Fig.
216
2a). The results show that the warming hole can be generated in the runs that include both
217
SST and radiative forcing (Fig. 2b), but only the SST forcing can produce the warming hole
218
as observed (Fig. 2c), while the radiative forcing by itself simulates warming over the entire
219
U.S. with no warming hole (Fig. 2d). This confirms earlier results from studies noted above
9
220
(e.g. Deser and Phillips, 2009, who did not show results for surface temperature, and the
221
analysis was for a somewhat shorter time period).
222
223
The remaining question is what is producing the decadal time scale variations of the SSTs
224
that appear to have forced the warming hole. The candidate for an internally-generated
225
phenomenon proposed by Meehl et al. (2009b) for the mid-1970s climate shift is the IPO.
226
Defined as the first EOF of low-pass filtered (13 year) Pacific Ocean SSTs in a long control
227
run with no changes in external forcings, the surface temperature signature of the IPO in that
228
multi-hundred year control run from CCSM3 is characterized by a warming hole over the
229
eastern part of the U.S. (Fig. 3a). This is calculated by first computing the IPO from the EOF
230
of detrended low-pass filtered SSTs, and then correlating the resulting PC time series with
231
SSTs.
232
233
In the observed trend (Fig. 1a) there are negative SST anomalies that extend into the western
234
subtropical Atlantic and high latitude North Atlantic associated with the southeastern U.S.
235
warming hole. The model IPO surface temperatures match these negative subtropical
236
features but with warming north of about 40°N (Fig. 3a). This suggests that the most
237
consistent SST signal for this model is coming from the tropical Pacific since the IPO signal
238
matches all but the high latitude North Pacific trend from observations.
239
240
To further investigate the relative importance of the tropical Pacific and North Atlantic SSTs
241
in generating the warming hole in the model, we look at the relationships in a long unforced
242
control simulation. We compute 451 overlapping 50-yr trends from a 500 year CCSM3
10
243
control run. By calculating pattern correlation values over the continental U.S. between each
244
of these SST trends and the warming hole SST anomalies in the single 20th century ensemble
245
member (Fig. 1c) and then picking the 42 cases for which the correlations are greater than
246
+0.5, we produce a composite of warming hole-like events generated by internal processes
247
(Fig. 3b). The strongest and most statistically significant (stippling indicates the 5% level)
248
connection for a warming hole trend over the southeastern U.S. is to a warm tropical Pacific,
249
cool northwest Pacific and southeast Pacific, and warm North Atlantic. There is no
250
connection to opposite-sign SST anomalies in the North Atlantic in this model. This
251
suggests that, in this model at least, a warming hole over the southeastern part of the U.S. is
252
primarily produced in association with warmer than normal tropical Pacific SSTs on the
253
decadal timescale.
254
255
To elucidate the mechanism for this connection, an additional experiment is considered in
256
which the CAM3 atmospheric component of CCSM3 is stimulated with a specified heating
257
anomaly centered at the equator and the Dateline. This is near the region where the tropical
258
SST anomalies associated with the warming hole in the southeastern U.S. are most
259
statistically significant in the composite of internally generated warming hole events (Fig.
260
3b). There is also a statistically significant correlation between the warm SSTs there and
261
positive local precipitation anomalies (e.g. the correlation is about +0.55 in this location in
262
the CCSM3 control run). Such positive precipitation anomalies correspond to anomalous
263
tropospheric heating that theory suggests will excite a train of Rossby waves which is
264
generally most pronounced in the winter hemisphere (Webster, 1982). The CAM3, with
265
repeating climatological SSTs, is run for 20 years with a specified heating anomaly with
11
266
maximum value of 5°C day-1. The heat source is 1500km in diameter and centered on the
267
equator. In the vertical it has the profile sin(.5 πσ), where σ is local pressure divided by
268
surface pressure.
269
270
We consider northern winter temperature anomalies for the specified heating runs, as this is
271
when the teleconnection to the equatorial central Pacific is strongest (e.g. Meehl et al.,
272
2006a), as suggested by theory. Fig. 3c shows negative surface air temperature anomalies of
273
about -2°C over the eastern U.S., with positive surface air temperature anomalies in the
274
western U.S. of about +1°C, and even larger positive anomalies across Alaska and northern
275
Canada. This pattern is comparable to the observed warming hole in Fig. 1a, and the one in
276
the single model ensemble member in Fig. 1c. Of course one does not expect a
277
correspondence in surface air temperature over the oceans between these experiments, with
278
specified SSTs, and the coupled experiments.
279
280
The mechanism that connects the convective heating anomaly in the central equatorial
281
Pacific is the anomalous Rossby wave response in the atmosphere (e.g. Hoskins and Karoly,
282
1981), with a barotropic tropospheric trough in the North Pacific, a broad ridge over North
283
America, and another trough in the northern Atlantic (Fig. 3d). These results demonstrate
284
that a warming hole can be generated from convective heating anomalies and the anomalous
285
Rossby wave response in the atmosphere associated with positive precipitation and SST
286
anomalies in the central equatorial Pacific connected to the internally generated IPO (Fig.
287
3a).
288
12
289
The main agent by which the remotely stimulated circulation anomalies influence annual
290
mean surface air temperature over the southeastern US is low level temperature convergence
291
as seen in Fig. 4a for the CCSM3 warming hole member, and Fig. 4b for the specified
292
convective heating anomaly experiment. That is, cold air advection from the northeast as
293
indicated by the vector arrows for low level wind anomalies at 850 hPa in Fig. 4 contribute to
294
surface temperature divergence (negative values in Fig. 4 over the eastern U.S.). Area
295
averages of surface temperature divergence in the area 24°N-40°N, 70°W-90°W are -2.44 x
296
10-5 K s-1 per 50 years for the warming hole CCSM3 member, and -1.67 x 10-5 K s-1 for the
297
specified convective heating anomaly experiment (note different units reflecting different
298
averaging). This pattern is associated with the anomalous ridge-trough pattern over the U.S.
299
connected to convective heating anomalies in the Pacific. This type of teleconnection pattern
300
is strongest in the winter, so the annual mean is influenced most from that season. But to
301
understand the nature of the warming hole, processes in the warm season must also be
302
considered.
303
304
3.
Seasonal results
305
306
Observations in Fig. 5 show that some manifestation of the warming hole is in evidence year
307
round, though, as noted earlier, its strength and position vary with season. The greatest
308
cooling trends over this time period occur in the southeastern U.S. in DJF (Fig. 5a) while in
309
JJA (Fig. 5b) negative temperature trend values spread from that area into the Great Plains.
310
As mentioned earlier, presumably the strongest forcing through anomalous Rossby wave
311
response from SST and convective heating anomalies in the tropical Pacific would be in the
13
312
cold season. Previous studies that showed the connection of the warming hole to tropical
313
Pacific SSTs noted a similar seasonal connection (Robinson et al., 2002; Kunkel et al, 2006;
314
Wang et al, 2009; Shin and Sardeshmukh, 2010). But the warming hole is also present in
315
the warm season, raising the question of whether there are different processes producing the
316
warming hole that are a function of season.
317
318
In the specified convective heating anomaly experiment, there is evidence of cooling over the
319
eastern U.S. in DJF, with the cooling centered more over the Great Plains in JJA (Fig. 6).
320
The 850 hPa streamfunction anomalies in the convective heating anomaly experiment show
321
that the strongest anomalous planetary wave response occurs in DJF (Fig. 7a, same as Fig.
322
3d), but in JJA, the pattern is mostly a tropical quadrupole. This JJA quadrupole, with
323
negative anomalies north of the equator and west of the Dateline and positive anomalies
324
north of the equator and east of the Dateline, and opposite signed anomalies south of the
325
equator, is reminiscent of the combined Rossby wave, Kelvin wave response to equatorial
326
heating found in Gill’s (1980) equatorial beta plane model. In association with this pattern
327
in winter there should be advection of cooler air into the eastern U.S., and in summer there
328
should be convergence of moisture on the north side of the anomalous high over southern
329
North America. Indeed anomalies of 850 hPa temperature convergence in Fig. 8 for the
330
specified convective heating anomaly experiment show, in winter, surface temperature
331
divergence (from advection of cool air) as a major contributor to the warming hole. In
332
summer, the small positive values of low level temperature convergence over the central U.S.
333
should contribute to warming, but there is cooling in Fig. 6b. Something else is going on in
334
summer to contribute to cooling of surface air temperatures. In fact, the pattern of
14
335
anomalous temperature convergence in Fig. 8 coincides with simultaneous convergence of
336
moisture in summer over the central U.S. (not shown) that likely overcomes the impact of the
337
temperature convergence and contributes to increased precipitation and cooling.
338
339
Fig. 9 shows that this positive low level moisture convergence contributes to increased
340
precipitation over most areas of the central U.S. in summer, with concomitant increases of
341
soil moisture, surface evaporation, and increased cloudiness (not shown) that all contribute to
342
cooler surface temperatures in summer. Thus, the specified convective heating anomaly
343
experiment suggests that the contribution from positive tropical Pacific SST anomalies to
344
positive convective heating anomalies produces a different set of responses over the U.S. in
345
cold and warm seasons. In winter, the anomalous Rossby wave response produces a ridge-
346
trough pattern that results in cold air advection from the northeast, surface temperature
347
divergence, and cooler surface air temperatures. In summer, there is more of a Gill-type
348
response with low level moisture convergence over the central U.S., increased precipitation
349
and cloudiness, greater soil moisture and evaporation, and cooler surface air temperatures.
350
351
Now we turn to the CCSM3 20th century simulation where it is hypothesized that the
352
warming hole in that ensemble member is internally generated in connection with positive
353
SST anomalies in the tropical Pacific that resemble the positive phase of the IPO.
354
We intend to use the results from the specified convective heating anomaly experiment
355
described above to help interpret the CCSM3 results.
356
15
357
As in the observations and the convective heating anomaly experiment, there is a warming
358
hole year round in CCSM3 in the eastern U.S., strongest in winter, weakest in summer (Fig.
359
10), though the CAM3 warming hole is a bit farther west than the warming hole in CCSM3.
360
Also, as in the specified convective heating anomaly experiment, 850 hPa height anomalies
361
(not shown, but similar to Fig. 7) produce a pattern of 850 hPa temperature divergence in
362
winter (Fig. 11a), with values averaged over the area 24°N-40°N, 70°W-90°W, of -1.05 x
363
10-4 K s-1 per 50 years, and weak area-averaged temperature convergence in summer over
364
that area of +0.06 x 10-4 K s-1 per 50 years (Fig. 11b). The latter should contribute to
365
warming, but there is surface cooling in summer as seen in Fig. 10. But also as in the
366
specified convective heating anomaly experiment, there is positive low level moisture
367
convergence in summer over the southeastern U.S. (Fig. 12), with an area-averaged value of
368
+0.85 x 10-10 s-1 per 50 years. This should contribute to increases of summer-time
369
precipitation, and indeed that is the case as shown in Fig. 13a with positive precipitation
370
anomalies over the southeastern U.S. amounting to about 15%, along with increases (on the
371
order of 10-20%) of soil moisture (Fig. 13b), and low cloud (Fig. 13c). Enhanced soil
372
moisture increases evaporation and surface cooling, and more low clouds reduce incoming
373
solar radiation. Both then combine to produce the summer-time warming hole.
374
375
4.
Possible future changes
376
377
In Fig. 1d the CCSM3 ensemble member with the warming hole shows a resumption of
378
warming over the southeastern U.S. as the model simulation moves into the 21st century.
379
That ensemble member shows slightly greater overall warming in western U.S. compared to
16
380
the eastern U.S. through the 21st century (Fig. 14a,b), with mostly positive east-west
381
differences later in the century (Fig. 14c). In fact, all the other ensemble members also show
382
greater future warming on average in the western U.S. compared to the east (Fig. 15a,b) with
383
a greater tendency for positive values of area-averaged west minus east temperatures (Fig.
384
15c). This is associated with a consistent wave-5 response to increasing CO2 (Meehl and
385
Teng, 2007), that has a regional wave 3 signature over North America (Teng and Branstator,
386
2011). Such a response pattern to increasing CO2 indicates that the ratio of record highs to
387
record lows will likely remain somewhat greater in the western U.S. compared to the eastern
388
U.S. throughout the 21st century but with internally generated fluctuations in amplitude from
389
decade to decade.
390
391
392
5.
Conclusions
393
394
A prominent feature of U.S. climate trends over the second half of the 20th century was
395
stronger warming in the west than in the east. The most dramatic aspect of this east-west
396
gradient was a complete lack of warming over parts of the eastern U.S., the so-called
397
“warming hole. The east-west gradient produced greater increases of the ratio of daily record
398
high temperatures to record low temperatures in the west compared to the east in about the
399
same proportion as average warming (about two to one). Climate model simulations tend to
400
also produce somewhat greater warming in the western U.S. compared to the east associated
401
with a warmer tropical Pacific mainly in response to increasing GHGs. The contrast between
402
west and east coincides with, and is likely caused by, ridging over the western U.S. that in
17
403
many climate change simulations is associated with a wave-5 pattern of upper troposphere
404
circulation that circles the Northern Hemisphere. However the warming in multi-ensemble
405
member averages in the west due to external forcing is only about 20% greater than the
406
warming in the east, and there is no warming hole. This produces more record high
407
maximum temperatures compared to daily record low minima with a ratio of 5 to 1 over the
408
second half of the 20th century, and that is more than the observed differential of about two to
409
one over that same time period. To understand the observed east-west proportion of the
410
changes of record temperatures, climate model experiments are analyzed to gain insight into
411
what has produced the warming hole.
412
413
We build on previous studies that showed some individual model ensemble members in
414
climate change experiments produced a warming hole. A case study of an individual
415
ensemble member in a CCSM3 20th century experiment that produces a warming hole
416
comparable to observations, in addition to an analysis of a multi-century control run with that
417
model, shows a dominant role of tropical Pacific SST decadal variability associated with the
418
IPO in the generation of the warming hole. A positive phase of the IPO (warmer tropical
419
Pacific SSTs) is connected with the warming hole over the eastern U.S.
420
421
An atmosphere-only experiment with a specified heating anomaly indicates that positive
422
local precipitation and heating anomalies produced by SST anomalies near the equator and
423
Dateline can generate a warming hole over the eastern U.S. not unlike what was observed,
424
and also like that in the CCSM3 single ensemble member. This occurs via an anomalous
425
atmospheric planetary scale wave response to the specified convective heating anomaly.
18
426
427
However, there is an interesting seasonal signature to the warming hole, with different
428
processes for cold and warm seasons, all connected to positive SST and convective heating
429
anomalies in the tropical Pacific. In winter, the midlatitude wave response is primarily by
430
Rossby waves that result in a ridge-trough pattern over the U.S., with cold air advection from
431
the northeast and low level temperature divergence being the main factors that produce
432
cooling in the eastern U.S. However, in summer, the same positive SST and convective
433
heating anomalies produce a Gill-like combined Rossby wave and Kelvin wave response
434
with low level moisture convergence over the southern U.S. that contributes to greater
435
precipitation and cloudiness, increased soil moisture and surface evaporation, and consequent
436
cooler temperatures.
437
438
This does not rule out a role for North Atlantic SSTs in producing a warming hole since, as
439
noted earlier, previous studies have demonstrated a contribution to North American climate
440
from SSTs in that region. Additionally, the single CCSM3 ensemble member that simulated
441
a warming hole had negative SST anomalies in the North Atlantic similar to the observations
442
(comparing Fig. 1a,c). However, the analysis of the model control run trends indicates that
443
the IPO in the Pacific is a dominant contributor to the warming hole, at least in this model.
444
445
Additionally, since the warming hole is mainly a product of internally generated decadal
446
timescale variability, as the IPO transitions back to negative polarity, the warming hole
447
should be replaced by a more rapid rate of warming. For example, in the CCSM3 ensemble
448
member that produces the warming hole, the future evolution of southeastern U.S.
19
449
temperatures in Fig. 1d (following the mid-range SRES A1B scenario) shows little trend
450
from about 2000 to 2020, then a warming jump of nearly 1°C from roughly 2020 to 2025,
451
then nearly no trend again from 2025 to 2045. However, long term forced trends from other
452
ensemble members indicate a preponderance of periods of relatively greater warming over
453
the western U.S. compared to the east. This greenhouse gas forced temperature trend is due
454
in part to a wave-5 circulation response of the midlatitude circulation that favors a ridge over
455
the western U.S. and a trough over the east.
456
457
458
459
460
Acknowledgments
461
462
Portions of this study were supported by the Office of Science, Biological and Environmental
463
Research, U.S. Department of Energy, Cooperative Agreement No. DE-FC02-97ER62402 and
464
grant DE-SC0005355, and the National Science Foundation. We thank Andy Mai for his
465
assistance with accessing the convective heating experiments.
466
467
468
469
470
471
20
472
References
473
474
Anderson, A., and A. Kostinski, 2010: Reversible record breaking and variability: Temperature
475
distributions across the globe. J. Appl. Meteor. and Climatol., 49, 1681—1691,
476
doi:10.1175/2010JAMC2407.1.
477
478
Collins, W.D., and co-authors, 2006: The Community Climate system Model version 3
479
(CCSM3). J. Climate, 19, 2122—2143.
480
481
Deser, C., and A.S. Phillips, 2009: Atmospheric Circulation Trends, 1950-2000: The Relative
482
Roles of Sea Surface Temperature Forcing and Direct Atmospheric Radiative Forcing. J.
483
Climate, 22, 396-413, doi:10.1175/2008JCLI2453.1.
484
485
Easterling DR, 2002: Recent changes in frost days and the frost-free season in the United States.
486
Bull. Amer. Meteorol. Soc. 83, 1327-1332
487
488
Gill, A.E., 1980: Some simple solutions for heat induced tropical circulation. Quart. J. Roy.
489
Meteor. Soc., 106, 447—462.
490
491
Hegerl, G. and co-authors, 2007: Understanding and Attributing Climate Change. In: Climate
492
Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth
493
Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M.
21
494
Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge
495
University Press, Cambridge, United Kingdom and New York, NY, USA, 663—745.
496
497
Hoskins, B.J., and D.J. Karoly, 1981: The steady linear response of a spherical atmosphere to
498
thermal and orographic forcing. J. Atmos. Sci., 38, 1179—1196.
499
500
Hoerling, M., A.Kumar, J. Eischeid, and B.Jha, 2008: What is causing the variability in global
501
mean land temperature. Geophys. Res. Lett., 35, doi:10.1029/2008GL035984.
502
503
Kunkel, K.E., X.-Z. Liang, J. Zhu, Y. Lin, 2006: Can CGCMs simulate the Twentieth-Century
504
“Warming Hole” in the Central United States? J. Climate, 19, 4137-4153.
505
506
Meehl, G. A., C. Tebaldi, and D. Nychka, 2004: Changes in frost days in simulations of 21st
507
century climate. Climate Dynamics, 23, 495–511. doi: 10.1007/s00382-004-0442-9.
508
509
Meehl, G. A., W. M. Washington, C. Amman, J. M. Arblaster, T. M. L. Wigley, and C. Tebaldi,
510
2004: Combinations of natural and anthropogenic forcings and 20th century climate. Journal of
511
Climate, 17, 3721–3727.
512
513
Meehl, G. A., H. Teng, and G. W. Branstator, 2006a: Future changes of El Niño in two global
514
coupled climate models. Climate Dynamics, 26, 549—566, doi: 10.1007/s00382-005-0098-0.
515
22
516
Meehl, G. A., W. M. Washington, B. D. Santer, W. D. Collins, J. M. Arblaster, A. Hu, D.
517
Lawrence, H. Teng, L. E. Buja, and W. G. Strand, 2006b: Climate change projections for
518
twenty-first century and climate change commitment in the CCSM3. Journal of Climate, 19,
519
2597–2616.
520
521
Meehl, G.A. and H. Teng, 2007: Multi-model changes in El Niño teleconnections over North
522
America in a future warmer climate. Clim. Dyn., 29, 779–790, DOI 10.1007/s00382-007-0268-3
523
524
Meehl, G.A., C. Tebaldi, G. Walton, D. Easterling, and L. McDaniel, 2009a: The relative
525
increase of record high maximum temperatures compared to record low minimum temperatures
526
in the U.S. Geophys. Res. Lett., 36, L23701, doi:10.1029/2009GL040736.
527
528
Meehl, G. A., A. Hu, and B.D. Santer, 2009b: The mid-1970s climate shift in the Pacific and the
529
relative roles of forced versus inherent decadal variability, J. Climate, 22, 780--792.
530
531
Pan, Z., R.W. Arritt, E.S. Takle, W.J. Gutowski Jr., C.J Anderson, and M. Sega., 2004: Altered
532
hydrologic feedback in a warming climate introduces a “warming hole”. Geophys. Res. Lett., 31,
533
L17109, doi:10.1029/2004GL020528.
534
535
Portmann, R.W., S. Solomon, and G.C. Hegerl, 2009: Spatial and seasonal patterns in climate
536
change, temperatures, and precipitation across the United States. Proc. Nat. Acad. Sci., 106,
537
7324—7329, doi:10.1073/pnas.0808533106.
538
23
539
Robinson, W.A., R. Reudy and J.E. Hansen, 2002: General circulation model simulations of
540
recent cooling in the east-central United States. J. Geophys. Res., 107, D24, 4748,
541
doi:10.1029/2001JD001577.
542
543
Schubert S., and Coauthors, 2009: A U.S. CLIVAR project to assess and compare the
544
responses of global climate models to drought-related SST forcing patterns: Overview and
545
results. J. Climate, 22, 5251—5272.
546
547
Shin, S.-I., and P.D. Sardeshmukh, 2010: Critical influence of the pattern of tropical ocean
548
warming on remote climate trends. Clim. Dyn., DOI:10.1007/s00382-009-0732-3.
549
550
Teng, H., and G. Branstator, 2011: A zonal wavenumber-3 pattern of the northern winter
551
planetary waves at high latitudes, J. Climate, submitted.
552
553
Trenberth, K.E., and J.W. Hurrell, 1994: Decadal atmosphere-ocean variations in the Pacific.
554
Clim. Dyn., 9, 303—319.
555
556
Wang, H., S. Schubert, M. Suarez, J. Chen, M. Hoerling, A. Kumar, P. Pegion, 2009: Attribution
557
of the seasonality and regionality in climate trends over the United States during 1950-2000. J.
558
Climate, 22, 2571-2590.
559
560
Wang, H., S. Schubert, M. Suarez, and R. Koster, 2010: The physical mechanisms by which the
561
leading patterns of SST variability impact U.S. precipitation, J. Climate, 23, 1815-1836
24
562
doi: 10.1175/2009JCLI3188.1
563
564
Weber, R.J. and co-authors, 2007: A study of secondary organic aerosol formation in the
565
anthropogenic-influenced southeastern United States. J. Geophys. Res.,
566
doi:10.1029/2007JD008404.
567
568
Webster, P.J., 1982: Seasonality of atmospheric response to sea-surface temperature anomalies.
569
J. Atmos. Sci., 39, 29-40.
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
25
585
Figure captions
586
587
Fig. 1: a) Observed surface air temperature trend, 1950-99, from the HadCRUT3 dataset; b)
588
same as (a) except for the seven member ensemble average from CCSM3; c) same as (a) except
589
for the single CCSM3 ensemble member with a warming hole; d) surface air temperature
590
anomalies averaged over the southeastern U.S. land points (24°N-40°N, 70°W-90°W), for the
591
CCSM3 ensemble member in (c), red dashed line, and for observations in (a), black solid line
592
(differences compared to the 1961-1990 base period).
593
594
Fig. 2: a) Observed SST trends for 1950-99 from the HadISST data, b) same as (a) except for
595
AMIP-type runs with time-varying observed SSTs, and anthropogenic and natural forcings; c)
596
same as (b) except model run using observed SSTs with time-invariant anthropogenic and
597
natural forcings; d) same as (b) except for model run with climatological average SSTs and
598
time-varying anthropogenic and natural forcings.
599
600
Fig. 3: a) The PC time series from the first EOF of low pass filtered SSTs in a multi-hundred
601
year control run (the IPO) correlated with surface air temperatures in the control run; b) taking
602
451 overlapping 50-yr trends from the 500 year CCSM3 control run, pattern correlation values
603
are calculated over the continental U.S. between each of these SST trends and the warming hole
604
SST anomalies in the single 20th century ensemble member (Fig. 1c), then picking the 42 cases
605
for which the correlations are greater than +0.5, this composite of warming hole-like events is
606
formed; c) for a convective heating anomaly at equator, Dateline (green circle), surface air
26
607
temperature anomalies for the DJF season, red are positive values, blue are negative; d) same as
608
(c) except for 850 hPa streamfunction (m2 s-1).
609
610
Fig. 4: a) 850 hPa annual surface temperature convergence trends (K sec-1/50yr) for the CCSM3
611
warming hole member, positive values (yellow and red) denote anomalous convergence,
612
negative values (blue) are anomalous divergence, vectors denote 850 hPa wind trends (m sec-1
613
/50 yr), scaling arrow at lower right; b) same as (a) except for 20 year average annual anomalies
614
in the specified convective heating anomaly experiment (K sec-1).
615
616
Fig. 5: a) same as Fig. 1a except for DJF; b) same as Fig. 1a except for JJA
617
618
Fig. 6: a) same as Fig. 3c but only for the U.S.; b) same as (a) except for JJA
619
620
Fig. 7: a) same as Fig. 3d; b) same as (a) except for JJA
621
622
Fig. 8: a) same as Fig. 4b except for DJF; b) same as (a) except for JJA
623
624
Fig. 9: same as Fig. 8b except for precipitation change (%)
625
626
Fig. 10: a) same as Fig. 1c except for DJF; b) same as (a) except for JJA
627
628
Fig. 11: a) same as Fig. 4a except for DJF; b) same as (a) except for JJA
629
27
630
Fig. 12: a) same as Fig. 11a except for moisture convergence (s-1/50 yrs); b) same as (a) except
631
for JJA
632
633
Fig. 13: a) same as Fig. 10b except for precipitation (%/50 yrs); b) same as (a) except for soil
634
moisture (kg m-2/50 yrs); c) same as (a) except for low cloud (%/50 yrs)
635
636
Fig. 14: a) same as Fig. 1d except for area average over the western U.S. (24°N-48°N, 127°W-
637
100°W); same as (a) except for eastern U.S. (24°N-48°N, 62°W-100°W); c) difference of (a)
638
minus (b)
639
640
Fig. 15: a) annual mean surface air temperature anomalies (°C) for the western U.S. area in Fig.
641
14 (compared to the 1961-1990 base period), the five CCSM3 20th century ensemble members
642
with corresponding 21st century A1B scenario simulations, the warming hole ensemble member
643
is red line, observations are black line; b) same as (a) except for eastern U.S. area; c)
644
differences, west minus east (a minus b).
645
646
647
648
649
650
651
652
28
653
654
655
656
657
658
659
660
Fig. 1: a) Observed surface air temperature trend, 1950-99, from the HadCRUT3 dataset; b)
same as (a) except for the seven member ensemble average from CCSM3; c) same as (a) except
for the single CCSM3 ensemble member with a warming hole; d) surface air temperature
anomalies averaged over the southeastern U.S. land points (24°N-40°N, 70°W-90°W), for the
CCSM3 ensemble member in (c), red dashed line, and for observations in (a), black solid line
(differences compared to the 1961-1990 base period).
29
661
662
663
664
665
666
667
668
Fig. 2: a) Observed SST trends for 1950-99 from the HadISST data, b) same as (a) except for
AMIP-type runs with time-varying observed SSTs, and anthropogenic and natural forcings; c)
same as (b) except model run using observed SSTs with time-invariant anthropogenic and
natural forcings; d) same as (b) except for model run with climatological average SSTs and
time-varying anthropogenic and natural forcings.
669
670
671
672
673
30
674
675
676
677
678
679
680
681
682
683
684
685
Fig. 3: a) The PC time series from the first EOF of low pass filtered SSTs in a multi-hundred
year control run (the IPO) correlated with surface air temperatures in the control run; b) taking
451 overlapping 50-yr trends from the 500 year CCSM3 control run, pattern correlation values
are calculated over the continental U.S. between each of these SST trends and the warming hole
SST anomalies in the single 20th century ensemble member (Fig. 1c), then picking the 42 cases
for which the correlations are greater than +0.5, this composite of warming hole-like events is
formed; c) for a convective heating anomaly at equator, Dateline (green circle), surface air
temperature anomalies for the DJF season, red are positive values, blue are negative; d) same as
(c) except for 850 hPa streamfunction (m2 s-1).
686
687
688
689
690
31
691
692
693
694
695
696
697
698
699
700
Fig. 4: a) 850 hPa annual surface temperature convergence trends (K sec-1/50yr) for the CCSM3
warming hole member, positive values (yellow and red) denote anomalous convergence,
negative values (blue) are anomalous divergence, vectors denote 850 hPa wind trends (m sec-1
/50 yr), scaling arrow at lower right; b) same as (a) except for 20 year average annual anomalies
in the specified convective heating anomaly experiment (K sec-1).
701
702
703
704
705
706
707
708
709
710
711
712
32
713
714
715
716
Fig. 5: a) same as Fig. 1a except for DJF; b) same as Fig. 1a except for JJA
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
33
733
734
735
736
737
Fig. 6: a) same as Fig. 3c but only for the U.S.; b) same as (a) except for JJA
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
34
753
754
755
756
757
758
759
Fig. 7: a) same as Fig. 3d; b) same as (a) except for JJA
760
761
762
763
764
765
766
767
768
769
770
771
35
772
773
774
775
776
777
Fig. 8: a) same as Fig. 4b except for DJF; b) same as (a) except for JJA
778
779
780
781
782
783
784
785
786
787
788
789
790
791
36
792
793
794
795
796
797
Fig. 9: same as Fig. 8b except for precipitation change (%)
798
799
800
801
802
803
804
805
806
807
37
808
809
810
811
812
813
814
Fig. 10: a) same as Fig. 1c except for DJF; b) same as (a) except for JJA
815
816
817
818
819
820
821
822
823
824
825
826
827
38
828
829
830
831
832
833
Fig. 11: a) same as Fig. 4a except for DJF; b) same as (a) except for JJA
834
835
836
837
838
839
840
841
842
843
844
845
846
847
39
848
849
850
851
852
853
854
855
Fig. 12: a) same as Fig. 11a except for moisture convergence (s-1/50 yrs); b) same as (a) except
for JJA
856
857
858
859
860
861
862
863
864
865
866
867
868
40
869
870
871
872
873
874
Fig. 13: a) same as Fig. 10b except for precipitation (%/50 yrs); b) same as (a) except for soil
moisture (kg m-2/50 yrs); c) same as (a) except for low cloud (%/50 yrs)
875
876
41
877
878
879
880
881
882
Fig. 14: a) same as Fig. 1d except for area average over the western U.S. (24°N-48°N, 127°W100°W); same as (a) except for eastern U.S. (24°N-48°N, 62°W-100°W); c) difference of (a)
minus (b)
883
884
885
886
887
888
889
890
42
891
892
893
894
895
896
897
898
899
Fig. 15: a) annual mean surface air temperature anomalies (°C) for the western U.S. area in
Fig. 14 (compared to the 1961-1990 base period), the five CCSM3 20th century ensemble
members with corresponding 21st century A1B scenario simulations, the warming hole ensemble
member is red line, observations are black line; b) same as (a) except for eastern U.S. area; c)
differences, west minus east (a minus b).
900
901
902
903
904
905
43