? .Global and Planetary Change 26 2000 445?465
www.elsevier.comrlocatergloplacha
Climate sensitivity to changes in land surface characteristics
Jacob O. Sewall a,), Lisa Cirbus Sloan a, Matthew Huber a, Scott Wing b
a Department of Earth Sciences, Uni?ersity of California, Santa Cruz, CA 95064, USA
b Department of Paleobiology, Smithsonian Institute, Washington, DC, USA
Received 10 December 1999; accepted 19 June 2000
Abstract
Using a recently developed global vegetation distribution, topography, and shorelines for the Early Eocene in conjunction
with the Genesis version 2.0 climate model, we investigate the influences that these new boundary conditions have on global
climate. Global mean climate changes little in response to the subtle changes we made; differences in mean annual and
seasonal surface temperatures over northern and southern hemispheric land, respectively, are on the order of 0.58C. In
contrast, and perhaps more importantly, continental scale climate exhibits significant responses. Increased peak elevations
and topographic detail result in larger amplitude planetary ;4 mmrday and decreases by 7?9 mmrday in the proto
Himalayan region. Surface temperatures change by up to 188C as a direct result of elevation modifications. Increased leaf
? .area index LAI , as a result of altered vegetation distributions, reduces temperatures by up to 68C. Decreasing the size of the
Mississippi embayment decreases inland precipitation by 1?2 mmrday. These climate responses to increased accuracy in
boundary conditions indicate that AimprovedB boundary conditions may play an important role in producing modeled
paleoclimates that approach the proxy data more closely. q 2000 Elsevier Science B.V. All rights reserved.
Keywords: Cenozoic; paleoclimate; climate modeling; land surface; North America
1. Introduction
An ongoing goal for the paleoclimate modeling
community is to reproduce global paleoclimate as
accurately as possible. General circulation models
? .GCMs are a commonly used tool in these endeav-
ors. In order to assess the accuracy of GCMs, results
from paleoclimate modeling experiments are often
compared to the only available means of validation:
climate estimates derived from proxy climate data.
) Corresponding author. Tel.: q1-831-459-3504; fax: q1-831-
459-3074.
? .E-mail address: jsewall@es.ucsc.edu J.O. Sewall .
Physical, biological, and geochemical proxy data are
frequently derived from limited localities, possibly
representing areas as small as several square meters.
Consequently, the information derived from proxy
data reflects local and regional climate conditions in
addition to the broader global climate overprint.
Fossils, the most common source of continental
paleoclimate proxy data, are found in sedimentary
rocks that formed predominantly in coastal lowlands
and interior basins. Large-scale climatic conditions
often differ from the prevailing regional conditions,
especially in interior basins. Consequently, proxy
data record local land surface characteristics as well
?as regional and global climates Greenwood, 1992;
0921-8181r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
? .PII: S0921-8181 00 00056-4
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465446
.Demko et al., 1998 . For example, the species com-
position and structure of vegetation adjacent to
streams and lakes is often markedly different from
that found even a few hundred meters away from the
water body. Vegetation close to fluvial or lacustrine
depositional environments is more likely to be fos-
silized, and, therefore, paleoclimatic proxy data are
likely to represent very local, in addition to regional
or global, climate conditions.
This inherent bias in the proxy data makes com-
parisons to results from global-scale models difficult.
Improved comparisons between model output and
proxy data can be made by downscaling from the
resolution of model output to the resolution of proxy
data or through the application of regional climate
?models e.g., Wilby et al., 1998; Sailor and Li, 1999;
.Reichert et al., 1999; Gebka et al., 1999 . Unfortu-
nately, the application of downscaling and regional
modeling to paleoclimate investigations is in its in-
fancy. Before either of these methods is attempted, it
is important to have an accurate global climate simu-
lation for the time interval in question ? one that
reproduces regional climates, as indicated by the
proxy record, as accurately as possible. We can then
either interpolate down from this global climate or
use it to force a regional-scale model. With the
objective of improving global paleoclimate simula-
tions, we investigate the effects of land surface
changes on the climate produced by a GCM. Specifi-
cally, we explore the influence that a more accurate
global vegetation distribution and a more realistic
?topography individual peaks and valleys, narrow
mountain ranges, overall increase in topographic de-
.tail have on model-produced climate.
An additional motivation for this work lies in a
more historical context. In recent years, the spatial
resolutions available in GCMs have increased greatly.
However, it is not clear that researchers are taking
full advantage of these resolution increases. When a
new, higher resolution model version becomes avail-
able, a common practice is to interpolate old, low-
resolution boundary conditions to new, higher resolu-
tions.
For example, with the advent of Genesis version
2.0, researchers moved from an atmospheric and
?land surface resolution of spectral R15 ;4.58 lati-
. ? .tude=7.58 longitude e.g., Sloan and Rea, 1995 to
?an atmospheric resolution of spectral T31 ;3.758
.latitude=3.758 longitude and a land surface resolu-
tion of 28 latitude=28 longitude. When this resolu-
tion increase occurred, some researchers interpolated
their R15 boundary conditions to the new, higher
? .resolutions e.g., Sloan and Pollard, 1998 .
Topography at a low resolution, such as spectral
resolution R15 or coarser, is highly smoothed. Indi-
vidual peaks, valleys, and narrow mountain ranges
? .are not resolved Fig. 1A . Interpolating low resolu-
?tion topography to a higher resolution e.g., R15
topography interpolated to a resolution of 28 latitude
.=28 longitude generates the same highly smoothed
topography, simply composed of more grid cells
? .Fig. 1B . It is possible that highly smoothed topog-
raphy is a realistic enough boundary condition for
simulations of global climate; however, it is equally
possible that more realistic topography will influence
model-produced climate in a way that produces a
more accurate global paleoclimate.
There have been many previous studies that ex-
amined the impact of vegetation upon past climates
?e.g., Bonan et al., 1992; Henderson-Sellers et al.,
.1993; Foley et al., 1994; Crowley and Baum, 1997 .
These studies were designed to gain a first-order
estimate of the impact of a specific vegetation change
upon climate. Our study focuses on the change in
model-produced climate that occurs when the Eocene
vegetation distribution is modified more subtly, based
on our evolving knowledge of that distribution, ra-
ther than an investigation of extreme vegetation
change.
Previous studies of the effect of topographic vari-
ation on climate changed topography globally by
? .fixed amounts e.g., Kutzbach and Gallimore, 1989 ,
or changed elevations only for mountains that were
?fixed in location between the model cases e.g.,
Sloan and Barron, 1992; Broccoli and Manabe, 1992;
.Kutzbach et al., 1993 . In contrast, we explore the
influence that increased topographic detail has on
model-produced climate. Our global mean elevation
varies little between cases, but maximum and mini-
mum elevations increase and decrease, respectively.
?The locations of specific topographic features e.g.,
.highest and lowest points change somewhat be-
tween cases. While the aforementioned studies fo-
cused on the effects that mountain evolution has on
climate, our study is motivated by the question of
accuracy in the portrayal of those mountains.
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465 447
? .Fig. 1. A Old Eocene topography at spectral resolution R15
? .;4.58 latitude=7.58 longitude , contours 0?2000 m at 100 m.
? .B Old Eocene topography at a resolution of 28 latitude=28
? .longitude, contours 0?2000 m at 100 m. C New Eocene topogra-
phy and shorelines at a resolution of 28 latitude=28 longitude,
contours from 3800 to 600 m at 800 m, from 600 to 100 m at 250
m, from 100 to 10 m at 30 m, and 0 m.
2. Methods
Based on published elevation estimates, descrip-
tions of regional tectonics, and geologic maps, we
created new topographic and shoreline boundary
conditions for the Eocene at a resolution of 28=28
? .Fig. 1C . We did this to include more detailed and
realistic boundary conditions in our models. The old
topographic and shoreline boundary conditions were
a simple rendition of Early Eocene paleogeography
that incorporated broad assumptions about locations
of continental highlands and lowlands. The new
topography was created region by region. First, ele-
vation estimates and tectonic histories of prominent
?features were gathered e.g., Molnar and Tapponier,
1975; Tapponier and Molnar, 1979; Plaziat, 1981;
Molnar et al., 1987; Taylor et al., 1990; Roehler,
1993; Fitzgerald, 1994; Maxson and Tikoff, 1996;
Allmendinger et al., 1997; Lamb and Hoke, 1997;
.Chorowicz et al., 1998; Gurnis et al., 1998 . Then,
based on knowledge of regional tectonics and possi-
ble maximum and minimum elevations, the shape
and extent of mountain belts were created. Working
from the major mountain belts and the tectonic
framework of a region, topography was hand
smoothed to sea level. For example, passive conti-
nental margins were given broader coastal plains
than active continental margins. Lithospheric plate
? .positions are modified from Scotese et al. 1988 .
Continental shorelines are based partially on global
?tectonics e.g., Greenland had not rifted from Scan-
? ..dinavia Scotese et al., 1988 and partially on the
location of contacts between Paleocene and Eocene
sedimentary rocks as presented on various geologic
maps.
In conjunction with the creation of a more realis-
tic Eocene topography and orography, a more accu-
rate global vegetation distribution was created. The
vegetation distribution used in previous Eocene stud-
ies was constructed nearly 10 years ago and was
based on limited paleobotanical information. The
new Eocene global vegetation distributions were
based on a larger database of macro- and microflora
?for the Early Eocene e.g., Wing, 1998a,b; Rull,
.1999; Wilf, 2000 . In addition, the inferred influence
of reduced latitudinal temperature gradients on global
vegetation distributions was considered in this recon-
struction.
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465448
In a trio of sensitivity studies, we investigate the
effects of a more accurate vegetation distribution,
more realistic, higher resolution topography, and
changes in shorelines on Early Eocene climate.
3. Model
The model used in these studies is the Genesis
?climate model version 2.0 Pollard and Thompson,
.1995; Thompson and Pollard, 1997 . This model
reproduces present-day climate as well as other
?GCMs currently in use Thompson and Pollard,
.1997 . The model contains a diurnal cycle and a full
seasonal solar cycle. The atmospheric component of
the model was run at a resolution of spectral T31
? .;3.758 latitude=3.758 longitude with 18 vertical
levels. The atmospheric component is coupled to a
?land surface with a resolution of 28=28 Pollard and
.Thompson, 1995 . The vegetation was described us-
? .ing the 12 categories of Dorman and Sellers 1989 .
Clouds are predicted using prognostic three-dimen-
sional water cloud amounts. Mixing ratios for the
greenhouse gases CO , CH , N O, CFC , and2 4 2 11
CFC can be prescribed in the model, and the12
infrared radiation component of Genesis version 2.0
explicitly models the effects of greenhouse gases.
? .Sea surface temperatures SSTs are prescribed.
4. Description of experiments
The control case for the trio of sensitivity studies,
OLDVT, contains Eocene topography, shorelines,
and vegetation distributions as defined in previous
?research efforts boundary conditions are described
.in Sloan and Rea, 1995 . Atmospheric carbon diox-
ide concentration was set at 300 ppm. Atmospheric
CH and N O concentrations were set at pre-in-4 2
? .dustrial levels 0.700 and 0.285 ppm, respectively .
The prescribed solar constant is the modern value of
1365 Wrm2. Soil is specified at the present global
? .average 43% sand, 39% silt, 18% clay . The SSTs
are zonally constant, seasonally varying SSTs that
are characteristic of the warmest Cenozoic intervals.
SST values were derived by fitting an energy bal-
ance model to a data set of Early Eocene SST
interpretations and producing an annual average SST
?gradient and monthly zonal SST gradients see full
.description and Fig. 1 in Sloan et al., in review . We
use zonally constant SSTs because there are insuffi-
cient paleo-SST interpretations to allow characteriza-
tion of longitudinal gradients in SST. The prescribed
SSTs prohibit the development of El NinorSouthern
Oscillation and other oceanic variability. Case
OLDVT was started from the 16th year of an equili-
?brated Eocene run control case of Sloan and Pollard,
.1998 and run for 10 years.
The first sensitivity study focuses on the effect
that a more accurate global vegetation distribution
has on model-produced climate. This case is desig-
nated NEWVOLDT. With the exception of the new
? .vegetation distribution Fig. 2B , boundary condi-
tions for NEWVOLDT are identical to those of
OLDVT. NEWVOLDT was started from year 8 of
OLDVT and run for 8 years.
The second sensitivity study investigates the ef-
fects of a more accurate, higher resolution topogra-
? .phy and orography Fig. 1C . This case is designated
NEWVT. NEWVT uses the new topography and
shorelines; all other boundary conditions are identi-
cal to NEWVOLDT. NEWVT was started from year
8 of OLDVT and run for 8 years. For each case, the
final 3 years of results were averaged for analyses.
The third sensitivity study examines climate sensi-
tivity of interior North America to a relatively small
change in shoreline definition of the Mississippi
embayment. This case is explained more fully below.
5. Results
5.1. Effect of ?egetation changes upon Eocene cli-
mate
Because the only change between the
NEWVOLDT and OLDVT cases is the inclusion of
the new vegetation distribution, differences between
the results of these two cases provide an assessment
of the importance of vegetation specifications in
modeling experiments. The vegetation distributions
? .are quite different from each other Fig. 2 , however,
the change in distributions has no significant effects
on the mean global climate state. In December,
? .January, and February DJF , temperatures in the
NEWVOLDT case are 0.518C cooler over all land;
Northern Hemisphere land is 0.188C cooler and
Southern Hemisphere land is 0.928C cooler. In June,
? .July, and August JJA , the NEWVOLDT case is
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465 449
? . ? .Fig. 2. A Old Eocene global vegetation distribution. B New Eocene global vegetation distribution. Vegetation categories are after
? .Dorman and Sellers 1989 .
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465450
0.648C cooler over global land; Northern Hemi-
sphere land is 0.868C cooler while Southern Hemi-
sphere land is 0.358C cooler. Although the new
vegetation distribution generates small responses in
the global mean state, there are significant regional
effects.
5.1.1. Surface temperatures
DJF surface temperatures decrease by 48C over
northern coastal Australia and regions of central
South America and by 68C over south, central Africa
? .Fig. 3A . In the NEWVOLDT case, surface temper-
atures in JJA are cooler by 48C over central North
? . ? .Fig. 3. A DJF surface temperature difference, NEWVOLDT minus OLDTV, contours from y88C to 48C at 28C, no zero contour. B JJA
? .surface temperature difference, NEWVOLDT minus OLDTV, contours from y88C to 48C at 28C, no zero contour. C DJF total clouds
? .difference, NEWVOLDT minus OLDTV, contours from y0.2 to 0.3 fraction at 0.1 fraction, no zero contour. D JJA total clouds
difference, NEWVOLDT minus OLDTV, contours from y0.2 to 0.3 fraction at 0.1 fraction, no zero contour. All shaded differences shown
?on this plot are significant to the 99% level. Significance was calculated using Chervin and Schneider?s t-test Chervin and Schneider,
.1976a,b .
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465 451
America and central Australia and by 68C over a
? .large area of central Asia Fig. 3B .
5.1.2. Total clouds
In DJF, the new vegetation generates a 20?30%
increase in total cloudiness over central South Amer-
ica and an increase of 25% over south central Africa
? .Fig. 3C . Total cloudiness in the JJA NEWVOLDT
case increases by 20?40% over central Asia and by
up to 25% over north central South America and
? .Europe Fig. 3D .
5.1.3. E?apotranspiration
With the new vegetation, DJF evapotranspiration
decreases by 2 mmrday over coastal regions of
Asia. DJF evapotranspiration increases by 1?2 mmr
day over central South America and north coastal
Australia. Evapotranspiration increases over southern
Africa and Southeast Asia for DJF are 2 mmrday.
Decreases in JJA evapotranspiration over the north
coast of Africa are 1?2 mmrday with the new
vegetation. An increase in JJA evapotranspiration of
1?2 mmrday is seen over central North America
and evapotranspiration over Asia and central Africa
increases by 2 mmrday.
5.1.4. Latent heating
In DJF, the new vegetation results in latent heat-
ing increases of up to 55 Wrm2 over most of South
? .America Fig. 4A . DJF latent heating increases of
up to 30 Wrm2 are found over north coastal Aus-
tralia, parts of Southeast Asia, and much of southern
? .and central Africa Fig. 4A . Latent heating in JJA
increases by up to 55 Wrm2 over regions in central
? .Africa and central Asia Fig. 4B .
5.2. New topography and shorelines ?s. old topogra-
phy and shorelines
The two different topographies presented here
represent changes in knowledge about the Eocene
land surface. Because the only change between the
NEWVOLDT and NEWVT cases is the inclusion of
the new topographic and shoreline boundary condi-
tions, differences between these two cases provide an
assessment of the importance of ArealisticB topogra-
phy for our modeling experiments.
5.2.1. Ele?ation differences
Despite the fact that the global mean elevations of
the two topographies are very similar, there are
significant differences between the old and new to-
? .pographies on almost every continent Fig. 5 . Global
mean elevation of the old topography is 137.43 m
and the global mean elevation of the new topography
is 141.28 m. Elevation increases of note in the new
topography are found over the North American
Cordillera, the Appalachian Mountains, the Andes,
parts of Southeast Asia, the Transantarctic Moun-
?tains, and rift shoulders of India and Greenland Fig.
.5 . Elevation decreases of note are associated with
the proto Altiplano, the South American Foreland,
and the proto Himalayas. The most significant shore-
line changes are the addition of Beringia connecting
North America and Asia and an increase in the size
? .of the Mississippi embayment Fig. 5 .
5.2.2. Surface temperatures
Because shorelines were changed between
? .NEWVOLDT and NEWVT Fig. 1B,C , surface
temperature differences between the two cases are
masked to remove points that have changed from
land to sea and vice versa. These experiments have
fixed SSTs and, therefore, surface temperature dif-
ferences in those regions are meaningless.
Over the North American Cordillera, mean annual
? . ?temperature MAT decreases by up to 168C Fig.
.6A . Decreases in MAT of 5?108C occur over the
Appalachian Mountains, the Andes, the Transantarc-
tic Mountains, parts of Southeast Asia, a few areas in
central Australia, and rift shoulders of India and
? .Greenland Fig. 6A . MAT over central Africa and
portions of the proto Himalayas increases by 58C or
? .more Fig. 6A in the NEWVT case. MAT over the
South American Foreland is up to 108C higher in the
NEWVT case.
Seasonal differences in surface temperature be-
tween the old and new topographies largely reflect
?the same pattern as differences in MAT Fig.
.6A,B,C . In both DJF and JJA, there is cooling of up
to 168C over the North American Cordillera and
cooling of 5?108C is seen over the Appalachian
Mountains, the Andes, the Transantarctic Mountains,
? .and rift shoulders of India and Greenland Fig. 6B,C .
In both DJF and JJA, there is warming of up to 108C
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465452
? . 2 2 ? .Fig. 4. A DJF latent heat difference, NEWVOLDT minus OLDTV, contours from y45 to 60 Wrm at 15 Wrm , no zero contour. B
JJA latent heat difference, NEWVOLDT minus OLDTV, contours from y45 to 60 Wrm2 at 15 Wrm2, no zero contour. All shaded
differences shown on this plot are significant to the 99% confidence level. Significance was calculated using Chervin and Schneider?s t-test
? .Chervin and Schneider, 1976a,b .
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465 453
Fig. 5. Topography differences, new Eocene topography at 28=28 minus old Eocene topography at 28=28. Contours from y1800 to
3000 m at 800 m.
over the proto Altiplano and South American Fore-
land, 5?108C of warming over the proto Himalayas,
?and increases of up to 58C over central Africa Fig.
.6B,C . The only seasonal surface temperature
changes of note are found over the interior of North
America. In DJF, interior North America is 5?158C
? .cooler with the new topography Fig. 6B . In JJA,
the foreland region of western interior North Amer-
? .ica warms by over 58C Fig. 6C .
5.2.3. Fi?e hundred millibar winds
In the upper atmosphere of the NEWVT case, an
enhanced trough and ridge structure is apparent over
the mid to high latitude Northern Hemisphere in
DJF, but is less pronounced in the Southern Hemi-
? .sphere Fig. 7A,B . The enhanced trough and ridge
structure is especially pronounced in JJA over the
mid to high latitude Northern Hemisphere and around
the Transantarctic Mountains and the AAustralian
? .NeckB in the Southern Hemisphere Fig. 7C,D .
5.2.4. Surface winds
DJF northerlies over southern Alaska are replaced
by southerlies in NEWVT. Southwesterly DJF flow
over western, coastal North America is replaced by
? .southerlies Fig. 8A,B . In JJA, the new topography
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465454
? . ? . ? .Fig. 6. A MAT difference, NEWVT minus NEWVOLDT. B DJF surface temperature difference, NEWVT minus NEWVOLDT. C JJA
surface temperature difference, NEWVT minus NEWVOLDT. Contours for all plots are from y208C to 158C at 58C, no zero contour. All
differences shown on this plot are significant to the 99% confidence level. Significance was calculated using Chervin and Schneider?s t-test
? .Chervin and Schneider, 1976a,b .
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465 455
? . ? . ? . .Fig. 7. A NEWVOLDT DJF 500 mbar winds, B NEWVT DJF 500 mbar winds, C NEWVOLDT JJA 500 mbar winds, D NEWVT
JJA 500 mbar winds.
?weakens westerly flow over Southeast Asia Fig.
.8C,D . The northerlies in JJA over central North
America are stronger and easterlies are introduced
over the front of the Cordillera . JJA southerlies over
the Mississippi embayment are strengthened in the
? .NEWVT case Fig. 8C,D .
5.2.5. Sea le?el pressure
The DJF sea level pressure patterns generated
with the old and new topographies are very similar
in the Southern Hemisphere, with the magnitude of
sea level pressure associated with the new topogra-
phy being slightly lower than that with the old
? .topography not shown . The only Southern Hemi-
sphere area where DJF sea level pressure with the
new topography exceeds that with the old topogra-
phy is immediately offshore of the Transantarctic
Mountains. In NEWVT DJF Northern Hemisphere
sea level pressure, the high over Asia is stronger and
penetrates farther north. In DJF, the Aleutian low is
stronger but restricted to Beringia. A large winter
high occurs over central North America and a low
?has replaced the high over the Labrador strait not
.shown .
In JJA, sea level pressure is significantly lower
with the new topography over central Siberia, north
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465456
? . ? . ? . ? .Fig. 8. A NEWVOLDT DJF surface winds, B NEWVT DJF surface winds, C NEWVOLDT JJA surface winds, D NEWVT JJA
surface winds.
central North America, the North American
?Cordillera, Australia, and portions of Antarctica not
.shown . JJA sea level pressure is higher over south
central North America and off the east coast of
South America. In JJA, the low over Asia is stronger
and extends farther north. The Aleutian high is
slightly stronger. With the new topography, there is a
summer low over the North American Cordillera and
? .a high over east central North America not shown .
5.2.6. Con?ecti?e precipitation
DJF convective precipitation with the new topog-
raphy decreases by 3 mmrday over central Africa
and by 2?4 mmrday over the South American
Foreland. Just north of the South American Foreland,
however, DJF convective precipitation increases by 2
? .mmrday Fig. 9A . In DJF, convective precipitation
increases by up to 6 mmrday over the southern
coast of Alaska and by 4 mmrday over western and
? .southeastern North America Fig. 9A .
JJA convective precipitation decreases by 3
mmrday along the west coast of North America and
by 7?9 mmrday in the proto Himalayan region. The
new topography generates JJA convective precipita-
tion increases of 3 mmrday northeast of the proto
Himalayas and by 2 mmrday along the east coast of
? .China Fig. 9B . In JJA, convective precipitation
increases by 3?5 mmrday over the proto Front
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465 457
? .Fig. 9. A DJF convective precipitation difference, NEWVT
minus NEWVOLDT, contours from y8 to 8 mmrday at 2
? .mmrday, no zero contour. B JJA convective precipitation differ-
ence, NEWVT minus NEWVOLDT, contours from y8 to 8
mmrday at 2 mmrday, no zero contour. All shaded differences
shown on this plot are significant to the 99% confidence level.
Significance was calculated using Chervin and Schneider?s t-test
? .Chervin and Schneider, 1976a,b .
Ranges of North America and by 2 mmrday over
? .central North America Fig. 9B .
5.2.7. Con?ecti?e clouds
In the DJF NEWVT results, convective cloudi-
ness decreases by up to 50% over the proto Front
Ranges and into central North America. Convective
cloudiness decreases by 30% over the South Ameri-
can Foreland.
In the NEWVT case, there is a 30% decrease in
JJA convective cloudiness at the base of the West
Antarctic Peninsula. JJA convective cloudiness de-
creases by 40% over central and northern North
America and by 10?30% over west central South
America. With the inclusion of the new topography,
JJA convective clouds increase by 30% off the west
coast of North America and by 30?40% over the
Mississippi embayment and the proto Front Ranges
of North America.
6. Discussion
6.1. Climate sensiti?ity to changes in the global
?egetation distribution
Significant responses to changes in the global
vegetation distribution are limited to the regional-
scale; this is an important result given that proxy
data reflect regional, as well as global, climate condi-
tions. The most significant responses to our specified
vegetation changes are regional increases in evapo-
transpiration, which in turn generate increases in
clouds and latent heating, and decreases in incoming
radiation, and surface temperature. As leaf area in-
? .dex LAI is the vegetation characteristic that most
influences evapotranspiration in the model and be-
cause responses are found predominantly in areas
where vegetation has been altered from shrublands to
? . ?tropical forests Figs. 2?5 a substantial increase in
.LAI , we hypothesize that the climatic changes out-
lined above are related to increases in LAI.
If the LAI of each vegetation type, as defined in
the climate model, is multiplied by the areal extent
of that vegetation type, the result is the areal equiva-
lent leaf area. The new vegetation distribution has a
substantially greater leaf area equivalent in both DJF
and JJA, 1.18=108 and 1.49=108 km2 greater,
? .respectively Table 1 . We find that these LAI
changes have the greatest climatic impact in midlati-
tudes of the summer hemisphere.
The increase in LAI results in a corresponding
increase in evapotranspiration . This is seen most
clearly over central Asia and interior North America
in JJA and over Africa around 258 south latitude,
along the northern coast of Australia, and in areas of
South America in DJF. The increased evapotranspi-
ration has a direct effect on surface temperature,
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465458
Table 1
? .LAI of each of the included Dorman and Sellers 1989 vegetation types multiplied by the areal extent of that vegetation type in the old and
new vegetation distributions
Vegetation type LAI=areal extent LAI=areal extent Difference in
?in old vegetation in new vegetation coverage new
2 2 2? . ? . . ? .distribution km distribution km minus old km
? .A JJA
8 8? .1 Tropical forest 0 2.16=10 2.16=10
7 7? .2 Broadleaf deciduous trees 0 7.90=10 7.90=10
8 8? .3 Broadleaf and needleleaf trees 0 1.41=10 1.41=10
8 8? .4 Needleleaf evergreen trees 1.58=10 0 y1.58=10
7 7? .5 Needleleaf deciduous trees 1.72=10 0 y1.72=10
8 7 7? .6 Savanna 1.10=10 6.76=10 y4.24=10
6 6? .7 Perennial groundcover 2.64=10 0 y2.64=10
7 6 7? .8 Broadleaf shrubs with perennial groundcover 7.13=10 6.83=10 y6.45=10
? .9 Broadleaf shrubs with bare soil 0 0 0
6 6? .10 Tundra 1.63=10 0 y1.63=10
8 8 8Total 3.61=10 5.10=10 1.49=10
? .B DJF
8 8? .1 Tropical forest 0 2.16=10 2.16=10
7 7? .2 Broadleaf deciduous trees 0 3.85=10 3.85=10
8 8? .3 Broadleaf and needleleaf Trees 0 1.01=10 1.01=10
8 8? .4 Needleleaf evergreen trees 1.34=10 0 y1.34=10
? .5 Needleleaf deciduous trees 0 0 0
7 7 7? .6 Savanna 7.13=10 4.6=10 y2.5=10
7 7? .7 Perennial groundcover 1.66=10 0 y1.66=10
7 7 7? .8 Broadleaf shrubs with perennial groundcover 6.78=10 1.08=10 y5.70=10
? .9 Broadleaf shrubs with bare soil 0 0 0
6 6? .10 Tundra 4.31=10 0 y4.31=10
8 8 8Total 2.94=10 4.12=10 1.18=10
The table was split into seasons because the LAI of deciduous trees varies seasonally. LAI multiplied by areal extent of vegetation type
peaks in JJA because that is Northern Hemisphere summer. Deciduous trees have their greatest LAI in the summer. The Northern
Hemisphere has a greater land area than the Southern Hemisphere and thus supports more vegetation. Consequently, the season in which
Northern Hemisphere deciduous trees have their leaves will have greater overall vegetative cover than the season in which Southern
Hemisphere deciduous trees have their leaves.
? .cooling it by 2?68C in these areas Fig. 3 . These
changes are the inverse of those seen when LAI and
fractional cover are decreased by changing tropical
? .rainforest to savanna Crowley and Baum, 1997 .
The increased evapotranspiration in our model has
an additional, indirect effect on surface temperature.
The increase in evapotranspiration provides more
? .moisture for clouds Fig. 3A,B ; as clouds form,
latent heat is released. Latent heat and total cloudi-
ness increase over central Asia, interior North Amer-
ica, Africa around 258 south latitude, the north coast
of Australia, and regions of South America in the
? .summer Figs. 3A,B and 5A,B . The increased
cloudiness decreases incoming solar radiation by up
2 ? .to 80 Wrm not shown and there is a consequent
summer cooling in these regions.
Based on this series of modeled feedbacks, it
appears that LAI is the most important vegetation
characteristic influencing climate. Vegetation
changes are, therefore, only climatologically signifi-
cant where there is a substantial change in LAI. It
should be noted that this response is dependent on
the operational definition of vegetation in our model;
in other climate models, depending on the vegetation
parameterizations, the effects of vegetation change
on climate may be different. The sensitivity of cli-
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465 459
mate to changes in the global vegetation distribution
may also change with different vegetation parameter-
ization schemes.
6.2. Climate sensiti?ity to changes in topography
Surface temperature changes in NEWVT, as com-
pared with NEWVOLDT, are, for the most part,
directly associated with regions of elevation change
? .Figs. 6 and 7 . In regions where elevation increased,
surface temperature decreased and vice versa. When
change in model-produced MAT is plotted against
the difference in elevation between the two topogra-
phies, it is found that temperature decreases by 6.18C
? .for every kilometer increase in elevation Fig. 10 .
This is very close to the moist adiabatic lapse rate of
6.58Crkm. Cooler temperatures in areas such as the
North American Cordillera and warmer temperatures
in areas including the South American Foreland are
almost entirely a direct effect of the elevation change
in those areas.
Model output from NEWVT does not match proxy
?data estimates of Early Eocene MAT e.g., MAT in
the Rocky Mountains is from y8 toy38C in
NEWVT, proxy data estimates of MAT for the same
?area are 9?248C Wing and Greenwood, 1993;
..Greenwood and Wing, 1995 . This is probably be-
cause proxy data are most likely to be preserved in
intermontane basins, and a 28=28 land surface reso-
lution is too coarse to resolve these basins that most
2 ?likely occur at a scale smaller than 40,000 km Fig.
.11 . Instead, these basins are represented as part of a
broad mountain feature. Consequently, the elevation
of these fossil locales is overestimated in the model,
and MAT in these regions is underestimated.
If the basin elevation where the abovementioned
proxy data were deposited is assumed to be ;750
? .m above sea level Greenwood and Wing, 1995 and
? .the 28=28 model elevation is ;3000 m Fig. 1C ,
? . ? .Fig. 10. MAT difference NEWVT minus NEWVOLDT vs. elevation difference new Eocene topography minus old Eocene topography .
Line slope is y6.158Crkm.
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465460
Fig. 11. Cross section of present day North America at 408 north latitude. Topography at a resolution of 28=28 over topography at a
resolution of 0.38=0.38.
the elevation difference is 2250 m. Thus, the model
calculates air temperature for air at an elevation 2250
m higher than the actual elevation of that parcel of
air. If a moist adiabatic lapse rate of y6.58Crkm is
applied, the model-calculated air temperature will be
14.68C lower than the actual temperature of that
parcel of air. If the model-calculated air temperature
?is corrected for the lapse rate effect add 14.68C to
.the model-produced temperature , MAT results from
NEWVT become 7?128C, much closer to the proxy
data estimates of MAT.
As has been previously stated, surface tempera-
? .ture differences of up to 188C Fig. 6 are generated
by the new topography. These surface temperature
differences are wholly attributable to gross sub-
? .gridscale parameterizations lapse rate, specifically .
The magnitude of these differences is so large that
subtler surface temperature differences driven by
?more dynamic, explicitly defined processes e.g.,
.advection, convection, and radiation are eclipsed.
Better correlation of model output and proxy data
will require much higher land surface resolutions, a
more sophisticated downscaling approach than sim-
ply utilizing the atmospheric lapse rate, or a combi-
nation of both.
The decreased MATs as a result of increased
elevation are responsible for perennial snow in the
North American Cordillera and the Transantarctic
? .Mountains in the NEWVT case not shown . Snow-
? .fall rate not shown increases substantially over the
North American Cordillera in DJF. There are no
significant changes in snowfall rate over JJA Antarc-
tica. Snow falls only in the winter hemisphere; how-
ever, summer temperatures are cool enough in iso-
? .lated regions not greater than one grid cell in area
for snow to be present year round. The areas of
perennial snow are in equilibrium with the climate;
the areal extent of the snowpack has not changed for
the last 4 years of the model run. For the North
American Cordillera, the presence of Eocene snow
pack is possibly supported by isotopic signatures in
?sediments of the Green River Formation Dettman
.and Lohmann, 1993, 2000; Norris et al., 1996 .
For the most part, seasonal changes in tempera-
ture mirror the changes in MAT and are directly
related to the changes in elevation. The only signifi-
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465 461
cant changes in seasonal surface temperature that are
not directly related to changing elevation are found
over central North America. In JJA, a 58C warming
?over the northern foreland of North America Fig.
.6C is linked to changes in incoming solar radiation.
With the inclusion of the new topography, total
? .clouds in JJA decrease by 40% not shown in this
region. Consequently, incoming solar radiation re-
? .ceived at the surface not shown increases by 60?
120 Wrm2. This generates summer warming of 58C.
In DJF, the presence of a large high-pressure system,
and the associated trough, brings Arctic air as far
?south as the Mississippi embayment Figs. 7B and
.9B . The result is a cooling of up to 158C relative to
the NEWVOLDT results, which have much more
zonal winter winds, and, therefore, less continental
? .interior penetration of Arctic air Figs. 6B and 8A .
This enhanced trough and ridge structure of the
upper atmosphere that is responsible for DJF cooling
in North America is an expected result of increasing
elevation and roughness of the topography. Other
?researchers e.g., Kutzbach et al., 1993; Ruddiman
.and Prell, 1997 have noted similar responses.
The primary effect of the enhanced trough and
ridge structure and associated pressure regime is to
generate surface wind patterns that influence changes
in convective precipitation as well as temperature.
While there are no appreciable changes in large-scale
stable precipitation, convective precipitation changes
significantly. Along the southern coast of Alaska in
DJF, surface winds in NEWVT are onshore, vs.
? .offshore in the NEWVOLDT case Fig. 8A,B . The
increase in onshore winds brings moisture-laden air
off the North Pacific. When this air reaches the
coastal mountains of Alaska, the moisture is dropped
as convective precipitation. Over southern Alaska in
DJF, there is a 20% increase in convective clouds
? .and a 6 mmrday precipitation increase Fig. 9A .
Increased DJF convective precipitation is also seen
over the west coast of North America where subtrop-
ical air is forced northward along the western moun-
tain front instead of moving zonally across the low
? .mountains of NEWVOLDT Fig. 8A,B . As the sub-
tropical air moves north and east, it rises and cools,
and, as a result, convective precipitation increases
? .Fig. 9A .
Convective precipitation changes due to alteration
of the surface wind patterns are much more pro-
nounced in JJA. Reduced onshore flow from Tethys
? .over the proto Himalayas Fig. 8C,D results in a
significant decrease in convective precipitation of
? .7?9 mmrday over this region Fig. 9B .
One of the most significant results of this sensitiv-
ity study is seen along the proto Front Ranges of
North America. Climatic changes in this area are
particularly important because this is the region where
?most of our proxy data are located e.g. Greenwood
.and Wing, 1995 . With the new topography, JJA
?precipitation increases significantly by 3?5 mmr
.day in this region for a total summer rainfall of 6?8
? .mmrday not shown . All of this summer precipita-
? .tion 6?8 mmrday is in the form of convective
precipitation. In this region, enhanced southerly sur-
face winds off the Mississippi embayment collide
?with enhanced northerlies off the Arctic Ocean Fig.
.8D . These enhanced surface winds are driven by
rising air over the North American Cordillera and
there is a strong increase in easterly winds over this
? .region Fig. 8D . When the cool, dry, Arctic air
meets the warm, moist, Mississippi embayment air,
the result is a 40% increase in convective cloudiness
and a corresponding increase in convective precipita-
tion. These regional changes in convective precipita-
tion may explain the much greater development of
coals in basins east of the proto Front Ranges
? .Powder River and Hanna Basins compared with
? .those to the west Bighorn and Green River Basins .
It is possible that summer precipitation was greater
east of the mountains due to elevation effects. The
greater precipitation would have promoted formation
of vegetation rich swamps and lead to greater coal
production. If so, we speculate that fossils preserved
in the coals of the Hanna and Powder River Basins
would, therefore, reflect a wetter climate than plant
fossils from the Bighorn and Green River Basins.
6.3. Climate sensiti?ity to changes in shorelines
We believe that one of the main controls on the
amount of convective precipitation received by the
proto Front Ranges is the extent of the Mississippi
embayment. To test this hypothesis, a third sensitiv-
ity study, designated EMB, was conducted. The
boundary conditions for EMB are identical to those
of NEWVT with the exception that the Mississippi
embayment has been adjusted to approximately its
(
)
J.O
.Sewall
et
al.r
G
lobal
a
nd
Planetary
Change
26
2000
445
?465
462
? . ? .Fig. 12. A MAT difference over North America, EMB minus NEWVT, contours from y148C to 148C at 28C, no zero contour. B DJF surface temperature difference over
? .North America, EMB minus NEWVT, contours aty288C,y208C,y108C,y28C, and 28C. C JJA surface temperature difference over North America, EMB minus NEWVT,
? .contours from y68C to 88C at 28C, no zero contour. D Mean annual convective precipitation difference over North America, EMB minus NEWVT, contours at y1 and 1
? . ? .mmrday. E DJF convective precipitation difference over North America, EMB minus NEWVT, contours at y4, y2, y1, 1, 2, and 4 mmrday. F JJA convective
precipitation difference over North America, EMB minus NEWVT, contours at y4, y2, y1, 1, 2, and 4 mmrday. All contoured differences presented in this plot are
? .significant to the 99% confidence level. Significance was calculated using Chervin and Schneider?s t-test Chervin and Schneider, 1976a,b .
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465 463
modern extent. The EMB case was started from year
8 of NEWVT and run for 5 years. Results from the
last 3 years of the run were averaged together for
analyses.
7. Discussion of EMB results
Changes in surface temperatures over the area
previously occupied by the Mississippi embayment
are the result of altering this area from ocean to land.
During the winter months, land is substantially cooler
than ocean, and in the summer, land warms more
? .than the ocean does Fig. 12B,C . Overall, the ex-
?treme cooling of the land surface relative to the sea
.surface in the winter dominates the annual signal.
The result is a MAT for this region that is cooler
? . ? .;108C than in the NEWVT case Fig. 12A .
Cooler DJF temperatures along the Cordilleran
Front and into the Canadian Rockies are the result of
? .changes in the surface wind patterns not shown .
Weakened westerly flow into northern Canada in the
EMB case decreases the transport of warm, moist air
from the North Pacific to the interior of North
America. As a result, surface temperatures in this
area are reduced.
The differences in mean annual, DJF, and JJA
? . ? .total not shown and convective Fig. 12D,E,F
precipitation over south central North America are
identical. Precipitation differences are, therefore,
wholly attributable to changes in convective precipi-
tation. The DJF convective precipitation change is
centered over the area that was converted from sea to
land and is interpreted to be a direct effect of this
conversion. The changes in JJA convective precipita-
tion are related more to subtle alterations in atmo-
spheric circulation and proximity to a moisture
source. The hook-shaped band of decreased precipi-
? .tation over central North America Fig. 12F is
adjacent to areas just inland of the NEWVT Missis-
sippi embayment. Infilling the Mississippi embay-
ment to approximately its present extent in the EMB
case moved the moisture source farther away from
these regions; it also weakened southwesterly flow
? .off of the embayment not shown . Increased dis-
tance from the moisture source results in more mois-
ture precipitating out of air masses before they reach
the Laramide foreland. A decrease in wind strength
reduces inland penetration of moisture-laden air off
the proto Gulf of Mexico. This combination results
in an overall decrease in summer precipitation in the
foreland region of the Laramide uplifts; yet, summer
precipitation in this area remains 5?6 mmrday.
Removing the proximal moisture source slightly
decreases the amount of monsoonal convective pre-
cipitation; however, the North American summer
monsoon is still a robust feature.
8. Conclusions
Overall, this series of sensitivity studies indicates
? .that relatively small on a global scale changes to
the land surface, specifically topography, orography,
and the global distribution of vegetation, have little
effect on global mean climate. However, changes to
the land surface have significant climatic effects at
the regional scale. Indeed, the only really AglobalB
signal, enhanced ridge and trough structure in the
upper atmosphere, is significant for its influence on
surface winds that help initiate, or shut down, re-
gional monsoonal circulations.
Changes in monsoonal circulation affect regional
moisture budgets. Changes in elevation affect re-
gional surface temperature, and changes in plant LAI
influence evapotranspiration in some areas. Infilling
the Mississippi embayment to approximately its pre-
sent extent results in increased seasonality in that
? .region warmer summers and cooler winters and
slightly decreases the amount of summer monsoonal
precipitation over the Laramide foreland.
These model-produced regional climate responses
to changes in the land surface are important because
proxy data-derived estimates of paleoclimate reflect
the regional climates in which those data existed. If
we hope to better correlate climate model output and
proxy data-derived estimates of paleoclimate, we
must be able to more completely and accurately
model climate responses to changes in the land
surface. The magnitude of our model-produced re-
gional climate responses indicates that a more accu-
rate global vegetation distribution and a more realis-
tic topography have significant effects on regional
climates. Consequently, we conclude that the inclu-
sion of more realistic boundary conditions is neces-
sary if we wish to produce a more realistic represen-
tation of paleoclimate with a GCM.
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465464
Acknowledgements
LCS thanks the National Science Foundation
? .NSF and the David and Lucile Packard Foundation
for funding in support of this research. JOS thanks
? .the Department of Defense DOD for fellowship
? .support NDESG Fellowship . Computing was car-
ried out at the National Center for Atmospheric
Research, which was funded by NSF.
References
Allmendinger, R.W., Jordan, T.E., Kay, S.M., an Isacks, B.L.,
1997. The evolution of the Altiplano-Puna plateau of the
Central Andes. Annu. Rev. Earth Planet. Sci. 25, 139?174.
Bonan, G.B., Pollard, D., Thompson, S.L., 1992. Effects of boreal
forest vegetation on global climate. Nature 359, 716?718.
Broccoli, A.J., Manabe, S., 1992. The effects of orography on
midlatitude Northern Hemisphere dry climates. J. Clim. 5,
1181?1201.
Chervin, R.M., Schneider, S.H., 1976a. A study of the response of
NCAR GCM climatological statistics to random perturbations:
estimating noise levels. J. Atmos. Sci. 33, 391?404.
Chervin, R.M., Schneider, S.H., 1976b. On determining the statis-
tical significance of climate experiments with general circula-
tion models. J. Atmos. Sci. 33, 405?412.
Chorowicz, J., Collet, B., Boravia, F.F., Mohr, P., Parrot, J.F.,
Korme, T., 1998. The Tana Basin, Ethiopia: intra-plateau
uplift, rifting and subsidence. Tectonophysics 295, 351?367.
Crowley, T.J., Baum, S.K., 1997. Effect of vegetation on and
ice-age climate model simulation. J. Geophys. Res. 102,
16463?16480.
Demko, T.M., Dubiel, R.F., Parrish, J.I., 1998. Plant taphonomy
in incised valleys: implications for interpreting paleoclimate
from fossil plants. Geology 26, 1119?1122.
Dettman, D.L., Lohmann, K.C., 1993. Seasonal change in Paleo-
gene surface water d 18O: fresh-water bivalves of western
North America. Climate Change in Continental Isotopic
Records. Geophys. Monogr., vol. 78. America Geophysical
Union, pp. 153?163.
Dettman, D.L., Lohmann, K.C., 2000. Oxygen isotope evidence
for high-altitude snow in the Laramide Rocky Mountains of
North America during the Late Cretaceous and Paleogene.
Geology 28, 243?246.
Dorman, J.L., Sellers, P.J., 1989. A global climatology of albedo,
roughness length and stomatal resistance for atmospheric gen-
eral circulation models as represented by the Simple Biosphere
? .Model SiB . J. Appl. Meteorol. 28, 833?855.
Fitzgerald, P.G., 1994. Thermochronologic constraints on post-
Paleozoic tectonic evolution of the central Transantarctic
Mountains, Antarctica. Tectonics 13, 818?836.
Foley, J.A., Kutzbach, J.E., Coe, M.T., Levis, S., 1994. Feedbacks
between climate and boreal forests during the Holocene epoch.
Nature 371, 52?54.
Gebka, M., Mosbrugger, V., Schilling, H.-D., Utescher, T., 1999.
Regional-scale palaeoclimate modelling on soft proxy-data
basis ? an example from the Upper Miocene of the Lower
Rhine Embayment. Palaeogeogr., Palaeoclimatol., Palaeoecol.
152, 225?258.
Greenwood, D.R., 1992. Taphonomic constraints on foliar phys-
iognomic interpretations of Late Cretaceous and Tertiary
palaeoclimates. Rev. Palaeobot. Palynol. 71, 149?190.
Greenwood, D.R., Wing, S.L., 1995. Eocene continental climates
and latitudinal temperature gradients. Geology 23, 1044?1048.
Gurnis, M., Muller, R.D., Moresi, L., 1998. Cretaceous vertical
motion of Australia and the Australian?Antarctic Discordance.
Science 279, 1499?1504.
Henderson-Sellers, A., Dickinson, R.E., Durbidge, T.B., Kennedy,
P.J., McGuffie, K., Pitman, A.J., 1993. Tropical deforestation:
modeling local- to regional-scale climate change. J. Geophys.
Res. 98, 7289?7315.
Kutzbach, J.E., Gallimore, R.G., 1989. Pangaean climates: mega-
monsoons of the megacontinent. J. Geophys. Res. 94, 3341?
3357.
Kutzbach, J.E., Prell, W.L., Ruddiman, Wm.F., 1993. Sensitivity
of Eurasian climate to surface uplift of the Tibetan Plateau. J.
Geol. 101, 177?190.
Lamb, S., Hoke, L., 1997. Origin of the high plateau in the
Central Andes, Bolivia, South America. Tectonics 16, 623?
649.
Maxson, J., Tikoff, B., 1996. Hit and run collision model for the
Laramide orogeny, western United States. Geology 24, 968?
972.
Molnar, P., Tapponier, P., 1975. Cenozoic tectonics of Asia:
effects of a continental collision. Science 189, 419?426.
Molnar, P., Burchfield, B.C., Ziyun, Z., K?uangyi, L., Shuji, W.,
Minmin, H., 1987. Geologic evolution of Northern Tibet:
results of an expedition to Ulugh Muztagh. Science 235,
299?305.
Norris, R.D., Jones, L.S., Corfield, R.M., Cartlidge, J.E., 1996.
Skiing in the Eocene Uinta Mountains? Isotopic evidence in
the Green River Formation for snow melt and large mountains.
Geology 24, 403?406.
Plaziat, J.C., 1981. Late Cretaceous to Late Eocene palaeogeo-
graphic evolution of southwest Europe. Palaeogeogr., Palaeo-
climatol., Palaeoecol. 36, 263?320.
Pollard, D., Thompson, S.L., 1995. Use of a land-surface transfer
? .scheme LSX in a global climate model: the response to
doubling stomatal resistance. Global Planet. Change 10, 129?
161.
?Reichert, B.K., Bengtsson, L., Akesson, O., 1999. A statistical
modeling approach for the simulation of local paleoclimatic
proxy records using general circulation model output. J. Geo-
phys. Res. 104, 19071?19083.
Roehler, H.W., 1993. Eocene climates, depositional environments,
and geography, greater Green River Basin, Wyoming, Utah,
and Colorado. U.S. Geol. Surv. Prof. Pap. vol. 1506-F. United
States Government Printing Office, Washington, DC, pp. F1?
F74.
Ruddiman, W.F., Prell, W.L., 1997. Introduction to the uplift-
? .climate connection. In: Ruddiman, W.F. Ed. , Tectonic Uplift
and Climate Change. Plenum, New York, NY, pp. 3?15.
( )J.O. Sewall et al.rGlobal and Planetary Change 26 2000 445?465 465
Rull, V., 1999. Palaeoflorisitc and palaeovegetational changes
across the PaleocenerEocene boundary in northern South
America. Rev. Palaeobot. Palynol. 107, 83?95.
Sailor, D.J., Li, X., 1999. A semiempirical downscaling approach
for predicting regional temperature impacts associated with
climatic change. J. Clim. 12, 103?114.
Scotese, C.R., Gahagan, L.M., Larson, R.L., 1988. Plate tectonic
reconstructions of the Cretaceous and Cenozoic ocean basins.
Tectonophysics 155, 27?48.
Sloan, L.C., Barron, E.J., 1992. Paleogene climatic evolution: a
climate model investigation of the influence of continental
elevation and sea-surface temperature upon continental cli-
? .mate. In: Prothero, D.R., Berggren, W.A. Eds. , Eocene?
Oligocene climatic and biotic evolution. Princeton University
Press, Princeton, NJ, pp. 202?217.
Sloan, L.C., Rea, D.K., 1995. Atmospheric carbon dioxide and
Early Eocene climate: a general circulation modeling sensitiv-
ity study. Palaeogeogr., Palaeoclimatol., Palaeoecol. 119,
275?292.
Sloan, L.C., Pollard, D., 1998. Polar stratospheric clouds: a high
latitude warming mechanism in an ancient greenhouse world.
Geophys. Res. Lett. 25, 3517?3520.
Sloan, L.C., Crowley, T.J., Huber, M., Sewall, J.O., Baum, S.K.,
Ewing, A. Palaeogeogr., Palaeoclimatol., Palaeoecol., in re-
view.
Tapponier, P., Molnar, P., 1979. Active faulting and Cenozoic
tectonics of the Tien Shan. J. Geophys. Res. 84, 3425?3459.
Taylor, G., Truswell, E.M., McQueen, K.G., Brown, M.C., 1990.
Early Tertiary palaeogeography, landform evolution, and
palaeoclimates of the Southern Monaro, N.S.W., Australia.
Palaeogeogr., Palaeoclimatol., Palaeoecol. 78, 109?134.
Thompson, S.L., Pollard, D., 1997. Greenland and Antarctic mass
balances for present and coupled atmospheric CO from the2
GENESIS Version-2 global climate model. J. Clim. 10, 871?
900.
Wilby, R.L., Wigley, T.M.L., Conway, D., Jones, P.D., Hewitson,
B.C., Main, J., Wilks, D.S., 1998. Statistical downscaling of
general circulation model output: a comparison of methods.
Water Resour. Res. 34, 2995?3008.
Wilf, P., 2000. Late Paleocene?Early Eocene climate changes in
southwestern Wyoming: paleobotanical analysis. Geol. Soc.
Am. Bull. 112, 292?307.
Wing, S.L., 1998a. Late Paleocene?Early Eocene floral and cli-
matic change in the Bighorn Basin, Wyoming. In: Berggren,
? .W., Aubrey, M.-P., Lucas, S. Eds. , Late Paleocene?Early
Eocene biotic and climatic events. Columbia University Press,
New York, NY, pp. 371?391.
Wing, S.L., 1998b. Tertiary vegetation of North America as a
context for mammalian evolution. In: Janis, C.M., Scott, K.M.,
? .Jacobs, L.L. Eds. , Evolution of Tertiary Mammals of North
America. Terrestrial Carnivores, Ungulates, and Ungulatelike
Mammals, vol. 1. Cambridge Univ. Press, Cambridge, pp.
37?60.
Wing, S.L., Greenwood, D.R., 1993. Fossils and fossil climate:
the case for equable continental interiors in the Eocene. Philos.
Trans. R. Soc. London 341, 243?252.