date: Mon, 18 Feb 2002 13:58:58 -0500
from: "Raymond S. Bradley" <rbradley@geo.umass.edu>
subject: fig captions
to: Keith Briffa <k.briffa@uea.ac.uk>

<x-flowed>
Figure Captions
Figure 6.1.  Seasonal and annual trends in surface air temperature, 
1901-1999, based on instrumental measurements.
Figure 6.2.  Composite of 18O and melt records from Arctic ice caps
Figure 6.3.  Composite of records showing late Holocene temperature changes
Figure 6.4.  Reconstructed northern hemisphere mean annual temperature with 
2 standard error uncertainties (Mann et al 1999).
Figure 6.5.  Tree ring density reconstruction of warm-season (April to 
September) temperature from all land north of 20N, with the 1 and 2 
standard error ranges shaded.  Units are C anomalies with respect to the 
1961-90 mean (dotted line) and the instrumental temperatures are shown by 
the thick line.  Both series have been smoothed with a 30-year 
Gaussian-weighted filter (from Briffa et al 2001).
Figure 6.6.  Northern Hemisphere surface temperature anomalies (C) 
referenced to the 196190 mean (dotted line). Annual-mean land and marine 
temperature from instrumental observations (black, 18561999), and as 
reconstructed by Mann et al. (red, 10001980, with 2 standard errors shown 
by pink shading) and Crowley and Lowery (purple, 10001987). 
April-to-September mean temperature from land north of 20N as 
reconstructed by Briffa et al. (green, 14021960, with 2 standard errors 
shown by green shading), and reconstructed by re-calibrating the Jones et 
al. estimate of summer northern hemisphere temperature by simple linear 
regression over the period 18811960 (blue, 10001991). All series have been 
smoothed with a 30-year Gaussian-weighted filter.
Figure 6.7.  A comparison of century-long ground temperature trends from 
boreholes with data from co-located grid-boxes, derived by Mann et al. (1998).
Figure 6.8.  Reconstructed mean annual temperatures for the northern 
hemisphere, the 19th and 20th centuries, from Mann et al. (2000) compared 
to the calibration data (1902-1980) and an independent period (1854-1901) 
for which instrumental data are available.
Figure 6.9.  Northern Hemisphere surface temperature (C anomalies with 
respect to the 196190 mean) reconstructed by Mann et al. and Briffa et al., 
with shading indicating only the minimum and maximum of the 2 standard 
error ranges of the two reconstructions, and compared to that simulated by 
the Hadley Centres HadCM3 coupled climate under increasing greenhouse gas 
and sulphate aerosol concentrations (SRES scenario A2) from 
1950-2099.  Thin line indicates the April-September temperature from all 
land north of 20N, while the thick line indicates the annual temperature 
average over the entire northern hemisphere.  All data have been smoothed 
with a 30-year Gaussian-weighted filter.
Figure 6.10.  Coral 18O data from Maiana Atoll (central Pacific) and 
evolutive spectrum of these data plotted on the same horizontal axis (Urban 
et al. 2000). The top panel shows bimonthly values with a 21-yr running 
mean superimposed.  The bottom panel maps the changing concentrations of 
variance revealed by evolutive spectral analysis, in which 40-year segments 
were analyzed offset by 4 years.  Colored regions are significant above the 
median (50%) level, and the dark line encloses variance significantly 
different from a red noise background spectrum at 90%.  Changes in the mean 
of the time series correspond to changes in the frequency domain 
characteristics of the record, particularly in the correspondence of strong 
decadal variance and weak inter-annual variance to cooler/drier background 
conditions in the 19th century.
Figure 6.11.  Common patterns of decadal variability in tropical 
Indo-Pacific coral records during the 19th century.  The records shown here 
all exhibit cool-dry events in the late 1850s, ~1870, and early 1880s 
(shaded bars).  Small age offsets may be real, or may reflect age-model 
uncertainties.  The top three and the bottom record correlate closely with 
ENSO in their calibration periods, but the remaining records are somewhat 
removed from ENSO centers of action or have competing climate influences on 
their 18O, so do not correlate as strongly to interannual ENSO 
changes.  The fact that they all reflect the decadal variance of the late 
19th century suggests similarities with the 1976 shift, whose extent is 
latitudinally broader than typical ENSO variability.
Figure 6.12.  Sites where annual coral isotope records span the interval 
1895-1990. Numbers indicate the inferred SST trend in C per 100 years, 
assuming all isotopic variability is due solely to SST changes and the 
slope of the SST-18O relationship is 0.22C per 1.  Site sensitivity 
issues (e.g. depth of coral, influence of rainfall) have not been taken 
into account in these calculations.  Large central Pacific values are 
almost certainly due to the influence of rainfall on seawater isotopic 
content. Background colors indicate mean SST field.
Figure 6.13.  Composite showing ice core records of recent warming in the 
Tropics.
Figure 6.14.  Comparison of records of North Atlantic trade-wind strength 
(inferred from G. bulloides abundance at the Cariaco Basin; Black et al. 
1999), Lake Naivasha level (inferred from sedimentological indicators; 
Verschuren et al. 2000), and solar radiation (inferred from the 14C of 
atmospheric CO2 [Stuiver and Reimer 1993] and for the past 400 years from a 
reconstruction by Lean et al. 1995). Several of the multidecadal changes in 
these records are coincident (highlighted by grey bars), suggesting the 
possibility of a common response to radiative forcing on this time scale.
Figure 6.15.  Summer Palmer drought severity index (PDSI) as reconstructed 
from a continental network of drought-sensitive tree ring width records 
(Cook et al., 1999a).  PDSI less than zero represents dry conditions.  A.D. 
1746 and 1752 were El Nio and La Nia years, respectively, as 
reconstructed by Stahle et al., (1998).  These maps show that summer soil 
moisture conditions resembled those associated with the same phases of the 
ENSO fluctuation in the instrumental period.
Figure 6.16.  Nevada division 3 precipitation, July-June, from a network of 
lower forest border stripbark bristlecone pine (after Hughes and 
Funkhouser, 1998).  The series has been smoothed with a 50-yr gaussian 
filter.  1 standard deviation unit equals 4.4cm, mean = 18.3cm.  Map shows 
the location of tree ring sites (red + signs) and of Nevada Division 3 
(green line).
Figure 6.17.  Cumulative severity of A.D. 1561-1600 growth reduction in 
moisture-limited trees (from Biondi et al., 2000). The location of each 
symbol indicates the location of a tree ring width index chronology. These 
are expressed as dimensionless indexes with a mean of 1.0. The size of each 
symbol is proportional to the sum of all departures for the chronology over 
the period. The two symbols in the lower left of the map indicate the range 
of values on the map as percentages. Growth is reduced throughout this 
region in comparison to the long-term mean .
Figure 6.18.  Cumulative excess of A.D. 1601-1640 tree growth in 
moisture-limited trees (from Biondi et al., 2000). As Figure 6.18, except 
that growth is enhanced in comparison to the long-term mean.

Figure 6.19.  a 20-year running mean of fire events in five giant sequoia 
groves compared with tree ring indices of temperature-responsive 
bristlecone pine from near upper tree limit in the nearby White Mountains 
(from Swetnam, 1993)
                 b.  Departures from A.D. 500-1850 mean of tree ring width 
index of precipitation-responsive bristlecone pine from near lowest forest 
border in the nearby White Mountains are used as an index of regional 
drought, but are unaffected by fire in the Sierra Nevada.  They are shown 
for 5 years before and after years of fire in 0 to 5 giant sequoia groves 
in the Sierra Nevada.  Small asterisks p<0.01, large asterisks 
p<0.001.  The most extensive fires are clearly associated with drought 
years recorded by tree rings.  (from Swetnam, 1993).
Figure 6.20.  Selected climate-related records for the last millennium 
around the region of the northern North Atlantic. All of the series are 
plotted as effective 10-year (thin line) and 50-year (thick line) smoothed 
and standardized values (with reference to the common base period 1659-1999).
a. Central England mean annual temperatures (Manley, 1974; updated by the 
U.K. Meteorological Office);
b. Pseudo annual temperatures for the Benelux countries (produced from data 
in van Engelen et al., 2001);
c. reconstructed winter (DJF) North Atlantic Oscillation indices 
(Luterbacher et al., 2001);
d. warm season (A-S) N. Swedish temperatures reconstructed from tree ring 
data (Briffa et al., 1992);
e. moisture index based on bog flora in western Britain (Barber et al., 2000);
f. Bermuda Rise SST reconstructed from Foraminifera oxygen isotope 
composition (Keigwin, 1996);
g. an index of the speed of deep current flow to the north west of Scotland 
(Bianchi and McCave, 1999);
h. foraminiferal abundance in the Cariaco Basin, off Venezuela, indicative 
of trade wind intensity and possible changes in temperature in the North 
Atlantic (Black et al., 1999);
i. oxygen isotope data from the North Grip site ice core (Hammer, 2000);
j. combined series of ice-core-derived oxygen isotopes from the GISP2 and 
GRIP sites in central Greenland (best refs?);
k. a composite series of several west Greenland ice-core oxygen isotope 
series (Fisher et al., 1994);
l. high-resolution melt-layer data in an ice core from northeast Canada 
(Fisher ?);
m. a lower-resolution eastern Canadian ice-core melt record.
Figure 6.21.  Differences between composites of (a,b) temperature or (c,d) 
precipitation from (a,c) winters or (b,d) summers with positive North 
Atlantic Oscillation Index (NAOI) and those with negative NAOI. Seasonal 
temperature and precipitation were standardised to have zero mean and unit 
variance at each location prior to analysis.
Figure 6.22.  Holocene orbital insolation anomalies versus month of year at 
1000 year intervals.
Figure 6.23.  Insolation anomalies for the past millennium, by latitude and 
month.
Figure 6.24.  The record of total solar irradiance variations as 
reconstructed by Lean et al (1992) together with 14C and 10Be data for the 
past millennium.
Figure 6.25. Composite of Holocene forcing.
Figure 6.26. Reconstructed mean annual temperatures for North America and 
Europe since A.D. 1760 (data from Mann et al. 2000).
Figure 6.27.  Reconstructed mean annual temperatures in 1838 (data from 
Mann et al. 2000).
Figure 6.28  Detection of significant 20th century temperature trends of 
varying length (all expressed as C/decade). All temperatures are annual 
means averaged over all land and marine areas of the northern hemisphere. 
Blue line shows observed temperature trends from various length periods, 
all finishing in 1999.  Red line shows equivalent, but taken from the mean 
of an ensemble of four simulations from the HadCM2 coupled climate model 
forced by historical increases in greenhouse gas and sulphate aerosol 
concentrations.  These can be compared against the estimates of the 
95thcentile of the various length trends that are possible due to natural 
climate variability.  The black line with open circles is computed from a 
1000-year control integration of HadCM2, with fixed external forcing.  The 
thick black line with solid dots is computed from the pre-1900 portion of 
the 1000-year Mann et al. reconstruction.

Raymond S. Bradley
Distinguished Professor and Head of Department
Department of Geosciences
University of Massachusetts
Amherst, MA 01003-5820

Tel: 413-545-2120
Fax: 413-545-1200
Climate System Research Center: 413-545-0659
Climate System Research Center Web Page: 
<http://www.geo.umass.edu/climate/climate.html>
Paleoclimatology Book Web Site (1999): 
http://www.geo.umass.edu/climate/paleo/html



</x-flowed>
