Supplementary Material for:
A tree-ring reconstruction of East Anglian (UK) hydroclimate variability
over the last millennium
Richard J. Cooper1, Thomas M. Melvin2, Ian Tyers3, Rob J. S. Wilson4, Keith R. Briffa2
1
School of Environmental Sciences, University of East Anglia, Norwich, NR4 7TJ
2
Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich, NR4 7TJ
3
Dendrochronological Consultancy Ltd., 65 Crimicar Drive, Sheffield, S10 4EF
School of Geography and Geosciences, University of St Andrews, Fife, KY16 9AL
4
Corresponding author: Richard.J.Cooper@uea.ac.uk
Contents:

Supporting Figures:
- Site separation distance versus correlation strength (Figure SM1)
- 10-year/100-year high-pass chronology climate sensitivity (Figure SM2)
- Spatial drought analysis (Figure SM3)
- Northern France vs East Anglian precipitation (Figure SM4)
- East Anglian oak NAO sensitivity (Figure SM5)
- Spatial sea level pressure (SLP) analysis (Figure SM6)
- Additional regression materials (Figure SM7 – SM12)

Supporting Tables:
- Summary information for the 30 modern and historical oak chronologies (Table SM3)
- Soil properties at modern chronology sites (Table SM2)
- Reconstructed vs. Instrumental extreme years (Table SM3 – SM6)

Supporting References
Supporting Figures
Correlation Coefficient
1
0.8
0.6
0.4
y = -0.0019x + 0.6575
R² = 0.3171
p<0.01
0.2
0
0
20
40
60
Separation Distance (km)
80
100
Figure SM1 Plot showing the relationship between site separation distance
(km) against inter-site correlation strength for the nine modern oak
chronologies over the 1879-1981 common period. All chronologies
detrended with a 10-year high pass smoothing spline. Site separation
distance is able to explain 31.7% of inter-site correlation variability, with
correlation reducing by 0.19 per 100 km increase in distance
0.6
Correlation Coefficient
0.5
(a)
0.4
Previous
Current
0.3
0.2
0.1
0
J
J
A
S
O
N
D
J
F
M
A
M
J
J
A
S
pJJA
pSON
DJF
MAM
JJA
Oct-Aug
MAMJJ
-0.1
Month/Season
0.6
0.4
(b)
Previous
Current
0.3
0.2
0.1
0
-0.1
J
J
A
S
O
N
D
J
F
M
A
M
J
J
A
S
pJJA
pSON
DJF
MAM
JJA
Oct-Aug
MAMJJ
Correlation Coefficient
0.5
Month/Season
Figure SM2 Correlation coefficients
for individual monthly and various
seasonal measures of East Anglia
precipitation variability calculated for
the period 1901-2009 against a 10-year
and b 100-year high-pass smoothed
chronologies. Significant correlations
(p<0.05)
are
shown
in
blue.
Correlations closely match those
derived using the RCS East Anglian
chronology, thereby emphasising the
robustness of these associations
regardless of standardisation procedure
1901-2002
1952-2002
1901-1951
Figure SM3 Spatial field correlations of 0.5o x 0.5o latitude/longitude gridded European scale MAMJJ
self-calibrating Palmer Drought Severity Index (scPDSI) (CRUTS2.1; Mitchell and Jones 2005; van
der Schrier et al. 2006) against the RCS East Anglian oak chronology for early, late, and full length
periods. All correlations significant at p<0.1. In common with the spatial patterns derived for
precipitation, positive association is established between East Anglian oak growth and the scPDSI
across England, northern France, Benelux, and southern Scandinavia
1
Correlation Coefficient
0.8
52-53N 0-2E
49-53N 0-2E
0.6
0.4
0.2
0
-0.2
1900
1920
1940
1960
1980
2000
Year (AD)
Figure SM4 31-year moving period correlations between the East Anglian
chronology and two gridded precipitation datasets. Stronger correlations, particularly
during the early period, are obtained when correlating against the gridded data that
includes part of northern France (49-53oN, 0-2oE) than when using precipitation data
solely for East Anglia (52-53oN, 0-2oE).
0.8
DJFM
JA
MAMJJ
Correlation Coefficient
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
1820
1840
1860
1880
1900
1920
1940
1960
1980
2000
Year (AD)
Figure SM5 31-year moving period correlations between the East Anglian chronology and
winter (DJFM), summer (JA), and MAMJJ Gibraltar-Iceland NAO index (Jones et al. 1997) for
the period 1822-2009. Grey shading represents non-significant correlation at the 95%
significance level. This figure reveals no statistically significant associations between East
Anglian oak growth and the NAO during winter, summer, or MAMJJ periods
(a)
(b)
Figure SM6 European spatial field correlations of gridded 5o x 5o latitude/longitude MAMJJ mean sealevel pressure (SLP) (Trenberth and Paolino 1980) against a gridded (49-53oN, 0-2oE) instrumental
precipitation and b the East Anglian oak precipitation reconstruction for the period 1899-2009.
Negative associations are demonstrated between SLP and both instrumental and reconstructed East
Anglian precipitation series. Thus, low SLP is associated with high precipitation and increased oak
growth. All correlations significant at p<0.1
400
300
250
200
150
100
50
1900
Actual Decadal
Actual Annual
Recon. Annual
Recon. Decadal
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010
Year (AD)
Figure SM7 Split period calibration and verification procedure for the scaled (Esper et al. 2005)
reconstruction of MAMJJ precipitation over the period 1901-2009. Calibration shown for both
annual and decadal smoothed timescales. Note the reduced underrepresentation of extreme years
compared to the OLS regression approach, but also now the overrepresentation of precipitation in
years highlighted by circles
MAMJJ Precipitation (mm)
400
350
(a)
300
250
200
Scaling
OLS
150
100
1900
1920
1940
1960
1980
2000
1980
2000
Year (AD)
350
(b)
MAMJJ Precipitation (mm)
MAMJJ Precipitation (mm)
350
Calibration Period
1956-2009
Verification Period
1901-1955
300
250
200
150
1900
Scaling
OLS
1920
1940
1960
Year (AD)
Figure SM8 Comparison of the OLS linear regression and scaling techniques at a annual
and b decadal smoothed timescales for the period 1901-2009. This figure demonstrates how
the regression based variance reduction associated with the OLS regression approach is
partially overcome by using the scaling approach, thereby reducing the underrepresentation
of extreme wet and dry years. However, this in turn increases the chances of over estimating
precipitation in the reconstruction
325
Reconstructed Precipitation (mm)
Reconstructed Precipitation (mm)
400
Scaling - Annual
y = 0.5914x + 99.938
R² = 0.3653
p<0.01
300
200
Scaling - Decadal
y = 0.6421x + 87.535
R² = 0.4447
p<0.01
300
275
250
225
200
100
50
150
250
200
350
225
300
OLS - Annual
Reconstructed Precipitation (mm)
Reconstructed Precipitation (mm)
350
y = 0.3653x + 160.92
R² = 0.3653
p<0.01
300
250
275
300
Actual Precipitation (mm)
Actual Precipitation (mm)
250
200
150
275
OLS - Decadal
y = 0.4447x + 140.84
R² = 0.4447
p<0.01
250
225
200
50
100
150
200
250
300
350
200
225
Actual Precipitation (mm)
250
275
300
Actual Precipitation (mm)
Figure SM9 Late period (1956-2009) annual and decadal smoothed calibration scatter plots for both the
OLS linear regression and scaling reconstruction approaches
150
Annual Residuals
Decadal Residuals
Linear (Annual Residuals)
(a)
(b)
25
Frequency Distribution
Precipitation Residual (mm)
100
30
y = 0.1351x - 264.06
R² = 0.0123
50
0
-50
Annual Residuals
Decadal Residuals
20
15
10
5
-100
0
-150
1900
-100 -80
1920
1940
1960
Year (AD)
1980
2000
-60
-40
-20
0
20
40
60
Residual (mm)
Figure SM10 OLS regression residuals. a Time series of annual and decadal regression residuals with
linear trend line demonstrating no significant trend in the annual residuals. b Percentage frequency
distribution plot of the annual and decadal regression residuals reveals the residual precipitation indices
to be normally distributed
80
100
35
Annual Residuals
Decadal Residuals
Linear (Annual Residuals)
(a)
Residual (mm)
100
y = 0.0868x - 169.69
R² = 0.004
50
0
-50
(b)
30
Frequency Distribution
150
Annual Residuals
Decadal Residuals
25
20
15
10
5
-100
0
-150
1900
1920
1940
1960
1980
-100 -80 -60 -40 -20 0 20 40
Residual (mm)
2000
60
Year (AD)
Figure SM11 Scaling residuals. a Time series of annual and decadal scaled residuals with linear trend
line demonstrating no significant trend in the annual residuals. b Percentage frequency distribution plot
of the annual and decadal residuals reveals the residual precipitation indices to be normally distributed.
Scaled residuals show improved normal distribution and reduced trend in the annual residuals compared
to the OLS regression approach
400
MAMJJ Precipitation (mm)
350
Verification Period
1901-1955
Annual R2 = 0.24
Decadal R2 = 0.07
CE = 0.20
CE = 0.00
Annual R2 = 0.36
Decadal R2 = 0.47
Calibration Period
1956-2009
300
250
200
150
100
50
1900
Actual Annual
Recon Annual
Recon Decadal
Actual Decadal
1910
1920
1930
1940
Full Period 1901-2009
Annual R2 = 0.32 DW = 1.93 Decadal R2 = 0.41
1950
1960
1970
1980
1990
2000
2010
Year (AD)
Figure SM12 Split period calibration and verification procedure for the OLS regression
reconstruction of MAMJJ precipitation for the 100-year high-pass spline oak chronology over the
period 1901-2009. Calibration is shown for both annual and decadal smoothed timescales.
Verification statistics reveal a robust annual, but non-robust decadal, reconstruction of
precipitation in East Anglia, emphasising that the RCS derived chronology is a better proxy for
past precipitation variability
80 100
Supporting Tables
Table 1 Summary information for the modern (bold) and historical (fine) East Anglian oak ring-width chronologies. Mean
RBAR refers to the mean inter-series correlation; EPS (0.85) shows the minimum number of trees required to achieve an
Expressed Population Signal value of 0.85
Site Name
County
No. of
series
Full
Length
Latitude
Longitude
Elevation
(m)
Mean
RBAR
N (EPS
0.85)
Mean
Age
Mean
TRW
(mm)
Babingley
Norfolk
19
1840-1994
52.48
0.29
10
0.38
10
130.32
2.86
Blickling
Norfolk
74
1670-2008
52.48
1.13
46
0.28
15
191.11
1.56
Bradfield
Suffolk
11
1879-1984
52.12
0.49
85
0.45
7
81
2.03
Felbrigg
Norfolk
28
1715-2009
52.54
1.16
55
0.33
12
153.36
2.44
Foxley
Norfolk
29
1797-2009
52.45
1.01
39
0.37
10
123.24
2.57
Hethersett
Norfolk
22
1828-2008
52.35
1.10
41
0.35
11
134.77
2.43
Hevingham
Norfolk
32
1806-1981
52.44
1.15
20
0.37
10
95.56
2.07
Sandringham
Norfolk
33
1758-1988
52.49
0.30
30
0.39
9
128.03
2.91
Sotterley
Suffolk
62
1590-2004
52.24
1.36
25
0.38
10
188.84
1.85
Ballingdon Bridge
Suffolk
12
1484-1790
52.02
0.43
26
0.38
10
93.17
2.23
Cambridge
Cambridgeshire
31
802-1760
52.12
0.07
16
0.43
8
94.23
1.34
Castle Acre Priory
Norfolk
26
1237-1386
52.42
0.40
30
0.31
13
79.81
2.09
Denton
Norfolk
14
1220-1568
52.26
1.21
35
0.37
10
86
1.82
Ely Cathedral
Cambridgeshire
11
891-1179
52.23
0.15
29
0.31
13
117.91
1.28
Felbrigg Hall
Norfolk
16
1536-1684
52.54
1.16
55
0.41
8
95.25
1.82
Hapton
Norfolk
6
1432-1533
52.31
1.12
30
0.53
5
74.67
2.03
Harleston
Norfolk
4
1693-1768
52.24
1.17
30
0.6
4
69.25
2.83
Ipswich
Suffolk
8
1008-1293
52.03
1.09
20
0.17
25
111
1.34
Kings Lynn
Norfolk
18
1244-1583
52.44
0.25
10
0.2
22
109.11
1.73
New Buckenham
Norfolk
17
1271-1683
52.28
1.04
50
0.35
11
101.76
1.59
Norwich
Norfolk
26
781-1426
52.37
1.18
10
0.33
12
99.54
1.66
Otely Hall
Suffolk
42
1386-1553
52.09
1.13
66
0.37
10
90.71
2.54
Oxburgh Hall
Norfolk
14
1221-1591
52.34
0.34
15
0.45
7
135.5
1.16
Paston
Norfolk
22
1356-1568
52.51
1.26
20
0.27
16
117.68
1.26
Peterborough
Cathedral
Cambridgeshire
127
887-1656
52.34
0.14
17
0.28
15
151.66
1.16
Semer Bridge
Suffolk
2
1501-1696
52.04
0.55
29
N/A
N/A
87.5
2.09
Sutton
Suffolk
5
1508-1615
52.03
1.21
12
0.5
6
80.4
2.41
Tacolneston
Norfolk
3
1497-1594
52.3
1.09
60
0.43
8
85
2.05
Thetford
Norfolk
3
1461-1530
52.24
0.45
20
0.75
2
58.67
2.15
Tibenham
Norfolk
5
1525-1640
52.27
1.08
50
0.51
6
100.2
2.28
Table SM2 Generalized soil properties for the nine modern oak chronology locations from the
(National Soil Resources Institute, 2010)
Site
Texture
Acidity/Basicity
Permeability
Fertility
Babingley
Sandy
Very acidic
Naturally Wet
Very Low
Blickling
Loamy
Slightly acidic
Freely draining
Low
Slightly impeded
ModerateHigh
Bradfield
Loamy/Clay
Slightly acidic
Felbrigg
Loamy
Slightly acidic
Freely draining
Low
Foxley
Loamy/Clay
Slightly acidic, but
base-rich
Seasonally-wet, impeded
drainage
Moderate
Hethersett
Loamy/Clay
Acidic
Slightly impeded
ModerateHigh
Hevingham
Loamy
Slightly acidic
Freely draining
Low
Sandringham
Sandy
Slightly acidic
Freely draining
Low
Sotterley
Loamy/Clay
Slightly acidic, but
base-rich
Seasonally-wet, impeded
drainage
Moderate
Table SM3 Comparison of the actual 20 driest MAMJJ seasons in the gridded
instrumental precipitation series and the 20 driest seasons reconstructed from the East
Anglian oak chronology. The reconstruction successfully estimates 10 of the 20 driest
seasons in the instrumental record (highlighted in orange)
Reconstructed
Season
1949
1996
1957
1976
1956
1972
1943
1944
1934
1907
1933
1959
1948
1938
1909
1935
1989
2006
1970
1995
Driest
Reconstructed
Precipitation (mm)
178.0
185.3
195.2
207.5
211.2
212.1
213.4
214.3
214.5
214.7
216.3
217.8
218.0
218.5
219.1
220.9
221.6
221.6
221.8
224.2
Instrumental
Season
1976
1921
1996
1990
1949
1961
1943
1938
1959
1944
1995
1929
1957
1952
1974
1904
1945
1962
1934
1955
Driest
Instrumental
Precipitation (mm)
97.8
121.9
134.3
148.8
149.9
164.9
170.9
172.7
178.9
181.7
188.0
191.5
192.8
194.6
196.5
198.5
202.4
202.8
203.4
203.4
Table SM4 Comparison of the actual 20 wettest MAMJJ seasons in the gridded
instrumental precipitation series and the 20 wettest seasons reconstructed from the
East Anglian oak chronology. The reconstruction successfully estimates 8 of the 20
wettest seasons in the instrumental record (highlighted in blue)
Reconstructed
Season
2001
1917
1979
1980
2000
1982
1947
1922
1937
2008
1942
1964
1999
1927
1958
1939
1901
1978
1966
1929
Wettest
Reconstructed
Precipitation (mm)
323.6
313.5
309.7
306.9
300.1
299.9
293.0
289.3
287.1
280.7
278.7
276.7
274.8
274.3
274.1
273.7
271.5
271.0
270.8
270.1
Instrumental
Season
1981
1987
2007
1985
2008
2001
1937
1978
1903
1969
1910
1980
1988
1983
1927
1964
1951
1931
1963
1979
Wettest
Instrumental
Precipitation (mm)
332.8
329.4
326.4
325.1
319.7
314.9
312.4
310
309.8
302.9
300.7
300.7
299.0
297.8
295.8
291.6
290.9
290.8
290.4
288.4
Table SM5 Comparison of the 20 highest instrumental MAMJJ mean sea level
pressure (SLP) seasons over East Anglia (Trenberth and Paolino 1980) and the 20 driest
seasons reconstructed from the East Anglian oak chronology. The reconstruction
successfully estimates 8 of the 20 highest pressure (i.e. driest) seasons (orange)
Reconstructed
Season
1949
1996
1957
1976
1956
1972
1943
1944
1934
1907
1933
1959
1948
1938
1909
1935
1989
2006
1970
1995
Driest
Reconstructed
Precipitation (mm)
178.0
185.3
195.2
207.5
211.2
212.1
213.4
214.3
214.5
214.7
216.3
217.8
218.0
218.5
219.1
220.9
221.6
221.6
221.8
224.2
Instrumental
Season
1938
1990
1945
1943
1949
1973
1997
1976
1921
1929
1955
1961
1982
1948
1996
1944
1967
1953
2003
1906
Highest
Instrumental Mean
SLP (mb)
1020.2
1019.8
1019.6
1019.5
1019.3
1019.1
1018.9
1018.8
1018.8
1018.4
1018.2
1018.1
1018.0
1018.0
1018.0
1017.9
1017.9
1017.8
1017.7
1017.7
Table SM6 Comparison of the 20 lowest instrumental MAMJJ mean sea level pressure
(SLP) seasons (Trenberth and Paolino 1980) over East Anglia and the 20 wettest
seasons reconstructed from the East Anglian oak chronology. The reconstruction
successfully estimates 6 of the 20 lowest pressure (i.e. wettest) seasons (blue)
Reconstructed
Season
2001
1917
1979
1980
2000
1982
1947
1922
1937
2008
1942
1964
1999
1927
1958
1939
1901
1978
1966
1929
Wettest
Reconstructed
Precipitation (mm)
323.6
313.5
309.7
306.9
300.1
299.9
293.0
289.3
287.1
280.7
278.7
276.7
274.8
274.3
274.1
273.7
271.5
271.0
270.8
270.1
Instrumental
Season
Lowest Instrumental
Mean SLP (mb)
1930
1916
1931
1937
1932
1936
1928
1934
2008
1924
1909
1947
1927
2001
1903
1910
1912
1951
1985
1922
1012.9
1012.9
1013.1
1013.1
1013.2
1013.2
1013.2
1013.4
1013.4
1013.6
1013.6
1014.0
1014.1
1014.1
1014.1
1014.1
1014.1
1014.2
1014.4
1014.5
Supporting Material References
Esper J, Frank DC, Wilson RJS, Briffa KR (2005) Effect of scaling and regression on reconstructed temperature
amplitude for the past millennium. Geophysical Research Letters 32: L07711
Jones PD, Jónsson T and Wheeler D (1997) Extension to the North Atlantic Oscillation using early instrumental
pressure observations from Gibraltar and South-West Iceland. International Journal of Climatology 17:
1433-1450
Mitchell TD, Jones PD (2005) An improved method of constructing a database of monthly climate observations
and associated high- resolution grids. International Journal of Climatology 25: 693–712
National Soil Resources Institute (2010) SoilscapesTM. Cranfield University, Accessed 26/08/2010 Online:
http://wwwlandisorguk/soilscapes/
Trenberth KE, Paolino DA (1980) The Northern Hemisphere sea level pressure data set: Trends, errors, and
discontinuities. Monthly Weather Review 108: 855-872
van der Schrier G, Briffa KR, Jones PD, Osborn TJ (2006) Summer moisture variability across Europe. Journal
of Climate 19: 2828-2834