BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit
publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to
critical research.
KNMI Climate Explorer: A Web-Based Research Tool for
High-Resolution Paleoclimatology
Author(s): Valerie Trouet and Geert Jan Van Oldenborgh
Source: Tree-Ring Research, 69(1):3-13.
Published By: Tree-Ring Society
DOI: http://dx.doi.org/10.3959/1536-1098-69.1.3
URL: http://www.bioone.org/doi/full/10.3959/1536-1098-69.1.3
BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the
biological, ecological, and environmental sciences. BioOne provides a sustainable online
platform for over 170 journals and books published by nonprofit societies, associations,
museums, institutions, and presses.
Your use of this PDF, the BioOne Web site, and all posted and associated content
indicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/
terms_of_use.
Usage of BioOne content is strictly limited to personal, educational, and non-commercial
use. Commercial inquiries or rights and permissions requests should be directed to the
individual publisher as copyright holder.
RESEARCH TOOLS
KNMI CLIMATE EXPLORER: A WEB-BASED RESEARCH TOOL FOR
HIGH-RESOLUTION PALEOCLIMATOLOGY
VALERIE TROUET1
and GEERT JAN VAN OLDENBORGH2
*
1
Laboratory of Tree-Ring Research, University of Arizona, Tucson, AZ, 85721, USA
2
Royal Netherlands Meteorological Institute KNMI, P.O. Box 201, NL-3730 AE De Bilt, Netherlands
ABSTRACT
Climate Explorer (www.climexp.knmi.nl) is a web-based application for climatic research that is
managed by the Royal Netherlands Meteorological Institute (KNMI) and contains a comprehensive
collection of climatic data sets and analysis tools. One of its fields of application is high-resolution
paleoclimatology. We show how Climate Explorer can be used to explore and download available
instrumental climate data and derived time series, to examine the climatic signal in uploaded highresolution
paleoclimate time series, and to investigate the temporal and spatial characteristics of
climate reconstructions. We further demonstrate the value of Climate Explorer for high-resolution
paleoclimate research using a dendroclimatic data set from the High Atlas Mountains in Morocco.
Keywords: KNMI Climate Explorer, paleoclimatology, dendroclimatology, time-series analysis,
climate data, composite analysis, correlation map, spectral analysis.
INTRODUCTION
Since 1999, the Royal Netherlands Meteorological
Institute KNMI has been managing and
updating the Climate Explorer (climexp.knmi.nl),
a web-based application for climatic research that
contains a comprehensive collection of climatic
data sets and analysis tools. After free registration,
researchers can explore and download an array of
climatic data sets, generate derived data, upload
their own time series, and run statistical analyses
to investigate and compare data sets. The KNMI
Climate Explorer (hereafter referred to as CE) was
originally intended for ENSO teleconnection
analysis (van Oldenborgh and Burgers 2005). It
has rapidly grown to become a useful tool for
other climate research fields such as climate
change research (van Oldenborgh et al. 2009)
and the paleoclimate research community currently
forms one of its largest user communities (e.g.
Abram et al. 2008; Camuffo et al. 2010; Poljansek
et al. 2012). Its primary uses for high-resolution
paleoclimatology include (1) exploring and downloading
available instrumental climate data and
derived time series, (2) examining the climatic
signal in uploaded high-resolution paleoclimate
time series, and (3) investigating the temporal and
spatial characteristics of climate reconstructions.
In this note, we demonstrate the abovementioned
uses of CE for dendroclimatic analysis,
but the techniques can easily be extrapolated to
other high-resolution paleoclimatic data sets (e.g.
coral data and varved sediments). A PDF document
that includes a step-by-step instruction
manual organized according to the same structure
as the main text is available from the journal
website as Supplementary Material. We restrict
our demonstration to a dendroclimatic data set,
but it will become clear that several applications
are also useful for other dendrochronological
fields, particularly event-based dendro-ecology
(e.g. Trouet et al. 2010). It is worth mentioning
that CE supports a comprehensive help overview
(http://climexp.knmi.nl/help.cgi) that can be consulted
for additional information. Throughout this
paper, we use an Atlas cedar (Cedrus atlantica)*Corresponding author: oldenborgh@knmi.nl
TREE-RING RESEARCH, Vol. 69(1), 2013, pp. 3–13
Copyright ’ 2013 by The Tree-Ring Society 3
tree-ring width (TRW) chronology from Jaffa
(JAF) in the High Atlas Mountains in Morocco
(Esper et al. 2007) as an example of a dendroclimatic
data set. The chronology is based on 52 treering
series from 32 trees that cross-correlate
significantly (average cross-correlation r-value 5
0.83) and cover the period A.D. 1021–2001. For
the purpose of this exercise, the individual
contributing series were detrended using a negative
exponential curve of any slope to remove
growth trends (Fritts 1976).
EXPLORATION OF INSTRUMENTAL
DATA SETS
CE provides a plethora of instrumental
climate data sets in the form of either time series
(station data and climate indices) or gridded fields
(observations and reanalysis fields). All data sets
can be accessed and investigated by selecting the
appropriate time series or fields format on the CE
web page. For dendroclimatic analysis, monthly
time series and fields are most useful, and we will
therefore focus on the monthly data sets. Note,
however, that specific daily climate data can easily
be transformed to monthly and lower resolution in
CE. The search for available instrumental data for
a certain location (e.g. a tree-ring site) can be
approached in two ways by selecting (1) station
data or (2) gridpoint data. When a time series or
derived time series (see below) of either kind is
selected, it is important to make it available for
further analysis by adding it to the list of userdefined
time series.
Selection of Available Station Data for a Site
Amongst other databases, CE provides access
to two NOAA/NCDC Global Historical Climatology
Network monthly (GHCN-M) v2 databases
of adjusted and unadjusted monthly station
data (Diamond and Lief 2009). Both data sets
contain monthly precipitation and mean temperature
data that are valuable for dendroclimatic
analysis. The adjusted set is corrected for urban
effects and other biases and is generally preferable,
but its availability is limited in time, space, and
variable. Comparing the outcome of both data set
selections is therefore recommended. It is worth
noting that monthly or daily time series of other
climatic variables are also available, which can be
useful for dendroclimatic and dendrohydrological
research, including river discharge (Woodhouse
et al. 2006), cloud cover (Young et al. 2010), and
snow depth (Pederson et al. 2011).
When selecting a time series of monthly
station data, the researcher is given the opportunity
to make a selection based on station name (if
the station of interest is known) or on vicinity to a
location. The second option includes searching for
a defined number of stations nearest to a given site
or all stations within a defined region. Furthermore,
a minimum number of years of data
availability can be defined as well as a range of
years for which data should be available and an
elevation range. As an example, we searched for
the five stations nearest to JAF (32u329N 4u559W;
2200 m a.s.l.) and applied a filter of a minimum of
30 years of monthly data availability. No adjusted
GHCN precipitation data are available for Morocco
or its neighboring countries (the nearest
station is in Chimay, Belgium, 49.98uN 4.35uE).
For further analysis, we thus selected the nonadjusted
station with the longest available time
series (Meknes; 33.90uN 5.50uW, 549 m a.s.l.;
1931–2011). In general, researchers will take
length of time series, proximity to site, and
elevation into account when selecting (a) station(s)
for dendroclimatic analyses, as well as availability
of various climatic variables. Selection is likely to
be more complex for precipitation data than for
temperature data because of the strong localized
effects on interannual precipitation variability that
arise from orography and other physical processes,
versus more broad patterns in interannual
temperature variability (Bu¨ntgen et al. 2010).
Upon selecting a station, the researcher is
offered three plots of the absolute value time series,
the annual cycle of the climatic variable at this site,
and the time series of anomalies with respect to this
annual cycle. Furthermore, the selected time series
can be high- and low-pass filtered: high-pass
filtering eliminates the influence of low-frequency
variability and trends when calculating crosscorrelations,
whereas low-pass filtering can be of
use when working with non-annually resolved
4 TROUET and VAN OLDENBORGH
paleoclimate data. Also lower resolution (monthly,
seasonal, and annual) time series of average,
maximum, or minimum values can be derived from
the selected data series. Using this function,
threshold-delimited time series can be created that
represent the sum above or below or the number of
values that reach a defined threshold. If daily data
are available for a site, this function can be used to
calculate monthly number of growing degree days
by creating a monthly time series of the number of
days with a mean temperature greater than 10uC.
Selection of Gridpoint Data
For regions where publicly available station
data are few or far between, gridpoint data can
provide a valuable alternative. An added advantage
of gridpoint data is that long time series are
available for many climate variables. These
datasets come in three flavors: analyses, reconstructions,
and reanalyses. An analysis (e.g. the
HADCRU observational surface temperature
data sets; Jones 1994; Jones and Moberg 2003;
Brohan et al. 2006; Morice et al. 2012) is simply an
interpolation of all available data, which often
includes non-public time series that are not
available for download. Areas and times without
data are left blank. A reconstruction (e.g. the
GISS surface temperature data sets, Hansen et al.
1981; Hansen and Lebedeff 1987; Hansen et al.
2010) functions in the same way, except that data
gaps are filled to the maximum extent possible.
These gridpoints/time steps often do not contain
useful information and should be avoided. A
reanalysis (e.g. the NCAR/NCEP reanalysis data
set, Kalnay et al. 1996) uses a weather model to
interpolate between the observations, which is the
optimal way to propagate information to datasparse
areas. However, this power comes at the
expense of model biases and inhomogeneities at
the time when new data sources become available
(e.g. satellite data in the late 1970s).
For dendroclimatology, the CRU TS3.10
data set of monthly gridded (0.5u, 1u, and 2.5u)
fields, developed by the Climate Research Unit of
the University of East Anglia (Mitchell and Jones
2005) is of particular interest because of its global
coverage, its long temporal extent (1901–2009),
and its abundance of climatic variables. These
variables include the self-calibrating Palmer Drought
Severity Index (scPDSI, Wells et al. 2004), a variant
on the original PDSI drought index (Palmer 1965)
that incorporates lagged time series of precipitation
and temperature, as well as soil characteristics, is
comparable for different climate regimes, and is a
useful drought metric in regions where drought
rather than temperature is the limiting factor for
tree growth (e.g. Cook et al. 2004). A data set
comparable in spatial and temporal resolution and
extent to the CRU TS3.10 data set, but including
precipitation data only, is produced by the Global
Precipitation Climatology Centre (GPCC; http://
gpcc.dwd.de) and provides gridded monthly precipitation
sets based on data from ca. 64,400
stations.
Additional climate variables, including sea
level pressure (SLP), geopotential height fields
at various elevations, and a variety of derived
variables (e.g. wind speed and relative humidity),
are available in the NCAR/NCEP Reanalysis data
set (Kalnay et al. 1996), which offers a 2u global
grid of modeled monthly variables. Because of its
relatively long time series (1948-present), this data
set is valuable for dendroclimatology and particularly
for the analysis of atmospheric circulation
patterns (e.g. a reconstruction of the Pacific North
American (PNA) pattern in Trouet and Taylor
2010; see also sections 2 and 3). However, it should
be kept in mind that the weather model used to
produce this reanalysis is of mid-1990s vintage and
has some large biases, particularly in surface air
temperature and precipitation (Kistler et al. 2001;
Harnik and Chang 2003; Sturaro 2003; Bromwich
and Fogt 2004). Recently, the Twentieth Century
Reanalysis data set (Compo et al. 2011) has been
uploaded to CE, which covers the same variables
over even longer time-scales (1878–2010).
Note that a large range of gridded data sets is
available in CE, which are not explicitly mentioned
here. The aim of a study and the role of
gridded data sets in this study will determine
which data set should best be used. It is therefore
recommended to read the information links
available for all data sets before making a
decision. These links also provide the necessary
citations for all data sets.
KNMI Climate Explorer 5
Upon selection of a field, the researcher is
given the choice to retrieve data for a single
gridpoint or for an area. Similar options to derive
high- and low-pass filtered and lower resolution
time series are provided for gridpoint data as
previously described for station data. Differences in
the intra- and inter-annual precipitation variability
for three potential instrumental precipitation data
sets for the JAF chronology are shown in Figure 1.
The three data sets include precipitation data from
the Meknes station data and the nearest gridpoint
data from the CRU TS3.0 (0.5u) and GPCC (0.5u)
databases. The intra- and interannual comparisons
demonstrate that the GPCC data for this location
generally have lower values than the two other data
sets, which reflect comparable absolute amounts of
precipitation. The Meknes station data on the other
hand has a shorter time-series length than the other
two data sets and includes a substantial amount of
missing data from 1984 onwards. Taking these
factors into account, the CRU TS3.0 data set
appears to be the most reliable record for calibration
of the JAF chronology.
EXAMINATION OF THE CLIMATE
SIGNAL IN DENDROCLIMATIC
TIME SERIES
Uploading a Tree-Ring Series
CE is a powerful tool for initial and exploratory
examination of dendroclimatic time series. It allows
uploading of a user-defined time series and running
a variety of statistical analyses. Uploaded time series
will remain accessible in CE for three days after last
use. After uploading, a plot of the data can be
viewed, as well as a plot of the anomalies with
respect to the average over a defined range of years.
When uploading a dendroclimatic time series,
it is important to be aware that CE only correlates
data sets with the same temporal resolution (i.e.
monthly with monthly, annual with annual).
Therefore, uploaded and existing data sets should
be converted to the same resolution. To correlate
an annually-resolved tree-ring series with annual
instrumental climate data, an annually-resolved
climate time series can be derived from monthly (or
daily) data as described above. When a comparison
of tree-ring data with monthly climate data (e.g. for
response function analysis) is desired, the tree-ring
series should be converted to monthly resolution.
An annual or seasonal time series can be uploaded
and reduced to monthly resolution in CE, which
also allows for a choice of which months the signal
is most representative. An alternative (and likely
more straightforward) method for calculating
seasonal comparisons is described in the next
subsection. It is worth noting that ‘‘downscaling’’
an annual tree-ring time series to monthly or
seasonal resolution, as described above, can redden
the spectrum of the time series and can thus
introduce artifacts in the spectral properties (see
next section) of the downscaled time series (Schulz
and Stattegger 1997; Schulz and Mudelsee 2002).
Correlation of Tree-Ring Time Series with Other
Time Series
CE allows the user to correlate an uploaded
time series with other user-defined time series as
well as with system-defined time series (including
various climate indices) of the same temporal
Figure 1. Precipitation climatology (A) and interannual annual
(Jan–Dec) precipitation variability (B) for three instrumental
data sets in the vicinity of Jaffa, Morocco (32.53uN, 4.9uW).
The data sets include the Meknes meteorological station
(33.90N, 5.50W) and gridpoint data from the CRUTS3.0 and
GPCC (0.5u; 33uN, 5uW) databases.
6 TROUET and VAN OLDENBORGH
resolution. We comment on the investigation of
monthly-resolved dendroclimatic time series, but
analyses of seasonally and annually resolved series
are comparable.
Analysis options include (rank) correlation or
(least-square) regression analysis (Press et al. 1992,
Ch. 14, 15) for individual or all months of the
calendar year (January-December). The user is
also given the option to calculate correlation
coefficients using the average of multiple months
of climate data. Furthermore, there is the option
of calculating lagged correlations, which allows
for the investigation of the influence of climatic
conditions during the previous year on current
year growth. Positive lags represent the tree-ring
series lagging the climate data by a defined number
of months. Given the difference in growth year
for the tree-ring data (roughly April-September for
the NH) and calendar year for the climate data
(January-December), caution is warranted when
interpreting the generated results. It is often best to
investigate one lag and season separately to verify
that the months (shown on the axes of the scatter
plot) are correct.
All results are accompanied by estimates of
the significance in the form of a p-value for a twosided
t-test assuming the underlying data are
normally distributed (Press et al. 1992, Ch. 14).
A 95% confidence interval on the correlation or
regression coefficient is computed using a nonparametric
bootstrap (Efron and Tibshirani 1998).
If the data have been low-pass filtered prior to the
correlation step, the user has to specify the
decorrelation length (i.e. the cut-off length of
the applied filter) explicitly for the computation of
the significance and error bounds.
Other selection options include a range of
years over which the correlation is run and upper
and lower thresholds for both tree-ring and
climate data. Selecting one or several threshold
values restricts the range of values considered for
correlation. Furthermore, running correlation and
regression analyses can be performed with a userdefined
time-window and the significance of
running correlations can be tested with a Monte
Carlo test (Gershunov 2001). CE also offers the
opportunity to detrend the tree-ring and climatic
series, but the detrending filters offered are
generally too coarse for dendroclimatic analysis
and we recommend detrending the tree-ring series
using dendro-specific software (e.g. ARSTAN,
Cook 1985) before uploading in CE.
Correlation analysis and regression analysis
require a normal variable distribution. When
variables are not normally distributed (e.g. snowpack,
fire occurrence, insect outbreaks), applying
a logarithmic or square root transformation to the
variable can bring the distribution closer to
normal and improve the results. Another option
is to compute rank correlation coefficients (Press
et al. 1992) that do not make assumptions about
the underlying distribution. Finally, CE provides
the option to fit non-linear (parabolic, cubic)
functions to the tree-ring time series.
Field Correlations
Time series uploaded in CE can be not only
compared to other time series, but also to gridded
fields of observational and reanalysis climate data.
These field analyses typically result in correlation,
regression, or composite maps (Von Storch and
Zwiers 2002; Mudelsee 2010) that can be used for
data exploration, but also to examine the spatial
extent of the climatic signal present in a tree-ring
series (Treydte et al. 2007; Fan et al. 2009;
Bu¨ntgen et al. 2010) and to investigate teleconnection
patterns (Buckley et al. 2007; D’Arrigo et al.
2008; Trouet et al. 2012). Correlation maps for
SLP and geopotential height fields are particularly
useful in the analysis of atmospheric circulation
patterns (Trouet et al. 2009; Trouet and Taylor
2010; Baker et al. 2011; D’Arrigo et al. 2011).
Once the appropriate field is selected, the
options for field correlation analysis are comparable
to those for the time-series correlation
analysis as described in the previous subsection.
Additional options refer to the map output: the
CE user can select map projection, boundaries,
and color scale and can define contours and
significance levels to be shown/masked out. The
resulting maps can be saved as pdf or eps files or
can be output as kml files for use in Google Earth
or as GeoTIFF files for GIS software. For
computational efficiency, the levels of significance
are computed using a two-sided t-test, taking serial
KNMI Climate Explorer 7
correlations into account wherever these are
statistically significant at p , 0.05. Again, if the
data are not normally distributed and cannot be
mapped into a normal distribution using a
logarithm or square root, a rank correlation can
be computed instead. Finally, it should be kept in
mind that even in the absence of a real physical
mechanism, generated maps will on average show
10% of the area colored with correlations with p ,
0.1 because of random fluctuations. CE estimates
roughly whether the area on the map is significantly
larger than expected by pure chance and
gives an indication of this ‘field significance’.
Interpretation of correlation maps also depends
strongly on the physics: a small area at the
expected position is more likely a true signal than
a large area at the other side of the globe.
Another important additional option is the
ability to create composite maps (e.g. Hastenrath
1976, 1978; Ropelewski and Halpert 1987; Galarneau
et al. 2010). In composite map analysis,
climatic field variables are averaged over selected
years and the significance of the difference
between this average and the average of the
variable over the full time series is estimated and
indicated on the resulting map (non-significant
fields are masked out by default). A comparison of
correlation and composite maps for the JAF treering
chronology with gridded precipitation fields
(Figure 2) demonstrates that composite maps for
continuous, normally distributed time series (most
dendroclimatic series) will largely resemble correlation
maps, albeit with more noise (less statistical
power). The composite map approach is more
appropriate for non-continuous, event-based time
series (e.g. wildfire activity reconstruction in
Trouet et al. 2010), for continuous time series
that are not normally distributed, or when a nonlinear
relationship is suspected.
DENDROCLIMATIC
TIME-SERIES ANALYSIS
CE includes a number of gridded climate
fields reconstructed from proxy and historical
climate data that include seasonal temperature
(A.D. 1500–2002; Luterbacher et al. 2004; Xoplaki
et al. 2005), precipitation (A.D. 1500–2000;
Figure 2. Correlation (A) and composite (B) maps for the JAF tree-ring chronology with February–June CRUTS3.0 gridded (0.5u)
precipitation fields (1901–2001). Composite maps are calculated for the 90th
percentile of the JAF chronology over the period 1901–
2001 (i.e. 10 years with highest TRW value). Colored contours represent significant correlation coefficients in A and significant
anomalies from the 1901–2001 average in B.
8 TROUET and VAN OLDENBORGH
Pauling et al. 2006), sea level pressure (SLP; A.D.
1750–2008; Luterbacher et al. 2002; Kuttel et al.
2010), and 500 hPa geopotential height (A.D.
1658–1999; Luterbacher et al. 2002) for Europe,
summer temperature and SLP (A.D. 1600–1877;
Briffa et al. 2002) for the circum-Arctic, and
summer PDSI for the contiguous U.S.A. (A.D.
1000–2003; Cook et al. 2004) and for Monsoon
Asia (A.D. 1300–2005; Cook et al. 2010).
Comparing uploaded tree-ring series to these
reconstructed climate fields allows for the development
of correlation, regression, and composite
maps with similar options as described above
(Figure 3).
To complete the spatiotemporal analysis of
climate reconstructions, CE provides the opportunity
to perform simple spectral, autocorrelation,
and wavelet analysis. CE can be used to compute
Figure 3. Correlation maps for the JAF tree-ring chronology with reconstructed (March–May) fields of (A) temperature (A.D.
1500–2001; Luterbacher et al. 2004); (B) precipitation (1500–2000; Pauling et al. 2006), (C) SLP (1659–1999; Luterbacher et al.
2002), and (D) 500 hPa geopotential height (1658–1999; Luterbacher et al. 2002). Only weak correlations are found with eastern
European temperature fields, but the precipitation field shows a consistent pattern of positive correlation with precipitation over the
Western Mediterranean and negative correlations over northwestern Europe. Circulation reconstructions (SLP and 500 hPa
geopotential height) show weak but significant signals over northwest Africa. The precipitation correlation map (B) furthermore
shows a strong resemblance with the correlation map for instrumental precipitation fields (Figure 2A). Statistically significant (p ,
0.1) correlations with absolute values lower than 0.1 are indicated by grey shading.
KNMI Climate Explorer 9
the autocorrelation function an~SXiXiznT=SXiXiT
and use this to examine the association between
current and previous values in a time series.
Autocorrelation coefficients are plotted as a
function of time-lag n and can be used to quantify
the nature of autoregressive linkage in the time
series (Fritts 1976), to estimate the decorrelation
length needed to compute significances (see above),
and to investigate oscillating patterns in time series.
Spectral analysis is a more powerful way than
autocorrelation function analysis to describe cyclicity
in time series by examining the strength of a
periodic signal (i.e. the spectral power) in time series
at different time frequencies (Jenkins and Watts
1968). It is worth noting that, in contrast to the
autocorrelation, there are many different definitions
of the spectrum. The evolution over time of the
periodic signal can be analyzed using wavelet
analysis (Torrence and Compo 1998). Performing
a spectral analysis in CE results in a periodogram or
spectrum in which the spectral power of the time
series is plotted per cycle length (or 1/frequency) and
significance levels of the spectral power based on
Monte Carlo permutation tests are provided
(Figure 4A). When producing a periodogram in
CE, which is a simple Fourier transform of the input
data, the spectral information from neighboring
frequencies (and thus cycle lengths) can be averaged
into bins, a simple boxcar spectrum. This reduces the
number of frequency bins, but also the uncertainty
inherent to each bin. The optimal number of bins to
average over is therefore a trade-off between the
error bars on the vertical (spectral power) and
horizontal (frequency) axes of the spectrum. More
sophisticated filtering options are not yet available.
The raw data for this periodogram or
spectrum are available in column format, including
the frequency in the first column, the spectral
power in the second column, the power of a firstorder
autoregression (AR(1)) process fitted to the
lag-1 autocorrelation (used for significance level
calculations) in the third, and the significance level
(p-value) in the fourth column. If there are more
peaks in the spectrum above red-noise AR(1)
process than expected by chance, the location of
these is noted in text above the plot. Figure 4A
shows the spectrum for JAF: we averaged
frequencies into 10-frequency bins, which resulted
in 49 (rather than the original 490) frequency bins.
JAF contains statistically significant cyclicity with
cycles of approximately 2.5, 4 (p , 0.01), and 12
(p , 0.05) years in length.
A wavelet analysis (Figure 4B) shows that the
2-to-5 year cyclicity is fairly constant over time,
whereas 10–20 year cycles are only prominent
prior to A.D. 1500 and in the 20th
Century.
Running a wavelet analysis in CE thus results in a
plot of the periodicity of the time series over time,
with the strength (and significance level) of the
spectral power represented by shading and a black
line indicating the cone of influence. CE provides
the option to select different types (Morlet, Paul,
or DOG) and orders of wavelet functions, but for
the purpose of paleoclimatology, this selection will
have little influence on the resulting wavelet
transform. Again, be aware that on average 10%
of the area of the plot will appear significant at
p , 0.1 even in the absence of any physical signal.
CONCLUSIONS
In this note, we provide a general introduction
to the use of CE for high-resolution paleoclimatology.
We describe a number of options for the
exploration of instrumental data sets, and for the
climatic and spatiotemporal analysis of user-defined
paleoclimatic time series. It is worth noting,
however, that the described options do not constitute
the full spectrum of options available in CE and
that CE is a dynamic operation that is regularly
updated with new data sets and analysis tools. We
therefore encourage researchers to further explore
methods for engaging the CE tool in answering
research-specific questions. Possible CE applications
that are not explicitly covered in this paper and
that could be further explored include the analysis
of event-based (not normally distributed) time series
and of lower-frequency (non-annual) time series
and the comparison of paleoclimate time series with
modeling simulations. Plans exist to link CE to the
NOAA Paleoclimatology Website (http://www.
ncdc.noaa.gov/paleo/data.html), which includes
the International Tree-Ring Database (ITRDB).
This collaboration will make a set of basic paleoclimatic
research skills attainable for researchers from
a wide range of disciplines.
10 TROUET and VAN OLDENBORGH
REFERENCES CITED
Abram, N. J., M. K. Gagan, J. E. Cole, W. S. Hantoro, and M.
Mudelsee, 2008. Recent intensification of tropical climate
variability in the Indian Ocean. Nature Geosci 1(12):849–853.
http://www.nature.com/ngeo/journal/v1/n12/suppinfo/ngeo357_
S1.html.
Baker, A., R. Wilson, I. J. Fairchild, J. Franke, C. Spotl, D. Mattey,
V. Trouet, and L. Fuller, 2011. High resolution d18
O and d13
C
records from an annually laminated Scottish stalagmite and
relationship with last millennium climate. Global and Planetary
Change 79(3–4):303–311. 10.1016/j.gloplacha.2010.12.007.
Briffa, K. R., T. J. Osborn, F. H. Schweingruber, P. D. Jones,
S. G. Shiyatov, and E. A. Vaganov, 2002. Tree-ring width
Figure 4. Spectrum (10 period bins) (A) and morlet 6 wavelet transform (B) of the JAF tree-ring chronology (A.D. 1021–2001).
KNMI Climate Explorer 11
and density data around the Northern Hemisphere: Part 1,
local and regional climate signals. Holocene 12(6):737–757.
10.1191/0959683602hl587rp.
Brohan, P., J. J. Kennedy, I. Harris, S. F. B. Tett, and P. D.
Jones, 2006. Uncertainty estimates in regional and global
observed temperature changes: A new data set from 1850.
Journal of Geophysical Research-Atmospheres 111(D12).
D12106. 10.1029/2005jd006548.
Bromwich, D. H., and R. L. Fogt, 2004. Strong trends in the
skill of the ERA-40 and NCEP-NCAR reanalyses in the high
and midlatitudes of the southern hemisphere, 1958–2001.
Journal of Climate 17(23):4603–4619. 10.1175/3241.1.
Buckley, B. M., K. Palakit, K. Duangsathaporn, P. Sanguantham,
and P. Prasomsin, 2007. Decadal scale droughts over
northwestern Thailand over the past 448 years: Links to the
tropical Pacific and Indian Ocean sectors. Climate Dynamics
29(1):63–71. 10.1007/s00382-007-0225-1.
Bu¨ntgen, U., J. Franke, D. Frank, R. Wilson, F. GonzalezRouco,
and J. Esper, 2010. Assessing the spatial signature of
European climate reconstructions. Climate Research 41(2):
125–130. 10.3354/Cr00848.
Camuffo, D., C. Bertolin, M. Barriendos, F. DominguezCastro,
C. Cocheo, S. Enzi, M. Sghedoni, A. della Valle, E.
Garnier, M. J. Alcoforado, E. Xoplaki, J. Luterbacher, N.
Diodato, M. Maugeri, M. F. Nunes, and R. Rodriguez, 2010.
500-year temperature reconstruction in the Mediterranean
Basin by means of documentary data and instrumental
observations. Climatic Change 101(1–2):169–199. 10.1007/
s10584-010-9815-8.
Compo, G. P., J. S. Whitaker, P. D. Sardeshmukh, N. Matsui,
R. J. Allan, X. Yin, B. E. Gleason, R. S. Vose, G. Rutledge,
P. Bessemoulin, S. Bronnimann, M. Brunet, R. I. Crouthamel,
A. N. Grant, P. Y. Groisman, P. D. Jones, M. C. Kruk,
A. C. Kruger, G. J. Marshall, M. Maugeri, H. Y. Mok, O.
Nordli, T. F. Ross, R. M. Trigo, X. L. Wang, S. D.
Woodruff, and S. J. Worley, 2011. The Twentieth Century
Reanalysis Project. Quarterly Journal of the Royal Meteorological
Society 137(654):1–28. 10.1002/qj.776.
Cook, E. R., 1985. A Time-Series Analysis Approach to TreeRing
Standardization. Ph.D. dissertation, University of
Arizona, Tucson; 171 pp.
Cook, E. R., C. A. Woodhouse, C. M. Eakin, D. M. Meko, and
D. W. Stahle, 2004. Long-term aridity changes in the western
United States. Science 306(5698):1015–1018. 10.1126/science.
1102586.
Cook, E. R., K. J. Anchukaitis, B. M. Buckley, R. D. D’Arrigo,
G. C. Jacoby, and W. E. Wright, 2010. Asian Monsoon
failure and megadrought during the last millennium. Science
328(5977):486–489. 10.1126/science.1185188.
D’Arrigo, R., P. Baker, J. Palmer, K. Anchukaitis, and G.
Cook, 2008. Experimental reconstruction of monsoon
drought variability for Australasia using tree rings and
corals. Geophysical Research Letters 35(12):L12709. 10.1029/
2008gl034393.
D’Arrigo, R., N. Abram, C. Ummenhofer, J. Palmer, and M.
Mudelsee, 2011. Reconstructed streamflow for Citarum River,
Java, Indonesia: Linkages to tropical climate dynamics. Climate
Dynamics 36(3–4):451–462. 10.1007/s00382-009-0717-2.
Diamond, H. J., and C. J. Lief, 2009. A comprehensive data
portal for global climate information. EOS, Transactions
American Geophysical Union 90(39):341. 10.1029/2009EO390001.
Efron, B., and R. Tibshirani, 1998. The problem of regions.
Annals of Statistics 26(5):1687–1718.
Esper, J., D. Frank, U. Bu¨ntgen, A. Verstege, J. Luterbacher,
and E. Xoplaki, 2007. Long-term drought severity variations
in Morocco. Geophysical Research Letters 34(17):L17702.
17710.11029/12007gl030844.
Fan, Z. X., A. Brauning, B. Yang, and K. F. Cao, 2009. Tree ring
density-based summer temperature reconstruction for the
central Hengduan Mountains in southern China. Global and
Planetary Change 65(1–2):1–11. 10.1016/j.gloplacha.2008.10.001.
Fritts, H. C., 1976. Tree Rings and Climate. Academic Press,
London; 567 pp.
Gershunov, A., N. Schneider, and T. Barnett, 2001. Lowfrequency
modulation of the ENSO-Indian Monsoon rainfall
relationship: Signal or noise? Journal of Climate 14:2486–2492.
10.1175/1520-0442(2001)014,2486:LFMOTE.2.0.CO;2.
Hansen, J., D. Johnson, A. Lacis, S. Lebedeff, P. Lee, D. Rind,
and G. Russell, 1981. Climate impact of increasing atmospheric
carbon-dioxide. Science 213(4511):957–966. 10.1126/
science.213.4511.957.
Hansen, J., and S. Lebedeff, 1987. Global trends of measured
surface air-temperature. Journal of Geophysical Research-Atmospheres
92(D11):13345–13372. 10.1029/JD092iD11p13345.
Hansen, J., R. Ruedy, M. Sato, and K. Lo, 2010. Global surface
temperature change. Reviews of Geophysics 48:Rg4004.
10.1029/2010rg000345.
Harnik, N., and E. K. M. Chang, 2003. Storm track variations
as seen in radiosonde observations and reanalysis data. Journal
of Climate 16(3):480–495. 10.1175/1520-0442(2003)016,0480:stvasi.
2.0.co;2.
Jenkins, G. M., and D. G. Watts, 1968. Spectral Analysis and Its
Applications. Holden-Day, San Francisco; 541 pp.
Jones, P. D., 1994. Hemispheric surface air-temperature
variations - a reanalysis and an update to 1993. Journal of
Climate 7(11):1794–1802. 10.1175/1520-0442(1994)007,1794:
hsatva.2.0.co;2.
Jones, P. D., and A. Moberg, 2003. Hemispheric and large-scale
surface air temperature variations: An extensive revision and
an update to 2001. Journal of Climate 16(2):206–223. 10.1175/
1520-0442(2003)016,0206:halssa.2.0.co;2.
Kalnay, E., M. Kanamitsu, R. Kistler, W. Collins, D. Deaven,
L. Gandin, M. Iredell, S. Saha, G. White, J. Woollen, Y.
Zhu, M. Chelliah, W. Ebisuzaki, W. Higgins, J. Janowiak, K.
C. Mo, C. Ropelewski, J. Wang, A. Leetmaa, R. Reynolds,
R. Jenne, and D. Joseph, 1996. The NCEP/NCAR 40-year
reanalysis project. Bulletin of the American Meteorological
Society 77(3):437–471.
Kistler, R., E. Kalnay, W. Collins, S. Saha, G. White, J. Woollen,
M. Chelliah, W. Ebisuzaki, M. Kanamitsu, V. Kousky, H. van
den Dool, R. Jenne, and M. Fiorino, 2001. The NCEP-NCAR
50-year reanalysis: Monthly means CD-ROM and documentation.
Bulletin of the American Meteorological Society 82(2):
247–267. 10.1175/1520-0477(2001)082,0247:tnnyrm.2.3.co;2.
Kuttel, M., E. Xoplaki, D. Gallego, J. Luterbacher, R. GarciaHerrera,
R. Allan, M. Barriendos, P. Jones, D. Wheeler, and
12 TROUET and VAN OLDENBORGH
H. Wanner, 2010. The importance of ship log data:
Reconstructing North Atlantic, European and Mediterranean
sea level pressure fields back to 1750. Climate Dynamics
34(7–8):1115–1128. 10.1007/s00382-009-0577-9.
Luterbacher, J., E. Xoplaki, D. Dietrich, R. Rickli, J. Jacobeit,
C. Beck, D. Gyalistras, C. Schmutz, and H. Wanner, 2002.
Reconstruction of sea level pressure fields over the Eastern
North Atlantic and Europe back to 1500. Climate Dynamics
18(7):545–561.
Luterbacher, J., D. Dietrich, E. Xoplaki, M. Grosjean, and H.
Wanner, 2004. European seasonal and annual temperature
variability, trends, and extremes since 1500. Science
303(5663):1499–1503.
Mitchell, T. D., and P. D. Jones, 2005. An improved method of
constructing a database of monthly climate observations and
associated high-resolution grids. International Journal of
Climatology 25(6):693–712. 10.1002/joc.1181.
Morice, C. P., J. J. Kennedy, N. A. Rayner, and P. D. Jones,
2012. Quantifying uncertainties in global and regional
temperature change using an ensemble of observational
estimates: The HadCRUT4 data set. Journal of Geophysical
Research-Atmospheres 117:D08101. 10.1029/2011jd017187.
Palmer, W. C., 1965. Meteorological Drought. Research paper,
U.S. Dept. of Commerce, Washington; 58 pp.
Pauling, A., J. Luterbacher, C. Casty, and H. Wanner, 2006.
Five hundred years of gridded high-resolution precipitation
reconstructions over Europe and the connection to largescale
circulation. Climate Dynamics 26(4):387–405. 10.1007/
s00382-005-0090-8.
Pederson, G. T., S. T. Gray, C. A. Woodhouse, J. L.
Betancourt, D. B. Fagre, J. S. Littell, E. Watson, B. H.
Luckman, and L. J. Graumlich, 2011. The unusual nature of
recent snowpack declines in the North American Cordillera.
Science 333(6040):332–335. 10.1126/science.1201570.
Poljansek, S., D. Ballian, T. A. Nagel, and T. Levanic, 2012. A 435year
long European black pine (Pinus nigra) chronology for the
central-western Balkan region. Tree-Ring Research 68(1):31–44.
Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery,
1992. Numerical Recipes in FORTRAN: The Art of Scientific
Computing. Cambridge University Press, New York; 963 pp.
Sturaro, G., 2003. A closer look at the climatological
discontinuities present in the NCEP/NCAR reanalysis
temperature due to the introduction of satellite data. Climate
Dynamics 21(3–4):309–316. 10.1007/s00382-003-0334-4.
Torrence, C., and G. P. Compo, 1998. A practical guide to
wavelet analysis. Bulletin of the American Meteorological
Society 79(1):61–78.
Treydte, K., D. Frank, J. Esper, L. Andreu, Z. Bednarz, F.
Berninger, T. Boettger, C. M. D’Alessandro, N. Etien, M. Filot,
M. Grabner, M. T. Guillemin, E. Gutierrez, M. Haupt, G. Helle,
E. Hilasvuori, H. Jungner, M. Kalela-Brundin, M. Krapiec,
M. Leuenberger, N. J. Loader, V. Masson-Delmotte, A. Pazdur,
S. Pawelczyk, M. Pierre, O. Planells, R. Pukiene, C. E.
Reynolds-Henne, K. T. Rinne, A. Saracino, M. Saurer, E.
Sonninen, M. Stievenard, V. R. Switsur, M. Szczepanek, E.
Szychowska-Krapiec, L. Todaro, J. S. Waterhouse, M. Weigl,
and G. H. Schleser, 2007. Signal strength and climate calibration
of a European tree-ring isotope network. Geophysical Research
Letters 34(24):L24302. 10.1029/2007gl031106.
Trouet, V., J. Esper, N. E. Graham, A. Baker, J. D. Scourse,
and D. C. Frank, 2009. Persistent positive North Atlantic
Oscillation mode dominated the Medieval Climate Anomaly.
Science 324(5923). 10.1126/science.1166349.
Trouet, V., and A. H. Taylor, 2010. Multi-century variability in
the Pacific North American circulation pattern reconstructed
from tree rings. Climate Dynamics 35(6):953–963. 10.1007/
S00382-009-0605-9.
Trouet, V., A. H. Taylor, E. R. Wahl, C. N. Skinner, and S. L.
Stephens, 2010. Fire-climate interactions in the American
West since 1400 CE. Geophysical Research Letters 37:L04702.
04710.01029/02009gl041695. L04702 10.1029/2009gl041695.
Trouet, V., M. Panayotov, A. Ivanova, and D. Frank, 2012. A
Pan-European summer teleconnection mode recorded by a new
temperature reconstruction from the eastern Mediterranean
(1768–2008). The Holocene. DOI: 10.1177/0959683611434225.
van Oldenborgh, G. J., and G. Burgers, 2005. Searching for
decadal variations in ENSO precipitation teleconnections.
Geophysical Research Letters 32(15):L15701. 15710.11029/
12005gl023110.
van Oldenborgh, G. J., S. Drijfhout, A. van Ulden, R.
Haarsma, A. Sterl, C. Severijns, W. Hazeleger, and H.
Dijkstra, 2009. Western Europe is warming much faster than
expected. Climate of the Past 5(1):1–12.
Woodhouse, C. A., S. T. Gray, and D. M. Meko, 2006.
Updated streamflow reconstructions for the Upper Colorado
River Basin. Water Resources Research 42(5):W05415.
W05415.
Xoplaki, E., J. Luterbacher, H. Paeth, D. Dietrich, N. Steiner,
M. Grosjean, and H. Wanner, 2005. European spring and
autumn temperature variability and change of extremes over
the last half millennium. Geophysical Research Letters 32(15):
L15713. 10.1029/2005gl023424.
Young, G. H. F., D. McCarroll, N. J. Loader, and A. J.
Kirchhefer, 2010. A 500-year record of summer near-ground
solar radiation from tree-ring stable carbon isotopes.
Holocene 20(3):315–324. 10.1177/0959683609351902.
Received 18 May 2012; accepted 14 November 2012.
Supplementary Material is available at http://
www.treeringsociety.org/TRBTRR/TRBTRR.htm
KNMI Climate Explorer 13