PR-Challenge Pseudo-Proxy and Pseudo-Instrumental Data
-----------------------------------------------------------------------
World Data Center for Paleoclimatology, Boulder
and
NOAA Paleoclimatology Program
-----------------------------------------------------------------------
NOTE: PLEASE CITE ORIGINAL REFERENCE WHEN USING THIS DATA!!!!!
NAME OF DATA SET: PR-Challenge Pseudo-Proxy and Pseudo-Instrumental Data
LAST UPDATE: 1/2014 (Addition of Theme 3 Mann experiments)
9/2012 (update documentation)
6/2011 (Addition of Esper 1/2 experiments)
3/2011 (Original receipt by WDC Paleo)
CONTRIBUTORS: N. Graham, K. Anchukaitis
IGBP PAGES/WDCA CONTRIBUTION SERIES NUMBER: N/A
WDC PALEO CONTRIBUTION SERIES CITATION:
Graham, N., et al. 2011.
Paleoclimate Reconstruction (PR) Challenge.
IGBP PAGES/World Data Center for Paleoclimatology
Data Contribution Series # N/A.
NOAA/NCDC Paleoclimatology Program, Boulder CO, USA.
GEOGRAPHIC REGION: Global
PERIOD OF RECORD: 800 - 2005 AD
FUNDING SOURCES: NOAA/OAR/Climate Program Office, PAGES
DESCRIPTION:
Special Note: As of August 16, 2011 - Please note that a coding
error was discovered for the monthly surface air temperature
(sites_sat) files in the following PR Challenge themes. These
have been corrected and updated as of this date.
THEME1/ESPER1 -- ERIK2
THEME1/ESPER2 -- ERIK2, MPI (all five simulations), NCAR
THEME2 -- ERIK2, MPI (all five simulations), NCAR
Design of the PR Challenge Pseudo Proxy Data
The organization of the Challenge pseudo proxy data is around
three primary themes. In each theme, a set of long (1000+ years)
radiatively forced global climate model (GCM) simulations is used
as the basis for formation of simulated proxy data ("pseudo-proxies")
and to provide the simulated "instrumental" target data for both
calibration and examination of reconstruction fidelity. A number
of GCM runs (from different ensemble members and different models)
are used for each theme, providing a range of simulated climate
evolutions that present different "reconstruction challenges"
(cf. Smerdon et al., 2010a, where differences between GCMs used
in reconstruction simulations are shown to have some impact on
experimental results). Description of the GCMs and development
of pseudo-proxies follows the description of the themes below.
Theme 1 - Reconstruction of Northern Hemisphere temperature index
with strongly limited proxy data set.
The proxy data set is designed to mimic that used by Esper et al. (2002),
updated by Cook et al. (2004), with 14 extratropical sites in the
Northern Hemisphere. In this theme, the simulated hemispheric-level
temperature (defined for the target data as 20-90° N) is provided
as a reconstruction target.
Specific details for Theme 1 are as follows
(GCM models are described below):
ESPER 1 – June 22, 2011 – This is the original suite of pseudo
proxy data as announced in April, 2011. It has been augmented to
include another GCM simulation, "ERIK2", and also to provide annual
growth of each simulated "tree" in the pseudo-proxy data and the
mean monthly temperature time series at the "instrumental" grid cell
nearest to each pseudo-proxy site.
MPI-M ensemble runs 1-4: proxy and instrumental data coverage 800-2005
MPI-M ensemble run 5: proxy data coverage 800-2005, instrumental data
coverage 1850-2005
NCAR run: proxy and instrumental data coverage 850-1999
ERIK2 run: proxy and instrumental data coverage 1000-1990
The annual growth outputs for the individual simulated "trees" that
make up the pseudo-proxy data at each site (described below) are also
available for the above runs, along with the mean monthly temperature
time series at the nearest "instrumental" grid cell (both covering
the period 1850-end of record). These data are provided in case
participants wish to examine individual simulated "trees" for their
relative fidelity to local subannual temperatures.
ESPER 2 – June 22, 2011 – This is a new set of data following
the same structure as ESPER 1, but with two minor changes in the
preparation of the pseudo-proxy data.
1) The "VSLite" tree growth model used to formulate the pseudo-proxies
(described below) has been updated to correct an error reported in an
exponential term in the portion of the moisture sub-model that computes
evapotranspiration. Note that the differences between the updated
and original VSLite outputs were found to be minimal for a sample
of cases checked.
2) Locations for three of the 14 pseudo-proxy sites were adjusted
with each GCM's own land mask, so that the surrounding grid cells
used to provide input for the VSLite model are all terrestrial sites
(details below). The three changed locations are listed here
(lon/lat pairs, <0 longitudes are West):
Site 4: -117.46/35.15 changed to -117.46/35.5
Site 5: -75.6/57.26 changed to -75.0/57.26
Site 8: 18.66/56.73 changed to 22.5/56.73
Note that users who would like to evaluate the Esper 1 proxy locations
with pseudo-proxies generated by the updated VSLite model can contact
Dr. Nicholas Graham at ngraham@hrc-lab.org to receive these data.
Theme 2 - Reconstruction of Northern Hemisphere temperature index
and spatial patterns with a richer, but still somewhat limited proxy
data set. The proxy data set is designed to
mimic that used by D'Arrigo et al (2006), with 62 sites in the Northern
Hemisphere at > 40° N (four sites were not implemented because they
were very close to other sites and would have required location
adjustments, similar to those described above for the Esper 2).
In this theme, the simulated hemispheric-level temperature (defined
for the target data as 20-90° N) is provided as a reconstruction target.
Spatial target data can be defined from the MPI-M, NCAR, and ERIK2
simulation data sets, available in the "Model Data" component of the
"Paleoclimate Network" developed by NOAA-Paleoclimatology/WDC for
Paleoclimatology, http://www.ncdc.noaa.gov/paleo/pubs/pcn/pcn.html.
Specific details for Theme 2 are as follows:
The Theme 2 data go by the name D'ARRIGO, to conform with the
nomenclature used for Theme 1 (first author's last name).
The structure of the data for this theme is the same as that
described for the Theme 1 Esper 1 data above.
For the D'Arrigo data a common land-sea mask was used across the three
GCMs employed (see below), and the corrected version of the VSLite tree
growth model (see above) was employed.
As with the Esper 1 and 2 sets of data, the annual growth outputs
for the individual simulated "trees" that make up the pseudo-proxy
data at each site are available, along with the mean monthly temperature
time series at the nearest "instrumental" grid cell (both covering
the period 1850-end of record).
Theme 3 - Reconstruction of Northern Hemisphere temperature index
and global spatial patterns using the richest and most spatially diverse
set proxy data set in the Challenge. The proxy data set is designed to
mimic that used by Mann et al (1998/1999) and in numerous reconstruction
simulation experiments, with 104 sites spread across the globe (e.g.,
Mann et al., 2007; Smerdon et al., 2010b). 77 individual sites are used
in Theme 3 to represent this data set. In this theme, the simulated
hemispheric-level temperature (defined for the target data as 0-90° N)
is provided as a reconstruction target. Spatial target data can be defined
from the MPI-M, NCAR, and ERIK2 simulation data sets, available in the
"Model Data" component of the "Paleoclimate Network" developed by
NOAA-Paleoclimatology/WDC for Paleoclimatology,
http://www.ncdc.noaa.gov/paleo/pubs/pcn/pcn.html.
Specific details for Theme 3 are as follows:
The Theme 3 data go by the name MANN, to conform with the nomenclature used
for Themes 1 and 2 (first author's last name). The structure of the data for
this theme is similar to that described for the Theme 1 Esper 1 data above,
with the addition that simulated coral d18O proxies are also included for
the ERIK2 and MPI-M Ensemble 1 runs. These are accompanied by simulated
ensemble coral time series for the individual sites and corresponding
"instrumental" grid cell monthly sea surface temperature and sea surface
salinity data over 1850-end of record (paralleling the detailed information
provided in the simulated tree growth ensemble data, see the "Characteristics
of pseudo-proxy data" section below).
The pseudo-tree ring data were calculated exactly as was done for the other
networks (e.g. Esper2 and D'Arrigo). There are 71 pseudo-chronologies in the
Mann network. This network is more widely distributed in space than the
D'Arrigo network (62 sites), including a few Southern Hemisphere chronology sites.
For the New Zealand sites, the information to drive interpolation for pseudo tree
production was handled differently (2-meter air temperature over the ocean was
employed) to obviate model grid land-sea definition problems.
As with the Esper 1 and 2 and D'Arrigo sets of data, the annual growth outputs
for the individual simulated "trees" that make up the pseudo-proxy data at each
site are available, along with the mean monthly temperature time series at the
nearest "instrumental" grid cell (both covering the period 1850-end of record).
For the Mann data a common land-sea mask was used across the three GCMs employed
(see below), and the corrected version of the VSLite tree growth model (see above)
was employed. [Gridding specifically related to construction of the coral pseudo
proxies is discussed below.]
GCM Simulations:
Results from millennium simulations with three climate models
have been used to produce pseudo-proxy and simulated instrumental data.
These include an ensemble of five simulations from the Max Planck Institute
for Meteorology (MPI-M) Earth System Model (ESM; Jungclaus et al. 2010),
composed of the ECHAM5 atmospheric model, the MPIOM ocean model, and a full
carbon cycle model composed of ocean and land-surface components.
These simulations include volcanic and solar forcing (reduced amplitude
compared with most earlier millennium simulations), prescribed land use changes,
orbital forcing, greenhouse gas changes (with interactive calculation of
CO2 concentrations), and aerosol forcing. The simulations cover 800-2005 AD.
The model atmosphere uses triangular-31 spectral truncation
(T31, approximately 3.75° resolution).
Other results come from a single 1150-year simulation (850-1999 AD) from
the National Center for Atmospheric Research (NCAR) Community Climate System
Model (CCSM) version 1.4 (Ammann et al. 2007). For this simulation,
the atmospheric component of CCSM was configured with the same ~3.75°
resolution as the MPI-M ESM. This simulation was forced with estimated changes
in solar irradiance (more than 2.5 times larger than those used in the MPI-M runs),
greenhouse gas concentrations, and volcanic aerosols. This run does not include
orbital forcing, land use changes, or a carbon cycle model. The simulation
covers 850-1999AD and had noticeable drift (cooling) through the first few
centuries. The effects of this drift were removed using splines
(see Ammann et al. 2007) for the pseudo-proxy results produced to date,
though results from the original (un-splined) data may be used in the future.
Results also come from a single 991-year simulation (1000-1990 AD) from HZG
(formerly (GKSS) in Germany, using the ECHO-G coupled model. This run is called
ERIK2, and uses the ECHAM4 atmospheric model at T30 (~4.0°) resolution,
coupled with the HOPE ocean model (Gonzalez-Rouco et al. 2006). The simulation
was forced with estimated changes in volcanic aerosols, solar irradiance,
and greenhouse gas concentrations; it does not include orbital forcing,
land use changes, or a carbon cycle model. ERIK2 uses the same forcing
as ERIK1 (Gonzales-Rouco et al. 2003), but with cooler initial conditions
to avoid drift during the early centuries noted in the ERIK1 simulation.
Characteristics of pseudo-proxy data:
Tree growth model
The pseudo-proxy data mimic annual tree growth increments ("trw"),
using the simplified Vaganov-Shashkin tree growth model (called "VSLite")
developed by Suz Tolwinski-Ward of the University of Arizona
(now at AIR Worldwide; Tolwinski-Ward et al., 2010). In the Challenge data,
this model, which incorporates a soil moisture content sub-component, was
applied using monthly temperature and precipitation from the GCM output
interpolated linearly to each proxy location. [Generally, these inputs
come from the surrounding 4 grid points, but with a restriction to 3
grid points for the pseudo-proxy locations listed above in the Esper 2
and D'Arrigo data sets to ensure that all inputs come from terrestrial
sites. Harmonic interpolation was used for Esper 1 and inverse distance
(great circle) interpolation for Esper 2 and D'Arrigo.] A suite of
ten simulated "trees" were grown at each site, each using the same
meteorological forcing but different (semi-randomized) values of
the parameters controlling the response to the temperature and precipitation.
As with real-world tree growth "chronologies", the growth outputs for each
of the individually simulated trees were combined to form one pseudo-proxy
chronology for each site.
Added non-climatic noise
A simulated chronology represents a "perfect"
biotic pseudo-proxy for each site, in the sense that its relationship
with the underlying climate is mediated only by the growth characteristics
embedded in the VSLite model. A set of increasing levels of white noise
(noise without frequency bias) are added to this baseline "perfect"
pseudo-proxy to mimic non-climatic factors that are also embedded
in tree growth characteristics. These levels are designed to make
the overall signal-to-noise (SN) ratios of the noise-added pseudo-proxies
be 1.0 and 0.25, by ratio of the variance of the original simulated
chronology to the variance of the added noise. A final set of
pseudo-proxies included additive "red noise" as a first-year
autoregressive process (AR1 = 0.32), again with an overall SN ratio
at 0.25 by variance ratio.
These four kinds of pseudo proxies are coded by the variable names
of the pseudo-proxy files, as listed below using the example of
the Esper 1 proxy set from Theme 1:
esper_trw_niter10_ens1
esper_trw_niter10_ens1.sn1.0
esper_trw_niter10_ens1.sn0.25
esper_trw_niter10_ens1.sn0.25rho0.32 .
The term "ens1" means the first of the five integrations of the MPI-M
paleoclimate model used to formulate Challenge data. MPI-M integrations
2-5 are similarly coded "ens2", etc.
Outputs from the NCAR and ERIK2 models are also available
for the same sequence of noise levels. The NCAR model run's
designation is "b056dtr" (b056 is the run family designator,
"dtr" indicates the spline detrending noted above).
The ERIK2 model run's designation is "erik2".
For the Esper 1 data, the pseudo-proxy data are contained in
folder paths entitled "Esper1/Esper_MPI/Esper_MPI_ensemble1(2-5)",
"Esper1/Esper_NCAR", and "Esper1/Esper_ERIK2". For the Esper 2 data,
the individual pseudo-proxy file names are similar, but they all
start with the designator "esper2". The folder path structure
is identical, except that first component of the path structure
is "Esper2".
Note that the pseudo-proxy data have been processed to have
zero mean and unit variance over the full record.
The Esper 1(2) growth output data for the individual simulated
"trees" are similarly contained in folders entitled
"Esper1(2)/Esper_MPI/Esper_MPI_trwz",
"Esper1(2)/Esper_NCAR/Esper_NCAR_trwz", and
"Esper1(2)/Esper_ERIK2/Esper_ERIK2_trwz".
These files have the following naming format,
using the MPI-M model case as an example:
esper(2)_mpiE1_trwz_ens01_01.1850.
The term "ens01_01" indicates the first of the five MPI-M integrations,
coupled with the first sequence number of the semi-randomized parameter
set used to drive the VSLite model (sequence ranges from 1-10).
Thus, there are 10 individual simulated "tree" files for each of the
five MPI-M integrations. The data cover 1850-2005. A file containing
metadata and format information is included in the
"Esper1/Esper_MPI/Esper_MPI_trwz" folder, and is named:
trwz.info.
The NCAR and ERIK2 runs each have ten individual simulated "tree"
files in the same format; the data cover 1850-1999 for the NCAR case
and 1850-1990 for the ERIK2 case.
The Esper1(2) mean monthly temperature time series at the nearest
"instrumental" grid cell to a pseudo-proxy location are contained
in folders entitled "Esper1(2)/Esper_MPI/Esper_MPI_sites_sat",
"Esper1(2)/Esper_NCAR/Esper_NCAR_sites_sat", and
"Esper1(2)/Esper_ERIK2/Esper_ERIK2_sites_sat".
These files have the following naming format,
using the MPI-M model case as an example:
esper(2)_mpiE1_sites_sat_01.1850.
The term "01" indicates the first of the five MPI-M integrations.
The data cover 1850-2005. A file containing metadata and format
information is included in the "Esper1/Esper_MPI/Esper_MPI_sites_sat"
folder, and is named:
sfc_air_temp.info.
The NCAR and ERIK2 runs each have one file in the same format;
the data cover 1850-1999 for the NCAR case and 1850-1990
for the ERIK2 case.
Data files for the Theme 2 D'Arrigo et al. network use the tag "dwj2"
in place of "esper", and folder paths for Theme 2 use the tag "DArrigo".
Data files and folder paths for the Theme 3 Mann et al. network
use the tag "manntree" or "manncoral". The data files and path structures for Themes
2 and 3 follow the formats described above for the Theme 1 data.
Coral model
The coral network provided in Theme 3 consists of 6 sites corresponding
to the following actual coral data sites, with their specific periods
of record, listed below. The sites are provided in this order in the
coral ensemble data files.
Malindi - Kenya (1697-1994)
Houtman Abrolhos Islands - western Australia (1795-1994)
Amedee Island - Noumea (1657-1992)
Palmyra Atoll – United States (928-1998, in sections)
Urvina Bay - Galapagos Islands, Ecuador (1657-1903)
Secas Island - Panama (1707-1984)
The two xls files provided (one in the MANN_ERIK2 folder and one in
the MANN_MPI_ensemble1 folder) give the locations of the pseudo-proxy
corals from the two climate model runs and the actual proxy locations.
Note that in the actual Mann network there are coral d18O records from
Noumea, Vanuatu, Galapagos and Panama. This set has been altered
and augmented for the purpose of producing the pseudo proxies in the
MANN data set for the PR Challenge. The simulated coral data cover the
entire millennium unlike the more limited periods of record noted above.
For the MPI-M simulation, the salinity data were taken from a 1x1 degree
data set and the surface temperature data from a T31 spectral resolution
data set (~3.75 degree resolution). Because of differences in grid
resolution and land/sea masking, the grid points for the salinity
and surface temperature data were selected manually to ensure that the
locations' data were as close as possible to the actual proxy locations
and, for surface temperature, not over land.
For the ERIK2 simulation, the salinity and SST data were supplied on a T42
spectral grid. These data were interpolated (inverse distance, no land points)
to the prescribed locations.
The coral pseudo-proxy d18O data were constructed as in Thompson et al. (2010)
using the relation -d18O' = a SST' + b SSS', where the primes indicate anomalies
from the long term averages (over the full simulation) in degrees C for SST,
psu for SSS, and per mil for d18O. The parameter "a" was set to -0.22 with
a standard deviation of 0.02. The parameter "b" was set to 0.27 (+/- 0.006)
for the tropical Pacific, 0.45 (+/- 0.28) for the South Pacific, and 0.16
(+/- 0.044) for the Indian Ocean.
[Graham and co-authors (in preparation) Pseudo-proxy and actual coral d18O data
for the past several centuries from two past millennium simulations: comparisons
and implications.]
The coral data sets are structured like the trw data sets; below is the listing
for the MPI-M Ensemble 1 coral data.
1) The following files parallel the individual trw series that make up the
"chronology" (i.e. each uses a different set of parameters). The parameters
were randomized using a normal distribution with the means and standard deviations
given above.
manncoral_mpiE1_d18_ens01_01.1850.txt
manncoral_mpiE1_d18_ens01_02.1850.txt
manncoral_mpiE1_d18_ens01_03.1850.txt
manncoral_mpiE1_d18_ens01_04.1850.txt
manncoral_mpiE1_d18_ens01_05.1850.txt
manncoral_mpiE1_d18_ens01_06.1850.txt
manncoral_mpiE1_d18_ens01_07.1850.txt
manncoral_mpiE1_d18_ens01_08.1850.txt
manncoral_mpiE1_d18_ens01_09.1850.txt
manncoral_mpiE1_d18_ens01_10.1850.txt
2) The following files parallel tree ring chronologies with various amounts/types
of added noise.
manncoral_mpiE1_d18_niter10_ens1.sn0.25rho0.32.txt
manncoral_mpiE1_d18_niter10_ens1.sn0.25.txt
manncoral_mpiE1_d18_niter10_ens1.sn1.0.txt
manncoral_mpiE1_d18_niter10_ens1.txt
3) The following files provide the SST and SSS data. In this case, these are
the actual data used to calculate the pseudo-d18O.
manncoral_mpiE1_sites_sss_01.1850.txt
manncoral_mpiE1_sites_sst_01.1850.txt
Simulated instrumental data:
The file names for the Northern hemispheric target data
always include the string "nhtemp", an example is given below
using the Esper 1 MPI-M data:
mpimil_nhtemp_ens1.
For the Esper 1 data, the simulated instrumental data are contained
in folder paths entitled "Esper1/Esper_MPI/Esper_MPI_ensemble1(2-5)",
"Esper1/Esper_NCAR", and "Esper1/Esper_ERIK2". For the Esper 2 data,
the individual simulated instrumental data file names are similar
to those in Esper 1, and the data values they contain are identical
across the Esper 1(2) datasets. The folder path structure is identical,
except that first component of the path structure is "Esper2".
The data files and path structures for Theme 2 (DArrigo) follow
the formats described above for the Theme 1 data, and the values
contained in the simulated instrumental data files are identical
to those in the Esper 1(2) data sets. The data files and path
structures for Theme 3 (Mann) follow the formats described above
for the Theme 1 data.
Note that the target data are provided in degrees Kelvin.
For more information, please see the PR Challenge Home Page at:
http://www.ncdc.noaa.gov/paleo/pr-challenge/
Accessing Data Sets for the Challenge Experiments:
The pseudo-proxy and simulated instrumental data sets for the
Challenge themes are available for downloading in an ftp structure
from NOAA Paleoclimatology/World Data Center of Paleoclimatology,
at the following URLs:
Theme 1 Esper 1:
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/pr-challenge/Esper1
Theme 1 Esper 2:
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/pr-challenge/Esper2
Theme 2 DArrigo:
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/pr-challenge/DArrigo
Theme 3 Mann:
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/pr-challenge/Mann
REFERENCES:
Ammann, C.M., F. Joos, D. Schimel, B.L. Otto-Bliesner, R. Tomas. 2007.
Solar influence on climate during the Last Millennium.
Proceedings of the National Academy of Science USA 104:3713-3715.
Cook, E.R., J. Esper J., and R. D'Arrigo. 2004.
Extra-tropical Northern Hemisphere land temperature variability
over the past 1000 years.
Quaternary Science Reviews, 23, 2063-2074.
D'Arrigo, R., R. Wilson, and G. Jacoby. 2006.
On the long-term context for late 20th century warming.
Journal of Geophysical Research, 111 D03103,
doi:10.1029/2005JD006352.
Esper, J., E.R. Cook, and F.H. Schweingruber. 2002.
Low-frequency signals in long tree-ring chronologies
for reconstructing of past temperature variability.
Science, 295, 2250–2253.
Gonzalez-Rouco, J.F., H. Beltrami, E. Zorita, and H. von Storch. 2006.
Simulation and inversion of borehole temperature profiles in surrogate
climates: spatial distribution and surface coupling.
Geophysical Research Letters, 33 L01703 (doi: 10.1029/2005GL024693).
Gonzalez-Rouco, J.F., H. von Storch, and E. Zorita. 2003.
Deep soil temperature as proxy for surface air-temperature
in a coupled model simulation of the last thousand years.
Geophysical Research Letters, 30 2116.
Jungclaus, J.H., S.J. Lorenz, C. Timmreck, C.H. Reick, V.Brovkin,
K. Six, J. Segschneider, M.A. Giorgetta, T.J. Crowley, J. Pongratz,
N.A. Krivova, L.E. Vieira, S.K. Solanki, D. Klocke, M. Botzet,
M. Esch, V. Gayler, H. Haak, T.J. Raddatz, E. Roeckner, R. Schnur,
H. Widmann, M. Claussen, B. Stevens, and J. Marotzke. 2010.
Climate and carbon-cycle variability over the last millennium.
Climate of the Past Discussions, 6, 1009-1044,
doi:10.5194/cpd-6-1009-2010.
Mann, M.E., R.S. Bradley, and M.K. Hughes. 1998.
Global-scale temperature patterns and climate forcing
over the past six centuries.
Nature, 392, 779-787.
Mann, M.E., R.S. Bradley, and M.K. Hughes. 1999.
Northern Hemisphere temperatures during the past millennium –
inferences, uncertainties, and limitations.
Geophysical Research Letters, 26, 759-762.
Mann, M.E., S. Rutherford, E. Wahl, and C. Ammann. 2007.
Robustness of proxy-based climate field reconstruction methods.
Journal of Geophysical Research, 112, D12109.
Smerdon, J.E., A. Kaplan, and M.N. Evans. 2010.
A pseudoproxy evaluation of the spectral fidelity
of reconstructed temperature fields.
Abstract PP43A-1667, presented at 2010 Fall Meeting,
AGU, San Francisco, Calif., 13-17 Dec.
Smerdon, J.E., A. Kaplan, D. Chang, and M.N. Evans. 2010.
A pseudoproxy evaluation of the CCA and RegEM methods for
reconstructing climate fields of the last millennium.
Journal of Climate, 23, 4856-4880, doi: 10.1175/2010JCLI3328.1.
Tolwinski-Ward, S.E., M.N. Evans, M.K. Hughes,
and K.J. Anchukaitis. 2010.
An efficient forward model of the climate controls
on interannual variation in tree-ring width.
Climate Dynamics, doi:10.1007/s00382-010-0945-5.