Eric J. Steig & David P. Schneider

University of WashingtonC. A. Shuman

NASA/Goddard

WAIS Workshop

September, 2003

Reconstruction of Antarctic climate through the

Holocene and beyond:methodology, and

implications for deep ice core site selection

Past climate variability and change

Satellite Observations

Numerical Weather Reanalyses and Mesoscale Modeling

Antarctic Weather Stations

Array of Shallow Ice Cores

Deep Ice Cores

Time

Space

Question #2. How has the climate …changed over WAIS during the past 200-500 years …?

• Goal: Use ITASE ice cores to answer this

• Problem: interpretation of ice cores requires reliable calibration against the instrumental record, but the instrumental record is too short and too sparse

• Solution: use statistical methods to extend an estimate of the instrumental record as far back as we can

Modern interannual temperature variability

Figure after Schneider, Steig, Comiso, in press (J. Climate)

Methodology

€

T = λ k

k=1

K

∑ ukTvk

Obtain basis functions (time series and associated spacial

patterns) of variability using orincipal component analysis of “known” climate

Determine empirical relationship between proxy variables (e.g. ice cores) and known climate data

€

Ux = yshort

Use that relationship to extend climate basis functions farther back in time

€

ˆ U x = y long

Methodology

€

ˆ T = λ k

n=1

NPCs

∑ ˆ u kTvk

Reconstruct the climate field by summing over the reconstructed PCs

“Known” Data Proxy Data

Satellites: complete spatial coverage to 1982

Weather stations:Continuous temporal recordsto 1961; back to 1901 onlimited basis

MethodologyAVHRR PCs

(“known”)

Weather StationsCalibration Interval“proxy”

Weather StationsFull Data Set“proxy”

ReconstructedPCs

Calibration/Verification statistics(correlation coefficients)

Calibration (1982-1999) = 0.77 (monthly) = 0.77 (annual) = 0.91 (5-year averages)

Verification (1961-1981) = 0.61 (monthly)(weather stations) = 0.54 (annual) = 0.71 (5-year averages)

Verification (2000-2002) = 0.66 (monthly) (Vostok) = 0.60 (annual)

How well does it work?

1982 1984

1982 1984

Reconstruction

AVHRR Data

Reconstruction of Vostok Temperatures

Reconstruction of Vostok Temperatures

Figure from Rutherford/Mann shows fraction of data availability 1960-1999.

Implications for ITASE cores

Reconstructed instrumental records provide larger data set for calibration of ice core proxy variables.

Reconstructed records are long enough to allow for decade-to-decade as well as interannual comparison.

Reconstructions are inherently “filtered” to emphasize large-scale variability.

Prediction: ice core records will better reflect the PCs of the temperature field than with raw temperature, due to uncorrelated noise in both.

Implications for Deep Drilling Site Selection

Modes of Variability: Modern vs. LGM

Figure from Camille Li et al. CCM3 experiments.

Control EOF1 of Z500 (21%)

10 m contours

Control EOF2 of Z500 (14%)

LGM EOF1 of Z500 (31%)

LGM EOF2 of Z500 (16%)

Figure after Lea, Science 297 (2002).

Super ENSO and Global Climate Oscillations at Millennial Time ScalesLowell Stott, Christopher Poulsen, Steve Lund, and Robert Thunell Science 2002 297: 222-226.

El Niño-Like Pattern in Ice Age Tropical Pacific Sea Surface TemperatureAthanasios Koutavas, Jean Lynch-Stieglitz, Thomas M. Marchitto, Jr., and Julian P. Sachs Science 2002 297: 226-230.

Long term ENSO changes?

SOI vs. AVHRR temperature

Calibration Statistics (r)

PCs vs. Variance