NSF ATM-9909201
P.I.: R. S. Bradley
Co-P.I.: M. Vuille
in collaboration with R. Healy
(WHOI) and M. Werner
(MPI)
Project summary
The goal of this project was to analyze both observed and simulated d18O
and dD variability in precipitation at low latitudes, with a special focus
on the interpretation of the isotopic signal in tropical ice cores. We used
observational data from the International Atomic Energy Agency – Global Network
of Isotopes in Precipitation (IAEA-GNIP) database, our own measurements from
two Automated Weather Stations (AWS) installed near the ice core drill sites
on Sajama and Illimani in Bolivia, annual snow samples collected from snow
pits dug near the drill sites, and two Atmospheric General Circulation Models
(AGCM’s) with incorporated stable isotopic tracers (GISS II and ECHAM-4)
to study the isotopic variability in precipitation at ice core locations.
Our GISS II experiments were based on a new version of the model, which includes a higher spatial resolution of 4° lat. * 5° lon., with 9 vertical layers based on sigma levels, running in improved double precision, and with improvements to the land surface radiation scheme and replacement of the radiation module from model II'. The ECHAM-4 isotope model simulations were based on a hybrid sigma-pressure coordinate system and were run with triangular truncation at both wavenumber 30 (T30 ~ 3.75° lat. 3.75° lon.) and 106 (T106 ~ 1.1° lat. 1.1° lon.), including 19 vertical layers from surface to 30 hPa. All three models experiments (GISS II, ECHAM-4 T30, ECHAM-4 T106) were run with modern boundary conditions and forced with observed global sea surface temperatures (SST). The GISS simulation is based on AMIP I (Atmospheric Model Intercomparison Project) and Reynolds Optimum Interpolation SST, while the ECHAM model uses GISST 2.2 (Global sea-Ice and Sea Surface Temperature) data. While greenhouse gas concentrations were kept at a constant modern level in the GISS II and the ECHAM-4 T106 run, these levels were adjusted annually in the ECHAM-4 T30 experiment. Our model runs were unprecedented in many ways: The ECHAM-4 run (1979-98) with T106 spectral resolution is the highest-resolution simulation ever performed with isotopic tracers; the ECHAM-4 T30 run (1903-94) spans the longest continuous time period simulated so far and the GISS-II (1980-97) simulation has a new more realistic topography over the Andes and East Africa, which significantly improved the simulation of d18O records from tropical high altitude ice cores. Both models further include isotopic tracer tagging capabilities, that is, water evaporating from a source is tagged upon evaporation and the tag is only lost when the water molecule falls as precipitation. Water molecules precipitating over the tropical Andes could thus be traced back to their evaporative source area. This allowed us to determine the relative contribution of different source regions to precipitation and to assess the d18O signature of the different source regions at the ice core drill sites.
Our simulation experiments were particularly challenging because of the
high and complex topography surrounding the ice core drill sites (Andes).
To account for isotopic ‘altitude effects’, we raised the Andean and East
African topography in the GISS model. Since such changes in topography can
create severe distortions in the modeled atmospheric circulation, we performed
two 7-year control runs with and without the new topography implemented,
to analyze the effects of changing topography on the Andean and East African
climate and on the spatiotemporal variability of stable isotopes in precipitation,
before applying the new boundary conditions to further simulations.
Since we used two different models (GISS II and ECHAM-4) at various horizontal
resolutions, we also performed an extensive model intercomparison and projected
the modern model output onto NCEP-NCAR reanalysis and IAEA-GNIP data to evaluate
the model performance.We had to invest much more time than anticipated in
the model setup and validation. Our analysis therefore focused on the modern
simulations and the tropical Americas only, where a number of ice cores were
available to test the model performance. Simulations for 6ky and 21 ky have
been run however, and will be analyzed as part of a new project in the near
future.
Our results indicate that both models (GISS and ECHAM) capture the essential features of surface climate over the tropical Americas in terms of both their spatial and temporal characteristics. Using a low-resolution model (GISS II), adjusted to provide a more realistic Andean topography, or a higher resolution model (ECHAM-4 T106) leads to an improved d18O distribution over the tropical Americas with an altitude effect comparable to observations. Water vapor transport and gradual rainout and increasingly depleted composition of water vapor along its trajectory are correctly simulated in both models, although the ECHAM model appears to underestimate the continentality effect over the Amazon basin.
A significant dependence of d18O on the precipitation amount is apparent in both models, in accordance with observations, while the influence of temperature seems to be less significant in most regions and is accurately reproduced by the ECHAM model only. Over most regions however, the d18O signal in precipitation is influenced by a combination of factors, such as precipitation amount, temperature, moisture source variability and atmospheric circulation changes. Over parts of the tropical Americas the d18O signal is therefore also significantly correlated with ENSO, because ENSO is an integrator of many factors affecting the d18O composition of precipitation.
We used two simulations run between 1979 and 1998, to simulate the d18O signal in three tropical Andean ice cores, from Huascarán (Peru), Quelccaya (Peru) and Sajama (Bolivia). We were able to show that in both models the simulated stable isotopic records compare favorably with the observational data, when the seasonality of precipitation and dry season loss due to sublimation and wind scour are taken into account. We were further able to demonstrate that the tropical Pacific exerts a dominant control on interannual timescales, even though the moisture originates over the Amazon basin and the tropical Atlantic. More enriched (depleted) d18O values are associated with periods of warm (cold) conditions in the tropical Pacific. The growing number of stable isotope records from tropical Andean ice cores may thus provide an important archive for reconstructing Pacific climate variability.
Our results have major ramifications for the interpretation of stable
isotope records from tropical ice cores. These records are usually interpreted
as indicators of past temperature change. Our research however, indicates
that the stable isotope composition is much rather an integrator of several
climatic controls, which combined act as a dominant mode of atmospheric circulation.
In the tropical Andes, for example, this dominant mode is related to sea
surface temperature anomalies (SST) in the tropical Pacific, and the stable
isotopic composition from Andean cores might hence serve as a proxy for past
tropical Pacific SST variability. These results ask for a re-assessment of
past interpretations of tropical ice core records, in particular during key
time periods such as the Last Glacial Maximum (LGM).
Related Publications
|
|
|
|
University of Massachusetts | Department of Geosciences | Climate System Research Center | Mathias' Homepage | Current research