2002 Denver Annual Meeting (October 27-30, 2002)

Paper No. 1
Presentation Time: 1:30 PM

OPTIMAL INTERPOLATION OF THE STABLE ISOTOPE COMPOSITION OF METEORIC PRECIPITATION


BOWEN, Gabriel J.1, REVENAUGH, Justin1 and WILKINSON, Bruce2, (1)Earth Sciences, Univ of California, Santa Cruz, CA 95064, (2)Geological Sciences, Univ of Michigan, 2534 C.C. Little, Ann Arbor, MI 48109-1063, gbowen@es.ucsc.edu

Point estimates of the stable oxygen and hydrogen isotope compositions of modern precipitation (d18Oppt, dDppt) are necessary in many climatological and hydrological studies, and the regional and global distributions of d18Oppt and dDppt provide means for testing a new generation of general circulation models equipped with isotope tracers. Thus far, annual or multiannual records of these values have been obtained for <400 locations worldwide, and no rigorous method for interpolating between sampled locations has been established. We propose a new method for the interpolation of mean annual d18Oppt and dDppt and report the confidence of estimates made using this method. The isotopic composition of precipitation at a location is modeled as the sum of global patterns related to temperature and regional patterns related to variation in the source and transport of water vapor. Substituting latitude and altitude for temperature in the model generates equally good predictions and expands the applicability of the method. The new interpolation method decreases the average error of interpolated d18Oppt and dDppt estimates by >0.2 and >1 per mil relative to previously used methods. In addition, the new method significantly decreases the abundance and magnitude of high-error estimates. The average error of point estimates depends strongly on the number and proximity of data stations, but only 150-200 stations are needed to produce relatively precise and accurate estimates. Global maps of interpolated mean annual d18Oppt and dDppt and the confidence intervals for these estimates provide a reference framework that can be rigorously applied in future studies. Seasonal variation in the form of the relation between d18Oppt or dDppt and latitude makes treatment of monthly or seasonal data more complex, but upon development of the appropriate model to describe this relation, interpolation of sub-annual data can be achieved using the method proposed here.