CARBONATE OXYGEN ISOTOPE PALEOALTIMETRY: CALIBRATING d18O VS. ALTITUDE GRADIENTS AND QUANTIFYING THE ASSOCIATED ERRORS

Carbonate oxygen isotope paleoaltimetry is based on the analysis of authigenic carbonates, which record the d¹⁸O value of surface water (d¹⁸O_w) from which the carbonates precipitated. Sampling of surface waters over a range of elevations in the region of interest may provide the best means of calibrating the modern d¹⁸O_w vs. altitude gradient; these waters are more representative of the water that is recorded by carbonate than rainfall. In the Nepal Himalaya, it has been shown that selective sampling of small streams over an elevation transect yields a curved d¹⁸O_w vs. altitude gradient (R²=0.96), which displays increasing depletion in ¹⁸O with increasing altitude (Garzione et al., 2000). A curved gradient suggests that Rayleigh fractionation is an important process in rainout, although other processes such as water-vapor exchange during rainfall or evaporation from raindrops can also modify the curve.

The d¹⁸O value of stream water reflects the range of elevations in a drainage basin over which rainfall occurs. In the Nepal Himalaya, we have estimated the median elevation of the drainage basin to be representative of average rainfall elevation. The curve that results from this correction has been shifted to fit a low elevation site at New Delhi, where the d¹⁸O value of average annual rainfall is known. The resulting relationship can be used to estimate paleoelevation if other climate variables, such as paleotemperature and changes in the d¹⁸O value of source moisture, are known.

Carbonate precipitation has another host of variables that must be considered in determining the error associated with paleoelevation estimates. The uncertainties that are considered here include: 1) the scatter in the d¹⁸O_w vs. altitude relationship, 2) an additional error of ±100 m on the estimation of the average elevation of the drainage basin, 3) analytical error of ±0.1‰, 4) ±5^oC uncertainty on the estimation of the temperature of carbonate precipitation, and 5) the scatter in the empirical relationship governing water-carbonate fractionation. These uncertainties are propagated using Taylor series expansion to yield complete errors of ±800 m at 0 m elevation and ±1100 m at 3000 m elevation. The latter 3 sources of uncertainty are inherent to all studies using the d¹⁸O value of carbonates as a proxy for paleoelevation.