Isotope paleohydrology has the potential to make a major contribution to the debate over the timing of the uplift of the Tibetan Plateau. If a long geochemical record reconstructing the d¹⁸O of meteoric water can be recovered from a basin on the plateau it may reflect large negative offsets associated with elevation change. However, because the d¹⁸O of rainfall responds to numerous factors (temperature, source region, rainfall amount, aridity and elevation, among others), a single record will likely remain ambiguous. A comparison with a low elevation record from a region adjacent to the plateau should remove many of the inherent ambiguities. The modern d¹⁸O-elevation gradient in this region (3.1‰ / km) predicts that large d¹⁸O differences would exist.

We present a 40 Ma record of carbonate cement d¹⁸O and d¹³C from fine grain sediments in the Linxia Basin, Gansu Province, China. This basin, currently at 2000 meters a.s.l., contains a relatively unbroken record of fluvial and lacustrine sedimentation from the latest Eocene into the Pleistocene (based on paleomagnetic data). These carbonate cements, precipitated from shallow ground water or lake water, are most likely primary or early diagenetic features. The d¹⁸O record is relatively stable from 40 to 12 Ma, ranging from –9 to –12‰ PDB. After 12 Ma the d¹⁸O baseline shifts to –8 to –9‰ until approximately 5 Ma, after which d¹⁸O values become much more variable with a mean value of –7‰. Major excursions to more positive values occur at 12 Ma, 8 to 7 Ma, 4.1 Ma and 1.8 Ma. A permanent change in the d¹⁸O of the carbonate record may reflect changes in aridity or moisture source for the Linxia basin, something that may have been affected by nearby Plateau uplift. Although the correlation of these features with other monsoon records is intriguing, this remains speculative.

The low elevation Linxia Basin d¹⁸O record is a baseline to which high elevation records can be compared. It should record broad changes in climate, aridity, and the source region of meteoric water at the northeastern margin of the Tibetan Plateau. If a well-dated high elevation record can be recovered from this region of the plateau, a major divergence in the two records would most likely reflect the creation of a large offset in elevation.