Palaeoclimatic interpretation of a topographic profile across middle Holocene regressive shorelines of Longmu Co (western Tibet)

Most lakes and closed basins in Tibet are surrounded by abandoned shorelines attesting for high-water levels in the past. Two such lakes, Longmu Co and Sumxi Co (Western Tibet), were the focus of detailed palaeoclimatic work during the 1989 Sine-French expedition based on the study of lake sediments and analysis of a piston core. Here it is shown that additional information may be deduced from the abandoned shorelines around Longmu Co. A 1800 m long topographic profile was levelled from 75 m up to 230 m above the present lake elevation (5008 m), on which the shorelines form a flight of horizontal terraces separated by more or less degraded scarps. The elevations of the terraces were spaced regularly about 1.35 m, except for 3-5 m high lacunae reccurring every 15-16 m. The potential annual evapo-transpiration was computed based on meteorological data collected at Tien Shui Hai from 13/08/89 to 11/09/90. Using a modified Penman formula a value of 1.60 +/- 0.3 m/yr was obtained under close system conditions (no outlet) and a synthetic curve of lake regression was derived which appears to be consistent with the measured shoreline heights. II is concluded that the 1.35 m terraces represent annual regression of the lake level. By contrast, the 3-5 m high lacunae might reflect the periodic return, about every 9-12 yr, of particular climatic conditions responsible for a perturbation of the annual process of shoreline formation. All the shorelines of this profile have formed probably within one century, during a continuous and abrupt regression of the lake. Such evolution has also been identified in the sediments of a piston core collected at Sumxi Co, and dated at about 6-5.5 kyr B.P. The lake level must have been at its highest stand, shortly before that major regression. The mechanism responsible for this middle Holocene high stand and for the major sudden drop of the lake level remains uncertain.