Spatial-temporal characteristics of lake level change in Tibet and Qinghai from two decades of altimeter observations

Abstract:

Lake levels in the Tibet and Qinghai Plateaus provide valuable records for climate change studies. Most lakes here are hard to access, having only few lake level gauges that give in situ measurements of changes in lake level and area. Remote sensing sensors, such as satellite imagery and satellite altimetry, are able to measure lake level variations with a dense spatial coverage and with a high temporal resolution. In this study, we use the TOPEX/Poseidon series of satellite altimeters (TP, Jason-1 and -2, 1992-2014) to observe lake level variations at 23 lakes along their repeat ground tracks every 10 days. We also use the Envisat altimeter (2002-2008) to observe lake levels in three lakes of Qinghai at an interval of 35 days. We use subwaveform retracking to improve the ranging precisions of satellite altimeters. We employ an optimal processing technique to obtain quality measurements, including outlier detection, space-time reduction of measurements to a common reference point, and optimal filtering. Jason-1 fails to deliver height measurements over most of the lakes. The waveforms are classified to ensure observations with a sufficiently good quality. Over 1992-2014, the lake levels of most lakes in eastern Tibet rose, while the lake levels in western Tibet dropped. In Qinghai, the lake levels dropped before 2005 and then rose afterwards, suggesting a sharp climate change in 2005, or that the measure to protect the Qinghai ecosystem (e.g., reducing livestocks) started to take effect in 2005. In general, the overall pattern of lake change in Tibet and Qinghai is related to the variation of the Indian monsoons and locations of lakes. Most lake levels show clear annual and inter-annual oscillations. In some lakes, the amplitudes of annual variation in the TP era (1992-2002) were large and then turned smaller in the Jason-2 era (2008-2014). In some lakes, the annual amplitudes were reversed between the TP and Jason-2 eras. Also, some lakes show phase shifts in the lake level highs. The lake level trends from 1992 to 2014 can be classified into three categories: (1) lakes with continuous dropping levels, (2) lakes with continuous rising levels, (3) lakes with rising or dropping levels before a major ENSO event and then with reversed trends afterwards.