Study of the recent disappearance of a tropical glacier in the Bolivian Andes with the help of the high resolution regional climate model MAR

This study provides a first evaluation of the MAR (Modèle Atmosphérique Régional) model over the Bolivian tropical Andes. MAR is currently developed at the ULg and allows dynamical downscaling up to 5 km of horizontal resolution. The purpose of this work is to model the recent changes in the climatic parameters which are thought to control the mass and energy balance of mountain glaciers in the outer tropics. We focus on the recently vanished Chacaltaya Glacier (16°S), which by virtue of its location and its environmental context is representative of many small-sized glaciers of the Bolivian Andes.

To evaluate our model, we first examine simulated precipitation and near-surface temperature against in situ observations from ground weather stations. Since observational data is very scarce in this part of the world, we also refer to qualitative information provided by the scientific literature. We compare the performance of the model forced with two different reanalyses, and with several corrections applied to the lateral boundary conditions (LBCs) impacting on temperature and humidity. MAR forced with the ERA-Interim reanalysis yields better results than with the NCEP/NCAR-v2 reanalysis. We then use the best ERA-driven simulation to assess the long-term climate change over 1960-2014 in the region of the Chacaltaya Mountain.

The regional atmospheric circulation is adequately simulated by MAR, which reproduces the prevailing seasonal features of the lower, middle and upper troposphere. The climatic anomalies associated to the El Niño Southern Oscillation (ENSO) events are also properly recreated. Remaining model biases include an overall year-round dry bias over the Altiplano region, due to the strong corrections applied to the LBCs (-15% for humidity). Over the highest elevations of the Andes, both precipitation and cloud cover are probably overestimated by MAR. The sporadic spatial distribution of convective rainfall also suggests numerical instability in the convective parameterization scheme. Modeled temperatures match very well the observations, but the reliability of the observed time series is highly questionable.

Between 1960 and 2014, the most significant trends concern precipitation and cloud cover which both decreased of about 35%. The surface radiation budget also changed as a result of the reduced cloudiness. Near-surface temperature increased by about 1.5°C. Those trends are believed to have been enhanced by the more frequent and more intense El Niño events during the warm Pacific Decadal Oscillation (PDO) phase between 1977 and 1999. The combination of the repeated droughts and the enhanced incoming short-wave radiation due to reduced low-level cloudiness are probably the main factors responsible for the acceleration of the Chacaltaya Glacier recession. Nevertheless, the ENSO signal is not always clear in the Bolivian Andes, because of the interference with strong local climatic processes. Moreover, additional forcings of non climatic nature may also be responsible for the rapid demise of the Chacaltaya Glacier.