Modeling time-averaged surface mass balance of the Elbrus glacial complex (Central Caucasus) and evaluation of the uncertainties caused by random weather fluctuations

The purpose of the research is to assess the influence of the random weather fluctuations on the estimates of the model-based surface mass balance (SMB) components of the mountain glacier. The common approach in the modeling studies is to use meteorological records (measured or modelled) – surface air temperature and precipitation rate – as weather forcing in numerical experiments. The results of the calculations are normally very sensitive to the parameter choice. The model should be carefully calibrated against measured SMB to obtain correct results. What is usually ignored within the frameworks of this approach is that forcing records at e.g. daily resolution contain internal weather variability which after being integrated by the model can result in spurious variations in surface mass balance and, as a result, in further misinterpretation of the results. To evaluate uncertainty in SMB calculations we force a spatially distributed energy balance model of the Elbrus glacial complex in the Central Caucasus with surrogate series of surface air temperature and precipitation rate. An ensemble of 50 surrogate series at a daily resolution of two model decades duration each were produced by a stochastic weather generator WGEN basing on the observed meteorological series at the weather stations Terskol located nearby. In WGEN, precipitation events are simulated by a first-order Markov chain, and the intensity of precipitation is represented using independent gamma distribution. Air temperature is calculated by fitting the appropriate distributions and harmonic functions separately for wet and dry days. Seasonality is reproduced by an estimate of individual sets of model parameters for different periods of the year. Statistical analysis of the generated ensemble of SMB components reveals that the standard deviation of SMB components (accumulation rate, melting, evaporation, runoff, melt water retention) vary within the limits several percents, but the range of variability within the ensemble is rather higher and cannot be ignored in retrospective studies. Our approach enables to filter out reaction of the modelled glaciers induced by the weather noise from systematic reply on climate change.