Australian snowpack in the NARCliM ensemble: evaluation, bias correction and future projections.

Di Luca, A., J.P. Evans, and F. Ji
Climate Dynamics, doi: 10.1007/s00382-017-3946-9, 2017.


In this study we evaluate the ability of an ensemble of high-resolution Regional Climate Model simulations to represent snow cover characteristics over the Australian Alps and go on to asses future projections of snowpack characteristics. Our results show that the ensemble presents a cold temperature bias and overestimates total precipitation leading to a general overestimation of the snow cover as compared with MODIS satellite data. We then produce a new set of snowpack characteristics by running a temperature based snow melt/accumulation model forced by bias corrected temperature and precipitation fields. While some positive snow cover biases remain, the bias corrected (BC) dataset show large improvements regarding the simulation of total amounts, seasonality and spatial distribution of the snow cover compared with MODIS products. Both the raw and BC datasets are then used to assess future changes in the snowpack characteristics. Both datasets show robust increases in near-surface temperatures and decreases in snowfall that lead to a substantial reduction of the snowpack over the Australian Alps. The snowpack decreases by about 15 and 60% by 2030 and 2070 respectively. While the BC data introduce large differences in the simulation of the present climate snowpack, in relative terms future changes appear to be similar to those obtained using the raw data. Future temperature projections show a clear dependence with elevation through the snow-albedo feedback effect that affects snowpack projections. Uncertainties in future projections of the snowpack are large in both datasets and are mainly dominated by the choice of the lateral boundary conditions.

Key Figure

Fig. 13 Snow cover mean extent (top panels) and changes relative to present values (bottom panels) for present, near future and far future climates. Left and right panels show results for the NARCliM raw and BC datasets respectively. Areal-mean values are calculated using grid points that have regular snow cover in all three periods. The NARCliM ensemble mean (red line), inter-quartile range (blue box) and total range (whisker) are shown for each period. Red circles represent values from individual members

UNSW    This page is maintained by Jason Evans | Last updated 23 January 2018