Modeling the surface mass balance of a high Arctic glacier using the ERA-40 reanalysis

Investigating glacier mass balance via the use of numerical models provides an important insight into glacier-climate interactions and allows for the assessment of future changes in sea level and local water resources. The application of mass balance models to unmonitored regions has previously been inhibited by the inadequate spatial coverage of local meteorological observations. One way to overcome this problem is through the use of climate reanalysis products. In this paper we evaluate the ability of the European Centre for Medium-Range Forecasts ERA-40 reanalysis to drive a surface mass balance model of Midre Lovenbreen, northwest Spitsbergen, Svalbard. The ERA-40 reanalysis is validated and bias-corrected against long-term observations from two nearby in situ weather stations. The model is calibrated over the period of mass balance observations (1968-2001) using a downhill simplex parameter optimization technique to minimize the error between observed and simulated mass balances. The calibrated model is then used to extend the mass balance record of Midre Lovenbreen back to the beginning of the reanalysis in 1958. Overall, the ERA-40 reanalysis is found to correspond sufficiently well with surface observations to be used for mass balance modeling. When driven using the ERA-40 reanalysis, the calibrated surface mass balance model performs very well. The area-averaged cumulative errors for winter, summer, and net balances are all very small over the period 1968-2001 (<0.3 m water equivalent (w.e.)). For the centerline stakes the correlation coefficients between observed and modeled net, winter, and summer balances are 0.83, 0.84, and 0.77, respectively. Using the hindcasted mass balances, a mean cumulative mass loss of -17.8 +/- 3.7 m w.e. is estimated over the 44 year period 1958-2001.