Are empirical glacier melt models robust under climate change conditions?
Assigned Session: FS 3.108: Glacier modelling and downstream impacts
Abstract ID: 3.13922 | Not reviewed | Requested as: Talk | TBA | TBA
Christophe Kinnard (1)
Lisa, Michaud (2)
(1) Université du Québec, 3351 Des Forges, G8Y7R7 Trois-Rivières, CA
(2) École de technologies supérieures (ETS)
Abstract
Empirical melt models based on the so-called degree-day method have been used for decades to model snow and ice melt within glacier mass balance models. They are easy to implement, require minimal data inputs, and showing usually good results. As such, these models have been used for climate impact projections on glaciers at the regional to global scale. However, these models rely heavily on calibration against observations, which calls into question their transferability and robustness under future climate conditions. Here we perform a ‘virtual world’ experiment by calibrating different empirical glacier melt models against mass balance simulated by a physically based (energy balance) models on Saskatchewan Glacier, Canada, for the period 1979-2010, and then test the transferability of the models against future mass-balance under different climate change scenarios. All empirical models show good performance during the calibration period, but different behaviour during the future climate periods. We find that all melt models suffer from strong parameter equifinality which hampers the unambiguous identification of model parameters. Calibrating on spatial observations (mass balance stakes) instead of on the mean glacier mass balance seems to reduce model equifinality and improve mass balance simulations under climate change scenarios. Our work highlights the challenges of using degree-day based models to project future glacier melt trajectories.
N/A | ||||||||
|