Conditioning temperature‐index model parameters on synoptic weather types for glacier melt simulations

Abstract

Temperature‐index models are widely favoured as a pragmatic means of simulating glacier melt because of their generally good performance, computational simplicity and limited demands for in situ data. However, their coefficients are normally treated as temporally stationary, unrealistically assuming a constancy of the prevailing weather. We address this simplification by prescribing model coefficients as a function of synoptic weather type, in a procedure that utilizes reanalysis data and preserves the minimal data requirements of temperature‐index models. Using a cross‐validation procedure at Vestari Hagafellsjökull, Iceland, and Storglaciären, Sweden, we demonstrate that applying transient model coefficients, for three temperature‐index models, results in statistically significant increases in the skill with which melt is modelled: Median simulation improvements in the Nash–Sutcliffe efficiency coefficient of 7.3 and 23.6% are achieved when hourly and daily melt totals are evaluated respectively. Our weather‐type modelling approach also yields insight to processes driving parameter variability, revealing dependence that is consistent with a priori considerations of the surface energy balance. We conclude that incorporating weather types into temperature‐index models holds promise for improving their performance, as well as enhancing understanding variability in coefficient values. Copyright © 2014 John Wiley & Sons, Ltd.

Impact of supraglacial deposits of tephra from Grímsvötn volcano, Iceland, on glacier ablation

“…K −1 d −1 (Guðmundsson and others, 2009) and 12.6-14.2 mm w.e. K −1 d −1 (Matthews and others, 2015) are reported.…”

The Influence of Weather Systems in Controlling Mass Balance in the Southern Alps of New Zealand

“…In particular, little is known about what controls variability in snow accumulation in winter (Purdie et al, 2011), which compared to our knowledge of the atmospheric drivers controlling ablation has been largely neglected and is hindering our ability to predict how glaciers will respond to future changes in climate (e.g., Hock et al, 2017). One approach that can be used to bridge the gap between atmospheric scales is to use synoptic weather types (e.g., Isaksen et al, 2016;Käsmacher & Schneider, 2011;Matthews et al, 2015;Romolo et al, 2006;Yarnal, 1984), with the Kidson (2000) weather types (Renwick, 2011) providing a time series of 12-hourly synoptic conditions that is suitable for a range of applications in the New Zealand region. Kidson weather types have been used to account for variations in the accumulation and redistribution of snow in mountainous terrain (Purdie et al, 2011;Webster et al, 2015), hydrological flows (McKerchar et al, 2010), and in paleoclimate research (Lorrey et al, 2007(Lorrey et al, , 2014, as well as in a range of other applications outside the snow and glacier hydrology space (e.g., Appelhans et al, 2013;Beentjes & Renwick, 2001;Gibson & Cullen, 2015;Sturman & Quénol, 2013).…”

Glacio-hydrological melt and runoff modelling: a limits of acceptability framework for model selection

“…The challenge here was to reconcile three signatures that characterise glacier melt over different spatial and temporal scales. This is not a straightforward task, particularly when using temperature index models that lump a number of spatially and temporally variable terms from the full 15 energy balance equation into a handful of calibration parameters which may lack robustness in space and time (MacDougall et al, 2011;Matthews et al, 2015;Gabbi et al, 2014). The inclusion of solar and topographic effects in the TIM 2 and TIM 3 melt model structures addressed some of these limitations.…”