Evaluating the transferability of empirical models of debris-covered glacier melt

• Published in: Journal of Glaciology
• Main Authors: Winter-Billington, A., Moore, R. D., Dadic, R.
• Format: Article in Journal/Newspaper
• Language: English
• Published: Cambridge University Press (CUP) 2020
• Subjects: Merra; Journal of Glaciology
• Online Access: http://dx.doi.org/10.1017/jog.2020.57; https://www.cambridge.org/core/services/aop-cambridge-core/content/view/S002214302000057X

Supraglacial debris is significant in many regions and complicates modeling of glacier melt, which is required for predicting glacier change and its influences on hydrology and sea-level rise. Temperature-index models are a popular alternative to energy-balance models when forcing data are limited, but their transferability among glaciers and inherent uncertainty have not been documented in application to debris-covered glaciers. Here, melt factors were compiled directly from published studies or computed from reported melt and MERRA-2 air temperature for 27 debris-covered glaciers around the world. Linear mixed-effects models were fit to predict melt factors from debris thickness and variables including debris lithology and MERRA-2 radiative exchange. The models were tested by leave-one-site-out cross-validation based on predicted melt rates. The best model included debris thickness (fixed effect) and glacier and year (random effects). Predictions were more accurate using MERRA-2 than on-site air temperature data, and pooling MERRA-2-derived and reported melt factors improved cross-validation accuracy more than including additional predictors such as shortwave or longwave radiation. At one glacier where monthly ablation was measured over 4 years, seasonal variation of melt factors suggested that heat storage significantly affected the relation between melt and energy exchange at the debris surface.