Fig. 1: Study area, its climatology, and evaluation of the model outputs.

a Upper Rio Santa catchment and its topography, including the hydrological (red dots) and meteorological stations (yellow dots) used in this study (AG is Artesonraju Glacier, CG is Cuchillacocha Glacier, and SG is Shallap Glacier). The inset map shows the location of the catchment in Peru. Glacier (blue) and debris cover (grey) areas are from ref. ⁶⁶, lake areas (turquoise) were provided by INAIGEM, and the digital elevation model is an ALOS PALSAR from 2000. b Modelled monthly averaged meteorological conditions and discharge (Q, yellow line) for the catchment, including near-surface air temperature (purple line), precipitation (Pr., dark blue bars for rain and light blue bars for snow), and incoming shortwave radiation (INSW, red line). Conditions are an average over the 2008-2018 modelling period. c Comparison of modelled (2015-2018, red dashes) and remote sensing (2015-2019, blue dashes) derived altitudinal glacier mass balance profile for the entire catchment. The dashed lines are the average mass balance for all the cells in each 100 m elevation band. The shading for the model results (pink) represents +/− the standard deviation of the mass balance per elevation band, whereas the shading for the remote sensing data (light blue) represents the uncertainty of the average. d Comparison of monthly averaged snowline elevations between MODIS (blue lines) and the model (red lines), calculated using an approach following Meier⁶⁷. Modelled cells are defined as snow-covered when the snow water equivalent is > 10 mm w.e., with the error bars representing when this is varied from >1 mm w.e. (minimum) to >20 mm w.e. (maximum). The MODIS error bars represent the snowline derived using alternate Normalised Difference Snow Index threshold values of 0.1 (minimum) and 0.45 (maximum).