Text S1: Supplemental Background
Each summer, the delivery of supraglacial meltwater to the GrIS bed causes a rapid initial rise in subglacial water pressure, which reduces basal traction and enhances ice sliding (e.g. Bartholomew et al., 2010, Hoffman et al., 2011; Zwally et al., 2002 ). A gradual slowdown in ice motion then occurs as increasing subglacial efficiency reduces regional subglacial pressure and increases basal traction (e.g.,Bartholomew et al., 2010; Hoffman et al., 2011; 2016 ). Superimposed upon this seasonal cycle are short-term accelerations lasting several hours to several days attributed to variations in meltwater input (Andrews et al., 2014; Schoof, 2010 ). In the lower ablation zone, brief increases in ice speed of up to ~300% (with lesser accelerations at higher elevations) are broadly attributed to the effect of diurnal surface melting on subglacial hydrology and water pressure (e.g. Andrews et al., 2014; Cowton et al., 2016; Davison et al., 2019; Hoffman et al., 2011; Shepherd et al., 2009 ).
At a process level, however, the interaction among subglacial cavity evolution, subglacial storage, and ice motion remains difficult to interpret across spatial and temporal scales despite extensive collection of on-ice surface measurements (e.g., Anderson et al., 2004; Andrews et al., 2018; Bindschadler, 1983; Cowton et al., 2016; Flowers et al., 2016; Hoffman et al., 2011; Howat et al., 2008; Iken et al., 1983; Jansson, 1996; Kamb, 1970; Schweizer & Iken, 1992 ). GPS-derived ice surface elevations, in particular, are typically noisy and partitioning the components of uplift is uncertain. This makes interpretation of melt-induced basal uplift and uplift rates in the context of ice motion challenging (e.g. Andrews et al., 2018, Cowton et al., 2016 ). Furthermore, there is a growing appreciation that meltwater surface routing through Greenland’s large supraglacial river catchments modulates the magnitude and timing of meltwater runoff entering moulins (Smith et al. 2017; Yang et al., 2018; 2020 ), which must surely influence observed variations in basal water pressure and associated ice velocity (e.g. Banwell et al. 2016; Clason et al., 2015; Palmer et al., 2011; Pitcher and Smith, 2019; Zwally et al.,2002 ). Yet, the influence of supraglacial river discharge on short-term subglacial water storage fluctuations and ice motion has received little observational study.
The purpose of this study is to examine the influence of moulin input (i.e. supraglacial river discharge) on localized, short-term accelerations in ice surface velocity. To achieve this, we explore temporal correlations between hourly time series of surface energy balance, ice ablation, supraglacial river discharge, and horizontal/vertical ice surface motion for Rio Behar, a moderately sized (~60.2 km2 in July 2016) mid-elevation (>1200 m a.s.l.) supraglacial river catchment in the southwest Greenland ablation zone (Smith et al., 2017 ). It represents a typical catchment of the snow-free, bare ice ablation zone (Cooper and Smith, 2019; Ryan et al., 2019 ), including intense melting and development of weathering crust development during the month of July (Cooper et al., 2018 ).
The novel datasets analyzed here are: (1) 168 high-quality Acoustic Doppler Current Profiler (ADCP) consecutive hourly measurements of supraglacial river discharge (i.e. catchment runoff flux, m3 s-1) acquired 6-13 July 2016 approximately 750 m upstream of the Rio Behar terminal moulin; and (2) simultaneous GPS measurements of horizontal and vertical ice surface motion (5-second sampling interval). We also use PROMICE KAN_M AWS data to estimate surface energy inputs and ablation; and compute proxies for subglacial storage (S) and its rate-of-change (ΔS) using both GPS and hydrographic methods. These data are freely available as Additional Supporting Information (Additional Supporting Information Datasets S1-S7 ).
Permanent discharge gauging stations are infeasible in the rapidly melting ablation zone environment. Owing to continuous thermal erosion of the ice bed, empirical stage-discharge rating curves rapidly obsolesce, necessitating that discharges be measured in situ rather than estimated from empirical rating curves relating occasional discharge measurements to continuously recorded water level changes. This requirement of hourly around-the-clock ADCP operations (together with non-trivial logistical challenges of camping and anchoring instruments in rapidly melting bare ice) explain the relative brevity (1 week) of our hourly supraglacial river discharge time series.
Simultaneous measurements of air temperature, radiation, and ice surface ablation were acquired from the nearby PROMICE KAN_M Automated Weather Station (Fausto and van As, 2019 ). Proglacial river discharges from two permanent gauging stations (Rennermalm et al., 2013b; 2017; van As et al., 2017; 2019 ) were also incorporated into this study. The 60.2 km2 July 2016 Rio Behar catchment boundary was delineated using a fixed-wing drone and WorldView satellite imagery following the methods of Smith et al. 2017 . This 2016 catchment boundary is presented for illustration purposes inFigure 1 , but is not otherwise used in this study.