Hourly surface meltwater routing for a Greenlandic supraglacial catchment across hillslopes and through a dense topological channel network

Abstract

Abstract. Recent work has identified complex perennial supraglacial stream/river networks in areas of the Greenland Ice Sheet (GrIS) ablation zone. Current surface mass balance (SMB) models appear to overestimate meltwater runoff in these networks compared to in-channel measurements of supraglacial discharge. Here, we constrain SMB models using the Hillslope River Routing Model (HRR), a spatially explicit flow routing model used in terrestrial hydrology, in a 63 km2 supraglacial river catchment in southwest Greenland. HRR conserves water mass and momentum and explicitly accounts for hillslope routing, and we produce hourly flows for nearly 10,000 channels given inputs of an ice surface DEM, a remotely sensed supraglacial channel network, SMB-modelled runoff, and an in situ discharge dataset used for calibration. Model calibration yields a Nash Sutcliffe Efficiency as high as 0.92 and physically realistic parameters. We confirm earlier assertions that SMB runoff exceeds the conserved mass of water routed to match measured flows in this catchment (by 12–59 %) and that large channels do not dewater overnight despite a diurnal shutdown of SMB runoff production. We further test hillslope routing and network density controls on channel discharge and conclude that explicitly including hillslope flow and routing runoff through a realistically fine channel network produces the most accurate results. Modelling complex surface water processes is thus both possible and necessary to accurately simulate the timing and magnitude of supraglacial channel flows, and we highlight a need for additional in situ discharge datasets to better calibrate and apply this method elsewhere on the ice sheet.

Supraglacial Drainage Efficiency of the Greenland Ice Sheet Estimated From Remote Sensing and Climate Models

“…However, VHR satellite imagery and DEMs have limited temporal and spatial coverage (King et al., 2016; Rippin et al., 2015; Smith et al., 2015; Yang & Smith, 2013) making time‐varying evolution of the supraglacial drainage patterns difficult to capture. Due to this limitation, current routing models commonly assume supraglacial drainage pattern and efficiency are temporally static, although some simple parameterizations of routing distance or velocity may be used to mimic the temporal evolution of meltwater routing (Gleason et al., 2021; Yang et al., 2020). Our finding of a direct, correlative R ‐ D d relationship suggests that a simpler, alternative approach may be to parameterize this evolution directly from climate model output (Figure 3).…”

“…Climate models provide daily or hourly simulations of surface meltwater runoff (Fettweis et al, 2020). Because supraglacial channels are fed by meltwater runoff, their pattern and abundance evolves temporally in response to surface meltwater production (Lampkin and VanderBerg, 2014;Smith et al, 2017;Yang et al, 2017;Decaux et al, 2018;Yang et al, 2018;Pitcher and Smith, 2019;Muthyala et al, 2020;Gleason et al, 2021).…”

“…Supraglacial catchment boundaries were extracted from 5 m ArcticDEM mosaic rasters following Karlstrom and Yang (2016). The topographic depression breaching method of Lindsay (2016) was used to determine supraglacial flow direction and flow accumulation following Gleason et al (2021). Supraglacial catchment outlets were manually identified and used as pour points to generate upstream catchment boundaries, following Karlstrom and Yang (2016).…”

“…ArcticDEM is the highest spatial resolution DEM available for the GrIS and has been widely used in ice sheet hydrology studies (Crozier et al., 2018; Lu et al., 2020; Smith et al., 2017; Yang et al., 2018, Yang, Smith, Sole, et al., 2019). Importantly, it can discriminate continuous supraglacial drainage networks that match the general pattern of satellite‐mapped supraglacial river/stream networks quite well (Gleason et al., 2021; Yang et al., 2018), although it may overestimate or underestimate streams depending on the choice of the channel initiation threshold A c (Lu et al., 2020; Yang, Smith, Sole, et al., 2019). The 5 m mosaic ArcticDEM product was used because of its sufficient resolution for representing interfluve routing distance (Yang et al., 2018), and its higher computational efficiency compared to the 2 m tile ArcticDEM v2.0 product.…”

“…Climate models provide daily or hourly simulations of surface meltwater runoff (Fettweis et al., 2020). Because supraglacial channels are fed by meltwater runoff, their pattern and abundance evolves temporally in response to surface meltwater production (Decaux et al., 2019; Gleason et al., 2021; Lampkin & VanderBerg, 2014; Muthyala et al., 2020; Pitcher & Smith, 2019; Smith et al., 2017; Yang et al., 2017, 2018). Previous studies have noted some qualitative relationships between channel planforms and modeled runoff, for example, becoming wider as surface runoff increases (Smith et al., 2015; Yang et al., 2017), migrating to higher elevations (Lampkin & VanderBerg, 2014), or growing in area as modeled runoff increases (Lu et al., 2020; Yang et al., 2021).…”