High-resolution ice sheet surface mass-balance and spatiotemporal runoff simulations: Kangerlussuaq, west Greenland

Sebastian H Mernild,Glen E Liston,Dirk Van As,Bent Hasholt,Jacob C Yde

doi:10.1080/15230430.2017.1415856

Abstract

ABSTRACT The spatiotemporal distribution of freshwater runoff from the Greenland Ice Sheet (GrIS) determines the hydrographic and circulation conditions in Greenlandic fjords. The distribution of GrIS first-order atmospheric forcings, surface mass-balance (SMB), including snow/ice melt, and freshwater river discharge from the Kangerlussuaq drainage catchment were simulated for the thirty-five-year period 1979/1980–2013/2014. ERA-Interim (ERA-I) products, together with the modeling software package SnowModel, were used with relatively high-resolutions of 3-h time steps and 5-km horizontal grid increments. SnowModel simulated and downscaled grid mean annual air temperature (MAAT) and SMB correspond well to point observations along a weather station transect (the K-transect). On average, simulated catchment runoff was, however, overestimated and subsequently adjusted against observed runoff. This overestimation could likely be because of missing multiyear firn processes, such as nonlinear meltwater retention, percolation blocked by ice layers, and refreezing. In the GrIS Kangerlussuaq catchment, the simulated thirty-five-year MAAT was −15.0 ± 1.4°C, with a mean 0° isotherm below 280 m a.s.l. near the ice sheet margin. At the ice sheet margin, on average, 45 percent of precipitation fell as snow. At 2,000 m a.s.l., snow constituted 98 percent of the total precipitation. At the catchment outlet of Watson River draining into the fjord Kangerlussuaq, 80 percent of the simulated runoff originated from GrIS ice melt, 15 percent from snowmelt, and 5 percent from rain.

Highlights

The Greenland Ice Sheet (GrIS) plays an essential role in the Arctic hydrological cycle and for the individual Greenland catchment water budgets (Bing et al 2016; Hanna et al 2009; Langen et al 2016), where freshwater runoff is the hydrological link between snowmelt and ice melt and hydrographic and circulation conditions in fjords and the surrounding ocean (e.g., Rahmstorf et al 2005; Cullather et al 2016; Hansen et al 2016)
The SnowModel/HydroFlow simulations over the GrIS Kangerlussuaq catchment show a detailed and physically realistic representation of surface snow and ice ablation, snowpack evolution, surface mass-balance (SMB), and runoff routing based on the established flow network at relatively high temporal and spatial resolution compared with previous studies by Mernild et al (2011, 2012)
SnowModel/HydroFlow capabilities present a contrast with studies that (1) largely rely on air temperature as a proxy for the energy available for melt, and (2) do not link between surface runoff production from snowmelt and ice-melt processes and the associated freshwater fluxes to downstream areas and surrounding oceans through an estimated flow network that combines the individual grid cells that make up the simulation catchment

Summary

Introduction

The Greenland Ice Sheet (GrIS) plays an essential role in the Arctic hydrological cycle and for the individual Greenland catchment water budgets (Bing et al 2016; Hanna et al 2009; Langen et al 2016), where freshwater runoff is the hydrological link between snowmelt and ice melt and hydrographic and circulation conditions in fjords and the surrounding ocean (e.g., Rahmstorf et al 2005; Cullather et al 2016; Hansen et al 2016).The GrIS net mass-balance loss has increased since the 1980s (e.g., Box and Colgan 2013; Church et al 2013; Hanna et al 2013; Langen et al 2015; Rignot et al 2011; Shepherd et al 2012; van den Broeke et al 2016), where 60 percent of the mass loss since 1991 was largely caused by the surface mass-balance (SMB) and the remainder by calving dynamics (Hurkmans et al 2014; van den Broeke et al 2016). Direct independently observed GrIS point MAAT and SMB, and catchment-integrated river runoff time series, which partly covered the period 1990–2014, were used to verify the performance of the SnowModel and HydroFlow simulations.

Results

Conclusion