Abstract. The root-mean-squared error (RMSE) and mean absolute error (MAE) are widely used metrics for evaluating models. Yet, there remains enduring confusion over their use, such that a standard practice is to present both, leaving it to the reader to decide which is more relevant. In a recent reprise to the 200-year debate over their use, Willmott and Matsuura (2005) and Chai and Draxler (2014) give arguments for favoring one metric or the other. However, this comparison can present a false dichotomy. Neither metric is inherently better: RMSE is optimal for normal (Gaussian) errors, and MAE is optimal for Laplacian errors. When errors deviate from these distributions, other metrics are superior.

Ocean-driven melting of ice shelves is a primary mechanism for ice loss from Antarctica.However, due to the difficulty in accessing the sub-ice shelf ocean cavity, the relationship between ice shelf melting and ocean conditions is poorly understood, particularly near the grounding zone, where the ice transitions from grounded to floating. We present the first borehole oceanographic observations from the grounding zone of the Ross Ice Shelf, Antarctica's largest ice shelf by area. Contrary to predictions that tidal currents near grounding zones mix the water column, we found that Ross Ice Shelf waters were vertically stratified. Current velocities at middepth in the ocean cavity did not change significantly over measurement periods at two different parts of the tidal cycle. The observed stratification resulted in low melt rates near this portion of the grounding zone, inferred from phase-sensitive radar observations. These melt rates were generally <10 cm/year, which is lower than average for the Ross Ice Shelf (~20 cm/year). Melt rates may be higher at portions of the grounding zone that experience higher subglacial discharge or stronger tidal mixing. Stratification in the cavity at the borehole site was prone to diffusive convection as a result of ice shelf melting. Since diffusive convection influences vertical heat and salt fluxes differently than shear-driven turbulence, this process may affect ice shelf melting and merits further consideration in ocean models of sub-ice shelf circulation.Plain Language Summary Ice shelf melting is an important player in ice loss from the Antarctic Ice Sheet, affecting sea level rise. Ice shelf melting is controlled by ocean properties and processes, but sparse observations of the sub-ice shelf ocean cavity limit our understanding of these controls and thus limit our ability to predict sea level rise. This study presents rare ocean observations deep below the largest ice shelf by area, the Ross Ice Shelf, far from the open ocean. The observed ocean setting is surprisingly quiescent, and waters are cold, around À2°C. This study also presents new, highly localized ice shelf melting measurements at the site that show that these ocean conditions lead to slow ice shelf melting of only centimeters per year. These observations reveal the ways in which the Ross Ice Shelf contrasts with rapidly melting ice shelves affected by warmer seawater elsewhere in West Antarctica. Thus, they adds nuance to our scientific understanding of ice-ocean interactions around the Antarctic continent.Ice shelf basal melting is driven by the flux of heat from sub-ice shelf cavity water masses to the ice shelf base. On broad spatial scales, basal melt rates increase with the thermal driving, the difference between water BEGEMAN ET AL.7438

25The hydrologic system beneath the Antarctic Ice Sheet is thought to influence both the 26 dynamics and distribution of fast flowing ice streams, which discharge most of the ice lost by the 27 ice sheet. Despite considerable interest in understanding this subglacial network and its affect on 28 ice flow, in situ observations from the ice sheet bed are exceedingly rare. Here we describe the 29 first sediment cores recovered from an active subglacial lake. The lake, known as Subglacial 30 Lake Whillans, is part of a broader, dynamic hydrologic network beneath the Whillans Ice 31Stream in West Antarctica. Even though "floods" pass through the lake, the lake floor shows no 32 evidence of erosion or deposition by flowing water. By inference, these floods must have 33 insufficient energy to erode or transport significant volumes of sediment coarser than silt. 34Consequently, water flow beneath the region is probably incapable of incising continuous 35 channels into the bed and instead follows preexisting subglacial topography. Sediment on the 36 lake floor consists of till deposited during intermittent grounding of the ice stream following 37 flood events. The fabrics within the till are weaker than those thought to develop in thick 38 deforming beds suggesting subglacial sediment fluxes across the ice plain are currently low and 39 unlikely to have a large stabilizing effect on the ice stream's grounding zone. 40Text 41 42

Liquid water occurs below glaciers and ice sheets globally, enabling the existence of an array of aquatic microbial ecosystems. In Antarctica, large subglacial lakes are present beneath hundreds to thousands of metres of ice, and scientific interest in exploring these environments has escalated over the past decade. After years of planning, the first team of scientists and engineers cleanly accessed and retrieved pristine samples from a West Antarctic subglacial lake ecosystem in January 2013. This paper reviews the findings to date on Subglacial Lake Whillans and presents new supporting data on the carbon and energy metabolism of resident microbes. The analysis of water and sediments from the lake revealed a diverse microbial community composed of bacteria and archaea that are close relatives of species known to use reduced N, S or Fe and CH4 as energy sources. The water chemistry of Subglacial Lake Whillans was dominated by weathering products from silicate minerals with a minor influence from seawater. Contributions to water chemistry from microbial sulfide oxidation and carbonation reactions were supported by genomic data. Collectively, these results provide unequivocal evidence that subglacial environments in this region of West Antarctica host active microbial ecosystems that participate in subglacial biogeochemical cyclingauthorsversionPeer reviewe

Abstract. The mean absolute error (MAE) and root mean squared error (RMSE) are widely used metrics for evaluating models. Yet, there remains enduring confusion over their use, such that a standard practice is to present both, leaving it to the reader to decide. Some of this confusion arises from a recent debate between Willmott and Draxler (2005) and Chai and Draxler (2014), in which either side presents their arguments for one metric over the other. Neither side was completely correct; however, because neither metric is inherently better: MAE is optimal for Laplacian errors, and RMSE is optimal for normal (Gaussian) errors. When errors deviate from these distributions, other metrics are superior.