Surface mass-balance gradients from elevation and ice flux data, Columbia River Basin, Canada

Abstract:

The mass balance-altitude relation for a given glacier is required for many numerical models of ice flow. Direct measurements of this relation using remote sensing methods are complicated by ice dynamics, so observations are currently limited to glaciers where surface mass balance measurements are routinely made. We use a mass conservation, flux-gate approach to estimate annual surface mass balance using ice flux and elevation data for three glaciers in the Columbia Mountains of British Columbia, Canada. Repeat airborne laser scanning (ALS) surveys, ice penetrating radar surveys and publicly available maps of bed topography provide data to estimate changes in surface elevation and ice flux. Our method yields modeled mass-balance gradients within one standard error of those obtained from direct measurement of surface mass balance. Low average mean error (–0.02 ± 0.14 m w.e.) and average percent bias (3.0 ± 15.6%) between observed and modeled mass balance over flux bins suggest mass balance is reliably modeled. Mass conservation, assessed with glaciological data, is respected for 81% of flux bins representing 85% of the total glacier area. Due to relatively sparse observations for some bins, it remains uncertain whether mass conservation is violated or whether our glaciological data insufficiently represent surface mass balance for some bins. Uncertain mass conservation and high average modeled mass balance uncertainty (0.55 m w.e.) cautions against over-interpreting individual flux-bin mass balance. Ice velocity uncertainty dominates ice flux uncertainty for gates with ice velocity below 10 m a^-1. We find that using modeled ice thicknesses produces similar modeled mass balances relative to using observations of ice thickness. Given the performance of modeled ice thickness and the increasing availability of ice velocity and surface topography data, we conclude that similar efforts to produce mass-balance gradients using modern high-resolution datasets are feasible for entire mountain ranges.