Uncertainty Assessment for Mean Snow Cover Depth Derived from Direct Measurements on Aldegondabreen Glacier (Svalbard)

Abstract

Received March 27, 2022; revised May 5, 2023; accepted June 27, 2023This study introduces an empirical equation allowing to estimate an uncertainty of area-averaged snow depth on the Aldegondabreen Glacier, computed from standard snow surveys and made by an avalanche probe or by similar equipment. The two-decade history of the ongoing mass-balance monitoring program on this glacier shows that the methodology of field work on snow-measuring survey varies somewhat from year to year: the number and location of measurement points change. To identify and quantify long-term trends and variations in snow cover, it is crucial to assess the inter-comparability of the data in the obtained measurement series. The proposed equation was intended to solve this task basing on the collected data only, allowing to estimate the uncertainty even retrospectively. To build this equation, we applied a bootstrap statistical approach to the results of snow surveys carried out in Svalbard in 2015–2021. After interpolating the field measurements, obtained rasters were sampled sequentially with different numbers of points, simulating the real snow survey. The points were initially located in a form of a quasiregular grid and then randomly shifted between the iterations. After a thousand simulations for each number of points, the standard deviations were calculated relative to the “true” values, derived from corresponding rasters. These standard deviations, which we admit to be a random error of the area-averaged snow depth value, expectedly decrease with the number of sampling points and increase with the coefficient of variation (\({{{\text{C}}}_{\user1{v}}}\)). The well-known \({{{\text{C}}}_{\user1{v}}}\) index indirectly characterizes the irregularity of snow cover. After approximating the bootstrap results, the authors derived an equation that yields a relative error. The equation includes only two predictors which are the probing density per area unit and the \({{{\text{C}}}_{\user1{v}}}\), which potentially allows using it for the other glaciers. However, the universality of the empirically obtained coefficients is debatable, since they may vary due toa glacier size, its morphology and other parameters.