Glacier melt is an important fresh water source. Seasonal changes can have impacting consequences on downstream water resources management. Today’s glacier monitoring lacks an observation-based tool for regional, sub-seasonal observation of glacier mass balance and a quantification of associated meltwater release at high temporal resolution. The snowline on a glacier marks the transition between the ice and snow surface, and is, at the end of the summer, a proxy for the annual glacier mass balance. It was shown that glacier mass balance model simulations closely tied to sub-seasonal snowline observations on optical satellite sensors are robust for the observation date. Recent advances in remote sensing permit efficient and extensive snowline mapping. Different methods automatically discriminate snow over ice on high- to medium-resolution optical satellite images. Other studies rely on lower ground resolution optical imagery to retrieve snow cover fraction at pixel level and produce regional maps of snow cover extent. However, state-of-the-art methods using optical sensors still have important shortcomings, such as cloud-cover related issues. Images acquired by Synthetic Aperture Radar (SAR), which are almost insensitive to cloud coverage, have proofed suitable for transient snowline delineation. The combination of SAR and optical data in a complementary way carries a unique potential for a better monitoring of snow depletion on high temporal and spatial resolution. The aim of this work is to map snow cover over glaciers by combining Sentinel-1 SAR, Sentinel-2 multispectral and lower resolution MODIS images. Consecutively, we developed an approach that can automatically handle classification of multi-source and multi-resolution satellite image stacks. This provides a unique solution for continuous snowline mapping since the beginning of the century. With the provided close-to-daily transient snow cover fractions on glacier level, we provide the basis for a new strategy to directly integrate multi-source satellite image classification into glacier mass balance monitoring.

This paper presents a novel data fusion technique for improving the snow cover monitoring for a mesoscale Alpine region, in particular in those areas where two information sources disagree. The presented methodological innovation consists in the integration of remote-sensing data products and the numerical simulation results by means of a machine learning classifier (support vector machine), capable to extract information from their quality measures. This differs from the existing approaches where remote sensing is only used for model tuning or data assimilation. The technique has been tested to generate a time series of about 1300 snow maps for the period between October 2012 and July 2016. The results show an average agreement between the fused product and the reference ground data of 96%, compared to 90% of the moderate-resolution imaging spectroradiometer (MODIS) data product and 92% of the numerical model simulation. Moreover, one of the most important results is observed from the analysis of snow cover area (SCA) time series, where the fused product seems to overcome the well-known underestimation of snow in forest of the MODIS product, by accurately reproducing the SCA peaks of winter season.

This paper presents a new concept to derive the snow water equivalent (SWE) based on the joint use of snow model (AMUNDSEN) simulation, ground data, and auxiliary products derived from remote sensing. The main objective is to characterize the spatial-temporal distribution of the model-derived SWE deviation with respect to the real SWE values derived from ground measurements. This deviation is due to the intrinsic uncertainty of any theoretical model, related to the approximations in the analytical formulation. The method, based on the k-NN algorithm, computes the deviation for some labeled samples, i.e., samples for which ground measurements are available, in order to characterize and model the deviations associated to unlabeled samples (no ground measurements available), by assuming that the deviations of samples vary depending on the location within the feature space. Obtained results indicate an improved performance with respect to AMUNDSEN model, by decreasing the RMSE and the MAE with ground data, on average, from 154 to 75 mm and from 99 to 45 mm, respectively. Furthermore, the slope of regression line between estimated SWE and ground reference samples reaches 0.9 from 0.6 of AMUNDSEN simulations, by reducing the data spread and the number of outliers.