Snow and precipitation estimates in high-mountain regions typically suffer from low temporal and spatial resolution and large uncertainties. Here, we present a two-step statistically based model to derive spatio-temporal highly resolved estimates of snow water equivalent (SWE) across the Swiss Alps. A multiple linear regression model (Step-1 MLR) was first used to combine the CombiPrecip radar-gauge product with the precipitation and wind speed (10 m from the ground) of the numerical weather prediction model COSMO-1 in order to adjust the precipitation estimates. Step-1 MLR was trained with SWE data from a cosmic ray sensor (CRS) installed on the Plaine Morte glacier and tested with SWE data from a CRS on the Findel glacier. Step-1 MLR was then applied to the entire area of eight Swiss glaciers and evaluated with scattered end-of-season in-situ manual SWE measurements. The cumulative estimates of Step-1 MLR were found to agree well with the end-of-season measurements. The observed differences can partially be explained by considering the radar visibility, melting processes and preferential snow deposition, which are dictated by the local topography and local weather conditions. To address these limitations of Step-1 MLR, several high-resolution topographical parameters and a solar radiation parameter were included in the subsequent MLR version (Step-2 MLR). Step-2 MLR was evaluated by means of cross-validation, and it showed an overall correlation of 0.78 and a mean bias error of 4 mm with respect to end-of-season in-situ measurements. Step-2 MLR was also evaluated for non-glacierized regions by evaluating it against twice-monthly manual SWE measurements at 44 sites in the Swiss Alps. In such a setting, the Step-2 model showed an overall weaker correlation (0.53) and a higher mean bias error (31 mm). On the other hand, negative variations of the measured SWE were removed because of the lower altitude of the sites, thereby leading to more pronounced melting periods, which again increased the correlation values to 0.63 and reduced the mean bias error to 12 mm. Such results confirm the high potential of the model for applications to other mountainous regions.

Abstract. Accurate and reliable solid precipitation estimates for high mountain regions are crucial for many research applications. Yet, measuring snowfall at high elevation remains a major challenge. In consequence, observational coverage is typically sparse, and the validation of spatially distributed precipitation products is complicated. This study presents a novel approach using reliable daily snow water equivalent (SWE) estimates by a cosmic ray sensor on two Swiss glacier sites to assess the performance of various gridded precipitation products. The ground observations are available during two and four winter seasons. The performance of three readily-available precipitation data products based on different data sources (gauge-based, remotely-sensed, and re-analysed) is assessed in terms of their accuracy compared to the ground reference. Furthermore, we include a data set, which corresponds to the remotely-sensed product with a local adjustment to independent SWE measurements. We find a large bias of all precipitation products at a monthly and seasonal resolution, which also shows a seasonal trend. Moreover, the performance of the precipitation products largely depends on in situ wind direction during snowfall events. The varying performance of the three precipitation products can be partly explained with their compilation background and underlying data basis.

Abstract. Although reanalysis products for remote high-mountain regions provide estimates of snow precipitation, this data is inherently uncertain and assessing a potential bias is difficult due to the scarcity of observations, thus also limiting their reliability to evaluate long-term effects of climate change. Here, we compare the winter mass balance of 95 glaciers distributed over the Alps, Western Canada, Central Asia and Scandinavia, with the total precipitation from the ERA-5 and the MERRA-2 reanalysis products during the snow accumulation seasons from 1981 until today. We propose a machine learning model to adjust the precipitation of reanalysis products to the elevation of the glaciers, thus deriving snow water equivalent (SWE) estimates over glaciers uncovered by ground observations and/or filling observational gaps. We use a gradient boosting regressor (GBR), which combines several meteorological variables from the reanalyses (e.g. air temperature, relative humidity) with topographical parameters. These GBR-derived estimates are evaluated against the winter mass balance data by means of a leave-one-glacier-out cross-validation (site-independent GBR) and a leave-one-season-out cross-validation (season-independent GBR). Both site-independent and season-independent GBRs allowed reducing (increasing) the bias (correlation) between the precipitation of the original reanalyses and the winter mass balance data of the glaciers. Finally, the GBR models are used to derive SWE trends on glaciers between 1981 and 2021. The resulting trends are more pronounced than those obtained from the total precipitation of the original reanalyses. On a regional scale, significant 41-year SWE trends over glaciers are observed in the Alps (MERRA-2 season-independent GBR: +0.4 %/year) and in Western Canada (ERA-5 season-independent GBR: +0.2 %/year), while significant positive/negative trends are observed in all the regions for single glaciers or specific elevations. Negative (positive) SWE trends are typically observed at lower (higher) elevations, where the impact of rising temperatures is more (less) dominant.

No abstract

No abstract