In snow-dominated river basins, floods often occur during early summer, when snowmelt-induced runoff superimposes with rainfall-induced runoff. An earlier onset of seasonal snowmelt as a consequence of a warming climate is often expected to shift snowmelt contribution to river runoff and potential flooding to an earlier date. Against this background, we assess the impact of rising temperatures on seasonal snowpacks and quantify changes in timing, magnitude and elevation of snowmelt. We analyse in situ snow measurements, conduct snow simulations and examine changes in river runoff at key gauging stations. With regard to snowmelt, we detect a threefold effect of rising temperatures: snowmelt becomes weaker, occurs earlier and forms at higher elevations. Due to the wide range of elevations in the catchment, snowmelt does not occur simultaneously at all elevations. Results indicate that elevation bands melt together in blocks. We hypothesise that in a warmer world with similar sequences of weather conditions, snowmelt is moved upward to higher elevation. The movement upward the elevation range makes snowmelt in individual elevation bands occur earlier, although the timing of the snowmelt-induced runoff stays the same. Meltwater from higher elevations, at least partly, replaces meltwater from elevations below.
Impact of rising temperatures on seasonal snowpacks.
Investigations on the point scale for elevation bands and from a catchment perspective.
Threefold effect: snowmelt becomes weaker, occurs earlier and forms at higher elevations.
Meltwater from higher elevations can, at least partly, replace meltwater from elevations below.
Alpine landscapes react particularly sensitively towards climatic changes. By the end of the century, glaciers in the European Alps most likely will be gone, and seasonal snowpacks downsized to a small fraction (Horton et al. 2006; Zemp et al. 2006; Huss 2011; Fatichi et al. 2014; Gobiet et al. 2014; Beniston et al. 2018; Hanzer et al. 2018). Cryospheric changes are ongoing and will fundamentally alter river runoff and water availability on regional and global scales (Stewart 2009; Clow 2010; Gillan et al. 2010; Viviroli et al. 2011; Laghari et al. 2012; Radić & Hock 2014).
Snowmelt is an important flood-generating process (Berghuijs et al. 2019; Parajka et al. 2019). In snow-dominated river basins, flooding often occurs during spring and early summer, when high baseflow due to snowmelt overlaps with strong precipitation (Wetter et al. 2011; Vormoor et al. 2015). For Switzerland, changes in mean and extreme regimes in both rainfall and melt-dominated regions are expected (Addor et al. 2014; Brunner et al. 2019). Recent studies investigating regional flood discharge trends for Europe indicate that climate change impact on flood timing and frequency is ongoing already (Blöschl et al. 2017, 2019; Bertola et al. 2020).
Studies analysing cryospheric and hydrological climate change impacts often focus on changes in timing and mean annual cycles in relatively small areas, i.e. individual valleys. Seldom, larger catchments covering a wide range of elevations and pre-alpine to high alpine areas are analysed. So far, only very few studies have resolved changes along the elevation range (e.g. Hunsaker et al. 2012; Kormann et al. 2015). However, in alpine settings, temperature is a function of both time and elevation. Elevation as an additional dimension with regard to changes in snowmelt often receives only little attention.
In this study, we analyse snow observations, conduct snow simulations and analyse discharge records from key gauging stations in the Rhine Basin upstream gauge Basel. We use Moderate Resolution Imaging Spectrometer (MODIS) snow cover maps to validate our model results. The main goal is to better understand the impact of rising temperatures on alpine snowpacks. We focus on the timing, magnitude and elevation of snowmelt events and assess how changes in snowmelt translate into changes in river runoff.
STUDY AREA AND DATA
The study area is the Rhine Basin upstream gauge Basel. The basin covers a total area of 3.59 × 104 km² and an elevation range of almost 4,000 m (Figure 1). Gauge Basel is located at 294 m a.s.l. The southern parts of the basin are of high-alpine character. The highest mountain peaks reach up to elevations above 4,000 m a.s.l. In winter, precipitation is often solid and accumulates in temporary snowpacks. Depending on the elevation, areas are covered by snow for weeks or even months. A considerable fraction of runoff originates from snowmelt (Stahl et al. 2016). In general, elevation is an important factor determining local climatic conditions, vegetation and land use. The basin also encompasses large parts of the Swiss Plateau, hilly to flat areas north of the alpine ridge, which mostly cover elevations between 300 and 1,000 m a.s.l. In recent decades, temperatures in the Swiss Alps have been rising at a very fast pace (e.g. Ceppi et al. 2012; Rottler et al. 2019). Studies inter alia hint at changes in forest growth, species distribution and phenology (e.g. Walther et al. 2005; Gehrig-Fasel et al. 2007; Mietkiewicz et al. 2017). Very prominent changes have been observed with regard to glaciers (Zemp et al. 2006; Huss & Farinotti 2012; Fischer et al. 2014; Beniston et al. 2018). Strong alteration of runoff due to rising temperatures is expected in glacierized headwater catchments (Huss et al. 2008; Junghans et al. 2011). However, the importance of ice melt with regard to runoff decreases with the distance to the glacier. At gauge Basel, only 2% of the total annual runoff originates from ice melt (Stahl et al. 2016). The investigations presented are based on daily records of temperature, rainfall, snow depth and discharge measurements, MODIS snow cover maps and in situ observations of snow water equivalent (SWE), as well as gridded temperature and rainfall datasets.
All station-based meteorological and cryospheric observational data were provided by the Federal Office of Meteorology and Climatology of Switzerland (MeteoSwiss) and the Institute for Snow and Avalanche Research (SLF) from the Swiss Federal Institute for Forest, Snow and Landscape Research (WSL) (Supplementary Table A1). Discharge time series were obtained from the Global Runoff Data Centre (GRDC). Daily and nearly cloud-free MODIS snow cover maps for the European Alps in a resolution of 250 m and from 2002 onwards were provided by the Institute for Earth Observation, Eurac Research, Bolzano/Bozen, Italy (Matiu et al. 2019, 2020). All data analysis was carried out using the free software R 3.6.1 (R Core Team 2019).
MeteoSwiss and WSL conduct monitoring and analysis programmes for the snowpack in Switzerland at numerous locations. In this study, we focus on a few selected stations. Selected records of snow depth stand out by their length and cover a broad range of elevations, i.e. 555–2,691 m a.s.l. To examine changes in river runoff, we exert a nested catchment approach in the two main branches of the Rhine River network: the Aare branch (left in Figure 1) and the Rhine branch including Lake Constance (right in Figure 1). The two branches flow together approximately 60 km upstream Basel. The Aare River is the main tributary of the High Rhine. We analyse discharge data from the Aare River recorded at gauges Bern, Brugg and Untersiggenthal. Between gauges Brugg and Untersiggenthal, the two rivers Reuss and Limmat merge with the Aare River increasing the size of the Aare River Basin by about 50%. The Rhine branch is subdivided into nested basins from gauges Diepoldsau, Neuhausen and Rekingen. The enclosing character of individual basins enables the backtracking and localisation of changes in runoff. Gridded temperature and precipitation data used to drive snow simulations originate from E-OBS v12 gridded datasets from the European Climate & Dataset (ECA&D) project (Hofstra et al. 2008; Cornes et al. 2018). Within the project ‘EDgE – End-to-end Demonstrator for improved decision making in the water sector in Europe’ by Copernicus, these data were further refined to a 5 km grid using external drift kriging (EDK). EDK addresses altitude effects and provides input data for hydrological modelling at high spatial resolutions (e.g. Zink et al. 2017; Marx et al. 2018; Samaniego et al. 2018; Thober et al. 2018).
Our analysis is based on the combination of both observed and modelled data. We analyse changes in snowpacks and snowmelt dynamics and assess the effect of changes in snowpacks on the timing and magnitude of snowmelt-induced runoff. An in-depth analysis of discharge data and governing flood drivers complements the investigations (Figure 2).
To get a first insight into observed changes in snowpacks, we display snow observations as raster graphs and determine the mean annual cycle for snow depth and accumulation/melt rates for two time windows: 1958–1987 and 1988–2017. We select the time windows in order to have two 30-year windows with maximum data availability at individual stations while keeping a sufficient overlap with the snow simulation period (1951–2014). Enhanced warming since the 1980s strongly affects Swiss snowpacks (Laternser & Schneebeli 2003; Scherrer et al. 2004; Marty 2008; Philipona et al. 2009; Rottler et al. 2019). By comparing two 30-year windows located before and after the onset of strong changes, we aim to capture signals of change in the snow cover in a robust way. Accumulation/melt rates are calculated as the difference in snow depth between two consecutive days. These mean annual cycles are calculated after applying a 30-day moving average window.
Snow simulations: model selection, adaption and verification
We conduct snow simulations in order to extend first insights gained by analysing snow observations. For our study, we need a flexible snow model that is able to easily operate on spatial scales of several thousand square kilometres and time frames of several decades. The main focus is on the effect of changing temperatures on the timing and magnitude of snowmelt events. As rain-on-snow can be an important driver of high melt rates, the energy input from rainfall needs to be represented. In view of the model requirements, we conduct snow simulations using the energy balance method of the modelling framework ECHSE (Eco-Hydrological Simulation Environment) (Kneis 2015). In comparison to empirical approaches, physically based model setups allow more detailed investigations of the snow cover processes and offer several possibilities to refine the model setup if necessary.
ECHSE is an open-source modelling framework developed to facilitate ‘the rapid development of new, re-usable simulation tools and, more importantly, the safe modification of existing formulations’ (Kneis 2015). The ECHSE environment provides mathematical representations of various hydrological processes, e.g. evapotranspiration, runoff concentration or the dynamics of a snow cover. Recent application of the ECHSE modelling framework includes Abon et al. (2016), Kneis et al. (2017) and Pilz et al. (2020). A key feature of ECHSE is its well-documented open-source software, which forms the basis for transparent and reproducible studies. In our analysis, we do not build an entire model engine but focus on the simulation of the Alpine snow cover. The standard ECHSE snow module couples mass and energy balances with SWE and snow energy content as primary state variables. Snow albedo is considered an auxiliary state variable. The meltwater outflow from the snowpack is computed from available meltwater and the current hydraulic conductivity in the snowpack. A detailed description of the snow module including all equations can be found in Kneis (2014). We use the original functions from the ECHSE online repository written in C ++ and conduct simulations in an R environment. To avoid instabilities in the melting process, we perform daily snow simulations in 24 intermediate steps, with daily mean temperature input values being disaggregated assuming sinusoidal diurnal temperature variations. Thereby the amplitude was set to 8 °C, and the timing of the maximum value to 14:00 h. Daily precipitation input is equally distributed among all intermediate time steps. An average yearly cycle of global radiation is calculated based on measurements (1981–2017) from the Swiss meteorological station Napf (1,404 m a.s.l) and used as radiation input. Values of air pressure, relative humidity, wind speed and cloud coverage are currently assumed to be constant with values of 1,000 hPa, 70%, 1 m/s and 50%, respectively.
As a first step, we validate the model setup and modifications using measured temperature and precipitation records from meteorological stations and a default parameter set for daily data from two sites in the German low mountain ranges (Kneis 2014). After ascertaining flawless model operation, we re-calibrate model parameters (see Supplementary Table B1) using the particle swarm optimisation algorithm from the R-package ‘ppso’ (https://github.com/TillF/ppso). This calibration aims to improve model performance by customising parameters to the alpine setting. We employ 5,000 model runs and assess model performance using the average normalised root-mean-square error (NRMSE) using the standard deviation of the observations (Zambrano-Bigiarini 2017) between simulated and measured SWE for the locations Andermatt, Davos, Zermatt and Weissfluhjoch within the time frame 1983–2014, i.e. the maximum overlap of available meteorological input for snow simulations. We use the NRMSE in the attempt to weight stations equally during calibration. We give the NRMSE in percentage. A value of 100% indicates that the RMSE is equal to the standard deviation. To facilitate the evaluation of attained values, we also calculate the percentage bias (PBIAS) (Sorooshian et al. 1993) and the Nash–Sutcliffe Efficiency (NSE) (Nash & Sutcliffe 1970).
Next, we use the resulting parameters and gridded datasets of temperature and precipitation (available for the time frame 1950–2014) to perform snow simulations for the entire study area upstream gauge Basel. The temperature grid used to drive the simulations was obtained by downscaling from 5 to 1 km resolution using a lapse rate-based approach. The lapse rates, i.e. changes of temperature with elevation, were determined every day individually based on the temperature grid and information on elevation from the EU-DEM v1.1 by the EU Copernicus Programme (resampled to a resolution of 1 km). Based on the information on temperature and elevation, we calculate lapse rates as linear trends using the Theil–Sen trend estimator (Theil 1950; Sen 1968; Bronaugh & Werner 2013). To downscale temperature to the 1-km grid, the temperate lapse rates are multiplied by the elevation differences calculated as the difference between the elevation of the 1 km resolution grid cells and the elevation of the respective cell in the 5-km grid.
In the following, we aggregate results into 50 m elevation bands and calculate annual cycles of mean average and trends in SWE, SWE area volume totals and accumulation/melt rates. Accumulation/melt rates are calculated as the difference in SWE volume between two consecutive days. In addition, we assess mean annual cycles and trends for temperature and precipitation. Trends are determined using the Theil–Sen trend estimator (Theil 1950; Sen 1968; Bronaugh & Werner 2013) after applying a 30-day moving average filter. Grid points showing a continuous accumulation of snow over the simulation period were classified as ‘glacier points’ and not included in this analysis (2.75% of the data points). To validate our modelling approach, we calculate annual average snow cover durations and compare with satellite-based snow cover maps. With regard to snow simulations, areas are considered covered by snow when SWE is equal to or exceeds 2 mm. All points classified as ‘water body’ in more than 90% of the MODIS maps were treated as lake surfaces and not included in the analysis.
Snowmelt as flood-driver
Interested in snowmelt as a flood-generating process, we calculate the total snowmelt within a 14-day moving window for all 50 m elevation bands from modelling results, i.e. snowmelt rates, and determine the annual maximum (max14) thereof and assess the two characteristics timing (T-max14) and magnitude (M-max14). In addition, we determine the mean elevation of defined max14 within a 30-day moving window (E-max14). To assess changes over time, we compare max14 characteristics between the two 30-year time frames 1954–1983 and 1984–2013. We also calculate the timing and magnitude of max14 on the catchment scale (T-max14-C and M-max14-C) for the total snow volume in the basin. We assess trend magnitude and significance of these variables using the Theil–Sen trend estimator and the Mann–Kendall trend test (Mann 1945; Kendall 1975).
To improve our understanding of the interplay between snowmelt-induced runoff and rainfall-induced runoff with regard to the annual maximum runoff events, we take a closer look at the 3-day sums of precipitation and 14-day sums of modelled snow accumulation/melt and visualise these variables along with observed discharge measured at gauge Basel. We differentiate between solid and liquid precipitation following the temperature threshold obtained during model calibration (see Supplementary Table B1). In addition, we add information on the timing T-max-14 for all elevation bands to identify elevations contributing most to the snowmelt. Measurements of snow depth from five selected stations are added to enable an additional visual comparison of snowmelt timing in different elevations. We visualise these variables for selected years. We select years with the aim of giving an insight into the different possible interactions between liquid/solid precipitation and the build-up and melt of the seasonal snow cover.
For characterising floods according to their genesis, we use a peak-over-threshold approach: we determine runoff peaks above the 95th percent quantile observed at gauge Basel for the time window 1951–2014 (i.e. time frame of snow simulations). To ensure the independence of detected peaks, two runoff peaks need to be at least 14 days apart. Similar to the approach presented in Vormoor et al. (2016), we distinguish rainfall- and snowmelt-driven events by estimating their snowmelt and rainfall contributions based on gridded data. We use the ratio of snowmelt of the preceding 14 days to snowmelt of the preceding 14 days plus liquid precipitation of the preceding 3 days and compare the two time windows 1951–1982 and 1983–2014 for all seasons and for events observed between February and June.
In addition to changes in snow cover, many other factors can cause changes in runoff in snow-dominated river basins (e.g. the modification of the river network, the construction and operation of reservoirs of high-head hydropower plants or changes in precipitation characteristics). The last step of our analysis aims at setting changes in snow cover in relation to other factors. Therefore, we analyse discharge observations for the simulation period 1951–2014 and an extended time frame of 1919–2016. At all gauging stations, continuous daily data are available at least since 1919. The extension of the study period enables the assessment of more robust signals of change less influenced by climate variability. With the aim of getting an overview of runoff seasonality and intra-/inter-annual runoff variability, we display discharge measured as raster graphs (Keim 2000; Koehler 2004). A horizontal line in the figure represents 1 year of measured discharge. Furthermore, we calculate quantile discharge values on a monthly level based on all daily values of a month for the time windows 1951–1982, 1983–2014, 1919–1967 and 1968–2016. These time windows have been chosen to locate two time windows with maximum length in the two time periods investigated, respectively. In the time window 1968–2016, quantiles for the month of January, for example, are based on 49 times 31 daily values. Probability levels investigated range from 0.01 to 0.99. Changes in quantile values between the two investigated time frames are determined by subtracting the value of the older time window from the recent one.
Seasonal snowpack characteristics are subject to strong inter-annual variability (Figure 3(a)). In recent decades (1988–2018), seasonal snowpacks are diminished at all stations compared to 1958–1988 (Figure 3(b)). Our results indicate that accumulation is reduced. The maximum snow depth is often already reached earlier in the year. For stations above 1,000 m, a pattern of increased melt rates at the beginning of the snowmelt period and a reduction in maximum melt rates show up (Figure 3(c)). Snowpacks have already thawed completely earlier in the year. Changes in measured snow depth only are an indication of changes in water stored in the snow cover. During the compaction of a snowpack, for example, snow depth can decrease, while water content stays the same.
Snow simulations: calibration, validation and application of the snowpack model
The snow simulations at the point scale indicate that the calibrated snow model captures the dynamic of high alpine snowpacks well (Figure 4). The average NRMSE of the four stations is 46%, the average PBIAS −5.8% and the NSE 0.78. Both timings of accumulation and melt are in good agreement with observed values. The attained NRMSE indicates that the RMSE is on average less than half a standard deviation of measured values. Furthermore, results indicate that highest SWE values observed at the peak of the snow season tend to be underestimated (e.g. Figure 4(c) and 4(d)). A factor possibly contributing to this uncertainty is a bias in the driving data caused by undercatch of snowfall (Savina et al. 2012) and, consequently, reduced precipitation input to the snow simulations.
At the catchment scale, SWE depth and snow cover duration increase with elevation (Figure 5(a)). The total SWE volume is distributed more uniformly along the elevation range, as the areal fraction an elevation band covers decreases with elevation (Figure 1 and 5(c)). Similar to results from snow observations, the simulations indicate that SWE depth and volume is reduced in recent decades, particularly before and during the melt season (Figure 5(b) and 5(d)). We detect a decline also in snow accumulation for elevation bands below approximately 2,000 m. Annual averages and trends in snowmelt/accumulation rates, i.e. snow volume changes between two consecutive days, are depicted in Figure 5(e) and 5(f). The time window where snow is accumulating widens with elevation (Figure 5(e)). At the highest elevation bands investigated, snow is accumulating from October until April. The snowmelt season lasts approximately from February at the lowest elevations to July at the highest elevation.
Negative values in Figure 5(f) indicate either decreases in accumulation or increases in snowmelt (initial negative values further decrease). Positive values in Figure 5(f), on the other hand, indicate either increases in accumulation or decreases in snowmelt. Simulation results hint at an increase in snowmelt at the beginning of the snowmelt period and at an earlier disappearance of the snow at the end of the melting season. For most of the year, decreases in snowmelt rates at the end of the snowmelt season (blue diagonal feature in Figure 5(f)) coincide with increases from higher elevations. In May, for example, snowmelt rates between approximately 1,500 and 2,000 m seem to decrease, while snowmelt rates above 2,000 m increase. Mean annual cycles of average temperatures point at strong differences in temperature depending on elevation and time of the years (Figure 5(g)). Trends in temperature are positive and subject to inter-annual variability (Figure 5(h)). With regard to precipitation, increases and decreases in precipitation show up (Figure 5(h)).
In general, patterns of average annual snow cover duration are similar for simulations and MODIS snow cover maps (Figure 6). The features of the terrain reflect in the duration of the seasonal snow cover. On the Swiss Plateau and on the valley bottoms, there are only few days with snow cover. On the other hand, the high alpine parts are characterised by a short snow-free period. However, differences show up, particularly for two types of areas: (1) in very high elevations, particularly in glaciated areas, satellite-based maps show more days with snow cover and (2) in the valleys of the Alpine Rhine in the South-East, the satellite-based maps show less days with snow cover as compared to the simulation results.
Snowmelt as flood-driver
In recent decades, max14 events occur earlier in the year and are weaker in magnitude (Figure 7(c)). Comparing E-max14 for the two 30-year time frames 1954–1983 and 1984–2013, our results suggest a shift upward the elevation range by about 170 m. On the basin scale, our analysis shows a wide temporal spread of max14 ranging from February to June (Figure 7(c)). No significant change in timing (T-max14-C) nor magnitude (M-max14-C) can be detected.
Figure 8 describes the interplay of precipitation and snowmelt with regard to the annual runoff maximum for four selected hydrological years. In the year 1969/1970, the highest discharge value is observed at the end of February after strong, mainly liquid, precipitation in the basin (Figure 8(a)). According to the model results, a part of the precipitation input is solid and stored in temporary snowpacks (23%). Our results indicate that the runoff event is exclusively rain-fed, and no snowmelt is contributing. In the year 1978/1979, the annual runoff maximum is recorded in June. Rainfall-induced runoff overlaps with high baseflow due to snowmelt from high elevations (Figure 8(b)). In the two years 1980/1981 and 1987/1988, the annual runoff maximum is recorded in March (Figure 8(c) and 8(d)). In March 1981, strong snowmelt from low elevations overlaps with a moderate rainfall event. In March 1988, heavy precipitation is the main driver of the runoff peak observed. A part of the rainfall is stored in high elevation snowpacks. The temporarily stored water is released in the following months. Snowmelt from low elevations (below 1,000 m a.s.l.), on the other hand, seems to contribute to the observed runoff peak.
In total, we detect 174 independent discharge peaks above the 95th percent quantile at gauge Basel for the time frame 1951–2014 (Figure 9). With regard to the events of all seasons, precipitation is the dominant flood-generating process. Between the months February and June, snowmelt-induced runoff is more important compared with the whole year. Our results indicate that for the time window February and June, 49 out of the 79 detected events for the time frame 1951–2014 potentially have snowmelt contributions of more than 25%. The total number of events and the number of precipitation-dominated events increase in the more recent time frame (1983–2014) with regard to both ‘All seasons’ and ‘February to June’. This overall increase goes along with a decrease in snowmelt contribution to detected events. The number of discharge peaks with potential snowmelt contributions of >50% decreases from 12 to 7. Absolute contributions of snowmelt/rainfall tend to decrease/increase (Figure 9(b)).
Seasonal snowpacks are an important factor redistributing runoff from winter to summer at all gauges investigated (Figure 10). Runoff seasonality is most pronounced in the alpine part of the catchment (Figure 10(f) and 10(g)). A strong weekly pattern of reservoir operations for hydropower production with higher runoff during weekdays than on the weekend is imprinted at gauge Diepoldsau since the 1960s (‘dashed pattern’ in Figure 10(g)).
In the Rhine branch including Lake Constance, our results hint at a decrease in runoff during summer (Figure 11(a5)–11(a7) and 11(b5)–11(b7)). At the same time, runoff increases during winter and spring. In the Aare branch including gauges Bern, Brugg and Untersiggenthal, decreases in runoff during the summer seem to be less pronounced (Figure 11(a3) and 11(b1)–11(b3)). The most prominent signal in the Aare branch arises at gauge Brugg: high runoff values increase from the beginning of spring until the beginning of summer (Figure 11(a3) and 11(b3)). At gauge Basel, signals from the two branches overlap, and a general redistribution of runoff from summer to winter superimposes with increases in high runoff values from spring until the beginning of the summer (Figure 11(a1) and 11(b1)). With increasing length of the time windows, patterns showing up get more stable, and differences between neighbouring quantiles and months are less abrupt.
Snowmelt event characteristics
Rising temperatures reduce seasonal snowpacks (Figures 3 and 5; Laternser & Schneebeli 2003; Scherrer et al. 2004; Marty 2008; Marty & Blanchet 2012; Klein et al. 2016). The effects seem most pronounced at elevations below 2,000 m. In those elevations, our simulation results hint at strong decreases during both snow accumulation and snowmelt (Figure 5(b) and 5(d)). In addition, we detect indications of potential elevation-dependent compensation effects for parts of the year (approximately March–June) (Figure 5(f)). Increases/decreases in meltwater outflow seem to be partly compensated by changes in meltwater from the snow cover from elevations below/above. The reduction in seasonal snowpacks goes along with decreases in maximum melt rates (Figure 7(c)). Detected changes go along with results from Musselman et al. (2017), who indicate that a ‘shallower snowpack melts earlier, and at lower rates, than deeper, later-lying snow-cover’ and ‘that the fraction of meltwater volume produced at high snowmelt rates is greatly reduced in a warmer climate’.
According to our analysis, rising temperatures do not just decrease the maximum melt rates, we identify a threefold effect: snowmelt becomes weaker, occurs earlier and originates from higher elevations (Figure 7(c)). If we refer to a fixed location (e.g. at an observational site), we can detect the shift forward in time. Conversely, if looking at a fixed time of year, the location of the melt event (i.e. contributing elevation bands) moves upward the elevation range (Figure 12(a)).
Snowmelt as flood-driver
In the Rhine Basin upstream gauge Basel, both snowmelt and precipitation seem to be important flood-drivers. Even moderate precipitation events can cause the annual runoff maximum when superimposing with snowmelt-induced runoff (e.g. Figure 8(c)). Investigations of snow observations and snow simulations demonstrate that snowmelt-induced runoff originates from a wide range of elevations. However, due to this wide elevation range, snowmelt does not occur simultaneously at all elevations. Our results indicate that elevation bands commonly melt together in blocks (Figure 8(a2)–8(d2)). The beginning and end of a snowmelt event occurring in the basin seem to be determined by the passage of warm air masses, the respective elevation range affected by accompanying temperatures and snow availability. We hypothesise that in a warmer climate with similar sequences of weather conditions, snowmelt is moved up to higher elevation, i.e. the block of elevation bands contributing most to the snowmelt-induced runoff is located at higher elevations (Figure 12(b)). The movement upward the elevation range makes snowmelt in individual elevation bands (e.g. T-max14 within an elevation band) occur earlier, and the timing of the snowmelt-induced runoff, however, stays the same. Snowmelt from higher elevation might, at least partly, replace meltwater from lower elevations. Investigating five peak discharge events in the mountain river basin of the Merced River in California, USA, Biggs & Whitaker (2012) were able to show that for each discharge peak, between 60 and 80% of the snowmelt originated from elevations that only covered between 22 and 38% of the total basin relief. They refer to the elevations contributing most to a runoff event as the ‘critical zone’.
Role of precipitation
The analysis of the annual runoff maximum of the hydrological year 1969–1970 indicates, in an exemplary way, how precipitation alone can cause high runoff values (Figure 8(a)). According to our analysis, no snowmelt is contributing to the runoff peak. On the contrary, part of the precipitation is solid and stored in temporary snowpacks. The accumulation of snow reduces the effective precipitation, i.e. precipitation translating into river runoff. In the following, we will refer to this effect of partial storage of precipitation in snowpacks as the protective effect of snow accumulation. Strong precipitation events do not have an immediate effect when their water is stored in snowpacks. However, with rising temperatures, the snowline moves upwards and larger fractions of the basin potentially receive liquid precipitation (Barnett et al. 2005; Scherrer & Appenzeller 2006; Allamano et al. 2009a, 2009b; Kormann et al. 2015; Beniston et al. 2018). Hence, in a warmer climate, the protective effect of snow accumulation is reduced. Effective precipitation might increase. A simple calculation supports this notion: without the protective effect (all precipitation liquid), the cold season (October until March) annual maxima 3-day rainfall events would have been stronger by 11.4% on average in our simulations. Recent studies point at the increasing importance of rainfall-induced runoff with regard to flooding in catchments influenced by a seasonal snow cover (e.g. Middelkoop et al. 2001; Addor et al. 2014; Vormoor et al. 2016; Brunner et al. 2019; Rottler et al. 2020). Our analysis of runoff peaks observed at gauge Basel further supports these investigations (Figure 9). The number of runoff peaks with most precipitation as flood-driver tends to increase, while at the same time, events with a majority of snowmelt-induced runoff decrease.
With the analysis of observed discharge, we complete our analysis by assessing changes induced by other factors. The extension of the time frame back to 1919, beyond the simulation period, helps to reduce the impact of inter-annual to decadal climate variability, e.g. due to varying frequencies of weather types (Rottler et al. 2019). We attribute detected increases in high runoff values during spring and early summer (Figure 11) to variations in rainfall characteristics. The nested catchment approach indicates that increases in high runoff values during spring and early summer are most prominent at gauge Brugg, whose catchment covers large parts of the Swiss Plateau and whose discharge is most strongly influenced by rainfall-induced runoff compared to the other gauges investigated. Another factor having a strong impact on the runoff in Rhine River Basin upstream gauge Basel are the numerous large reservoirs that have been constructed for hydropower production during the course of the 20th century (e.g. Wildenhahn & Klaholz 1996; Bosshard et al. 2013). In the Alpine Rhine Basin until gauge Diepoldsau, for example, the total storage volume is about 10% of the total annual runoff (Bosshard et al. 2013). The redistribution of water from summer to winter and a weekly pattern with high/low discharge on weekdays/weekends are well-known effects of reservoir operations (Figure 10(g); Verbunt et al. 2005; Belz et al. 2007; Meile et al. 2011; Pérez Ciria et al. 2019; Rottler et al. 2020).
Model performance and limitations
We are confident that our modelling approach enables investigations of snow cover changes. Temporal dynamics and absolute values have been reproduced for the stations, patterns realistically generated on the catchment scale. Elevation-dependent differences in snowpack characteristics are represented well. However, our modelling approach includes several assumptions and simplifications that require caution and limit the explanatory value of our simulations. In the framework of our snow model, we do not address changes in incoming solar radiation. Instead, the simple mean annual cycle we use as input inter alia does not supply information on recent regional brightening effects (Norris & Wild 2007; Ruckstuhl et al. 2008; Ruckstuhl & Norris 2009; Duethmann & Blöschl 2018). Thus, our modelling setup is solely driven by variations and changes in temperature and precipitation. However, the energy-based concept enables us to address energy input during rain-on-snow events, which is a phenomenon poorly reflected by empirical approaches. Rain-on-snow is an important flood-generating process in upland basins (e.g. Sui & Koehler 2001; Freudiger et al. 2014; Rössler et al. 2014). The current modelling setup does not include the redistribution of snow by wind or avalanches (e.g. Kerr et al. 2013; Warscher et al. 2013; Musselman et al. 2015; Freudiger et al. 2017). Missing redistribution processes increase the uncertainty with regard to snow volume distribution along the elevation range. Furthermore, aspect and slope can be important parameters controlling snow accumulation and snowmelt.
With regard to the detected differences between simulated annual snow cover durations and MODIS snow cover maps, we identify several possible factors contributing to these differences: first, the current model setup does not resolve very high elevations. The resolution of 1 km in combination with the simple lapse rate-based approach to downscale temperature does not capture the highest elevation in the basin. Likewise, it introduces a smoothing effect to the very low-lying cells, resulting in an overestimation of snow cover there. In addition, we suspect that the precipitation input is not capturing the full altitudinal gradient. It seems that climatic conditions, particularly in the valley bottoms in the South-East, are not well captured in the meteorological input used to drive the snow simulations. Precipitation patterns seem too coarse to capture the strong differences over short distances in this complex terrain (Freudiger et al. 2016). In addition, the parameters fitted for the observations at the four snow gauges possibly do not capture the entire range of conditions in the basin. Particularly, the snow simulated in the comparatively dry valleys in the South-East and in the highest altitudes might need an adapted set of parameters to describe its properties more accurately.
We analyse snow and discharge observations and simulate the Alpine snow cover in order to get a better understanding of how changes in snowmelt timing translate into changes in river runoff. The focus of the study is the Rhine River Basin upstream gauge Basel. We are confident that the physically based snow model setup used represents the snow cover dynamics in the basin well. However, several assumptions and simplification advice caution, most importantly the lapse rate-based approaches, applied to downscale gridded temperature input, simple seasonal cycle used as radiation input and the neglect of snow redistribution processes.
Our results point at strong decreases in seasonal snowpacks in recent decades. With regard to snowmelt events, we detect a threefold effect of rising temperatures: snowmelt becomes weaker, occurs earlier and forms at higher elevations (Figure 7(c)). Investigations on the catchment scale indicate that snowmelt-induced runoff can originate from a wide range of elevations. Due to the wide range of elevations covered in the basin, snowmelt does not occur simultaneously at all elevations. Elevation bands rather melt together in blocks (Figure 8). The beginning and end of the release of meltwater seem to be determined by the passage of warm air masses, and the respective elevation range affected by accompanying temperatures and snow availability. We hypothesise that in a warmer world with similar sequences of weather conditions, snowmelt is moved upward to higher elevations, i.e. the block of elevation bands providing most water to the snowmelt-induced runoff is located at higher elevations (Figure 12(b)). The movement upward the elevation range makes snowmelt in individual elevation bands occur earlier, and the timing of the snowmelt-induced runoff, however, stays the same. Meltwater from higher elevations, at least partly, replaces meltwater from elevations below.
Furthermore, our analysis points at the protective effect of snow accumulation with regard to flooding. Precipitation in the form of snow in part of the catchment reduces the effective precipitation, i.e. precipitation directly translating into river runoff. With rising temperatures, the snowline moves upwards and an increased fraction of the precipitation is possibly liquid instead of solid (Scherrer & Appenzeller 2006; Allamano et al. 2009b; Vormoor et al. 2015). The analysis of discharge time series enabled the assessment of other factors modifying river runoff, i.e. more intense rainfall increasing high runoff quantiles in spring and early summer and the construction of reservoirs for hydropower production redistributing runoff from summer to winter (Meile et al. 2011; Bosshard et al. 2013; Pérez Ciria et al. 2019; Rottler et al. 2020).
This study represents a further step towards understanding how changes in alpine snowpacks translate into changes in river runoff. Future studies need to further investigate the proposed hypotheses describing elevation-dependent compensation effects. Investigations using meso-scale hydrological modelling frameworks in combination with satellite data-derived snow cover maps seem predestined for such a task.
We thank the Federal Office of Meteorology and Climatology of Switzerland (MeteoSwiss), the Institute for Snow and Avalanche Research (SLF) from the Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), the Institute for Earth Observation, Eurac Research and the Global Runoff Data Centre, 56068 Koblenz, Germany (GRDC) for providing meteorological, snow cover and hydrological data, respectively. We acknowledge the datasets generated in the EDgE proof-of-concept project performed under a contract for the Copernicus Climate Change Service (http://edge.climate.copernicus.eu, last accessed: 27 October 2020). ECMWF implements this service and the Copernicus Atmosphere Monitoring Service on behalf of the European Commission.
This research was supported by the Deutsche Forschungsgemeinschaft (GRK 2043/1-P2) within the NatRiskChange research training group at the University of Potsdam.
DATA AVAILABILITY STATEMENT
Data cannot be made publicly available. Readers should contact the corresponding author for details.