Operational and experimental snow observation systems in the upper Rofental : data from 2017-2020

According to the living data process in ESSD, this publication presents extensions of a comprehensive hydrometeorological and glaciological data set for several research sites in the Rofental (1891–3772 m a.s.l., Ötztal Alps, Austria). Whereas the original dataset has been published in a first original version in 2018 (https://doi.org/10.5194/essd-10-151-2018), the new time series presented here originate from meteorological and snow-hydrological recordings that have been collected from 2017 to 2020. Some data sets represent continuations of time series at existing locations, others come from new installa5 tions complementing the scientific monitoring infrastructure in the research catchment. Main extensions are a fully equipped automatic weather and snow monitoring station, as well as extensive additional installations to enable continuous observation of snow cover properties. Installed at three high Alpine locations in the catchment, these include automatic measurements of snow depth, snow water equivalent, volumetric solid and liquid water content, snow density, layered snow temperature profiles, and snow surface temperature. One station is extended by a particular arrangement of two snow depth and water equivalent 10 recording devices to observe and quantify wind-driven snow redistribution. They are installed at nearby wind-exposed and sheltered locations and are complemented by an acoustic-based snow drift sensor. The data sets represent a unique time series of high-altitude mountain snow and meteorology observations. We present three years of data for temperature, precipitation, humidity, wind speed, and radiation fluxes from three meteorological stations. The continuous snow measurements are explored by combined analyses of meteorological and snow data to show typical 15 seasonal snow cover characteristics. The potential of the snow drift observations are demonstrated with examples of measured wind speeds, snow drift rates and redistributed snow amounts in December 2019 when a tragic avalanche accident occurred in the vicinity of the station. All new data sets are provided to the scientific community according to the Creative Commons Attribution License by means of the PANGAEA repository (https://www.pangaea.de/?q=%40ref104365).

The slope behind and below the instruments faces south, and from the location of the station one has a panoramic view of the 110 total area of the Hintereisferner. The new instruments are located approximately 20 m east of the totalizing rain gauge that has been in operation since 1952. The centroid of the instrumentations is located at 46.82951°N, 10.82407°E. For all stations, the height of the sensors above ground is at least 1.5 m; in winter, the distance between the snow surface and the sensors can become much smaller, and in extreme snow-rich periods the instruments even can become completely snow-covered. Such periods can be recognized in the data by typical recordings of zero wind speed and increasing dampening of the other meteoro-115 logical variables. All three stations Bella Vista, Latschbloder, and Proviantdepot have been equipped with extensive automatic snow cover measurement systems. These systems are presented in Sect. 3.2. The data at all three stations is recorded in 10 min. intervals and transmitted by means of GSM. In Tabs. 1, 2, and 3 the technical sensor specifications are listed in detail.

Automatic snow cover measurements
An important aim in the conceptual development of the Rofental measurement network was the extensive and operational 120 observation of the snow cover and its properties, for which characteristic locations in the high Alpine terrain of the catchment were chosen. Therefore, the three AWS Bella Vista, Latschbloder, and Proviantdepot include extensive automatic measurements of various snow cover properties which are recorded continuously in a 10 min. interval. These comprise observations of snow depth (SD), snow water equivalent (SWE), layered snow temperature profiles, snow surface temperature, liquid and solid water content of the snowpack, as well as snow drift (see Tab. 1, 2, and 3). In the following, these are named automatic weather 125 and snow stations (AWSS). The data of the AWSSs are operationally used by the European Avalanche Warning Services EAWS and visualized in real-time at https://avalanche.report/weather/measurements and https://www.lawis.at/station/.

Bella Vista
The Bella Vista AWSS (2805 m a.s.l., 46.78284°N, 10.79138°E) is located in close vicinity to the "Schöne Aussicht Schutzhütte". It is located exactly at the central ridge of the Central Eastern Alps, the weather divide between the North-130 ern and Southern Eastern Alps. Generally, the region belongs to the rather dry inner Alpine climate zone. The station is built in small scale heterogeneous terrain at a gentle slope in a barren, rocky landscape. It is affected by wind and frontal systems from both, northern and southern directions. The AWSS has been extended by an additional snow measurement system in October 2019 and September 2020. A Sommer SSG-2 snow scale to measure SWE, an ultrasonic SD sensor, and a snow temperature profiler have been installed in a depressed location near the main station that is prone to snow accumulation by wind-drift 135 (Fig. 2). The new instrument complements the existing measurements of SWE (by means of a snow pillow), SD (by means of an ultrasonic ranger) and snow temperature profile (by means of a series of temperature sensors at different height levels) located at the main station. The location of the main station is rather exposed and therefore prone to snow erosion by wind.
The relation between the exposed and sheltered snow measurements allows for an assessment of the timing and amount of wind-driven snow redistribution. This technique is illustrated by the data analysis of a blowing snow event that was related to   (top, Sep 30, 2020). The red arrow marks the main AWS and "exposed" snow measurements. The blue arrow marks the additional snow measurements (SD, SWE, and snow temperatures) in the slight depression ("sheltered" location). Close ups of the main AWSS (bottom left) and the secondary snow measurements (bottom right). the tragic avalanche accident in December 2019 (Sect. 4.2.4). In addition, a new instrument to directly measure the particle flux of drifting snow by means of an acoustic sensor (Sommer SND -Snow Drift Sensor) has been installed in September 2020.

Latschbloder
The Latschbloder AWSS (2919 m a.s.l., 46.80106°N, 10.80659°E) is located on a gently sloped plateau below the "Rofenbergköpfe" (3229 m a.s.l.) and was chosen as a meteorologically representative measurement for the regional climate that is not 145 largely influenced by steep surrounding slopes and the corresponding local wind systems (Fig. 3). It is located near a totalizing rain gauge that was installed in 1965 and has been equipped with an ultrasonic snow depth sensor (Sommer USH-8) in 2017.
In September 2020 an automatic snow temperature profiler that records temperature in the snow cover at the base level and in 20, 40, 60, and 100 cm from the ground has been installed.   Analyzer (SPA). The snow scale is buried right in front of the photographer besides the SPA. There is a second snow depth sensor at the main mast (not visible from this angle). Behind the main mast the old totalizing rain gauge can be seen, in the background the Kesselwandferner.

150
The Proviantdepot AWSS (2.737 m a.ssl., 46.82951°N, 10.82407°E, 4) is located on a flat section of a south-facing slope underneath the "Guslarspitzen" (3126 m a.s.l.) halfway between the summit and the "Hochjoch-Hospiz" (2413 m a.s.l.). The station comprises a large set of operational snow cover sensors. SD is measured by a USH-9 ultrasonic device, SWE by means of a SSG-2 snow scale. The temperature of the snow surface is continuously measured by an infrared sensor (Sommer SIR).
The layered snow temperatures are recorded analog to the other stations by a SCA temperature profiler. A SPA-2 Snow Pack 155 Analyzer records volumetric contents of solid and liquid water of the snow cover based on measuring the dielectric constants of different frequencies along a flat strap sensor that is spanned within the snowpack. (all three) -suffer from precipitation gauge undercatch that is typical for high mountain observations with high wind speeds and a large amount of solid precipitation. The two totalizing rain gauges that are located close to the stations recorded long-term annual precipitation totals of 1012 mm (Latschbloder, 1965(Latschbloder, -2016, and 941 mm (Proviantdepot, 1952(Proviantdepot, -2016, respectively.

Snow cover data
In the following section, we analyze the first data sets obtained by the various snow observation sensors. Snow depth, snow    period October 2017 to December 2020. Generally winter SD varies between 0.5 and 2 m depending on station and year. At the Bella Vista exposed site usually less SD is measured than at the Latschbloder station. The Bella Vista sheltered site data is available as of autumn 2020 and already shows large SD values in November and December compared to the exposed site.

195
Maximum SD of close to 2 m were measured at Latschbloder in May 2019. Maximum SD at the Bella Vista exposed site was significantly lower at that time (1 m). The winter 2019/2020 shows large differences in maximum SD between Latschbloder (50 cm) and Proviantdepot (160 cm), indicating that the Proviantdepot received a lot of snowfall from northern frontal systems that did not reach the Latschbloder location further in the south.

200
A snow temperature profile recorded by the Sommer SCA temperature profiler for the whole snow covered period in winter 2019/2020 at the station Proviantdepot is shown in Fig. 8 b). Heights above the ground are 0, 20, 40, 60, 80 and 100 cm, respectively. The corresponding snow depth (Fig. 8 a) shows a typical course over the season. The first large snowfall events in the beginning of November result in a snow depth of 40 cm that grows to 140 cm with more snowfall until the end of the month. In the following, the snow cover settles slowly and periodically increases again with single snow precipitation events. 205 Snow depth reaches its seasonal peak of 160 cm at the end of December. A long period with only small sporadic snowfall amounts and constant snow settling follows during January and February where SD varies between 110 and 130 cm. Another large snowfall at the end of February leads to a SD of 150 cm. After some settling and little new snow periods, the melting period starts at the beginning of April. Snow melts constantly -only interrupted by some late snowfall at the beginning of May -until SD is 0 cm at the end of May. The described behaviour and distinguishing melt and settling periods can well 210 be reconstructed using the SWE measurements in Fig. 8 a). SWE steadily increases during snowfall events and stagnates in midwinter when snow is settling but not melting. SWE decreases with the start of the melting period in April. The difference of the time points of SWE and SD reaching zero can be explained by the ultrasonic SD sensor not being directly over the snow scale at the Proviantdepot station (Fig. 4). This is due to constraints in the space around the main mast where the ultrasonic device is installed. By means of webcam images (not shown here) we identified that there was a wind slab snow patch covering 215 half of the scale while the surrounding was already free of snow. Snow temperature at the base (0 cm) is at 0°C throughout the whole snow covered period (Fig. 8 b). The elevated temperature sensors at 20, 40, 60, 80, and 100 cm obviously show strong diurnal variations when they are not covered by snow, i.e. measure air temperature at the beginning and end of the snow https://doi.org/10.5194/essd-2021-68  season. As soon as the sensors are covered by snow, the temperature signal is dampened. A clear snow temperature stratigraphy with warmer (negative) temperatures in deeper levels (closer to the ground) develops and is retained throughout the season.

220
The layer closest to the surface (100 cm) is influenced the most by prevailing air temperatures and cools down to -8.1°C on January 21. This cold minimum in the layer approximately 20 cm below the snow surface follows air temperature with a time lag of 1 day. Very cold air temperatures of -17°C were measured in the night preceding January 20. A minimum snow surface temperature of -36.6°C was recorded after that cold night at 10:40pm (Fig. 8 c). This time lag is carried on into the deeper snow layers. This dampening effect can be observed in the data in both directions, i.e. when the snowpack is cooling or 225 warming. The data show a very sharp point in time when the snowpack becomes and stays isotherm, i.e. all layers are at 0°C (April 9, 2020). Snow starts to melt two days later, i.e. SWE starts to decrease. After the first snowmelt, SWE increases again, but SD steadily decreases, which is a clear indicator for rain falling and percolating into the snow cover. After the small last snowfall at the beginning of May snow melts away steadily.

Liquid and solid water content (Snow Pack Analyzer)
230 Fig. 9 shows measurements of the Snow Pack Analyzer (SPA) at the Proviantdepot station. Due to a logger failure, SPA data is not available before October 2020. Here we show the data from November 25, 2020 to December 31, 2020, i.e. the start of the snow covered period of the winter season 2020/2021. It is characterized by two days of heavy snowfall on December 4 and 5 where SD increases from 10 to 120 cm (Fig. 9 d). Thereafter that the snowpack settles slowly with a next significant snowfall at the end of December. Fig. 9 a) shows measured solid and liquid volumetric water content measured by the SPA, 235 and the corresponding snow density measurement in Fig. 9 b). It is clearly shown that ice content slowly increases with settling of the new snow, and therefore, snow density increases from 100 kg/m 3 to 260 kg/m 3 in late December. Liquid water content https://doi.org/10.5194/essd-2021-68  Webcam images before (left, Dec 3, 12:00pm) and shortly after the blowing snow event (right, Dec 6, 2020, 9:00am). The main AWSS (exposed location) can be identified in the left side of the pictures, the complementing, second SWE and SD measurements are located at the mast on the right side in the pictures (sheltered location). potential of this semi-quantitative approach for avalanche warning applications. The accuracy of the sensor in quantifying snow transport rates and its relation to redistributed snow on the ground and at the SD and SWE measurement points will further be investigated when a longer time series is available. Additional field campaigns will be carried out using a mobile terrestrial laser scanning device to measure the spatial distribution of snow depth in the small-scale heterogeneous terrain around the AWSS.

295
In the following, we present an additional example of data analysis in the winter season 2019/2020 where an avalanche accident happened in the region. On December 28, 2019 a large avalanche buried several skiers in the ski resort at the Val Senales Glacier, South Tyrol. During the tragic accident three skiers were fatally injured. On the day of the accident, the avalanche report assigned a "considerable" level of danger for the area. The main source of danger was identified to be windblown snow. Strong northerly winds redistributed fresh snow from the previous day to form wind slabs that poorly bonded to 300 the old snowpack. The scene of the accident is located in close proximity to the station Bella Vista (Fig. 13). The instrument setup described above allowed for an assessment of meteorological conditions and wind-driven snow redistribution preceding the avalanche.
Fig. 14 shows wind speed and snow conditions preceding the avalanche. The winter season began with a series of snowfall events in November. Snow depth increased from 0 to 40 cm from November 3 to November 10 at the exposed location of 305 the Bella Vista station (Fig. 14 a). Snow depth and SWE at the exposed location shows only little variation until December 22 with only small amounts of snowfall in between. However, SWE at the sheltered site shows strong increases during that period resulting in 260 mm SWE compared to less than 100 mm SWE at the exposed location at the beginning of December.
This offset between the nearby measurements grows even larger in the days preceding the time of the avalanche at the end of December (450 mm SWE to 100/150 mm SWE) and corresponds to 400% larger snow masses at the sheltered site compared 310 to the exposed location. The horizontal distance between the two devices is only about 30 m. Fig. 14 c) shows daily mean wind speed and gusts during the corresponding period. It is well discernable that the increase in SWE at the sheltered site occurs when high mean wind speeds including strong gusts prevail. When looking closely at single events, e.g. the time before the avalanche (Fig. 14 b and d), transport of snow from wind-exposed to -sheltered locations can be observed. On December 27 in the afternoon, mean wind speed began to strongly increase to 8 m/s with gusts reaching 18 m/s and decreased again at 00:00.