European In-Situ Snow Measurements: Practices and Purposes

Abstract

In-situ snow measurements conducted by European institutions for operational, research, and energy business applications were surveyed in the framework of the European Cooperation in Science and Technology (COST) Action ES1404, called “A European network for a harmonised monitoring of snow for the benefit of climate change scenarios, hydrology, and numerical weather prediction”. Here we present the results of this survey, which was answered by 125 participants from 99 operational and research institutions, belonging to 38 European countries. The typologies of environments where the snow measurements are performed range from mountain to low elevated plains, including forests, bogs, tundra, urban areas, glaciers, lake ice, and sea ice. Of the respondents, 93% measure snow macrophysical parameters, such as snow presence, snow depth (HS), snow water equivalent (SWE), and snow density. These describe the bulk characteristics of the whole snowpack or of a snow layer, and they are the primary snow properties that are needed for most operational applications (such as hydrological monitoring, avalanche forecast, and weather forecast). In most cases, these measurements are done with manual methods, although for snow presence, HS, and SWE, automatized methods are also applied by some respondents. Parameters characterizing precipitating and suspended snow (such as the height of new snow, precipitation intensity, flux of drifting/blowing snow, and particle size distribution), some of which are crucial for the operational services, are measured by 74% of the respondents. Parameters characterizing the snow microstructural properties (such as the snow grain size and shape, and specific surface area), the snow electromagnetic properties (such as albedo, brightness temperature, and backscatter), and the snow composition (such as impurities and isotopes) are measured by 41%, 26%, and 13% of the respondents, respectively, mostly for research applications. The results of this survey are discussed from the perspective of the need of enhancing the efficiency and coverage of the in-situ observational network applying automatic and cheap measurement methods. Moreover, recommendations for the enhancement and harmonization of the observational network and measurement practices are provided.

1. Introduction

Snow is considered the largest gap in the current knowledge of the global water budget [1]. It is a critical component of the Earth’s ecosystem, it is very sensitive to climate change [2], and causes strong climate feedbacks. In-situ snow observations are sparse; vast snow covered areas are hardly accessible, and therefore snow monitoring mostly relies on satellite snow observations. However, these have several limitations, which depend on the parameters that we plan to investigate. For instance, snow cover maps mainly rely on optical sensors, which are affected by the presence of clouds; active microwave sensors have demonstrated their capability in mapping dry/wet status of snow, but do not have a high temporal coverage; snow water equivalent (SWE) can be derived by microwave passive sensors, which have a coarse resolution (i.e., tens of kilometres), have a limited operability on mountain regions, and have a high uncertainty for high SWE values (i.e., higher than 120 mm). In the ongoing climate change, the snow cover extent is decreasing [3], modifying the Northern Hemisphere atmospheric circulation [4]. Thus, snow measurements are and will be more and more crucial for freshwater management, mitigation of climate changes, adaptation to new climate conditions, and risk assessments (such as avalanches and floods). Lastly, the evolution of the snow cover is a major concern in some specific mountainous locations (e.g., alpine regions), where the economic activity depends on it (e.g., [5,6]). Snow can indeed be a critical resource for tourism (e.g., ski resorts), water supply, and hydropower energy.

The advances in the modelling of the snow-electromagnetic interaction (e.g., [7,8,9,10]) and in the observational capabilities of the satellite-based sensors (e.g., the sensors on board ESA Sentinel missions) have pushed the development of new in-situ instrumentation, which are able to provide suitable reference and ground-truth data for the validation of snow satellite products and of earth system models. In particular, instruments that can measure snow microstructure properties in the field have appeared, such as the IceCube [11], ASSSAP [12], or the SnowMicroPen [13], although they are available only for a few expert users. The testing of physically-based snow models and retrieval algorithms requires, on one hand, that the basic snow properties (such as snow depth, SWE, snow density) are measured with better precision and temporal resolution and, on the other hand, that new snow properties (such as, the surface roughness and snow correlation length) become observable. Having this objective, different institutions and research teams have developed different snow measurement practices and instrumentations that are customized to their purposes. While these efforts constitute significant steps forward in the understanding of snow physical processes, the lack of coordination in the measurements procedure and measured parameters limits the capability to share knowledge and data in the community. It is worth noting that, besides the new mentioned measurements capabilities, the instruments and methodologies for measuring basic snow parameters from ground (e.g., snow density, snow depth, SWE) also strongly differ among different institutions.

Indeed, while the International Classification for Seasonal Snow on the Ground [14] provides the definition of most of the observable snow properties, the classification of snow grain shapes, and a guideline to perform traditional snowpit measurements, it does not include a systematic description of the instruments that are used for snow monitoring. A recent review of the in-situ measurement techniques and devices that are used to determine the physical properties of the seasonal snowpack was done by Kinar and Pomeroy [15]. Their historical perspective illustrates the development and state-of-the-art of the field, with a focus on the practices that are applied in North America and Canada, which, however, do not entirely correspond to the current practices that are in use by European institutions. Moreover, the review by Kinar and Pomeroy [15] does not address the effective use of the devices.

In order to fill this gap and harmonize the snow monitoring procedure, the European Cooperation in Science and Technology (COST) promoted the Action ES1404 called “Harmosnow” (“A European network for a harmonised monitoring of snow for the benefit of climate change scenarios, hydrology, and numerical weather prediction”). The aim of the action is to coordinate the effort of harmonizing the snow monitoring practices, by promoting new observing strategies, bringing together different communities, facilitating data transfer, upgrading and enlarging knowledge through networking, exchange and training, and linking them to activities in international agencies and global networks.

One of the first activities of Harmosnow was to carry out a survey to obtain an updated picture of the existing variety of snow measurement practices and instrumentations in use by the European institutions. The survey collected a compilation of the measured snow properties and associated measurement techniques that are applied by the participating European countries for a large variety of applications. This paper aims to summarize the results of this survey, providing an overview on the European snow measurements that are carried out for a large variety of applications. Synthesis from the overview enables us to draw recommendations on the best measurement practices and on strategies to increase the effectiveness and the extension of the snow monitoring network.

2. European Survey on Snow Measurement Practices and Applied Instrumentation

A survey on the measured snow properties and applied instrumentation was carried out among the European institutions through a questionnaire that was designed with the contribution of several partners of Harmosnow with experience on snow observations. The questionnaire only addressed general information, as each respondent was asked to fill it in just once.

To get unambiguous answers, despite the absence of standard terminology for many snow properties, and even more importantly for devices or techniques, most of the questions were formulated as multiple choices. This required an investigation on the existing measurable snow properties and measurement instruments. Snow properties were grouped into five categories: snow macrophysical properties, snow microphysical properties, snow electromagnetic properties, precipitating and suspended snow, and snow composition. The measurable parameters that were associated to the snow macrophysical properties, microphysical properties, electromagnetic properties, solid precipitation and snow composition, as well as the existing instruments to measure them, are listed and defined in Appendix A in Table A1, Table A2, Table A3, Table A4 and Table A5, respectively. Table A6 in Appendix B is complementary to the tables in Appendix A, as it includes a description of the instruments that are applied to measure or derive the snow parameters.

Snow macrophysical properties describe the bulk characteristics of the whole snowpack or of a snow layer. Some of them, such as snow presence, snow depth (HS), and SWE, are the most important input information for hydrological and numerical weather prediction models, and therefore, they have the most established and wide spread observational network. Others (such as snow hardness, penetrability of snow surface, and snow strength) are mostly measured for snow avalanche forecast, thus in mountain regions only.

Snow microphysical properties describe the characteristics of the snow microstructure, such as snow grain shape, snow specific surface area, snow correlation length, etc. (a comprehensive list of these properties is given in Table A2). The estimation of these properties is needed, for instance, to simulate the interaction of the snow with electromagnetic radiation, with the purpose of developing algorithms to retrieve important variables, such as SWE or near-surface snow microstructure properties from the satellite observations. The snow microphysical properties are also utilized in the snow physical models that support avalanche forecasting [16,17].

Snow electromagnetic properties characterize the interaction of snow with the electromagnetic radiation. In the visible spectral region, the snow albedo controls the amount of solar energy that is reflected back to the atmosphere and is absorbed by the snowpack, being thus a key parameter for the snow energy and mass budget, and in particular, for meltwater runoff estimations and forecasts. In the near-infrared wavelengths, the albedo of snow is lower and strongly depends on the microstructural characteristics of the near-surface layer. Thus, the snow infrared albedo is applied to retrieve the snow microstructural properties (e.g., [18]). In the microwave spectral region, the signal that is captured by radiometers and radars over snow-covered surfaces (i.e., the electromagnetic radiation emitted or reflected by it) depends on SWE, snow microstructure, melting/freezing state, and background (e.g., soil, rocks, shrubs, etc.). By using appropriate retrieval algorithms, these variables can be hence derived remotely (e.g., [19,20,21,22]). Note that the ground-based instruments that are used to measure snow electromagnetic properties are similar (as frequencies, polarizations, etc.) to those that are installed on board satellites, and that these can be used to calibrate and validate satellite-derived snow data and extend the analysis over a broader area.

Solid precipitation has a variety of forms, including snowflakes, snow crystals, and graupel (see definitions in the Global Cryosphere Watch site: http://globalcryospherewatch.org/about/about_solidprecip.html). The amount of precipitation is an essential input for all of the models that simulate the snow cover and produce forecasts of weather, avalanches, melt water amount, and floods. Drifting and blowing snow is horizontally transported by the wind through saltation (i.e., in periodic contact with and directly above the snow surface) [23]. The snow crystals that are transported by the wind have a different shape and smaller size than the solid precipitation [24], as during the saltation process, they break and sublimate.

In our survey, the physical-chemical composition of snow mainly concerns light absorbing impurities as well as water and heavy metal isotopes. Snow impurities affect the snow albedo, and therefore the absorption of solar energy by the snowpack. The isotopic composition of snow is related to the atmospheric conditions occurring during the snow formation in the atmosphere, as well as to the degree of snow metamorphism.

In the questionnaire, for each category of snow property, the respondents were first asked to specify if any of the snow properties that were listed in that category were measured or not. Only if the response was “yes”, the questions concerning that category were asked. Each snow property was associated with one of the instruments that were available to measure it, thus, respondents needed to select all of the combinations of property/instrument that they apply in the field. This strategy facilitated the respondents, as no free text was required (except if the applied property/instrument combination was missing from the list) and it also ensured a consistent terminology, which was necessary for the interpretation and analysis of the results. The questionnaire also included a question to find out whether a written protocol (published or unpublished, written in any language) is used when doing the snow measurements, as well as questions to identify the landscape where the measurements are carried out and the purpose of the measurements. The questionnaire is reported in Appendix C. It was conducted between December 2015 and November 2017 and it was distributed among the COST member countries, through their national representatives within Harmosnow, who were invited to send it also to their national contacts (colleagues doing snow measurements, and snow specialists in the private sector). As there was no significant coordination of national snow measurements in any of the countries that were participating in the survey, the distribution of the questionnaire mostly relied on personal effort and on the contact network of the individual Harmosnow representatives. No instructions were given concerning the profile or professional level of the respondents, and it was not defined whether the answers should represent the measurement practices that are applied by a single person, a group, or the whole affiliated institution. Therefore, the collected answers are largely heterogeneous in terms of institutional representativeness. We did not apply filters or weight to the answers to balance these differences of representativeness, but we discussed the answers case by case when needed, and we accounted for the potential limitations in the representativeness in our conclusions. All of the received answers were analysed manually (i.e., the presented statistical calculations were not automatically generated by the software that was used to create the questionnaire).

3. Results

This section is divided by subheadings. It provides a concise and precise description of the results of the questionnaire and their interpretation.

3.1. Participating Countries and Institutions

The questionnaire was answered by 125 participants from 99 operational bodies, research institutions, and energy providers, from 38 European countries (Table A7 in Appendix D). Generally, most of the answers were given by those countries that provide weather-related services through several institutions, that are possibly distributed in different areas of the national territory, or that promote snow-related research in various universities, research institutes, or research groups. For most of the participating countries, at least one answer was given by the institution that provides the national weather service. Exceptions were Andorra, Greece, Luxembourg, and The Netherlands, from which the only received answers were from the Snow and Mountain Research Center (Andorra), the University of Thessaly (Greece), the Agricultural Service (Luxembourg), and the Utrecht University (The Netherland). As illustrated in Figure 1, the largest number of answers was from Finland and Italy (twelve answers each, together corresponding to about 19% of the total). In the case of Finland, the respondents were mostly from the Finnish Meteorological Institute, which includes the national operational services for weather, aviation, marine, and terrestrial traffic, as well as a research division with a lot of focus on polar areas. In the case of Italy, most of the answers (9/12) were from operational services, eight of them were regional and one was nation-wide (the Technical Centre for Meteorology of the Italian Air Force). A relatively large number of answers were given also from France, Spain, the United Kingdom, Slovenia, and Poland (nine, eight, seven, seven, and six answers, respectively). In those countries, most of the answers came from research institutes or universities, and except for Slovenia, one answer was provided by the respective national meteorological offices. Other answers were given by water management agencies (from Spain and the U.K.), environment protection agencies (from Spain, Slovenia, and the U.K.), avalanche/mountain rescue services (from Slovenia and the U.K.), and electrical companies (from France). Switzerland, Austria, Turkey, Bulgaria, Czech Republic, Hungary, Germany, Iceland, Norway, Russia, Slovakia, and Sweden provided three to five answers, generally distributed between operational services (related to weather, hydrology, and avalanches) and research institutions, with the addition of an energy company from Iceland. The twenty-two remaining countries provided only one or two answers, generally from the national weather service. Only in the case of the Netherlands and Greece, the single answer was given by a university. While not all of the European research and operational institutions that are involved in snow measurements responded to the survey, we believe, based on the numbers that were obtained and the qualification of the respondents, that it well represents the status of the methodologies that are adopted for snow monitoring at a European level.

Figure 1. Map of Europe showing the countries that participated in the survey, and the number of answers per country (through the color categories).

3.2. Purpose of the Measurements and Measurement Environment

Figure 2 illustrates the fraction of responses declaring research only, operational only, or both research and operational purposes of the snow measurements. The research purpose was dominating, with the largest percentage of responses (43%) attributing the exclusive purpose of the snow measurements to research. However, responses from operational oriented institutions (71 responses) were more numerous than the responses from the research only institutions (54 responses), but 55% of them had also research objectives. On the contrary, research institutions generally did not have any operational purposes.

Figure 2. Purpose of the snow measurements, and number of respondents selecting them. The width of each slice of the pie represents the fraction of responses declaring ‘Research’, ‘Operational’, or both.

The application areas of snow measurements are illustrated in Figure 3. The respondents could select more than one of the listed application areas, and eventually add extra applications (in the option ‘other’), if relevant. Hence, the total number of selected application areas is the sum of the numbers that are reported in Figure 3. Of the respondents, 86 (69% of the total number of selected application areas) listed ‘climatology’ among their application areas, 84 respondents (67% of the total) listed ‘meteorology’, and 91 respondents (65% of the total) listed ‘hydrology’, although these may not be the main scope of the operational oriented institutions (in the questionnaire, it was not asked to provide a hierarchic order of the relevance of the application areas). Specific applications, such as avalanche risk forecast, water management, or agriculture and forestry, are often handled in dedicated institutions, thus, the number of respondents selecting those areas were necessarily lower than the number of respondents selecting more general purposes. Among these specific application areas, the most frequent one was water management (48 respondents, corresponding to 38% of the total), followed by flooding forecast (41 respondents), avalanche risk forecast, agriculture and forestry, traffic, and health and sport. It is also worth mentioning that two participants stated ‘energy production’ in the option ‘other’, and other four stated ‘permafrost’, ‘environment protection’, ‘geodesy and gravity’, and ‘remote sensing of temperature and humidity’, respectively.

Figure 3. Application areas of the snow measurements, and the number of respondents selecting them. The width of each slice corresponds to the fraction each application area has, relative to the total number of the application areas reported (i.e., the sum of the numbers given in the figure).

Figure 4 illustrates the occurrence of measurement sites typologies. As in the case of the application areas, the respondents could select more than one of the listed site typologies and eventually add extra ones (in the option other). The multi-choice option was allowed because a single respondent/institution may perform measurements in different site types, and because a single site may belong to more than one typology (e.g., a forest covered mountain site could have been classified as ‘mountain’ and ‘forest’, an Alpine glacier as ‘mountain’ and ‘glacier’, etc.). The total number of the selected site typologies was the sum of the numbers that are reported in Figure 4. The figure clearly demonstrated the diversity of the snow conditions across Europe. In Southern and Western Europe, seasonal snow is present mostly over the mountains and only occasionally at sea level, while in Northern and in the more continental Eastern Europe, the occurrence of seasonal snow at the sea level is more frequent. Mountains are present in all of the European countries, except the Netherlands, Denmark (without Greenland), Estonia, Latvia, Lithuania, Belarus, Moldova, Malta, and Finland (if we exclude the area close to the Norwegian border) and are covered by seasonal snow, thus quite expectedly, they were the most frequent types of environment (selected by 85 respondents, corresponding to 68% of the total). On the other hand, the snow measurements sites are preferably chosen on flat open areas, so that the observations are not affected by nearby obstacles or slopes that would cause local redistribution or heterogeneity, compromising the representativeness of the site or the simple interpretation of the data. This explains why plains (open areas) had the second largest occurrence among the selected environment types. The landscape in the Northern and Eastern Europe is characterized by a large presence of forests, while in Southern and Western Europe, forests are concentrated into the mountainous areas, and the landscape is more urban and is dominated by cultivated fields. Glaciers are present in 13 of the 38 countries that were participating in the questionnaire. Frozen lakes and sea ice, over which snow could accumulate, cover significant areas only in Northern Europe (Scandinavia and northern Russia) during winter. Tundra is the characteristic landscape of the subpolar Europe (north of Scandinavia, Iceland, and North Russia), while bogs are present in a few European countries (Czech Republic, Estonia, Finland, Germany, Ireland, Norway, Sweden, Switzerland, and the United Kingdom). Some of the research institutions that were participating in the survey also have field measurement activities over ice sheets and ice shelves (four responses). These institutions apply their snow measurement practices also in the polar regions outside Europe (e.g., in Antarctica), implying a natural transfer of the gained knowledge that is beneficial for the measurements that are carried out in the European polar areas (Greenland, Svalbard, Northern Scandinavia, Northern Russia, and, in some respects, the high Alps). The category ‘other’ in Figure 4 included airports and some laboratories that produce artificial snow. These differences in the environment types obviously affect the measured snow properties and the applied measurement techniques.

Figure 4. Types of measurement sites reported by the respondents. The width of each slice corresponds to the fraction each site type has, relative to the total number of site types reported (i.e., the sum of the numbers given in the figure).

3.3. Measured Properties and Applied Instrumentation

Kinar and Pomoroy [15] provided a classification of the instrumentation, distinguishing between ‘portable’ or ‘stationary’, ‘invasive’ or ‘non-invasive’, and ‘active’ or ‘passive’. Non-invasive measurements, which do not modify the snowpack, can be more easily automated than the invasive measurements, and can be made stationary for long term monitoring. Active or passive measurements refer to devices that send and receive (active), or only receive (passive) electromagnetic or sound waves. Here, we also classified the instruments distinguishing between ‘manual’ and ‘automatic’, as this classification better illustrates the possible application of the measurements: sustainable observations for monitoring usually require automatic devices, which also imply that the instruments are stationary and non-invasive. Among the manual instruments, we also distinguished between ‘manual without electronics’ and ‘manual with electronics’, the latter having the potential of being further developed into automatic devices (when they are operated remotely and not in a snowpit). We also distinguished the devices that are unique, developed by a single group or company, and are usually known by their name (written in italics in the plots and the tables), versus the generic techniques or devices which obviously encompass a larger diversity than the former ones. All of the instruments that were listed in the survey and are cited in the present paper are described in Appendix B in Table A6, with references to the literature illustrating their application. Table A6 also includes the classification of the instruments and the snow properties that are measured by/derived from each of them.

3.3.1. Snow Macrophysical Properties

Snow macrophysical properties are the most commonly measured, as 117 respondents (93% of the total) measure one or (more often) several of them. Figure 5 illustrates how frequently each macrophysical property is measured among the respondents, with respect to the total number of measured macrophysical properties. HS is the most measured property (measured by 113 respondents), followed by snow presence, SWE, and snow bulk density. These four properties provide the basic knowledge on the snowpack, and the occurrences of the measurement techniques that are applied to measure them are illustrated in Figure 6 (see Table A6 for a description of the instruments). It should be pointed out that, in many cases, multiple instruments and techniques are often used, as it occurs in particular for HS and SWE. This means that more instruments are simultaneously deployed in the same location (for instrument inter-comparison and calibration), or that different instruments are used by the same respondent in different measurement campaigns/stations/time periods.

Figure 5. Measured snow macrophysical properties. The width of each slice corresponds to the fraction each macrophysical property has relative to the total number of reported macrophysical properties (i.e., the sum of the numbers given in the figure).

Figure 6. Occurrences of the applied instruments to measure snow macrophysical properties, such as snow depth (a), snow presence (b), snow water equivalent (c), and snow density (d). Different colours denote manual instrument without electronics (blue), manual instrument with electronics (red), manual instrument without electronic or fully automatic depending on the version of the instrument (blue with green contour), and fully automatic instrument (green).

The most common measurement methods for HS, snow presence (snow on/off), SWE, and snow density are not automated. Often, however, both manual and automatic methods are applied by the same respondents, highlighting the complementarity of the techniques. Indeed, although 74% of the respondents measuring snow presence do visual observations, 78% of them perform also automatic measurements, retrieving snow presence from, for example, snow depth sensors, camera, web-cam, or indirectly estimating it from the thermometer automatic record. About 73% of the respondents measuring HS apply manual ruler and/or stakes. In the questionnaire, snow stakes and rulers were given as two distinct instrument options to measure the HS. However, their distinction is not always clear, especially in the case of the stationary installations. Therefore, in Figure 6b, we merged the responses in which either one of them or both of them were selected, the latter corresponding to about half of the cases. About 57% of the respondents measuring HS do automatic ultrasonic or laser depth sensors. Of these, 82% utilize also one or more manual methods, such as rulers, stakes, and snow probes. Although it was not specifically mentioned among the possible HS measurement methods, GPS reflectometry (see definition in Table A6) is an emerging technique that takes advantage of the large number of GPS satellites as a free timestamped electromagnetic source, and it is used by the Slovakian Institute of Hydrology and the Geodetic Institute of Slovenia. The latter institute also applies tacheometry to measure HS and the snow covered area over slopes. Another interesting approach that is adopted by the Federal Hydrometeorological Institute of Bosnia and Herzegovina is to infer snow cover bulk properties from the soil measurements (mainly temperature). This approach cannot provide accurate quantitative estimations of HS or SWE, nevertheless, it allows the detection of snow presence and a rough estimation of HS in a very cheap and automatic way, and therefore it is potentially applicable over large areas. An example of the relationship between snow presence and soil temperature is given by Mackiewicz [25].

A large number of techniques are applied to measure the SWE (Figure 6c). While most of them measure the SWE directly targeting the entire snowpack, the snow gauge measures the daily SWE of the new snow. From the cumulative snow gauge measurements, the SWE of the snowpack can be obtained with assumptions on (or independent estimations of) snow melting and sublimation, hence the accuracy of the measurement is lower than in the case of the other techniques. For the estimation of HS, snow presence, and SWE, several automatic instruments are available, some of them quite sophisticated (such as gamma and cosmic ray sensors, or ground penetrating radars), but others quite mature in technology and easily available and affordable (such as ultrasonic depth sensor for HS, camera or web-cam for snow presence, snow pillow for SWE, and automated snow gauge for daily SWE of new snow). While the old snow gauges were manual, hence requiring the manual emptying of the gauge to measure the accumulated snow volume, in many stations they were recently replaced by automatic devices (e.g., in Finland). Snow gauges are employed to measure the SWE of new snow in Albania, Bosnia and Herzegovina, Bulgaria, Czech Republic, Finland, France, Hungary, Iceland, Ireland, Italy, Poland, Republic of Macedonia, Romania, Russia, Slovakia, Spain, and Ukraine. SWE is measured with the Snowpack Analyzer in Bulgaria, Czech Republic, Italy, Poland, Slovakia, Sweden, Turkey, and the U.K.; and with the snow pillow in Bulgaria, Czech Republic, Italy, Norway, Poland, Spain, Sweden, Switzerland, and Turkey. Ground penetrating radars are employed in Austria, France, Iceland, Norway, Poland, Spain, Sweden, and Switzerland. Gamma and cosmic ray sensors are utilized only in Finland, France, Iceland, Norway, and Spain, and the acoustic sensor are used in Germany. Automatic instruments to measure snow presence (such as camera or web-cam, snow depth sensors, and infrared sensors) are utilized by all countries, except Bosnia and Herzegovina, Bulgaria, Croatia, Lithuania, Portugal, Republic of Macedonia, Romania, Serbia, and Ukraine.

The bulk density of the snowpack cannot be directly measured with automatic methods, but it can be indirectly inferred from the collocated automatic measurements of SWE (e.g., using a scale, a snow pillow, a neutron probe, or a Snowpack Analyser) and HS (e.g., using ultrasonic or laser depth sensors), or from the snow penetration resistance measured with the SnowMicroPen. Snow density can also be measured with the Snow Fork and the Denoth Meter, but these instruments were listed in the questionnaire only among the possible measurement methods for liquid water content (Figure 7), as this is the quantity for which they are mostly designed.

Figure 7. As in Figure 6, but for snow temperature (a), snow layer thickness (b), snow cover area (c), snow hardness (d), penetrability of the snow surface (e), snow strength (f), liquid water content (g), and surface roughness (h). Different colours denote manual instrument without electronics (blue), manual instrument with electronics (red), manual instrument with or without electronics depending on the version of the instrument (blue with red contour), manual instrument with electronic or fully automatic depending on the version of the instrument (red with green contour), and fully automatic instrument (green).

The instruments that are used to measure the other macrophysical properties are shown in Figure 7. These properties are much less frequently measured by the respondents than those listed in Figure 6 (see Figure 5). Snow temperature (the vertical temperature profile of the snowpack or just the temperature of the surface layer) is measured by 46% of the respondents, 74% of whom do manual measurements of the snow temperature profile using a thermistor probe. However, several automatic techniques are also applied; the vertical thermistor strings that are used to measure the vertical temperature profile of the snowpack had the highest occurrence after thermistor probe, followed by infrared sensors and pyrgeometers (applied to measure the surface temperature), and iButtons and probes with infrared sensors (applied to measure the vertical temperature profile of the snowpack). The latter instruments are based on a more advanced technology than the previous and are used by two respondents only. In the literature, also other thermometers such as thermocouples (handheld or in fixed arrays) have been applied to measure snow temperature (e.g., [15,26]) but they were neither listed among the possible choices in the questionnaire nor reported by the respondents in the option ‘other’, as they are evidently not commonly used by European institutions.

Although snow covered area is a fundamental property comparable to those shown in Figure 6, it is mostly derived from satellite observations, and hence, it is measured in the field only by 32% of the respondents. Excluding temperature and snow cover area, the other properties in Figure 7 are only manually measured. Snow hardness, penetrability of the snow surface, and snow strength (or shear resistance) are applied in the assessment of avalanche risk, therefore they are mostly measured by the operational institutions or research centres that are responsible for avalanche warnings and/or research. Among the listed tests to assess the snow strength (Figure 7f), only the shear frame really measures the snow strength (see Table A6). The other methods provided an estimation of the stability of the snowpack, and, hence, a qualitative evaluation of its strength. Measurements of liquid water content in snow, which are used for forecasting wet snow avalanches and melt-water runoff [27], and for snow mass and energy budget estimations, are quite challenging, and are performed by 14% of the respondents only. For these measurements, the Snow Fork is used in Finland, Lithuania, and Slovakia; the Denoth meter in Austria, Italy, and Switzerland; the dilution method in Croatia, Russia, and Spain; the melting/freezing calorimetry in France, Iceland, and Slovenia; while the time domain reflectometry is used in Romania. Surface roughness measurements are made to better interpret the optical and microwave signature of the snow, and are performed by a very limited group of scientists (15 respondents, corresponding to 13% of the total). To estimate the surface roughness, photography is applied in Finland, France, Italy, Poland, Russia, Spain, and the U.K., while laser scanner is applied in Finland, France, Italy, and Spain. Another measured snow macrophysical property that was not included in Figure 6 and Figure 7 is the snow thermal conductivity that is measured by 10 participants using heat flux plate (Ireland, Poland, and Switzerland), transient needle probe (France and Switzerland), or derived from micro-computed tomography of casted snow samples (France and Switzerland) or thermistor strings (Finland, Iceland, Poland, and the U.K.). Also not shown in Figure 6 and Figure 7 are meltwater runoff, measured by eight respondents with the lysimeter (in Czech Republic, France, Latvia, Norway, Russia, Slovakia, Spain, Switzerland, and Turkey), and the snow aerodynamic roughness (roughness length) that is obtained by five respondents from the vertical profile of the wind speed (this quantity, however, is mostly applied in meteorological applications, and therefore it is mostly measured by a community different from the one that was targeted by this survey).

3.3.2. Snow Microphysical Properties

Of the respondents, 51 (41% of the total) measured snow microphysical properties, generally for research purposes. The number of respondents measuring each microphysical property, and the fraction of each measured property over the total amount of measured microphysical property is given in Figure 8. It is worth noting that only two parameters, namely snow grain shape and size, are observed by most of the respondents (48 and 43, respectively, corresponding to 94% and 84% of the respondents measuring microphysical properties), the majority of whom applies traditional visual methods (Figure 9). Others parameters are measured by few groups only. Indeed, Figure 9 reveales that only a few research groups apply macro-photography with image processing (Belgium, Finland, France, Ireland, Italy, Norway, Poland, Russia, and Switzerland) or near-infrared (NIR) photography and image processing (Italy, Switzerland, and the U.K.), and even fewer use snow sample casting, micro-computed tomography, and image processing—a sophisticated method that requires specialized laboratories (only present at the Centre d’Etudes de la Neige, Meteo-France/CNRS, France, and at the WSL Institute for Snow and Avalanche Research, Switzerland). The snow specific surface area (SSA) is measured only by six respondents (from Finland, France, Italy, Switzerland, and the U.K.), each of them generally applying several techniques that were developed only recently. This is the case for the IceCube that was developed in France [11] and was utilized in Finland, France, Switzerland, and the U.K.; the SnowMicroPen was developed in Switzerland [13] and was applied in France, Switzerland, and the U.K.; the NIR-photography method was developed in Switzerland [28] and is applied in Italy (Insubria University), Switzerland (WSL Institute for Snow and Avalanche Research), and the U.K. (Northumbria University); the ASSSAP [12] was developed and is applied in France; and the InfraSnow [29] was developed and is applied in Switzerland. These new techniques are much more practical and ‘portable’ than the more traditional methods that are based on gas absorption or micro-computed tomography, which require specialized laboratories. The methods that are applied to measure SSA are used to derive also other microstructural properties; the SnowMicroPen is applied to estimate the snow correlation length and density, while the micro-computed tomography of casted samples is applied to derive snow tortuosity, porosity, stickiness, anisotropy, and air permeability.

Figure 8. Measured snow microphysical properties. The width of each slice corresponds to the fraction each snow microphysical property has, relative to the total number of reported microphysical properties (i.e., the sum of the numbers given in the figure).

Figure 9. Occurrences of the applied instruments to measure snow microphysical properties, such as grain shape (a), grain size (b), specific surface area (SSA) (c), and thermal conductivity (d). The abbreviations IP and MCT stand for image processing and micro-computed tomography, respectively. Different colors denote manual instrument without electronics (blue) and manual instrument with electronics (red).

The optical- and microwave-equivalent grain size are measured by five and three respondents, respectively (optical-equivalent grain size from Finland, France, Romania, and Switzerland, and the microwave-equivalent grain size from Finland, France, and the U.K.).

In the questionnaire, only the spectro-radiometers were listed among the instruments that are available to derive the optically equivalent grain size (applying the radiometric measurements as the input to the inverted radiative transfer models). However, the optically-equivalent grain size is related to SSA (see Table A2), hence, in practice, all of the SSA measurements can be converted to the optical equivalent grain size and vice versa. Moreover, an estimation of the optical equivalent grain size can also be obtained from the visual observations and macro-photography of snow crystals (e.g., [30,31]).

3.3.3. Snow Electromagnetic Properties

Figure 10 illustrates how frequently each electromagnetic property is measured among the respondents, with respect to the total number of measured electromagnetic properties. Of the respondents, 33, which corresponded to 26% of the total, measure snow electromagnetic properties, and the majority of them (26 respondents) measure the snow broadband albedo (Figure 10), mostly applying pyranometers (in 22 of the twenty-six responses). These numbers seemed small, in view of the high relevance of snow albedo for surface energy budget and water runoff calculation, and for the strong feedbacks that its variability has on climate. One explanation is that, although the pyranometers are fully automatic, they need daily manual cleaning and have demanding requirements in terms of instrument calibration/characterization (re-calibration is needed every year and the deviation from the ideal angular response need to be measured) and installation (the holding structure should be designed to minimize shadows and reduction of diffuse illumination, the target surface should be flat, and the field of view should be free from obstacles and from reflecting tilted surfaces) to ensure the correct interpretation and good accuracy of the measurements [32,33].

Figure 10. Measured snow electromagnetic properties. The width of each slice corresponds to the fraction each snow electromagnetic property has, relative to the total number of reported electromagnetic properties (i.e., the sum of the numbers given in the figure).

As for the other snow properties, most of the respondents measure more than one electromagnetic parameter. For instance, among the respondents who apply spectro-radiometers to measure spectral albedo (from Finland, France, Norway, and Switzerland), four derive the broadband albedo from the same instrument/dataset. In the microwave spectral region, the snow backscatter and snow brightness temperature are measured through ground-pointing radars and microwave radiometers by ten and eight respondents, respectively; only the backscattering coefficient is measured by the University of Pavia (Italy), the Norwegian Water Resource and Energy Directorate (Norway), the Swedish Meteorological and Hydrological Institute (Sweden), and University of Zurich (Switzerland), and only the brightness temperature is measured by the Grenoble-Alps University (France), the University of Oslo (Norway), the Middle East Technical University (Turkey), and Met Office (the U.K.), while the both of them are measured at the Finnish Meteorological Institute (Finland); Centre d’etudes de la neige, Meteo-France/CNRS (France); University College of Dublin (Ireland); and Northumbria University (the U.K.). From the measurements of the backscattering coefficient and brightness temperature, and by using advanced microwave models and inversion techniques, the SWE could be estimated and, potentially, information on liquid water content could be extracted. More specialized measurements, which require experimental instruments that are not commercially available, include the bidirectional reflectance distribution function (BRDF), the snow e-folding depth, and the snow optical transmittance. BRDF is measured at the Centre d’etudes de la neige, Meteo-France/CNRS (France), the Regional Environmental Protection Agency (ARPA) of Lombardia region (Italy), the University of Oslo (Norway), and the University of Zurich (Switzerland) mostly using gonio-spectro-radiometers (but in one case also applying a camera on board an Unmanned Aerial Vehicle and Lambertian-reflecting targets at the snow surface), while snow e-folding depth and optical transmittance are measured in France with fibre-optic probes. Snow dielectric permittivity can be directly measured with several instruments, which generally apply the measurements to automatically derive other snow bulk properties (see Table A3). However, a few respondents (from Finnish Meteorological Institute, Finland, and Northumbria University, the U.K.) measured the dielectric permittivity as the targeted snow property (later applied to calculate snow macrophysical properties).

3.3.4. Precipitating and Suspended Snow

Of the respondents, 93 (74% of the total) measure one of the measurable properties characterizing the precipitating and suspended snow. This group of measurements includes properties, such as the height of new snow (HN, also called depth of snowfall) and precipitation intensity, which are used by operational monitoring services, and others such as the fall velocity of hydrometeors and their particle size distribution, the occurrence of drifting/blowing snow, the number flux, and particle size distribution of drifting/blowing snow, which are mainly utilized in research applications. The first properties are therefore much more frequently measured than the second ones (Figure 11).

Figure 11. Measured precipitating and suspended snow properties. The width of each slice corresponds to the fraction each precipitating and suspended snow property has, relative to the total number of reported precipitating and suspended snow properties (i.e., the sum of the numbers given in the figure).

Most of the respondents measure the HN (83 respondents, corresponding to 89% of those measuring precipitating and suspended snow properties), with a variety of instruments (Figure 12a), the most popular ones being rulers/stakes and ultrasonic/laser depth sensors (used by 66 and 50 respondents, respectively). To measure the HN, the recommended practice is to cover the old snow surface with a wooden or plastic board after the last measurement, as a marker for the top layer (Table A4). According to the World Meteorological Organization (WMO) guideline, “if there is a layer of old snow, it would be incorrect to calculate the depth of a snowfall from the difference between two consecutive measurements of the total depth of the fresh and old snow, because of the continuous settling of the old snow” [34]. In our questionnaire, we did not ask to specify if the reference board above the old snow layer is applied or not. In Finland, for instance, such a board is not used, which imply that the HN measurements are not reported by the environmental and meteorological snow monitoring agencies. In Germany, on the contrary, the reference board is used for both manual and automatic HN measurements, and an optimized board has recently been designed and applied to the observational network [35]. As in the case of HS, the measurement of HN is mostly done using manual instruments, the only automatic devices being ultrasonic/laser depth sensors and digital cameras that are set to collect lapse-rate photos.

Figure 12. Occurrences of the applied instruments to measure the precipitating and suspended snow properties, such as the height of new snow (a), precipitation intensity (b), number/mass flux of drifting/blowing snow (c), and particle size of drifting/blowing snow (d). Different colors denote manual instrument without electronics (blue), manual instrument with electronics (red), manual instrument without electronic or fully automatic depending on the version of the instrument (blue with green contour), manual instrument with electronic or fully automatic depending on the version of the instrument (red with green contour), and fully automatic instrument (green).

The precipitation intensity is recorded by 47 respondents, mostly using a snow gauge (Figure 12b). This snowfall property is generally measured automatically; however, traditional manual gauges are still utilized. The hydrometeor fall velocity and size distribution are measured with optical disdrometers by eight respondents. For research applications, disdrometers are also used to provide an estimate of the particle size and type. Present Weather Detectors (PWDs) can observe the precipitation intensity and type, as well as the visibility. Optical probes, which are developed for airborne use, are also, in some cases, used on the ground for measuring particle size precipitation. Of the 39 respondents measuring drifting/blowing snow properties, all but one also measure the precipitating snow properties, and 32 do visual measurements of the occurrence of drifting/blowing snow. The number or mass flux of drifting snow is measured by thirteen participants applying a large variety of techniques that are all automatic (Figure 12c), except the traditional manual version of the mechanical trap (See Table A6 in Appendix B). While the FlowCapt and the mechanical trap measure the mass fluxes of snow particles, the Snow particle counter (SPD), the optical disdrometer, and the camera system detect the number fluxes of snow particles. The particle size of the drifting/blowing snow is measured by ten respondents, eight of them applying automatic devices. Only two respondents (from Russia and Slovenia) use sticking slides, which need to be manually removed and photographed (see Table A6 in Appendix B).

3.3.5. Snow Composition

The snow composition properties are measured by only 16 respondents (corresponding to 13% of the total). The snow impurities are measured by ten respondents (from Austria, Finland, France, Italy, Poland, Switzerland, and Ukraine), with ion chromatography being the most applied measurement method (Figure 13a). A visual, qualitative estimation of the snow impurities was carried out only in Ukraine. All of the other methods rely on the collection of snow samples that are then analysed later, with equipment that can only operate inside laboratories. Thus, these measurement methods are all classified as ‘manual with electronics’ in Figure 13a (see the description in Table A6 of Appendix B). Different instruments are applied to measure different impurities; ion chromatography is used for dust, SP2 for soot or small black carbon (BC) particles, and the organic and elemental carbon (OC/EC) analyzer for a wider spectrum of organic components. Water and heavy metal isotopes measurements also require the collection of snow samples and their successive laboratory analysis. They are performed by nine respondents (from Austria, Iceland, Italy, Poland, Slovakia, Slovenia, and Spain), most of whom apply a mass spectrometer (Figure 13b).

Figure 13. Occurrences of the applied instruments to measure snow impurities (a) and snow isotopes (b). Different colors denote manual instrument without electronics (blue) and manual instrument with electronics (red).

3.4. Use of Measurement Protocols

Of the respondents, 75 (60% of the total) use a written protocol for some of the measured snow properties, and 22 of them (all but one from the operational oriented institutions) use it for all of the measured properties (Table 1). For all of the well-established measurement techniques, a written protocol exists, to ensure the homogeneity and inter-comparability of the measurements and to reduce the error sources. This is especially important for those properties that are measured for the operational applications. Nevertheless, among the 71 respondents from operational oriented institutions, 18 do not apply any written protocol. Different environmental conditions, however, may require the adaptation of the protocol to the local needs. In case of the devices that are used for research applications, there may not be any written measurement protocol, especially when the instruments are developed by the research groups and are not commercially available.

Table 1. Optional answers available to the question “Did you apply a written protocol (published or unpublished, written in any language) when measuring the snow parameters?”, and the number of given answers per option.

N° of Answers	Answer
22	Yes, for all of the measured parameters I applied a written protocol, which describes both the use of the instrumentation and its set up in various environmental conditions.
53	Yes, for some of the measured parameters I applied a written protocol, which describes both the use of the instrumentation and its set up in various environmental conditions. For some parameters I only followed the general instructions of the instrument and applied measurement strategy based on my own experience or on the oral teaching of senior scientists/operators.
37	No, I only followed the general instructions of the instrument and applied measurement strategy based on my own experience or on the oral teaching of senior scientists/operators.
8	No, I do not use any written protocol, and I apply the instrument’s instructions using the instrument as I learned to, by myself.
5	Other

4. Summary and Discussion

The survey was answered by most of the European national weather agencies and by several national institutions that are responsible for water/environmental management and avalanche forecasts, and by numerous universities and research institutes. However, not all of the European research and operational institutions that are involved in the snow measurements responded to the survey. This is mainly because of the lack of centralized and institutionalized coordination of the snow measurement activities at national level. Moreover, some of the participants provided answers that cover the activities that are carried out in the whole institution they represented, while others just described the measurements that are done by themselves or their group. Indeed, there were no clear instructions on how comprehensive the answers should have been, and the position of the respondents in their own institution varied a lot, from technician and PhD student to professor, and to the head of weather service or of measurement division. Despite these limitations, the survey managed to capture the state-of-the-art of the European in-situ snow operational monitoring and research-driven snow measurements.

Strictly speaking, the percentages of respondents that measure each of the snow properties do not directly indicate how diffusely or frequently those properties are measured, but rather illustrate the measurement capacities of the respondents (what is measured and how it is measured). However, the received answers provided a quite strong indication on what are the most commonly measured properties and applied techniques. The respondents from the operational oriented institutions were more numerous than the respondents from the research only institutions, but the research objectives were, in most cases, shared by both, thus research was the dominant measurement purpose that was marked in the survey. On the other hand, snow research is largely driven by the needs of the operational services, therefore, the most widely measured snow properties are the few ones that are utilized by the operational monitoring services. These include snow macrophysical properties, such as snow presence, HS, SWE, and snow density, which are measured by 93% of the respondents, and the snow precipitation quantified as the height of new snow, which is measured by 66% of the respondents. A large variety of manual and automatic instruments are applied in order to collect these properties, with the manual instruments being the most diffuse and with the longest application history (e.g., [36]). The manual measurements of the HS are mostly reported only in presence of snow, without distinguishing the cases of zero HS and of missing observations. This reporting practice leads to a bias in the HS data records, which is an issue for the data assimilation applications. This issue was addressed by the WMO 69th Executive council in May 2017, to improve operational exchange of snow data and reporting practices.

Our analysis showed that most of the respondents that apply the manual method also apply automatic techniques. Important drivers for the automatization of the snow macrophysical properties are the snow and avalanches services in mountain areas. In order to be able to provide the public with timely information about the snow situation and the avalanche hazard, automatic stations with a variety of reliable meteorological and nivological sensors are needed. These data are often used to drive snow models, which provide even more information in high temporal resolution. Nevertheless, manual observation of, for example, the height of new snow, amount of wind transported snow, or possible weak layers, are still an important part of a high quality snow and avalanche information service. Such in-situ snow observational networks need to achieve a high level of automatization to ensure the harmonization, the distribution across institutions, and the long term sustainability of the measurements, with direct manual measurements ideally performed only for the calibration and validation of the automatic methods. The results of this survey demonstrated that many European institutions are engaged in an ongoing effort toward this goal.

Some new automatic technologies that are developed and exploited by the European community have been transferred or have the potential to be transferred from the research to the operational domain, for instance the Snowpack Analyzer [37] and GPS reflectometry [38,39,40] to derive the snow bulk properties, and the web cam network to retrieve the snow cover fraction [41]. While some of the new automatic methods (e.g., those based on cameras) have a low price and, therefore, can possibly reach a more widespread usage, others that are based on expensive technology (e.g., Neutron probe, GPR Radar, and Gamma and cosmic ray sensors) cannot be adopted by a large community. An increase in the measurements of the snow bulk properties across institutions is envisaged also in view of the on-going efforts made by ECMWF and WMO to define the SWE data exchange protocol for operational NWP applications. Indeed, currently there is not an exchange protocol for the SWE measurements among the operational networks, thus SWE is not used in the NWP data assimilation systems. Present-day operational NWP systems only rely on HS measurements, which are exchanged on the operational networks supported by WMO.

In addition to the data exchange protocols, written measurement protocols are recommended, especially for the operationally relevant snow properties, to ensure the continuity of the measurements, the transfer of knowledge among different operators of the instruments, and the cooperation among the scientific community that use or develop similar instruments. In principle, the measurement protocols should be internationally agreed and shared, to ensure the comparability of the data and to facilitate their transfer. In practice, protocols need to be tailored to the specific measurement sites, because the measurement issues are dependent on the sites and their meteorological conditions (e.g., average wind speed and amount of snowfall), as observed in the Solid Precipitation Intercomparison Experiment (SPICE) that was organized by WMO in 2012–2015 [42].

The survey also clearly illustrated that the automatization of the instruments is dependent on the measured variable. Snow microphysical and composition variables, for instance, cannot yet be easily measured by fully automatic instruments (Figure 9 and Figure 13). This is a major limitation for the operational services, as they would need to put significant effort into the collection of data if it requires personnel for field work. In practice, the consequence is that snow microphysical and composition variables will not be widely monitored in Europe until a higher degree of automatization is reached. However, the monitoring of snow microphysical properties would be highly beneficial for many operational services. The availability of measurements of snow microstructure (optical- and microwave-equivalent grain size, SSA, correlation length) would support the development of assimilation schemes that use satellite-based radiances as input. The model analysis field could be produced using satellite-based snow microstructure, snow extent, albedo, SWE, and HS, and by applying in-situ observations of HS and snow microstructure to constrain the satellite data. The benefits of this physically consistent assimilation approach were recently demonstrated in the case of a snowpack model that was used for avalanche risk forecasting [43], but they most likely exist also in numerical weather prediction and hydrological models. Microstructural properties, such as snow stickiness, porosity, and tortuosity, which determine the elasticity of the snow [44], are derived from the extremely laborious X-ray tomography, which can be performed in a couple of European laboratories only. More observations of these quantities would be needed to develop models that describe them, and that could have many operational applications (warning on road slipperiness, forecasting snow surface conditions for winter sports and snow loads on trees, buildings, and wind turbines, etc.).

The in-situ measurements of the most common snow electromagnetic properties (broadband albedo, backscatter, and brightness temperature) are done with fully automatic instruments (see Table A6). In the microwave spectral region, the signal received from the radars and radiometers is used to derive the snow extent, SWE, and snow melting/freezing state. These measurements are mainly applied for research purposes, such as the ground truth for satellite observations, and the development/validation of snow retrieval algorithms and of models for the snow-electromagnetic interaction. The in-situ monitoring of the snow extent, SWE, and snow melting/freezing state is delegated to the cheaper and more practical instruments shown in Figure 6 and discussed in the paragraphs above. Although the snow broadband albedo is a crucial variable for the climate, weather forecasting, and snow mass and water runoff estimation, it is measured by 21% of the respondents only. A larger number of broadband albedo measurements is probably hampered by the demanding maintenance and installation requirements of the pyranometers (as discussed in Section 3.3.3). A clearer measurement protocol would possibly help in increasing the number of the in-situ broadband albedo measurements, which are now mainly used for the development and validation of models that simulate the radiative transfer in snow, snow albedo schemes, and snow retrieval algorithms, and for the validation of snow satellite products.

Among the snow properties related to precipitating and suspended snow, the HN and precipitation intensity are the only ones applied as input in numerical weather prediction, hydrological models, and snow models used for avalanche risk forecast (e.g., [45,46]). The microphysical properties of precipitating and suspended snow (hydrometeor fall velocity and size distribution, particle size and number flux of drifting snow) are mainly measured with automatic devices, mostly for research purposes, such as the validation of polarization radar measurements and the assessment of snow transport in avalanche terrain [47,48] and polar regions [49,50,51,52]. These measurements would also be very important input parameters for the physical snow models that simulate the snow albedo, to constrain the optical equivalent grain size and albedo of fresh snow. From the operational perspective, observations of drifting snow would be extremely beneficial in the management of traffic along roads, rails, and runaways in locations dominated by strong winds, and to assess the risks related to the transport of snow over infrastructures and on avalanche-prone terrains. The transfer of microphysical measurements of precipitating and drifting snow from research institutions to operational services would require proper instrument calibration (see, for example, [53]) and the automatization and standardization of the software applied to interpret the signals acquired by the automatic optical and acoustic sensors (disdrometer, FlowCapt, and snow particle counter).

5. Conclusions

The survey on the practices and purposes of the European in-situ snow measurements, carried out in the framework of Harmosnow, enabled the collection of a compilation of the state-of-the-art of in-situ snow operational and research-based snow measurements conducted by the European countries, in a time when new technological solutions are tested and developed. As such, it constitutes the basis for the upgrade and harmonization of the in-situ observation methods, which is one of the main goals of Harmosnow, and for the enlargement of the communities that applied automatized measurements for operational monitoring and frontier techniques for research purposes. The mapping of snow monitoring in Europe obtained through this survey is the needed pre-requisite for the identification of observational gaps and the planning of future long-term snow measurement strategies. Although the snow operational monitoring largely relies on satellite-based observations, in-situ measurements are an essential component of it, as both satellite and ground-based observations are applied in the snow assimilation schemes (e.g., [54,55,56]). Hence, the enhancement and harmonization of the in-situ monitoring will improve the description and the assimilation of the snow cover information into hydrological, land surface, meteorological, and climate models. This is critical for addressing the impact of snow on various phenomena, to predict local snow water resources, and to warn about snow-related natural hazards. The results of this compilation are therefore discussed, also, from the perspective of the need of enhancing the efficiency and coverage of the in-situ observational network by applying automatic measurement methods. The results showed that (1) some of the automatic instruments that are used to measure bulk snow properties have good potential for being applied to monitoring large and remote areas (GPS reflectometry, photography, etc.); (2) some of the snow properties that are presently measured only manually for research applications would be greatly beneficial for operational services (snow stickiness, SSA or snow optical/microwave equivalent grain size, and snow correlation length); (3) the snow drift properties (number flux and particle size distribution) that are automatically measured and are mostly used in the research applications would already be (or close to be) mature to be applied operationally to provide specific services (e.g., information on snow drift over roads, runaways, and rails). This leads to few final recommendations, namely: automatic instrumentation should be adopted, especially when cheap and practical solutions are available, and the properties that are presently measured for research applications would be needed and should be acquired also by operational networks, particularly when they are collected with automatic instruments. Moreover, the development and the increase in the use of internationally agreed measurement protocols for each of the applied measurement techniques are strongly encouraged, as their application will enhance the harmonization of the measurements. This is in line with the recent Global Cryosphere Watch Initiative for improving in-situ HS and SWE observations [57].

These results are relevant for the planning of snow observation infrastructures and monitoring programs, as well as for the scientific community—the specialists in specific applications who are familiar with a certain set of snow properties, will generally benefit from knowing the measurement practices that are applied by different communities and in different environments. The measurement practices that are developed on the basis of the knowledge acquired in local institutions and for targeting specific environments will get visibility among a wider audience, eventually broadening their applicability also in other contexts. Moreover, snow properties that are measured for certain purposes may be very relevant for other applications too. Indeed, this work is in line with the recommendation of sharing and comparing techniques between the snow monitoring teams provided by the recent review on the changing Arctic snow cover [58], which highlighted the need for an interdisciplinary approach to resolve the current snow measurement and modelling limitations.

The effort spent to send the survey to as many institutions as possible, and the difficulties encountered in order to identify the key responsible persons for snow measurements, evidenced that a much better intra-national collaboration and knowledge-exchange of snow measurement activities is necessary in Europe. The current fragmentation and peculiarities represent a major bottle neck for the necessary harmonization of snow measurements practices in Europe.